2022-12-07 04:36:17,366 INFO [train.py:941] (0/4) Training started 2022-12-07 04:36:17,370 INFO [train.py:951] (0/4) Device: cuda:0 2022-12-07 04:36:17,416 INFO [lexicon.py:168] (0/4) Loading pre-compiled data/lang_char/Linv.pt 2022-12-07 04:36:17,423 INFO [train.py:962] (0/4) {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 100, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'b2ce63f3940018e7b433c43fd802fc50ab006a76', 'k2-git-date': 'Wed Nov 23 08:43:43 2022', 'lhotse-version': '1.9.0.dev+git.97bf4b0.dirty', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'ali_meeting', 'icefall-git-sha1': 'f13cf61-dirty', 'icefall-git-date': 'Tue Dec 6 03:34:27 2022', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r8n04', 'IP address': '10.1.8.4'}, 'world_size': 4, 'master_port': 12354, 'tensorboard': True, 'num_epochs': 20, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp/v1'), 'lang_dir': 'data/lang_char', 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'context_size': 2, 'prune_range': 5, 'lm_scale': 0.25, 'am_scale': 0.0, 'simple_loss_scale': 0.5, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 5000, 'keep_last_k': 10, 'average_period': 200, 'use_fp16': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'manifest_dir': PosixPath('data/manifests'), 'enable_musan': True, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'max_duration': 300, 'max_cuts': 100, 'num_buckets': 50, 'on_the_fly_feats': False, 'shuffle': True, 'num_workers': 8, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'blank_id': 0, 'vocab_size': 3290} 2022-12-07 04:36:17,424 INFO [train.py:964] (0/4) About to create model 2022-12-07 04:36:17,818 INFO [zipformer.py:179] (0/4) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2022-12-07 04:36:17,861 INFO [train.py:968] (0/4) Number of model parameters: 75734561 2022-12-07 04:36:22,744 INFO [train.py:983] (0/4) Using DDP 2022-12-07 04:36:23,008 INFO [asr_datamodule.py:357] (0/4) About to get AMI train cuts 2022-12-07 04:36:23,011 INFO [asr_datamodule.py:204] (0/4) About to get Musan cuts 2022-12-07 04:36:23,011 INFO [asr_datamodule.py:208] (0/4) Enable MUSAN 2022-12-07 04:36:24,362 INFO [asr_datamodule.py:232] (0/4) Enable SpecAugment 2022-12-07 04:36:24,362 INFO [asr_datamodule.py:233] (0/4) Time warp factor: 80 2022-12-07 04:36:24,362 INFO [asr_datamodule.py:246] (0/4) About to create train dataset 2022-12-07 04:36:24,362 INFO [asr_datamodule.py:259] (0/4) Using DynamicBucketingSampler. 2022-12-07 04:36:24,710 INFO [asr_datamodule.py:268] (0/4) About to create train dataloader 2022-12-07 04:36:24,711 INFO [asr_datamodule.py:381] (0/4) About to get AliMeeting IHM eval cuts 2022-12-07 04:36:24,712 INFO [asr_datamodule.py:300] (0/4) About to create dev dataset 2022-12-07 04:36:24,902 INFO [asr_datamodule.py:315] (0/4) About to create dev dataloader 2022-12-07 04:36:55,258 INFO [train.py:873] (0/4) Epoch 1, batch 0, loss[loss=5.204, simple_loss=4.723, pruned_loss=4.799, over 14231.00 frames. ], tot_loss[loss=5.204, simple_loss=4.723, pruned_loss=4.799, over 14231.00 frames. ], batch size: 60, lr: 2.50e-02, grad_scale: 2.0 2022-12-07 04:36:55,259 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 04:37:02,329 INFO [train.py:905] (0/4) Epoch 1, validation: loss=4.832, simple_loss=4.375, pruned_loss=4.552, over 857387.00 frames. 2022-12-07 04:37:02,330 INFO [train.py:906] (0/4) Maximum memory allocated so far is 8765MB 2022-12-07 04:37:05,260 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:37:18,873 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:37:32,955 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=220.79 vs. limit=5.0 2022-12-07 04:37:40,096 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6259, 5.6266, 5.6244, 5.6211, 5.2224, 5.5785, 5.5937, 5.5956], device='cuda:0'), covar=tensor([0.0015, 0.0012, 0.0011, 0.0019, 0.0023, 0.0018, 0.0009, 0.0013], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([9.1956e-06, 9.2484e-06, 9.0974e-06, 9.3481e-06, 9.2369e-06, 9.2645e-06, 9.1997e-06, 9.3760e-06], device='cuda:0') 2022-12-07 04:37:44,010 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=23.84 vs. limit=2.0 2022-12-07 04:37:44,449 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=5.85 vs. limit=2.0 2022-12-07 04:37:48,807 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=104.53 vs. limit=5.0 2022-12-07 04:38:01,925 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:38:03,488 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=12.11 vs. limit=2.0 2022-12-07 04:38:13,698 INFO [train.py:873] (0/4) Epoch 1, batch 100, loss[loss=0.5109, simple_loss=0.4364, pruned_loss=0.5877, over 14247.00 frames. ], tot_loss[loss=0.9669, simple_loss=0.8694, pruned_loss=0.8957, over 832406.35 frames. ], batch size: 32, lr: 3.00e-02, grad_scale: 0.125 2022-12-07 04:38:16,996 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.426e+01 1.096e+02 2.039e+02 4.080e+02 7.158e+03, threshold=4.078e+02, percent-clipped=0.0 2022-12-07 04:38:28,327 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=9.23 vs. limit=2.0 2022-12-07 04:38:34,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=8.34 vs. limit=2.0 2022-12-07 04:38:42,230 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:38:50,196 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3419, 3.3547, 4.2978, 4.2581, 4.2744, 4.2799, 1.8189, 4.2821], device='cuda:0'), covar=tensor([0.0205, 0.0158, 0.0019, 0.0046, 0.0037, 0.0030, 0.0217, 0.0023], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([8.7688e-06, 8.8157e-06, 8.7562e-06, 8.9692e-06, 8.8314e-06, 8.8501e-06, 8.7994e-06, 8.9325e-06], device='cuda:0') 2022-12-07 04:39:03,230 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4226, 3.4132, 3.4236, 3.4155, 3.4284, 3.4169, 3.4113, 3.4002], device='cuda:0'), covar=tensor([0.0019, 0.0023, 0.0018, 0.0019, 0.0011, 0.0024, 0.0031, 0.0021], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([8.9646e-06, 8.9072e-06, 9.1700e-06, 8.9149e-06, 9.1003e-06, 9.0624e-06, 9.0270e-06, 9.0327e-06], device='cuda:0') 2022-12-07 04:39:21,524 INFO [train.py:873] (0/4) Epoch 1, batch 200, loss[loss=0.4403, simple_loss=0.3717, pruned_loss=0.4583, over 13957.00 frames. ], tot_loss[loss=0.6777, simple_loss=0.5968, pruned_loss=0.6603, over 1326163.76 frames. ], batch size: 19, lr: 3.50e-02, grad_scale: 0.25 2022-12-07 04:39:24,001 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=3.66 vs. limit=2.0 2022-12-07 04:39:24,210 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3551, 4.3575, 4.3514, 4.3571, 4.3540, 4.3576, 4.3552, 4.3576], device='cuda:0'), covar=tensor([1.1183e-04, 2.3555e-04, 1.2298e-04, 2.5742e-04, 1.7327e-04, 9.6900e-05, 1.8445e-04, 1.7528e-04], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([9.5295e-06, 9.4062e-06, 9.4614e-06, 9.0863e-06, 9.5359e-06, 9.1151e-06, 9.3046e-06, 9.2061e-06], device='cuda:0') 2022-12-07 04:39:24,804 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.748e+01 6.194e+01 9.239e+01 1.604e+02 3.274e+02, threshold=1.848e+02, percent-clipped=0.0 2022-12-07 04:40:10,960 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.33 vs. limit=2.0 2022-12-07 04:40:22,372 WARNING [optim.py:389] (0/4) Scaling gradients by 0.06466581672430038, model_norm_threshold=184.7753143310547 2022-12-07 04:40:22,527 INFO [optim.py:451] (0/4) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.6.weight with proportion 0.54, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=4.438e+06, grad_sumsq = 3.527e+09, orig_rms_sq=1.258e-03 2022-12-07 04:40:27,572 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:40:30,371 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 04:40:31,029 INFO [train.py:873] (0/4) Epoch 1, batch 300, loss[loss=0.4682, simple_loss=0.3971, pruned_loss=0.4289, over 14225.00 frames. ], tot_loss[loss=0.5762, simple_loss=0.5011, pruned_loss=0.5526, over 1568460.49 frames. ], batch size: 94, lr: 4.00e-02, grad_scale: 0.5 2022-12-07 04:40:34,363 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.885e+01 5.232e+01 7.085e+01 9.806e+01 2.857e+03, threshold=1.417e+02, percent-clipped=2.0 2022-12-07 04:40:55,960 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.59 vs. limit=5.0 2022-12-07 04:41:08,580 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=6.25 vs. limit=2.0 2022-12-07 04:41:12,144 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:41:14,450 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7099, 3.7121, 3.7133, 3.7127, 3.7101, 3.7107, 3.7071, 3.7066], device='cuda:0'), covar=tensor([0.0012, 0.0011, 0.0008, 0.0012, 0.0011, 0.0013, 0.0023, 0.0009], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([8.9418e-06, 9.0808e-06, 8.6428e-06, 8.8648e-06, 8.7596e-06, 8.7066e-06, 8.8322e-06, 8.7467e-06], device='cuda:0') 2022-12-07 04:41:19,778 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 04:41:33,247 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:41:43,792 INFO [train.py:873] (0/4) Epoch 1, batch 400, loss[loss=0.4616, simple_loss=0.3901, pruned_loss=0.3944, over 8604.00 frames. ], tot_loss[loss=0.5247, simple_loss=0.4497, pruned_loss=0.491, over 1783239.29 frames. ], batch size: 100, lr: 4.50e-02, grad_scale: 1.0 2022-12-07 04:41:46,262 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2022-12-07 04:41:47,371 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.807e+01 5.021e+01 6.661e+01 9.478e+01 2.713e+02, threshold=1.332e+02, percent-clipped=6.0 2022-12-07 04:42:00,281 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=5.04 vs. limit=2.0 2022-12-07 04:42:10,532 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:42:17,127 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 04:42:38,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.57 vs. limit=2.0 2022-12-07 04:42:55,428 INFO [train.py:873] (0/4) Epoch 1, batch 500, loss[loss=0.4803, simple_loss=0.3906, pruned_loss=0.4261, over 14306.00 frames. ], tot_loss[loss=0.4993, simple_loss=0.4225, pruned_loss=0.4538, over 1897294.99 frames. ], batch size: 44, lr: 4.99e-02, grad_scale: 1.0 2022-12-07 04:42:58,964 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.30 vs. limit=2.0 2022-12-07 04:42:59,042 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.328e+01 4.979e+01 6.873e+01 9.259e+01 1.851e+02, threshold=1.375e+02, percent-clipped=7.0 2022-12-07 04:43:17,014 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2022-12-07 04:43:29,463 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.01 vs. limit=5.0 2022-12-07 04:43:36,912 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.50 vs. limit=2.0 2022-12-07 04:43:38,489 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:43:42,864 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:43:57,725 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:44:05,295 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:44:05,743 INFO [train.py:873] (0/4) Epoch 1, batch 600, loss[loss=0.452, simple_loss=0.369, pruned_loss=0.3738, over 11978.00 frames. ], tot_loss[loss=0.4838, simple_loss=0.4051, pruned_loss=0.4268, over 1866644.36 frames. ], batch size: 100, lr: 4.98e-02, grad_scale: 1.0 2022-12-07 04:44:09,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.092e+01 6.985e+01 9.438e+01 1.366e+02 4.481e+02, threshold=1.888e+02, percent-clipped=23.0 2022-12-07 04:44:20,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.05 vs. limit=5.0 2022-12-07 04:44:21,323 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:44:25,208 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:44:30,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.46 vs. limit=2.0 2022-12-07 04:44:34,826 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9344, 3.8031, 3.2974, 3.9765, 3.6555, 3.7086, 3.7436, 3.4314], device='cuda:0'), covar=tensor([0.0334, 0.0576, 0.1260, 0.0276, 0.0987, 0.0725, 0.0630, 0.1352], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0016, 0.0017, 0.0017, 0.0018, 0.0017, 0.0018, 0.0018], device='cuda:0'), out_proj_covar=tensor([1.5088e-05, 1.6331e-05, 1.6382e-05, 1.5616e-05, 1.6700e-05, 1.6241e-05, 1.6372e-05, 1.7141e-05], device='cuda:0') 2022-12-07 04:44:39,002 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:44:41,268 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:44:41,841 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:44:50,879 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.05 vs. limit=2.0 2022-12-07 04:45:15,576 INFO [train.py:873] (0/4) Epoch 1, batch 700, loss[loss=0.4429, simple_loss=0.3607, pruned_loss=0.3501, over 14233.00 frames. ], tot_loss[loss=0.4704, simple_loss=0.3896, pruned_loss=0.4019, over 1892069.95 frames. ], batch size: 69, lr: 4.98e-02, grad_scale: 1.0 2022-12-07 04:45:19,021 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.942e+01 9.279e+01 1.429e+02 2.357e+02 5.770e+02, threshold=2.857e+02, percent-clipped=36.0 2022-12-07 04:45:42,055 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:45:45,177 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:46:15,751 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:46:25,997 INFO [train.py:873] (0/4) Epoch 1, batch 800, loss[loss=0.4295, simple_loss=0.3489, pruned_loss=0.3267, over 7799.00 frames. ], tot_loss[loss=0.4584, simple_loss=0.3773, pruned_loss=0.3774, over 1893260.30 frames. ], batch size: 100, lr: 4.97e-02, grad_scale: 2.0 2022-12-07 04:46:29,402 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.458e+01 1.291e+02 1.818e+02 2.513e+02 6.152e+02, threshold=3.636e+02, percent-clipped=18.0 2022-12-07 04:46:58,653 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:47:11,145 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6243, 4.0426, 3.8466, 4.3906, 3.9655, 4.0729, 3.9372, 4.0973], device='cuda:0'), covar=tensor([0.3055, 0.1408, 0.2238, 0.0958, 0.1852, 0.1629, 0.1462, 0.1701], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0026, 0.0028, 0.0026, 0.0029, 0.0025, 0.0028, 0.0026], device='cuda:0'), out_proj_covar=tensor([2.7705e-05, 2.4549e-05, 2.5864e-05, 2.2659e-05, 2.4425e-05, 2.2401e-05, 2.4828e-05, 2.3803e-05], device='cuda:0') 2022-12-07 04:47:36,963 INFO [train.py:873] (0/4) Epoch 1, batch 900, loss[loss=0.3277, simple_loss=0.269, pruned_loss=0.2361, over 2574.00 frames. ], tot_loss[loss=0.4454, simple_loss=0.3658, pruned_loss=0.351, over 1963995.06 frames. ], batch size: 100, lr: 4.96e-02, grad_scale: 2.0 2022-12-07 04:47:40,263 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.004e+01 1.837e+02 2.554e+02 3.837e+02 9.711e+02, threshold=5.109e+02, percent-clipped=29.0 2022-12-07 04:47:41,920 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 04:47:48,608 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:47:52,433 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.94 vs. limit=5.0 2022-12-07 04:47:52,711 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:47:59,252 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 04:48:08,935 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:48:13,005 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 04:48:46,725 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:48:47,389 INFO [train.py:873] (0/4) Epoch 1, batch 1000, loss[loss=0.4084, simple_loss=0.3352, pruned_loss=0.2853, over 14390.00 frames. ], tot_loss[loss=0.4316, simple_loss=0.3547, pruned_loss=0.3266, over 1983403.89 frames. ], batch size: 53, lr: 4.95e-02, grad_scale: 2.0 2022-12-07 04:48:50,982 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.449e+02 3.112e+02 4.355e+02 1.197e+03, threshold=6.224e+02, percent-clipped=18.0 2022-12-07 04:48:57,980 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7517, 3.6381, 3.5757, 3.6657, 3.6137, 3.4448, 2.3004, 3.5138], device='cuda:0'), covar=tensor([0.0472, 0.0684, 0.0679, 0.0574, 0.0751, 0.0782, 0.2025, 0.0776], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0025, 0.0025, 0.0024, 0.0025, 0.0023, 0.0029, 0.0027], device='cuda:0'), out_proj_covar=tensor([1.9942e-05, 2.1133e-05, 2.0381e-05, 2.0117e-05, 2.1510e-05, 1.9913e-05, 2.8662e-05, 2.2736e-05], device='cuda:0') 2022-12-07 04:49:17,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:49:51,007 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:49:58,565 INFO [train.py:873] (0/4) Epoch 1, batch 1100, loss[loss=0.2725, simple_loss=0.2246, pruned_loss=0.1842, over 2635.00 frames. ], tot_loss[loss=0.4177, simple_loss=0.344, pruned_loss=0.305, over 1919067.00 frames. ], batch size: 100, lr: 4.94e-02, grad_scale: 2.0 2022-12-07 04:50:02,057 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.129e+02 2.767e+02 3.669e+02 5.185e+02 1.491e+03, threshold=7.337e+02, percent-clipped=11.0 2022-12-07 04:50:17,649 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 04:50:25,577 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5050, 2.9791, 2.8880, 2.0115, 2.2170, 2.3644, 2.6401, 2.7837], device='cuda:0'), covar=tensor([0.0974, 0.0775, 0.0954, 0.3642, 0.1274, 0.1244, 0.0935, 0.0788], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0019, 0.0020, 0.0024, 0.0019, 0.0021, 0.0021, 0.0018], device='cuda:0'), out_proj_covar=tensor([1.5441e-05, 1.5219e-05, 1.4600e-05, 2.4199e-05, 1.4567e-05, 1.7127e-05, 1.6890e-05, 1.3378e-05], device='cuda:0') 2022-12-07 04:50:35,545 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:50:56,551 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4091, 3.6229, 3.3936, 2.2207, 2.6224, 2.8275, 3.2932, 3.4075], device='cuda:0'), covar=tensor([0.0850, 0.0745, 0.0993, 0.4495, 0.1443, 0.1475, 0.1048, 0.0725], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0020, 0.0021, 0.0026, 0.0019, 0.0023, 0.0022, 0.0019], device='cuda:0'), out_proj_covar=tensor([1.6303e-05, 1.5794e-05, 1.5426e-05, 2.5966e-05, 1.5014e-05, 1.8696e-05, 1.7704e-05, 1.3876e-05], device='cuda:0') 2022-12-07 04:51:09,750 INFO [train.py:873] (0/4) Epoch 1, batch 1200, loss[loss=0.2394, simple_loss=0.1968, pruned_loss=0.1588, over 1268.00 frames. ], tot_loss[loss=0.4047, simple_loss=0.3342, pruned_loss=0.2851, over 1917912.56 frames. ], batch size: 100, lr: 4.93e-02, grad_scale: 4.0 2022-12-07 04:51:11,113 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:51:12,882 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.090e+02 2.727e+02 3.697e+02 4.642e+02 1.114e+03, threshold=7.394e+02, percent-clipped=5.0 2022-12-07 04:51:18,457 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 04:51:21,871 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:51:25,912 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:51:38,741 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.85 vs. limit=5.0 2022-12-07 04:51:41,254 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:51:46,182 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.39 vs. limit=2.0 2022-12-07 04:51:55,345 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:51:57,813 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 04:51:59,312 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:52:03,962 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6119, 3.7318, 3.6783, 3.6352, 3.7581, 3.3157, 3.7954, 3.7708], device='cuda:0'), covar=tensor([0.0473, 0.0446, 0.0337, 0.0499, 0.0392, 0.0484, 0.0293, 0.0305], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0025, 0.0027, 0.0024, 0.0025, 0.0025, 0.0025, 0.0024], device='cuda:0'), out_proj_covar=tensor([2.6545e-05, 2.5393e-05, 2.4676e-05, 2.4824e-05, 2.4123e-05, 2.5244e-05, 2.5694e-05, 2.3218e-05], device='cuda:0') 2022-12-07 04:52:15,066 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:52:20,508 INFO [train.py:873] (0/4) Epoch 1, batch 1300, loss[loss=0.3846, simple_loss=0.3239, pruned_loss=0.2423, over 14429.00 frames. ], tot_loss[loss=0.3924, simple_loss=0.3253, pruned_loss=0.2673, over 1920156.89 frames. ], batch size: 41, lr: 4.92e-02, grad_scale: 4.0 2022-12-07 04:52:23,972 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.849e+02 3.755e+02 4.812e+02 1.034e+03, threshold=7.510e+02, percent-clipped=8.0 2022-12-07 04:53:29,340 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2322, 2.5295, 3.1497, 2.8852, 3.1569, 2.9481, 2.2398, 2.8156], device='cuda:0'), covar=tensor([0.1355, 0.0870, 0.0578, 0.0737, 0.0490, 0.0611, 0.1239, 0.0660], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0025, 0.0023, 0.0023, 0.0022, 0.0021, 0.0024, 0.0024], device='cuda:0'), out_proj_covar=tensor([2.5982e-05, 2.0118e-05, 1.8659e-05, 1.7634e-05, 1.7724e-05, 1.6485e-05, 2.1184e-05, 2.0687e-05], device='cuda:0') 2022-12-07 04:53:33,567 INFO [train.py:873] (0/4) Epoch 1, batch 1400, loss[loss=0.328, simple_loss=0.275, pruned_loss=0.2045, over 14469.00 frames. ], tot_loss[loss=0.3841, simple_loss=0.3193, pruned_loss=0.2541, over 1952884.62 frames. ], batch size: 18, lr: 4.91e-02, grad_scale: 4.0 2022-12-07 04:53:36,992 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 3.437e+02 4.813e+02 6.365e+02 1.224e+03, threshold=9.626e+02, percent-clipped=13.0 2022-12-07 04:53:57,848 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:54:06,980 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.96 vs. limit=5.0 2022-12-07 04:54:23,081 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0677, 3.1769, 3.0117, 3.1367, 3.1395, 2.9691, 3.2246, 3.1285], device='cuda:0'), covar=tensor([0.0565, 0.0600, 0.0517, 0.0453, 0.0495, 0.0532, 0.0430, 0.0532], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0029, 0.0032, 0.0027, 0.0031, 0.0029, 0.0029, 0.0029], device='cuda:0'), out_proj_covar=tensor([2.9974e-05, 2.9998e-05, 3.0404e-05, 2.8133e-05, 3.0273e-05, 2.9474e-05, 2.9257e-05, 2.7972e-05], device='cuda:0') 2022-12-07 04:54:23,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2022-12-07 04:54:40,829 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.42 vs. limit=2.0 2022-12-07 04:54:41,769 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 04:54:46,082 INFO [train.py:873] (0/4) Epoch 1, batch 1500, loss[loss=0.2427, simple_loss=0.2107, pruned_loss=0.1432, over 1230.00 frames. ], tot_loss[loss=0.3759, simple_loss=0.3133, pruned_loss=0.2429, over 1914713.11 frames. ], batch size: 100, lr: 4.89e-02, grad_scale: 4.0 2022-12-07 04:54:47,686 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:54:49,749 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.091e+01 2.709e+02 3.903e+02 5.151e+02 1.116e+03, threshold=7.805e+02, percent-clipped=3.0 2022-12-07 04:54:52,091 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 04:54:52,358 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2022-12-07 04:55:22,614 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:55:33,697 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:55:39,192 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.67 vs. limit=2.0 2022-12-07 04:55:59,954 INFO [train.py:873] (0/4) Epoch 1, batch 1600, loss[loss=0.397, simple_loss=0.3249, pruned_loss=0.2464, over 14276.00 frames. ], tot_loss[loss=0.3706, simple_loss=0.3092, pruned_loss=0.2345, over 1894101.05 frames. ], batch size: 76, lr: 4.88e-02, grad_scale: 8.0 2022-12-07 04:56:00,346 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.47 vs. limit=5.0 2022-12-07 04:56:03,632 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 3.341e+02 4.121e+02 5.397e+02 1.445e+03, threshold=8.242e+02, percent-clipped=8.0 2022-12-07 04:56:07,842 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 04:56:19,517 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:56:22,763 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 04:56:52,699 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={0, 2} 2022-12-07 04:57:13,819 INFO [train.py:873] (0/4) Epoch 1, batch 1700, loss[loss=0.2672, simple_loss=0.2189, pruned_loss=0.1634, over 2621.00 frames. ], tot_loss[loss=0.3653, simple_loss=0.3055, pruned_loss=0.2265, over 1983852.78 frames. ], batch size: 100, lr: 4.86e-02, grad_scale: 8.0 2022-12-07 04:57:17,699 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 3.504e+02 4.735e+02 5.720e+02 1.294e+03, threshold=9.470e+02, percent-clipped=11.0 2022-12-07 04:57:35,448 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.21 vs. limit=5.0 2022-12-07 04:57:39,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8514, 2.7203, 2.9095, 3.0109, 2.8967, 3.0176, 3.0453, 2.6332], device='cuda:0'), covar=tensor([0.0495, 0.0514, 0.0526, 0.0364, 0.0335, 0.0224, 0.0317, 0.0610], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0033, 0.0033, 0.0029, 0.0032, 0.0027, 0.0028, 0.0036], device='cuda:0'), out_proj_covar=tensor([2.7431e-05, 2.7512e-05, 2.8657e-05, 2.3324e-05, 2.8638e-05, 2.5006e-05, 2.3201e-05, 3.3919e-05], device='cuda:0') 2022-12-07 04:57:54,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2022-12-07 04:58:15,166 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 04:58:21,535 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 04:58:29,843 INFO [train.py:873] (0/4) Epoch 1, batch 1800, loss[loss=0.3859, simple_loss=0.3261, pruned_loss=0.227, over 14267.00 frames. ], tot_loss[loss=0.3596, simple_loss=0.302, pruned_loss=0.2189, over 1992650.45 frames. ], batch size: 76, lr: 4.85e-02, grad_scale: 8.0 2022-12-07 04:58:33,678 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.780e+01 3.438e+02 4.789e+02 6.464e+02 1.407e+03, threshold=9.578e+02, percent-clipped=4.0 2022-12-07 04:58:36,009 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:58:57,946 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:59:12,100 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 04:59:12,201 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:59:21,937 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 04:59:33,481 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=15.15 vs. limit=5.0 2022-12-07 04:59:34,660 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0310, 1.6404, 2.1950, 2.1234, 2.1223, 2.0972, 1.1629, 1.8619], device='cuda:0'), covar=tensor([0.0395, 0.0742, 0.0298, 0.0274, 0.0282, 0.0300, 0.1691, 0.0690], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0032, 0.0033, 0.0029, 0.0029, 0.0029, 0.0041, 0.0036], device='cuda:0'), out_proj_covar=tensor([2.3350e-05, 2.4624e-05, 2.5622e-05, 2.2758e-05, 2.3497e-05, 2.1891e-05, 4.0931e-05, 2.9015e-05], device='cuda:0') 2022-12-07 04:59:43,724 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 04:59:45,046 INFO [train.py:873] (0/4) Epoch 1, batch 1900, loss[loss=0.3663, simple_loss=0.3061, pruned_loss=0.2153, over 14282.00 frames. ], tot_loss[loss=0.3562, simple_loss=0.2992, pruned_loss=0.214, over 1982151.22 frames. ], batch size: 35, lr: 4.83e-02, grad_scale: 8.0 2022-12-07 04:59:48,578 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.566e+01 3.394e+02 4.356e+02 5.869e+02 1.320e+03, threshold=8.711e+02, percent-clipped=5.0 2022-12-07 04:59:58,569 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 05:00:01,456 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:00:08,188 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 05:00:35,443 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:01:01,970 INFO [train.py:873] (0/4) Epoch 1, batch 2000, loss[loss=0.3459, simple_loss=0.2926, pruned_loss=0.1996, over 14238.00 frames. ], tot_loss[loss=0.3539, simple_loss=0.2976, pruned_loss=0.2099, over 1954418.58 frames. ], batch size: 89, lr: 4.82e-02, grad_scale: 8.0 2022-12-07 05:01:05,829 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.042e+02 3.554e+02 4.725e+02 6.724e+02 1.490e+03, threshold=9.451e+02, percent-clipped=5.0 2022-12-07 05:01:25,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2022-12-07 05:01:36,445 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8067, 1.8640, 2.1315, 1.9291, 1.7541, 2.0609, 1.9112, 1.9258], device='cuda:0'), covar=tensor([0.0173, 0.0236, 0.0113, 0.0172, 0.0228, 0.0156, 0.0274, 0.0188], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0024, 0.0021, 0.0020, 0.0025, 0.0021, 0.0020, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.9247e-05, 1.9728e-05, 1.7634e-05, 1.5930e-05, 2.0788e-05, 1.7474e-05, 1.8399e-05, 2.0460e-05], device='cuda:0') 2022-12-07 05:01:52,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3054, 4.2048, 3.9882, 4.4168, 4.3197, 4.1076, 4.1915, 4.1899], device='cuda:0'), covar=tensor([0.0229, 0.0248, 0.0307, 0.0172, 0.0247, 0.0225, 0.0229, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0033, 0.0028, 0.0025, 0.0028, 0.0027, 0.0029, 0.0031], device='cuda:0'), out_proj_covar=tensor([2.3986e-05, 3.1846e-05, 2.5639e-05, 2.4104e-05, 2.5980e-05, 2.8066e-05, 2.7297e-05, 2.9423e-05], device='cuda:0') 2022-12-07 05:02:14,783 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 05:02:23,809 INFO [train.py:873] (0/4) Epoch 1, batch 2100, loss[loss=0.3384, simple_loss=0.2902, pruned_loss=0.1933, over 10349.00 frames. ], tot_loss[loss=0.3444, simple_loss=0.2921, pruned_loss=0.2014, over 1870587.04 frames. ], batch size: 100, lr: 4.80e-02, grad_scale: 16.0 2022-12-07 05:02:27,743 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.809e+01 3.029e+02 3.794e+02 4.821e+02 1.470e+03, threshold=7.588e+02, percent-clipped=1.0 2022-12-07 05:02:53,139 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:02:58,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-07 05:03:05,651 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1783, 2.0325, 2.4269, 2.2193, 2.1505, 2.3637, 1.9117, 2.2448], device='cuda:0'), covar=tensor([0.0289, 0.0536, 0.0263, 0.0298, 0.0427, 0.0257, 0.0550, 0.0312], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0024, 0.0022, 0.0019, 0.0025, 0.0021, 0.0020, 0.0023], device='cuda:0'), out_proj_covar=tensor([1.8657e-05, 2.1198e-05, 1.8655e-05, 1.5864e-05, 2.1657e-05, 1.7572e-05, 1.9005e-05, 2.0850e-05], device='cuda:0') 2022-12-07 05:03:38,914 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:03:45,039 INFO [train.py:873] (0/4) Epoch 1, batch 2200, loss[loss=0.2612, simple_loss=0.2128, pruned_loss=0.1548, over 2573.00 frames. ], tot_loss[loss=0.3392, simple_loss=0.289, pruned_loss=0.1965, over 1873059.74 frames. ], batch size: 100, lr: 4.78e-02, grad_scale: 16.0 2022-12-07 05:03:49,157 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 3.028e+02 4.377e+02 7.241e+02 2.014e+03, threshold=8.755e+02, percent-clipped=23.0 2022-12-07 05:03:54,619 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:01,887 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={1, 2} 2022-12-07 05:04:04,963 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:10,686 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.23 vs. limit=2.0 2022-12-07 05:04:17,600 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6952, 4.9132, 5.2572, 5.6447, 5.4183, 5.1612, 5.2007, 5.3735], device='cuda:0'), covar=tensor([0.0157, 0.0342, 0.0256, 0.0209, 0.0229, 0.0251, 0.0208, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0036, 0.0029, 0.0027, 0.0029, 0.0030, 0.0032, 0.0033], device='cuda:0'), out_proj_covar=tensor([2.6961e-05, 3.5660e-05, 2.7052e-05, 2.6201e-05, 2.8216e-05, 3.1498e-05, 2.9748e-05, 3.2747e-05], device='cuda:0') 2022-12-07 05:04:38,667 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:04:41,028 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:04:58,571 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8546, 2.7204, 2.7972, 2.8586, 2.7720, 2.8376, 3.0164, 2.9659], device='cuda:0'), covar=tensor([0.0283, 0.0396, 0.0297, 0.0246, 0.0355, 0.0344, 0.0276, 0.0275], device='cuda:0'), in_proj_covar=tensor([0.0049, 0.0051, 0.0047, 0.0046, 0.0052, 0.0048, 0.0045, 0.0045], device='cuda:0'), out_proj_covar=tensor([5.0846e-05, 4.5287e-05, 4.3737e-05, 4.3071e-05, 4.6489e-05, 4.2596e-05, 4.0718e-05, 4.3683e-05], device='cuda:0') 2022-12-07 05:05:06,463 INFO [train.py:873] (0/4) Epoch 1, batch 2300, loss[loss=0.3251, simple_loss=0.2822, pruned_loss=0.184, over 14287.00 frames. ], tot_loss[loss=0.3307, simple_loss=0.2842, pruned_loss=0.1897, over 1866914.85 frames. ], batch size: 44, lr: 4.77e-02, grad_scale: 16.0 2022-12-07 05:05:10,739 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.291e+01 3.360e+02 4.447e+02 6.275e+02 1.931e+03, threshold=8.894e+02, percent-clipped=10.0 2022-12-07 05:05:18,534 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:05:38,650 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:06:00,948 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2022-12-07 05:06:28,672 INFO [train.py:873] (0/4) Epoch 1, batch 2400, loss[loss=0.2784, simple_loss=0.2589, pruned_loss=0.1489, over 14348.00 frames. ], tot_loss[loss=0.3258, simple_loss=0.2821, pruned_loss=0.1854, over 1947502.90 frames. ], batch size: 55, lr: 4.75e-02, grad_scale: 16.0 2022-12-07 05:06:28,876 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:06:32,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 3.236e+02 4.365e+02 6.045e+02 1.327e+03, threshold=8.731e+02, percent-clipped=8.0 2022-12-07 05:06:59,532 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.31 vs. limit=2.0 2022-12-07 05:07:07,353 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 05:07:44,783 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:07:50,096 INFO [train.py:873] (0/4) Epoch 1, batch 2500, loss[loss=0.276, simple_loss=0.2155, pruned_loss=0.1682, over 1239.00 frames. ], tot_loss[loss=0.3231, simple_loss=0.2807, pruned_loss=0.1831, over 1957953.41 frames. ], batch size: 100, lr: 4.73e-02, grad_scale: 16.0 2022-12-07 05:07:54,361 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.013e+02 3.632e+02 4.712e+02 6.167e+02 1.692e+03, threshold=9.425e+02, percent-clipped=8.0 2022-12-07 05:08:00,239 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:11,297 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:23,745 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:24,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8683, 1.0512, 1.3705, 1.2902, 1.3272, 1.2852, 1.0506, 1.2994], device='cuda:0'), covar=tensor([0.0424, 0.0427, 0.0109, 0.0183, 0.0137, 0.0126, 0.0380, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0027, 0.0020, 0.0022, 0.0021, 0.0021, 0.0024, 0.0026], device='cuda:0'), out_proj_covar=tensor([2.8100e-05, 2.3917e-05, 1.6506e-05, 1.6468e-05, 1.6493e-05, 1.6763e-05, 2.2510e-05, 2.2249e-05], device='cuda:0') 2022-12-07 05:08:39,741 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:08:47,874 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 05:08:50,008 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:09:13,572 INFO [train.py:873] (0/4) Epoch 1, batch 2600, loss[loss=0.2158, simple_loss=0.1726, pruned_loss=0.1295, over 1311.00 frames. ], tot_loss[loss=0.3213, simple_loss=0.2796, pruned_loss=0.1817, over 1965414.54 frames. ], batch size: 100, lr: 4.71e-02, grad_scale: 16.0 2022-12-07 05:09:16,729 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3402, 4.0837, 3.8467, 4.3735, 4.2813, 4.2199, 4.1620, 3.8471], device='cuda:0'), covar=tensor([0.0211, 0.0313, 0.0356, 0.0289, 0.0301, 0.0291, 0.0265, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0043, 0.0033, 0.0032, 0.0034, 0.0035, 0.0038, 0.0039], device='cuda:0'), out_proj_covar=tensor([3.1921e-05, 4.4781e-05, 3.2237e-05, 3.3480e-05, 3.4583e-05, 3.8896e-05, 3.9373e-05, 4.1227e-05], device='cuda:0') 2022-12-07 05:09:17,459 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.018e+02 2.998e+02 4.312e+02 6.612e+02 1.607e+03, threshold=8.624e+02, percent-clipped=4.0 2022-12-07 05:09:53,466 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.10 vs. limit=2.0 2022-12-07 05:10:11,871 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1052, 5.0357, 4.7743, 5.2071, 5.1283, 4.3995, 5.4850, 5.0625], device='cuda:0'), covar=tensor([0.0540, 0.0406, 0.0595, 0.0454, 0.0417, 0.0457, 0.0387, 0.0507], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0041, 0.0059, 0.0049, 0.0055, 0.0043, 0.0045, 0.0054], device='cuda:0'), out_proj_covar=tensor([6.0357e-05, 4.7539e-05, 6.5190e-05, 5.4094e-05, 6.0238e-05, 4.7578e-05, 5.5687e-05, 6.0029e-05], device='cuda:0') 2022-12-07 05:10:23,704 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:10:32,509 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:10:36,427 INFO [train.py:873] (0/4) Epoch 1, batch 2700, loss[loss=0.2533, simple_loss=0.2106, pruned_loss=0.1481, over 2620.00 frames. ], tot_loss[loss=0.3172, simple_loss=0.2771, pruned_loss=0.1788, over 1949687.45 frames. ], batch size: 100, lr: 4.69e-02, grad_scale: 16.0 2022-12-07 05:10:40,371 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 3.041e+02 4.444e+02 6.160e+02 1.367e+03, threshold=8.889e+02, percent-clipped=4.0 2022-12-07 05:11:11,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3114, 1.5722, 2.4018, 1.8909, 2.3859, 2.2583, 2.1932, 1.7627], device='cuda:0'), covar=tensor([0.0679, 0.4563, 0.0622, 0.1168, 0.0554, 0.0679, 0.0855, 0.4251], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0076, 0.0035, 0.0044, 0.0037, 0.0038, 0.0029, 0.0074], device='cuda:0'), out_proj_covar=tensor([2.2905e-05, 7.0341e-05, 1.9632e-05, 3.0316e-05, 2.2173e-05, 2.1806e-05, 1.9872e-05, 6.2706e-05], device='cuda:0') 2022-12-07 05:11:14,218 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={0, 1} 2022-12-07 05:11:41,753 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 05:11:44,771 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1855, 1.3085, 1.7171, 1.5674, 1.5810, 1.7733, 1.1486, 1.5243], device='cuda:0'), covar=tensor([0.0870, 0.0605, 0.0417, 0.0262, 0.0291, 0.0196, 0.0673, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0030, 0.0021, 0.0025, 0.0023, 0.0021, 0.0026, 0.0026], device='cuda:0'), out_proj_covar=tensor([3.3350e-05, 2.7815e-05, 1.8291e-05, 1.9714e-05, 1.9939e-05, 1.7738e-05, 2.6081e-05, 2.3002e-05], device='cuda:0') 2022-12-07 05:12:00,033 INFO [train.py:873] (0/4) Epoch 1, batch 2800, loss[loss=0.2977, simple_loss=0.254, pruned_loss=0.1707, over 3868.00 frames. ], tot_loss[loss=0.3146, simple_loss=0.2754, pruned_loss=0.177, over 1882553.03 frames. ], batch size: 100, lr: 4.67e-02, grad_scale: 8.0 2022-12-07 05:12:05,006 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 3.640e+02 4.591e+02 6.544e+02 1.997e+03, threshold=9.181e+02, percent-clipped=12.0 2022-12-07 05:12:16,242 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4870, 1.9733, 3.7472, 3.5934, 3.4358, 3.4799, 2.4665, 3.7184], device='cuda:0'), covar=tensor([0.0319, 0.1626, 0.0253, 0.0318, 0.0282, 0.0306, 0.0834, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0040, 0.0027, 0.0029, 0.0033, 0.0028, 0.0027, 0.0026], device='cuda:0'), out_proj_covar=tensor([2.7958e-05, 4.1197e-05, 2.7714e-05, 2.6988e-05, 3.1797e-05, 2.6176e-05, 2.7534e-05, 2.6266e-05], device='cuda:0') 2022-12-07 05:12:18,674 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:13:09,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 05:13:22,398 INFO [train.py:873] (0/4) Epoch 1, batch 2900, loss[loss=0.3377, simple_loss=0.2889, pruned_loss=0.1932, over 12791.00 frames. ], tot_loss[loss=0.3132, simple_loss=0.2751, pruned_loss=0.1757, over 1966413.46 frames. ], batch size: 100, lr: 4.65e-02, grad_scale: 8.0 2022-12-07 05:13:25,436 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.13 vs. limit=2.0 2022-12-07 05:13:27,205 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 3.059e+02 4.261e+02 6.986e+02 2.302e+03, threshold=8.522e+02, percent-clipped=12.0 2022-12-07 05:13:52,960 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:14:43,004 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:14:45,377 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 05:14:48,188 INFO [train.py:873] (0/4) Epoch 1, batch 3000, loss[loss=0.3128, simple_loss=0.2657, pruned_loss=0.1799, over 9500.00 frames. ], tot_loss[loss=0.3093, simple_loss=0.2732, pruned_loss=0.1728, over 1947708.14 frames. ], batch size: 100, lr: 4.63e-02, grad_scale: 8.0 2022-12-07 05:14:48,189 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 05:14:54,716 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5137, 1.7064, 1.8957, 1.7267, 2.1403, 1.9856, 1.6767, 1.9472], device='cuda:0'), covar=tensor([0.2813, 0.1548, 0.1411, 0.1440, 0.0869, 0.1118, 0.1052, 0.0853], device='cuda:0'), in_proj_covar=tensor([0.0042, 0.0035, 0.0026, 0.0029, 0.0028, 0.0024, 0.0030, 0.0031], device='cuda:0'), out_proj_covar=tensor([4.3147e-05, 3.3952e-05, 2.4162e-05, 2.4522e-05, 2.4741e-05, 2.2668e-05, 3.1321e-05, 2.7526e-05], device='cuda:0') 2022-12-07 05:14:56,536 INFO [train.py:905] (0/4) Epoch 1, validation: loss=0.2054, simple_loss=0.2366, pruned_loss=0.08706, over 857387.00 frames. 2022-12-07 05:14:56,536 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17099MB 2022-12-07 05:15:01,308 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.332e+02 3.340e+02 4.389e+02 5.804e+02 1.068e+03, threshold=8.777e+02, percent-clipped=8.0 2022-12-07 05:15:11,978 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:15:30,345 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:15:32,900 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:15:34,123 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=23.23 vs. limit=5.0 2022-12-07 05:16:04,058 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:16:21,383 INFO [train.py:873] (0/4) Epoch 1, batch 3100, loss[loss=0.3083, simple_loss=0.2762, pruned_loss=0.1702, over 14234.00 frames. ], tot_loss[loss=0.3066, simple_loss=0.2718, pruned_loss=0.1707, over 1943975.36 frames. ], batch size: 25, lr: 4.61e-02, grad_scale: 8.0 2022-12-07 05:16:26,604 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 3.492e+02 4.643e+02 6.451e+02 1.266e+03, threshold=9.287e+02, percent-clipped=10.0 2022-12-07 05:16:34,220 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3714, 4.0620, 4.7845, 4.0917, 4.3513, 4.4904, 3.6153, 4.5983], device='cuda:0'), covar=tensor([0.0272, 0.0402, 0.0231, 0.0510, 0.0289, 0.0099, 0.0932, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0044, 0.0042, 0.0038, 0.0047, 0.0032, 0.0056, 0.0048], device='cuda:0'), out_proj_covar=tensor([4.8328e-05, 5.2198e-05, 5.1802e-05, 4.2171e-05, 5.6730e-05, 3.6026e-05, 6.6157e-05, 5.6899e-05], device='cuda:0') 2022-12-07 05:17:27,784 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:17:45,612 INFO [train.py:873] (0/4) Epoch 1, batch 3200, loss[loss=0.3165, simple_loss=0.2822, pruned_loss=0.1754, over 14507.00 frames. ], tot_loss[loss=0.3062, simple_loss=0.2715, pruned_loss=0.1705, over 2000668.99 frames. ], batch size: 49, lr: 4.59e-02, grad_scale: 8.0 2022-12-07 05:17:50,388 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 3.868e+02 5.118e+02 7.409e+02 2.192e+03, threshold=1.024e+03, percent-clipped=11.0 2022-12-07 05:18:06,360 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8761, 2.0224, 2.7977, 2.5922, 2.5476, 2.4739, 1.6950, 2.7499], device='cuda:0'), covar=tensor([0.0875, 0.0553, 0.0163, 0.0206, 0.0176, 0.0252, 0.0503, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0036, 0.0024, 0.0027, 0.0027, 0.0027, 0.0028, 0.0026], device='cuda:0'), out_proj_covar=tensor([4.6654e-05, 3.6287e-05, 2.4007e-05, 3.0060e-05, 2.5221e-05, 2.5841e-05, 2.8589e-05, 2.5810e-05], device='cuda:0') 2022-12-07 05:18:33,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9861, 1.9729, 1.7321, 1.7309, 1.7587, 1.8294, 1.9908, 1.9218], device='cuda:0'), covar=tensor([0.1053, 0.1002, 0.1510, 0.1366, 0.1347, 0.0794, 0.0966, 0.1562], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0055, 0.0075, 0.0065, 0.0071, 0.0051, 0.0058, 0.0070], device='cuda:0'), out_proj_covar=tensor([7.9106e-05, 6.6432e-05, 8.6591e-05, 7.7754e-05, 8.0531e-05, 5.8692e-05, 7.8480e-05, 8.7088e-05], device='cuda:0') 2022-12-07 05:18:57,697 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:19:01,663 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:19:08,886 INFO [train.py:873] (0/4) Epoch 1, batch 3300, loss[loss=0.3199, simple_loss=0.2791, pruned_loss=0.1803, over 14635.00 frames. ], tot_loss[loss=0.3009, simple_loss=0.2682, pruned_loss=0.1668, over 1964532.95 frames. ], batch size: 23, lr: 4.57e-02, grad_scale: 8.0 2022-12-07 05:19:14,104 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.921e+02 3.689e+02 4.978e+02 9.002e+02, threshold=7.379e+02, percent-clipped=0.0 2022-12-07 05:19:42,742 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:19:49,899 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={0, 3} 2022-12-07 05:19:57,491 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.47 vs. limit=5.0 2022-12-07 05:20:00,495 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8910, 3.6686, 3.3568, 3.7349, 3.5420, 2.6694, 4.1019, 3.8200], device='cuda:0'), covar=tensor([0.0621, 0.0698, 0.0897, 0.0715, 0.0830, 0.0972, 0.0546, 0.1115], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0053, 0.0073, 0.0063, 0.0068, 0.0051, 0.0058, 0.0068], device='cuda:0'), out_proj_covar=tensor([7.2902e-05, 6.4476e-05, 8.5868e-05, 7.5669e-05, 7.6808e-05, 5.8518e-05, 7.8973e-05, 8.3732e-05], device='cuda:0') 2022-12-07 05:20:06,815 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 05:20:11,328 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:20:24,033 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:20:34,758 INFO [train.py:873] (0/4) Epoch 1, batch 3400, loss[loss=0.2693, simple_loss=0.2441, pruned_loss=0.1472, over 6937.00 frames. ], tot_loss[loss=0.3005, simple_loss=0.2683, pruned_loss=0.1664, over 2010949.92 frames. ], batch size: 100, lr: 4.55e-02, grad_scale: 8.0 2022-12-07 05:20:39,711 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.829e+02 3.330e+02 4.669e+02 8.149e+02 2.778e+03, threshold=9.337e+02, percent-clipped=27.0 2022-12-07 05:20:44,111 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2979, 3.9597, 3.8537, 4.1809, 4.0793, 3.2586, 4.4452, 4.1758], device='cuda:0'), covar=tensor([0.0716, 0.1106, 0.0883, 0.0825, 0.0697, 0.0871, 0.0660, 0.1191], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0054, 0.0073, 0.0063, 0.0067, 0.0050, 0.0057, 0.0066], device='cuda:0'), out_proj_covar=tensor([7.2718e-05, 6.6242e-05, 8.6634e-05, 7.6356e-05, 7.6509e-05, 5.7674e-05, 7.8590e-05, 8.2759e-05], device='cuda:0') 2022-12-07 05:20:45,374 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2022-12-07 05:21:05,559 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8365, 0.8954, 1.1684, 1.0183, 1.2928, 1.2931, 0.9059, 1.2064], device='cuda:0'), covar=tensor([0.0899, 0.0579, 0.0193, 0.0157, 0.0218, 0.0130, 0.0418, 0.0144], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0042, 0.0027, 0.0032, 0.0032, 0.0028, 0.0036, 0.0037], device='cuda:0'), out_proj_covar=tensor([6.2443e-05, 4.3044e-05, 2.8083e-05, 3.0823e-05, 3.0527e-05, 2.7194e-05, 3.9090e-05, 3.5966e-05], device='cuda:0') 2022-12-07 05:21:08,231 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.13 vs. limit=5.0 2022-12-07 05:21:36,977 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6530, 1.9722, 1.6923, 2.1534, 2.1802, 1.8394, 2.1342, 2.2242], device='cuda:0'), covar=tensor([0.0093, 0.0107, 0.0186, 0.0093, 0.0119, 0.0115, 0.0097, 0.0083], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0014, 0.0012, 0.0012, 0.0012, 0.0011, 0.0012], device='cuda:0'), out_proj_covar=tensor([9.8758e-06, 9.5908e-06, 1.3101e-05, 1.0212e-05, 1.0146e-05, 1.0489e-05, 9.0904e-06, 1.0160e-05], device='cuda:0') 2022-12-07 05:21:40,705 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 05:21:43,557 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:22:01,459 INFO [train.py:873] (0/4) Epoch 1, batch 3500, loss[loss=0.2488, simple_loss=0.2407, pruned_loss=0.1284, over 14303.00 frames. ], tot_loss[loss=0.299, simple_loss=0.2674, pruned_loss=0.1653, over 1996173.54 frames. ], batch size: 39, lr: 4.53e-02, grad_scale: 8.0 2022-12-07 05:22:06,643 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.311e+02 3.200e+02 4.145e+02 5.923e+02 9.666e+02, threshold=8.289e+02, percent-clipped=1.0 2022-12-07 05:22:09,734 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.12 vs. limit=2.0 2022-12-07 05:22:24,422 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:22:30,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 05:22:40,257 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:22:54,629 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6035, 1.6092, 2.3386, 1.7026, 2.0796, 1.9972, 1.8454, 1.7216], device='cuda:0'), covar=tensor([0.3102, 0.1290, 0.0803, 0.2371, 0.0655, 0.0759, 0.0933, 0.0775], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0043, 0.0028, 0.0032, 0.0032, 0.0027, 0.0037, 0.0037], device='cuda:0'), out_proj_covar=tensor([6.7441e-05, 4.4956e-05, 3.0292e-05, 3.2060e-05, 3.1530e-05, 2.8270e-05, 4.0579e-05, 3.6511e-05], device='cuda:0') 2022-12-07 05:23:19,099 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 05:23:21,059 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:23:28,041 INFO [train.py:873] (0/4) Epoch 1, batch 3600, loss[loss=0.3604, simple_loss=0.2942, pruned_loss=0.2133, over 9518.00 frames. ], tot_loss[loss=0.2988, simple_loss=0.2679, pruned_loss=0.1648, over 2040997.60 frames. ], batch size: 100, lr: 4.50e-02, grad_scale: 8.0 2022-12-07 05:23:33,317 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 3.769e+02 4.638e+02 5.846e+02 1.150e+03, threshold=9.276e+02, percent-clipped=6.0 2022-12-07 05:23:33,582 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={1, 3} 2022-12-07 05:24:02,791 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:24:05,413 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:24:33,191 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:24:34,015 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:24:39,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.49 vs. limit=5.0 2022-12-07 05:24:55,836 INFO [train.py:873] (0/4) Epoch 1, batch 3700, loss[loss=0.3294, simple_loss=0.2888, pruned_loss=0.185, over 12739.00 frames. ], tot_loss[loss=0.2975, simple_loss=0.267, pruned_loss=0.1639, over 2028806.42 frames. ], batch size: 100, lr: 4.48e-02, grad_scale: 8.0 2022-12-07 05:25:00,987 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 3.561e+02 4.722e+02 6.251e+02 1.502e+03, threshold=9.443e+02, percent-clipped=7.0 2022-12-07 05:25:14,813 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:25:26,867 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={2, 3} 2022-12-07 05:25:36,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2022-12-07 05:25:48,865 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7459, 1.5401, 1.8614, 1.7406, 2.2328, 1.7808, 1.8472, 2.2489], device='cuda:0'), covar=tensor([0.0297, 0.1801, 0.0428, 0.2218, 0.0299, 0.1432, 0.0344, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0202, 0.0107, 0.0201, 0.0076, 0.0214, 0.0105, 0.0083], device='cuda:0'), out_proj_covar=tensor([6.8657e-05, 1.5803e-04, 7.4325e-05, 1.6239e-04, 5.2339e-05, 1.6423e-04, 7.1873e-05, 5.6982e-05], device='cuda:0') 2022-12-07 05:26:22,261 INFO [train.py:873] (0/4) Epoch 1, batch 3800, loss[loss=0.2668, simple_loss=0.2271, pruned_loss=0.1532, over 4945.00 frames. ], tot_loss[loss=0.2948, simple_loss=0.265, pruned_loss=0.1624, over 2013483.64 frames. ], batch size: 100, lr: 4.46e-02, grad_scale: 8.0 2022-12-07 05:26:25,040 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5581, 1.8574, 2.2411, 2.1041, 2.6443, 1.9801, 2.4310, 2.7217], device='cuda:0'), covar=tensor([0.0433, 0.3287, 0.0550, 0.3248, 0.0261, 0.2407, 0.0439, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0197, 0.0107, 0.0200, 0.0075, 0.0214, 0.0105, 0.0086], device='cuda:0'), out_proj_covar=tensor([6.9004e-05, 1.5491e-04, 7.4541e-05, 1.6131e-04, 5.2596e-05, 1.6386e-04, 7.1704e-05, 5.8650e-05], device='cuda:0') 2022-12-07 05:26:27,325 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 3.207e+02 4.617e+02 6.095e+02 1.543e+03, threshold=9.233e+02, percent-clipped=5.0 2022-12-07 05:27:14,403 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7808, 1.3447, 1.6168, 1.2355, 1.8123, 0.9464, 1.3044, 1.6302], device='cuda:0'), covar=tensor([0.0259, 0.1726, 0.0425, 0.0958, 0.0305, 0.0848, 0.0612, 0.0269], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0075, 0.0042, 0.0056, 0.0034, 0.0040, 0.0036, 0.0041], device='cuda:0'), out_proj_covar=tensor([3.1344e-05, 7.4389e-05, 3.9914e-05, 5.6530e-05, 3.3494e-05, 4.1576e-05, 3.9598e-05, 3.8879e-05], device='cuda:0') 2022-12-07 05:27:50,295 INFO [train.py:873] (0/4) Epoch 1, batch 3900, loss[loss=0.2925, simple_loss=0.2759, pruned_loss=0.1546, over 14601.00 frames. ], tot_loss[loss=0.294, simple_loss=0.2644, pruned_loss=0.1618, over 1995958.09 frames. ], batch size: 22, lr: 4.44e-02, grad_scale: 8.0 2022-12-07 05:27:51,248 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 05:27:55,147 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 3.646e+02 5.236e+02 6.626e+02 1.873e+03, threshold=1.047e+03, percent-clipped=8.0 2022-12-07 05:28:11,775 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:28:20,869 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.54 vs. limit=2.0 2022-12-07 05:28:27,593 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:29:05,128 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:29:09,492 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:29:17,376 INFO [train.py:873] (0/4) Epoch 1, batch 4000, loss[loss=0.306, simple_loss=0.2752, pruned_loss=0.1684, over 11989.00 frames. ], tot_loss[loss=0.2931, simple_loss=0.2643, pruned_loss=0.161, over 2016774.35 frames. ], batch size: 100, lr: 4.42e-02, grad_scale: 8.0 2022-12-07 05:29:22,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 3.697e+02 5.328e+02 7.328e+02 1.359e+03, threshold=1.066e+03, percent-clipped=6.0 2022-12-07 05:29:45,731 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:30:22,699 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9011, 3.1744, 3.2513, 3.7425, 3.8403, 3.9144, 3.1387, 2.8929], device='cuda:0'), covar=tensor([0.0355, 0.2083, 0.0209, 0.0476, 0.0225, 0.0266, 0.0259, 0.1796], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0095, 0.0043, 0.0050, 0.0045, 0.0051, 0.0042, 0.0097], device='cuda:0'), out_proj_covar=tensor([3.2348e-05, 7.3009e-05, 2.3724e-05, 3.1186e-05, 2.5853e-05, 3.0001e-05, 2.6943e-05, 7.0684e-05], device='cuda:0') 2022-12-07 05:30:45,997 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2022-12-07 05:30:47,313 INFO [train.py:873] (0/4) Epoch 1, batch 4100, loss[loss=0.2458, simple_loss=0.2184, pruned_loss=0.1366, over 3848.00 frames. ], tot_loss[loss=0.2911, simple_loss=0.2627, pruned_loss=0.1598, over 1908241.94 frames. ], batch size: 100, lr: 4.40e-02, grad_scale: 8.0 2022-12-07 05:30:52,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.129e+02 3.295e+02 4.605e+02 5.758e+02 1.082e+03, threshold=9.210e+02, percent-clipped=1.0 2022-12-07 05:31:22,927 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:31:33,075 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2022-12-07 05:31:33,123 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.20 vs. limit=5.0 2022-12-07 05:31:50,039 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0631, 2.0263, 2.0348, 1.8328, 1.9938, 1.9652, 1.9821, 1.9433], device='cuda:0'), covar=tensor([0.0326, 0.0386, 0.0408, 0.0607, 0.0375, 0.0375, 0.0555, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0070, 0.0081, 0.0056, 0.0079, 0.0073, 0.0075, 0.0068], device='cuda:0'), out_proj_covar=tensor([8.5636e-05, 8.1717e-05, 8.6337e-05, 6.2660e-05, 8.7928e-05, 8.0035e-05, 9.2121e-05, 8.2582e-05], device='cuda:0') 2022-12-07 05:32:16,950 INFO [train.py:873] (0/4) Epoch 1, batch 4200, loss[loss=0.2797, simple_loss=0.2612, pruned_loss=0.1491, over 14303.00 frames. ], tot_loss[loss=0.2896, simple_loss=0.2617, pruned_loss=0.1588, over 1937953.34 frames. ], batch size: 76, lr: 4.38e-02, grad_scale: 8.0 2022-12-07 05:32:17,760 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:32:17,810 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:32:21,867 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 3.430e+02 4.467e+02 5.944e+02 1.498e+03, threshold=8.934e+02, percent-clipped=5.0 2022-12-07 05:32:56,494 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.20 vs. limit=5.0 2022-12-07 05:33:00,182 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:33:20,537 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.85 vs. limit=5.0 2022-12-07 05:33:28,865 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:33:34,070 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:33:45,947 INFO [train.py:873] (0/4) Epoch 1, batch 4300, loss[loss=0.2543, simple_loss=0.2086, pruned_loss=0.15, over 1257.00 frames. ], tot_loss[loss=0.2884, simple_loss=0.2612, pruned_loss=0.1578, over 1958969.72 frames. ], batch size: 100, lr: 4.35e-02, grad_scale: 8.0 2022-12-07 05:33:48,642 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0028, 0.8207, 0.8203, 0.8146, 0.8802, 0.6769, 0.9347, 0.8832], device='cuda:0'), covar=tensor([0.0080, 0.0067, 0.0097, 0.0054, 0.0110, 0.0091, 0.0083, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0017, 0.0014, 0.0015, 0.0015, 0.0013, 0.0014], device='cuda:0'), out_proj_covar=tensor([1.2567e-05, 1.3427e-05, 1.8142e-05, 1.3525e-05, 1.4528e-05, 1.4049e-05, 1.2524e-05, 1.5081e-05], device='cuda:0') 2022-12-07 05:33:51,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.040e+01 3.141e+02 3.918e+02 5.765e+02 1.012e+03, threshold=7.835e+02, percent-clipped=6.0 2022-12-07 05:34:02,548 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.35 vs. limit=2.0 2022-12-07 05:34:05,499 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5902, 4.3423, 4.4473, 4.5324, 4.6372, 4.5374, 4.4244, 4.4384], device='cuda:0'), covar=tensor([0.0507, 0.0495, 0.0526, 0.0250, 0.0388, 0.0521, 0.0736, 0.0818], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0066, 0.0080, 0.0053, 0.0077, 0.0069, 0.0071, 0.0066], device='cuda:0'), out_proj_covar=tensor([8.3905e-05, 7.7156e-05, 8.7445e-05, 6.0531e-05, 8.6732e-05, 7.7029e-05, 8.9464e-05, 8.0983e-05], device='cuda:0') 2022-12-07 05:34:13,529 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 05:34:29,069 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:34:55,476 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 05:34:56,531 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:35:15,296 INFO [train.py:873] (0/4) Epoch 1, batch 4400, loss[loss=0.3275, simple_loss=0.2826, pruned_loss=0.1862, over 8575.00 frames. ], tot_loss[loss=0.2864, simple_loss=0.2601, pruned_loss=0.1564, over 1999950.73 frames. ], batch size: 100, lr: 4.33e-02, grad_scale: 8.0 2022-12-07 05:35:20,266 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.063e+02 3.285e+02 4.190e+02 5.766e+02 1.366e+03, threshold=8.380e+02, percent-clipped=8.0 2022-12-07 05:35:55,131 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4763, 4.7541, 4.6434, 5.5085, 5.4760, 5.0283, 5.2389, 5.0051], device='cuda:0'), covar=tensor([0.0192, 0.0541, 0.0349, 0.0302, 0.0256, 0.0316, 0.0322, 0.0284], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0063, 0.0051, 0.0048, 0.0045, 0.0055, 0.0068, 0.0063], device='cuda:0'), out_proj_covar=tensor([5.6314e-05, 8.3409e-05, 6.5270e-05, 6.8426e-05, 6.0637e-05, 7.8379e-05, 1.0224e-04, 8.7716e-05], device='cuda:0') 2022-12-07 05:36:40,444 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:36:44,319 INFO [train.py:873] (0/4) Epoch 1, batch 4500, loss[loss=0.2559, simple_loss=0.2471, pruned_loss=0.1324, over 13888.00 frames. ], tot_loss[loss=0.2827, simple_loss=0.2572, pruned_loss=0.1541, over 1920002.82 frames. ], batch size: 20, lr: 4.31e-02, grad_scale: 8.0 2022-12-07 05:36:48,738 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3614, 1.0364, 1.1656, 1.0964, 1.4753, 1.4518, 1.1621, 1.0077], device='cuda:0'), covar=tensor([0.0399, 0.0754, 0.0473, 0.1001, 0.0302, 0.0337, 0.0677, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0027, 0.0031, 0.0025, 0.0025, 0.0029, 0.0025, 0.0027], device='cuda:0'), out_proj_covar=tensor([2.6980e-05, 3.0486e-05, 3.3500e-05, 2.7672e-05, 2.4581e-05, 3.0489e-05, 2.6684e-05, 2.7422e-05], device='cuda:0') 2022-12-07 05:36:49,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 3.529e+02 4.865e+02 6.809e+02 1.533e+03, threshold=9.730e+02, percent-clipped=12.0 2022-12-07 05:37:41,178 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8329, 2.7263, 2.7075, 2.7748, 2.8355, 2.7854, 2.8638, 2.6842], device='cuda:0'), covar=tensor([0.0624, 0.0694, 0.0777, 0.0407, 0.0623, 0.0587, 0.0690, 0.1037], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0070, 0.0085, 0.0057, 0.0080, 0.0074, 0.0076, 0.0067], device='cuda:0'), out_proj_covar=tensor([8.8758e-05, 8.2374e-05, 9.4208e-05, 6.5919e-05, 9.2914e-05, 8.3135e-05, 9.7811e-05, 8.1755e-05], device='cuda:0') 2022-12-07 05:37:54,635 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:38:09,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8229, 1.8512, 2.6758, 2.1372, 2.8553, 2.0074, 2.7647, 2.6033], device='cuda:0'), covar=tensor([0.0417, 0.3399, 0.0463, 0.4278, 0.0224, 0.2539, 0.0436, 0.0323], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0210, 0.0130, 0.0272, 0.0091, 0.0235, 0.0128, 0.0099], device='cuda:0'), out_proj_covar=tensor([9.0870e-05, 1.7032e-04, 9.3834e-05, 2.1961e-04, 6.8490e-05, 1.8398e-04, 9.3032e-05, 7.1932e-05], device='cuda:0') 2022-12-07 05:38:11,655 INFO [train.py:873] (0/4) Epoch 1, batch 4600, loss[loss=0.2914, simple_loss=0.2678, pruned_loss=0.1575, over 14283.00 frames. ], tot_loss[loss=0.2833, simple_loss=0.2581, pruned_loss=0.1543, over 1946853.45 frames. ], batch size: 44, lr: 4.29e-02, grad_scale: 8.0 2022-12-07 05:38:16,960 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 3.319e+02 4.792e+02 5.771e+02 3.822e+03, threshold=9.585e+02, percent-clipped=7.0 2022-12-07 05:38:25,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 05:38:37,723 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:38:49,782 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:39:10,722 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5235, 3.4667, 3.4386, 2.9512, 2.9308, 2.9546, 3.5008, 2.0461], device='cuda:0'), covar=tensor([0.0285, 0.0418, 0.0259, 0.0603, 0.0328, 0.1146, 0.0135, 0.1338], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0043, 0.0030, 0.0038, 0.0043, 0.0066, 0.0030, 0.0059], device='cuda:0'), out_proj_covar=tensor([2.8551e-05, 3.1189e-05, 2.1747e-05, 2.8167e-05, 3.0428e-05, 5.2964e-05, 1.9699e-05, 4.8241e-05], device='cuda:0') 2022-12-07 05:39:36,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2022-12-07 05:39:41,707 INFO [train.py:873] (0/4) Epoch 1, batch 4700, loss[loss=0.2873, simple_loss=0.2685, pruned_loss=0.153, over 14207.00 frames. ], tot_loss[loss=0.2819, simple_loss=0.2575, pruned_loss=0.1532, over 1973299.92 frames. ], batch size: 89, lr: 4.27e-02, grad_scale: 8.0 2022-12-07 05:39:46,822 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 3.103e+02 4.134e+02 5.921e+02 2.901e+03, threshold=8.268e+02, percent-clipped=6.0 2022-12-07 05:40:34,430 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 05:41:06,488 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:41:10,380 INFO [train.py:873] (0/4) Epoch 1, batch 4800, loss[loss=0.2537, simple_loss=0.2109, pruned_loss=0.1482, over 2636.00 frames. ], tot_loss[loss=0.2797, simple_loss=0.2556, pruned_loss=0.1519, over 1961855.71 frames. ], batch size: 100, lr: 4.25e-02, grad_scale: 16.0 2022-12-07 05:41:15,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 3.055e+02 4.514e+02 6.112e+02 1.452e+03, threshold=9.028e+02, percent-clipped=8.0 2022-12-07 05:41:49,172 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:42:05,644 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6870, 1.4044, 2.2143, 2.3382, 2.2319, 2.0180, 1.4270, 2.3042], device='cuda:0'), covar=tensor([0.0604, 0.0844, 0.0143, 0.0148, 0.0101, 0.0141, 0.0432, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0070, 0.0036, 0.0044, 0.0038, 0.0040, 0.0038, 0.0037], device='cuda:0'), out_proj_covar=tensor([9.3967e-05, 9.2019e-05, 4.5000e-05, 6.4714e-05, 4.6238e-05, 5.2922e-05, 5.1714e-05, 4.5860e-05], device='cuda:0') 2022-12-07 05:42:14,664 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9068, 1.0333, 0.8629, 1.7786, 1.1877, 1.9356, 1.7790, 1.3990], device='cuda:0'), covar=tensor([0.0545, 0.0261, 0.0675, 0.0198, 0.0329, 0.0138, 0.0274, 0.0210], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0017, 0.0020, 0.0017, 0.0017, 0.0017, 0.0016, 0.0017], device='cuda:0'), out_proj_covar=tensor([1.5821e-05, 1.7293e-05, 2.4041e-05, 1.6431e-05, 1.8054e-05, 1.6768e-05, 1.7418e-05, 1.8121e-05], device='cuda:0') 2022-12-07 05:42:30,774 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9852, 1.0175, 1.2243, 0.7921, 1.2454, 0.8234, 0.8457, 1.0361], device='cuda:0'), covar=tensor([0.0333, 0.0535, 0.0337, 0.0601, 0.0338, 0.0370, 0.0473, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0028, 0.0031, 0.0026, 0.0025, 0.0030, 0.0023, 0.0026], device='cuda:0'), out_proj_covar=tensor([3.0645e-05, 3.2884e-05, 3.4724e-05, 3.0611e-05, 2.6481e-05, 3.3182e-05, 2.6772e-05, 2.8029e-05], device='cuda:0') 2022-12-07 05:42:39,196 INFO [train.py:873] (0/4) Epoch 1, batch 4900, loss[loss=0.2974, simple_loss=0.2656, pruned_loss=0.1646, over 6926.00 frames. ], tot_loss[loss=0.2795, simple_loss=0.2555, pruned_loss=0.1517, over 1916594.74 frames. ], batch size: 100, lr: 4.23e-02, grad_scale: 16.0 2022-12-07 05:42:44,141 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 3.107e+02 4.475e+02 6.071e+02 1.419e+03, threshold=8.951e+02, percent-clipped=8.0 2022-12-07 05:42:59,972 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0084, 1.5085, 1.8640, 1.2404, 1.6495, 1.0919, 1.5321, 1.8823], device='cuda:0'), covar=tensor([0.0237, 0.2781, 0.0402, 0.1392, 0.0838, 0.0722, 0.1279, 0.0395], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0088, 0.0043, 0.0063, 0.0040, 0.0047, 0.0037, 0.0044], device='cuda:0'), out_proj_covar=tensor([4.0371e-05, 9.9086e-05, 4.9496e-05, 7.1425e-05, 5.0211e-05, 5.5884e-05, 5.0101e-05, 4.9800e-05], device='cuda:0') 2022-12-07 05:43:16,681 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:43:44,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8833, 0.7487, 1.3023, 0.8248, 1.2412, 0.8736, 0.8597, 1.0500], device='cuda:0'), covar=tensor([0.0659, 0.1074, 0.0402, 0.0943, 0.0446, 0.0475, 0.0664, 0.0362], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0030, 0.0032, 0.0028, 0.0026, 0.0031, 0.0024, 0.0027], device='cuda:0'), out_proj_covar=tensor([3.2060e-05, 3.7011e-05, 3.6074e-05, 3.4581e-05, 2.7101e-05, 3.4232e-05, 2.8004e-05, 2.9649e-05], device='cuda:0') 2022-12-07 05:43:56,960 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6942, 1.5812, 1.6384, 1.7772, 1.9519, 1.1837, 1.6177, 1.9204], device='cuda:0'), covar=tensor([0.2784, 0.1234, 0.1247, 0.0649, 0.0400, 0.1458, 0.0615, 0.0506], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0054, 0.0033, 0.0033, 0.0039, 0.0033, 0.0042, 0.0041], device='cuda:0'), out_proj_covar=tensor([1.0682e-04, 7.0063e-05, 4.8417e-05, 4.8883e-05, 4.7157e-05, 4.7514e-05, 5.7194e-05, 5.0906e-05], device='cuda:0') 2022-12-07 05:43:58,609 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:43:59,272 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 05:44:05,892 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-5000.pt 2022-12-07 05:44:10,009 INFO [train.py:873] (0/4) Epoch 1, batch 5000, loss[loss=0.3035, simple_loss=0.2651, pruned_loss=0.171, over 14156.00 frames. ], tot_loss[loss=0.2766, simple_loss=0.2538, pruned_loss=0.1497, over 1940351.29 frames. ], batch size: 99, lr: 4.20e-02, grad_scale: 16.0 2022-12-07 05:44:15,206 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 3.215e+02 4.595e+02 5.686e+02 1.097e+03, threshold=9.191e+02, percent-clipped=3.0 2022-12-07 05:44:16,132 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2994, 3.6230, 4.5026, 3.8643, 4.2185, 3.9156, 2.3163, 4.0840], device='cuda:0'), covar=tensor([0.0160, 0.0416, 0.0216, 0.0327, 0.0192, 0.0254, 0.2106, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0058, 0.0060, 0.0054, 0.0070, 0.0046, 0.0087, 0.0068], device='cuda:0'), out_proj_covar=tensor([8.8761e-05, 9.8801e-05, 9.2444e-05, 8.5465e-05, 1.1060e-04, 7.1433e-05, 1.2579e-04, 1.0188e-04], device='cuda:0') 2022-12-07 05:44:56,697 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7618, 3.0292, 3.2821, 3.0194, 3.5436, 3.3404, 3.1167, 3.1460], device='cuda:0'), covar=tensor([0.0190, 0.1984, 0.0130, 0.0684, 0.0165, 0.0333, 0.1026, 0.1707], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0146, 0.0060, 0.0080, 0.0068, 0.0075, 0.0067, 0.0162], device='cuda:0'), out_proj_covar=tensor([5.6234e-05, 1.1054e-04, 3.9521e-05, 5.6967e-05, 4.6355e-05, 5.1975e-05, 5.0571e-05, 1.1902e-04], device='cuda:0') 2022-12-07 05:45:07,423 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1915, 4.7624, 4.7137, 5.1771, 5.1423, 5.0689, 5.1928, 4.8352], device='cuda:0'), covar=tensor([0.0217, 0.0455, 0.0230, 0.0369, 0.0189, 0.0266, 0.0246, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0067, 0.0054, 0.0051, 0.0050, 0.0059, 0.0071, 0.0066], device='cuda:0'), out_proj_covar=tensor([6.8056e-05, 9.2153e-05, 7.1987e-05, 7.7159e-05, 6.9415e-05, 8.8384e-05, 1.1357e-04, 9.7138e-05], device='cuda:0') 2022-12-07 05:45:39,009 INFO [train.py:873] (0/4) Epoch 1, batch 5100, loss[loss=0.3036, simple_loss=0.2752, pruned_loss=0.166, over 14524.00 frames. ], tot_loss[loss=0.2776, simple_loss=0.2548, pruned_loss=0.1502, over 1936016.17 frames. ], batch size: 34, lr: 4.18e-02, grad_scale: 16.0 2022-12-07 05:45:43,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 3.085e+02 4.030e+02 5.035e+02 8.863e+02, threshold=8.060e+02, percent-clipped=0.0 2022-12-07 05:47:03,131 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0770, 1.9588, 2.0450, 1.8925, 1.9590, 1.9221, 2.0063, 1.9911], device='cuda:0'), covar=tensor([0.0290, 0.0438, 0.0399, 0.0527, 0.0428, 0.0385, 0.0511, 0.0367], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0072, 0.0089, 0.0064, 0.0079, 0.0076, 0.0082, 0.0072], device='cuda:0'), out_proj_covar=tensor([9.5372e-05, 8.7828e-05, 1.0393e-04, 7.4769e-05, 9.4595e-05, 8.8890e-05, 1.1121e-04, 9.0761e-05], device='cuda:0') 2022-12-07 05:47:06,723 INFO [train.py:873] (0/4) Epoch 1, batch 5200, loss[loss=0.332, simple_loss=0.2871, pruned_loss=0.1884, over 10372.00 frames. ], tot_loss[loss=0.2789, simple_loss=0.2558, pruned_loss=0.151, over 1969199.14 frames. ], batch size: 100, lr: 4.16e-02, grad_scale: 16.0 2022-12-07 05:47:11,952 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 3.667e+02 4.771e+02 6.312e+02 1.162e+03, threshold=9.542e+02, percent-clipped=12.0 2022-12-07 05:48:14,988 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.30 vs. limit=2.0 2022-12-07 05:48:35,987 INFO [train.py:873] (0/4) Epoch 1, batch 5300, loss[loss=0.2788, simple_loss=0.2111, pruned_loss=0.1732, over 1297.00 frames. ], tot_loss[loss=0.2774, simple_loss=0.2548, pruned_loss=0.15, over 1977406.49 frames. ], batch size: 100, lr: 4.14e-02, grad_scale: 16.0 2022-12-07 05:48:39,166 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2022-12-07 05:48:40,962 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 3.215e+02 4.180e+02 5.101e+02 1.112e+03, threshold=8.360e+02, percent-clipped=0.0 2022-12-07 05:48:41,961 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0802, 3.8196, 3.9814, 3.9401, 4.0486, 4.0028, 4.2214, 4.0556], device='cuda:0'), covar=tensor([0.0372, 0.0520, 0.0352, 0.0248, 0.0229, 0.0396, 0.0331, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0076, 0.0095, 0.0065, 0.0083, 0.0082, 0.0088, 0.0076], device='cuda:0'), out_proj_covar=tensor([1.0290e-04, 9.4724e-05, 1.1317e-04, 7.5876e-05, 9.8552e-05, 9.6052e-05, 1.2110e-04, 9.6527e-05], device='cuda:0') 2022-12-07 05:48:44,487 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2090, 2.8924, 3.3943, 2.8733, 3.0519, 2.4449, 1.7220, 3.2848], device='cuda:0'), covar=tensor([0.0268, 0.0579, 0.0368, 0.0468, 0.0365, 0.0901, 0.2327, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0061, 0.0065, 0.0055, 0.0075, 0.0048, 0.0091, 0.0069], device='cuda:0'), out_proj_covar=tensor([9.4381e-05, 1.0883e-04, 1.0445e-04, 9.0410e-05, 1.2531e-04, 7.7044e-05, 1.3240e-04, 1.0802e-04], device='cuda:0') 2022-12-07 05:50:04,680 INFO [train.py:873] (0/4) Epoch 1, batch 5400, loss[loss=0.3001, simple_loss=0.2605, pruned_loss=0.1699, over 11157.00 frames. ], tot_loss[loss=0.2764, simple_loss=0.2541, pruned_loss=0.1493, over 1968050.87 frames. ], batch size: 100, lr: 4.12e-02, grad_scale: 16.0 2022-12-07 05:50:09,835 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 3.355e+02 4.207e+02 5.481e+02 1.330e+03, threshold=8.415e+02, percent-clipped=4.0 2022-12-07 05:50:14,397 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8879, 0.7590, 0.9082, 0.9534, 0.5544, 0.7982, 0.8895, 0.9194], device='cuda:0'), covar=tensor([0.0106, 0.0180, 0.0196, 0.0087, 0.0208, 0.0116, 0.0128, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0021, 0.0024, 0.0019, 0.0020, 0.0023, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([2.1669e-05, 2.2163e-05, 3.0639e-05, 1.9359e-05, 2.3597e-05, 2.4161e-05, 2.4090e-05, 2.2008e-05], device='cuda:0') 2022-12-07 05:51:33,375 INFO [train.py:873] (0/4) Epoch 1, batch 5500, loss[loss=0.2452, simple_loss=0.1966, pruned_loss=0.1469, over 1242.00 frames. ], tot_loss[loss=0.2738, simple_loss=0.2524, pruned_loss=0.1476, over 1932252.18 frames. ], batch size: 100, lr: 4.10e-02, grad_scale: 16.0 2022-12-07 05:51:38,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.070e+02 3.601e+02 4.745e+02 6.042e+02 1.360e+03, threshold=9.490e+02, percent-clipped=8.0 2022-12-07 05:51:40,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4388, 2.4219, 2.8804, 2.6853, 2.4707, 2.2379, 2.1822, 2.2782], device='cuda:0'), covar=tensor([0.0269, 0.0280, 0.0146, 0.0266, 0.0242, 0.0855, 0.0131, 0.0793], device='cuda:0'), in_proj_covar=tensor([0.0051, 0.0052, 0.0038, 0.0050, 0.0052, 0.0086, 0.0036, 0.0078], device='cuda:0'), out_proj_covar=tensor([4.2552e-05, 4.4766e-05, 3.1921e-05, 4.3799e-05, 4.1564e-05, 7.4052e-05, 2.7520e-05, 6.8015e-05], device='cuda:0') 2022-12-07 05:51:41,563 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 05:53:01,581 INFO [train.py:873] (0/4) Epoch 1, batch 5600, loss[loss=0.1924, simple_loss=0.168, pruned_loss=0.1084, over 2658.00 frames. ], tot_loss[loss=0.2723, simple_loss=0.2517, pruned_loss=0.1464, over 1954978.72 frames. ], batch size: 100, lr: 4.08e-02, grad_scale: 16.0 2022-12-07 05:53:06,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 3.374e+02 4.886e+02 6.824e+02 1.449e+03, threshold=9.773e+02, percent-clipped=6.0 2022-12-07 05:53:26,826 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 05:53:44,200 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:54:00,411 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2631, 0.6608, 1.2840, 1.2920, 1.5386, 0.9427, 0.7321, 1.2786], device='cuda:0'), covar=tensor([0.0973, 0.1040, 0.0498, 0.0549, 0.0381, 0.0510, 0.1053, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0030, 0.0033, 0.0026, 0.0026, 0.0033, 0.0023, 0.0025], device='cuda:0'), out_proj_covar=tensor([3.4309e-05, 3.9895e-05, 4.0809e-05, 3.5489e-05, 3.0567e-05, 4.0120e-05, 3.0094e-05, 3.1426e-05], device='cuda:0') 2022-12-07 05:54:00,784 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2022-12-07 05:54:01,558 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.27 vs. limit=2.0 2022-12-07 05:54:29,449 INFO [train.py:873] (0/4) Epoch 1, batch 5700, loss[loss=0.2856, simple_loss=0.222, pruned_loss=0.1746, over 1253.00 frames. ], tot_loss[loss=0.2741, simple_loss=0.2528, pruned_loss=0.1477, over 1993717.88 frames. ], batch size: 100, lr: 4.06e-02, grad_scale: 16.0 2022-12-07 05:54:34,444 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 3.498e+02 4.626e+02 6.502e+02 1.133e+03, threshold=9.251e+02, percent-clipped=2.0 2022-12-07 05:54:38,245 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 05:55:02,310 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9842, 0.5272, 1.0790, 1.0402, 1.2369, 0.6475, 0.6510, 1.2626], device='cuda:0'), covar=tensor([0.0862, 0.1621, 0.0771, 0.1050, 0.0473, 0.0609, 0.0960, 0.0518], device='cuda:0'), in_proj_covar=tensor([0.0028, 0.0030, 0.0033, 0.0027, 0.0027, 0.0034, 0.0024, 0.0027], device='cuda:0'), out_proj_covar=tensor([3.6893e-05, 4.0684e-05, 4.1819e-05, 3.7324e-05, 3.2625e-05, 4.1956e-05, 3.1706e-05, 3.4859e-05], device='cuda:0') 2022-12-07 05:55:14,459 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.77 vs. limit=5.0 2022-12-07 05:55:16,772 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:55:20,850 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.08 vs. limit=2.0 2022-12-07 05:55:58,100 INFO [train.py:873] (0/4) Epoch 1, batch 5800, loss[loss=0.2583, simple_loss=0.2415, pruned_loss=0.1375, over 14032.00 frames. ], tot_loss[loss=0.2714, simple_loss=0.2507, pruned_loss=0.146, over 1905744.45 frames. ], batch size: 26, lr: 4.04e-02, grad_scale: 16.0 2022-12-07 05:56:02,977 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.20 vs. limit=2.0 2022-12-07 05:56:03,252 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 3.658e+02 4.622e+02 6.453e+02 1.135e+03, threshold=9.244e+02, percent-clipped=6.0 2022-12-07 05:56:11,688 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:56:16,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 05:56:16,869 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2022-12-07 05:56:45,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4979, 4.1726, 3.9109, 4.4137, 4.4413, 4.4665, 4.3772, 3.9664], device='cuda:0'), covar=tensor([0.0278, 0.0450, 0.0354, 0.0379, 0.0280, 0.0321, 0.0457, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0050, 0.0075, 0.0060, 0.0057, 0.0056, 0.0067, 0.0079, 0.0074], device='cuda:0'), out_proj_covar=tensor([8.0998e-05, 1.0735e-04, 8.6624e-05, 8.8708e-05, 8.6478e-05, 1.0521e-04, 1.3105e-04, 1.1493e-04], device='cuda:0') 2022-12-07 05:56:48,653 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:57:26,652 INFO [train.py:873] (0/4) Epoch 1, batch 5900, loss[loss=0.2603, simple_loss=0.2469, pruned_loss=0.1369, over 14469.00 frames. ], tot_loss[loss=0.2701, simple_loss=0.25, pruned_loss=0.1451, over 1964579.96 frames. ], batch size: 49, lr: 4.02e-02, grad_scale: 16.0 2022-12-07 05:57:29,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8337, 1.4011, 3.1000, 3.0165, 2.8559, 2.8847, 2.3629, 3.1978], device='cuda:0'), covar=tensor([0.1418, 0.1646, 0.0160, 0.0174, 0.0162, 0.0212, 0.0277, 0.0133], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0087, 0.0044, 0.0053, 0.0045, 0.0047, 0.0043, 0.0041], device='cuda:0'), out_proj_covar=tensor([1.2536e-04, 1.2238e-04, 6.4544e-05, 8.5576e-05, 6.1914e-05, 6.9117e-05, 6.6439e-05, 5.8483e-05], device='cuda:0') 2022-12-07 05:57:31,833 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.800e+02 3.450e+02 4.399e+02 5.936e+02 1.198e+03, threshold=8.798e+02, percent-clipped=6.0 2022-12-07 05:57:34,668 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6401, 0.6537, 0.7184, 0.8985, 1.1768, 0.9553, 0.7306, 0.6921], device='cuda:0'), covar=tensor([0.0450, 0.0307, 0.0218, 0.0163, 0.0131, 0.0197, 0.0230, 0.0333], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0018, 0.0020, 0.0018, 0.0017, 0.0021, 0.0018, 0.0019], device='cuda:0'), out_proj_covar=tensor([2.1353e-05, 2.0729e-05, 2.5456e-05, 1.8498e-05, 1.9967e-05, 2.3299e-05, 2.3802e-05, 2.2396e-05], device='cuda:0') 2022-12-07 05:57:42,970 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:58:39,802 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6631, 0.5824, 0.5532, 0.4484, 0.5048, 0.3658, 0.5386, 0.5215], device='cuda:0'), covar=tensor([0.0076, 0.0152, 0.0136, 0.0393, 0.0105, 0.0140, 0.0178, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0029, 0.0032, 0.0028, 0.0026, 0.0033, 0.0026, 0.0027], device='cuda:0'), out_proj_covar=tensor([3.5128e-05, 3.9962e-05, 4.0308e-05, 3.9529e-05, 3.1957e-05, 4.2469e-05, 3.4725e-05, 3.4058e-05], device='cuda:0') 2022-12-07 05:58:41,056 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2022-12-07 05:58:44,784 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:58:54,696 INFO [train.py:873] (0/4) Epoch 1, batch 6000, loss[loss=0.2827, simple_loss=0.2631, pruned_loss=0.1511, over 14514.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.2511, pruned_loss=0.1463, over 2014857.24 frames. ], batch size: 51, lr: 4.00e-02, grad_scale: 16.0 2022-12-07 05:58:54,696 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 05:59:00,607 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9304, 1.7961, 2.5560, 2.2408, 2.3405, 2.4322, 1.8118, 2.3407], device='cuda:0'), covar=tensor([0.0488, 0.0383, 0.0081, 0.0145, 0.0130, 0.0096, 0.0356, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0045, 0.0044, 0.0043, 0.0045, 0.0041, 0.0071, 0.0047], device='cuda:0'), out_proj_covar=tensor([5.2228e-05, 4.9575e-05, 5.0136e-05, 3.9218e-05, 4.4485e-05, 4.0644e-05, 8.3081e-05, 5.2927e-05], device='cuda:0') 2022-12-07 05:59:02,828 INFO [train.py:905] (0/4) Epoch 1, validation: loss=0.159, simple_loss=0.1938, pruned_loss=0.06211, over 857387.00 frames. 2022-12-07 05:59:02,829 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17446MB 2022-12-07 05:59:02,932 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9098, 1.4729, 2.8922, 2.8557, 2.8389, 2.6673, 2.0793, 3.0292], device='cuda:0'), covar=tensor([0.1234, 0.1401, 0.0141, 0.0174, 0.0124, 0.0200, 0.0300, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0088, 0.0043, 0.0054, 0.0046, 0.0048, 0.0042, 0.0043], device='cuda:0'), out_proj_covar=tensor([1.2815e-04, 1.2386e-04, 6.3014e-05, 8.8207e-05, 6.4139e-05, 7.0500e-05, 6.5930e-05, 6.1053e-05], device='cuda:0') 2022-12-07 05:59:03,848 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9976, 4.7005, 4.7077, 5.2456, 5.1108, 4.3544, 5.4494, 5.2180], device='cuda:0'), covar=tensor([0.0471, 0.0453, 0.0612, 0.0432, 0.0478, 0.0416, 0.0499, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0063, 0.0087, 0.0083, 0.0085, 0.0056, 0.0082, 0.0081], device='cuda:0'), out_proj_covar=tensor([1.1253e-04, 9.5030e-05, 1.2533e-04, 1.1719e-04, 1.1950e-04, 7.9498e-05, 1.2408e-04, 1.1838e-04], device='cuda:0') 2022-12-07 05:59:04,174 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 05:59:07,197 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 05:59:07,936 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 3.896e+02 5.286e+02 6.584e+02 1.638e+03, threshold=1.057e+03, percent-clipped=9.0 2022-12-07 05:59:19,574 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.89 vs. limit=2.0 2022-12-07 05:59:35,670 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.08 vs. limit=5.0 2022-12-07 05:59:39,727 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 05:59:46,794 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:30,193 INFO [train.py:873] (0/4) Epoch 1, batch 6100, loss[loss=0.2914, simple_loss=0.2639, pruned_loss=0.1594, over 14169.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.2505, pruned_loss=0.1464, over 1912673.47 frames. ], batch size: 84, lr: 3.98e-02, grad_scale: 16.0 2022-12-07 06:00:33,336 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:35,714 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 3.351e+02 5.105e+02 6.275e+02 1.538e+03, threshold=1.021e+03, percent-clipped=3.0 2022-12-07 06:00:39,099 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:00:43,139 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2022-12-07 06:00:47,851 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5729, 1.9932, 3.8842, 2.6752, 3.4236, 2.2683, 3.5090, 3.7284], device='cuda:0'), covar=tensor([0.0252, 0.3825, 0.0224, 0.5609, 0.0168, 0.3027, 0.0284, 0.0158], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0239, 0.0129, 0.0330, 0.0103, 0.0265, 0.0158, 0.0110], device='cuda:0'), out_proj_covar=tensor([1.1248e-04, 2.0402e-04, 1.0317e-04, 2.7296e-04, 8.3191e-05, 2.1407e-04, 1.2156e-04, 8.6244e-05], device='cuda:0') 2022-12-07 06:01:00,388 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:06,488 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 06:01:12,150 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 06:01:31,123 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:53,481 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:01:58,121 INFO [train.py:873] (0/4) Epoch 1, batch 6200, loss[loss=0.2601, simple_loss=0.2562, pruned_loss=0.132, over 14023.00 frames. ], tot_loss[loss=0.2708, simple_loss=0.2499, pruned_loss=0.1458, over 1975900.35 frames. ], batch size: 22, lr: 3.96e-02, grad_scale: 16.0 2022-12-07 06:02:00,040 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:03,437 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.672e+01 3.177e+02 4.421e+02 6.121e+02 1.475e+03, threshold=8.841e+02, percent-clipped=2.0 2022-12-07 06:02:07,203 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:07,493 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.34 vs. limit=5.0 2022-12-07 06:02:09,684 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:24,745 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:54,083 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:02:55,555 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:03:00,788 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:03:06,621 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.35 vs. limit=5.0 2022-12-07 06:03:26,922 INFO [train.py:873] (0/4) Epoch 1, batch 6300, loss[loss=0.2546, simple_loss=0.2445, pruned_loss=0.1323, over 14107.00 frames. ], tot_loss[loss=0.27, simple_loss=0.25, pruned_loss=0.145, over 1973789.64 frames. ], batch size: 19, lr: 3.94e-02, grad_scale: 16.0 2022-12-07 06:03:31,255 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:03:31,913 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 3.956e+02 5.182e+02 6.845e+02 1.642e+03, threshold=1.036e+03, percent-clipped=11.0 2022-12-07 06:03:49,221 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:03:53,175 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 06:04:05,560 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:04:12,433 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:04:16,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2786, 4.0291, 3.7811, 4.2440, 4.3298, 4.2632, 4.1301, 3.7592], device='cuda:0'), covar=tensor([0.0280, 0.0518, 0.0383, 0.0394, 0.0309, 0.0395, 0.0524, 0.0542], device='cuda:0'), in_proj_covar=tensor([0.0049, 0.0078, 0.0064, 0.0059, 0.0060, 0.0070, 0.0084, 0.0082], device='cuda:0'), out_proj_covar=tensor([8.0521e-05, 1.1636e-04, 9.1252e-05, 9.3271e-05, 9.5544e-05, 1.1367e-04, 1.4402e-04, 1.3009e-04], device='cuda:0') 2022-12-07 06:04:22,855 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-07 06:04:39,123 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 06:04:51,703 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:04:52,105 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2022-12-07 06:04:53,609 INFO [train.py:873] (0/4) Epoch 1, batch 6400, loss[loss=0.2645, simple_loss=0.2517, pruned_loss=0.1387, over 14023.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.2481, pruned_loss=0.1433, over 1947189.86 frames. ], batch size: 19, lr: 3.92e-02, grad_scale: 8.0 2022-12-07 06:05:00,167 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 3.375e+02 4.095e+02 5.340e+02 1.039e+03, threshold=8.189e+02, percent-clipped=1.0 2022-12-07 06:05:03,115 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:35,041 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9691, 1.8914, 2.3288, 2.4139, 1.9287, 1.7476, 2.2988, 1.8125], device='cuda:0'), covar=tensor([0.0208, 0.0289, 0.0183, 0.0196, 0.0250, 0.0689, 0.0081, 0.0582], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0064, 0.0047, 0.0063, 0.0065, 0.0106, 0.0042, 0.0100], device='cuda:0'), out_proj_covar=tensor([5.2791e-05, 6.0413e-05, 4.3698e-05, 6.2641e-05, 5.9639e-05, 9.9465e-05, 3.6006e-05, 9.3814e-05], device='cuda:0') 2022-12-07 06:05:44,989 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:47,694 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:52,015 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:05:57,681 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9669, 1.9019, 1.9619, 1.9380, 1.9021, 1.7103, 1.2462, 1.8429], device='cuda:0'), covar=tensor([0.0321, 0.0340, 0.0315, 0.0236, 0.0335, 0.0520, 0.1554, 0.0343], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0065, 0.0069, 0.0056, 0.0087, 0.0060, 0.0105, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:06:13,023 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:21,240 INFO [train.py:873] (0/4) Epoch 1, batch 6500, loss[loss=0.2118, simple_loss=0.173, pruned_loss=0.1253, over 1249.00 frames. ], tot_loss[loss=0.2682, simple_loss=0.2487, pruned_loss=0.1438, over 1954972.06 frames. ], batch size: 100, lr: 3.90e-02, grad_scale: 8.0 2022-12-07 06:06:27,500 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.203e+01 3.294e+02 4.343e+02 5.711e+02 1.098e+03, threshold=8.686e+02, percent-clipped=4.0 2022-12-07 06:06:32,748 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:40,903 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:06:43,493 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:45,668 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:06:49,794 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5901, 1.6050, 2.9700, 2.2555, 2.8739, 1.6793, 2.5348, 2.5317], device='cuda:0'), covar=tensor([0.0376, 0.3572, 0.0245, 0.3384, 0.0172, 0.2742, 0.0537, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0249, 0.0132, 0.0339, 0.0101, 0.0272, 0.0163, 0.0118], device='cuda:0'), out_proj_covar=tensor([1.2305e-04, 2.1513e-04, 1.0982e-04, 2.8244e-04, 8.4772e-05, 2.2340e-04, 1.2844e-04, 9.4920e-05], device='cuda:0') 2022-12-07 06:07:12,822 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:15,386 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:20,393 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:07:50,112 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.43 vs. limit=5.0 2022-12-07 06:07:50,327 INFO [train.py:873] (0/4) Epoch 1, batch 6600, loss[loss=0.2659, simple_loss=0.2458, pruned_loss=0.143, over 13542.00 frames. ], tot_loss[loss=0.2666, simple_loss=0.2474, pruned_loss=0.1429, over 1932067.09 frames. ], batch size: 100, lr: 3.89e-02, grad_scale: 8.0 2022-12-07 06:07:56,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 3.480e+02 4.603e+02 5.913e+02 1.121e+03, threshold=9.206e+02, percent-clipped=7.0 2022-12-07 06:08:08,508 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:08:13,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.50 vs. limit=5.0 2022-12-07 06:08:14,814 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:08:29,782 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:08:30,347 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:09,354 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:12,787 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:16,417 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5076, 1.4101, 1.8155, 1.7223, 1.4667, 1.4149, 1.7411, 1.2400], device='cuda:0'), covar=tensor([0.0108, 0.0107, 0.0065, 0.0077, 0.0101, 0.0244, 0.0041, 0.0175], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0068, 0.0048, 0.0066, 0.0069, 0.0107, 0.0044, 0.0105], device='cuda:0'), out_proj_covar=tensor([5.6448e-05, 6.6315e-05, 4.6634e-05, 6.6404e-05, 6.4362e-05, 1.0382e-04, 3.8843e-05, 1.0059e-04], device='cuda:0') 2022-12-07 06:09:17,226 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:09:18,765 INFO [train.py:873] (0/4) Epoch 1, batch 6700, loss[loss=0.2322, simple_loss=0.1782, pruned_loss=0.1431, over 1272.00 frames. ], tot_loss[loss=0.268, simple_loss=0.2489, pruned_loss=0.1436, over 1985190.89 frames. ], batch size: 100, lr: 3.87e-02, grad_scale: 8.0 2022-12-07 06:09:24,613 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 3.291e+02 4.382e+02 6.181e+02 1.245e+03, threshold=8.764e+02, percent-clipped=7.0 2022-12-07 06:09:34,342 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2733, 2.7078, 2.1619, 2.7053, 2.5726, 2.4920, 2.4478, 2.1543], device='cuda:0'), covar=tensor([0.0318, 0.0593, 0.2142, 0.0245, 0.0227, 0.0226, 0.0599, 0.2297], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0134, 0.0219, 0.0109, 0.0101, 0.0109, 0.0131, 0.0255], device='cuda:0'), out_proj_covar=tensor([6.9415e-05, 8.8281e-05, 1.3795e-04, 6.7155e-05, 6.5938e-05, 6.8643e-05, 8.5338e-05, 1.5737e-04], device='cuda:0') 2022-12-07 06:09:55,819 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6630, 1.6433, 2.9188, 2.1702, 3.0248, 1.7433, 2.5167, 2.8597], device='cuda:0'), covar=tensor([0.0312, 0.4697, 0.0267, 0.5847, 0.0227, 0.3225, 0.0654, 0.0261], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0254, 0.0133, 0.0353, 0.0105, 0.0279, 0.0171, 0.0121], device='cuda:0'), out_proj_covar=tensor([1.2561e-04, 2.2228e-04, 1.1122e-04, 2.9461e-04, 8.9200e-05, 2.3182e-04, 1.3818e-04, 9.8653e-05], device='cuda:0') 2022-12-07 06:09:58,739 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:10:37,328 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:10:44,981 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 06:10:46,419 INFO [train.py:873] (0/4) Epoch 1, batch 6800, loss[loss=0.2282, simple_loss=0.2328, pruned_loss=0.1118, over 14285.00 frames. ], tot_loss[loss=0.2646, simple_loss=0.2472, pruned_loss=0.141, over 2031406.85 frames. ], batch size: 63, lr: 3.85e-02, grad_scale: 8.0 2022-12-07 06:10:52,798 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.483e+01 3.173e+02 4.494e+02 5.972e+02 9.591e+02, threshold=8.989e+02, percent-clipped=5.0 2022-12-07 06:11:01,518 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:11:05,695 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:08,555 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:19,535 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:37,256 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:44,216 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:49,843 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:11:53,601 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:12:13,451 INFO [train.py:873] (0/4) Epoch 1, batch 6900, loss[loss=0.2506, simple_loss=0.2379, pruned_loss=0.1317, over 14019.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.2478, pruned_loss=0.1414, over 2037686.63 frames. ], batch size: 20, lr: 3.83e-02, grad_scale: 8.0 2022-12-07 06:12:18,840 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:19,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 3.106e+02 4.451e+02 6.203e+02 1.044e+03, threshold=8.902e+02, percent-clipped=8.0 2022-12-07 06:12:20,619 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:21,079 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=14.45 vs. limit=5.0 2022-12-07 06:12:25,624 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:32,031 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:12:46,831 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:12:57,686 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.08 vs. limit=2.0 2022-12-07 06:13:03,675 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5836, 1.1834, 1.9705, 1.6826, 1.7875, 1.7833, 0.8427, 1.5641], device='cuda:0'), covar=tensor([0.0258, 0.0683, 0.0136, 0.0182, 0.0178, 0.0197, 0.0744, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0050, 0.0047, 0.0047, 0.0049, 0.0048, 0.0085, 0.0051], device='cuda:0'), out_proj_covar=tensor([5.8106e-05, 6.0393e-05, 5.8038e-05, 4.9425e-05, 5.2433e-05, 5.4119e-05, 1.0211e-04, 6.2554e-05], device='cuda:0') 2022-12-07 06:13:04,490 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5948, 1.5877, 3.3016, 1.9042, 3.5490, 3.3515, 2.4344, 3.5878], device='cuda:0'), covar=tensor([0.0162, 0.2309, 0.0319, 0.1781, 0.0227, 0.0248, 0.0651, 0.0188], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0099, 0.0055, 0.0099, 0.0073, 0.0057, 0.0057, 0.0057], device='cuda:0'), out_proj_covar=tensor([9.4051e-05, 1.5849e-04, 1.0584e-04, 1.5878e-04, 1.2871e-04, 9.9931e-05, 1.0413e-04, 1.0121e-04], device='cuda:0') 2022-12-07 06:13:09,446 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2022-12-07 06:13:11,735 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1995, 2.3621, 1.9827, 2.6948, 2.5429, 2.4615, 2.2893, 2.0628], device='cuda:0'), covar=tensor([0.0312, 0.0761, 0.2624, 0.0254, 0.0259, 0.0325, 0.0615, 0.2669], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0140, 0.0230, 0.0112, 0.0100, 0.0112, 0.0134, 0.0258], device='cuda:0'), out_proj_covar=tensor([7.2860e-05, 9.1465e-05, 1.4406e-04, 6.8905e-05, 6.6901e-05, 7.3807e-05, 8.8512e-05, 1.6114e-04], device='cuda:0') 2022-12-07 06:13:13,999 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:14,154 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:18,705 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 06:13:22,612 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6947, 4.0698, 4.6139, 3.8718, 4.4324, 4.6319, 2.2587, 4.5502], device='cuda:0'), covar=tensor([0.0133, 0.0247, 0.0362, 0.0308, 0.0243, 0.0115, 0.2691, 0.0159], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0064, 0.0071, 0.0059, 0.0091, 0.0062, 0.0110, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:13:27,095 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:13:41,154 INFO [train.py:873] (0/4) Epoch 1, batch 7000, loss[loss=0.2549, simple_loss=0.2076, pruned_loss=0.1512, over 1220.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.2464, pruned_loss=0.1397, over 2021011.75 frames. ], batch size: 100, lr: 3.81e-02, grad_scale: 8.0 2022-12-07 06:13:47,944 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 2.871e+02 4.067e+02 4.755e+02 1.195e+03, threshold=8.134e+02, percent-clipped=1.0 2022-12-07 06:14:05,702 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 06:14:14,545 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-07 06:14:23,435 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:15:09,088 INFO [train.py:873] (0/4) Epoch 1, batch 7100, loss[loss=0.2817, simple_loss=0.2537, pruned_loss=0.1548, over 7774.00 frames. ], tot_loss[loss=0.263, simple_loss=0.2462, pruned_loss=0.1399, over 2036534.93 frames. ], batch size: 100, lr: 3.79e-02, grad_scale: 8.0 2022-12-07 06:15:15,338 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 3.113e+02 4.115e+02 5.668e+02 1.201e+03, threshold=8.229e+02, percent-clipped=4.0 2022-12-07 06:15:16,485 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:15:23,769 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:15:28,163 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:03,237 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:05,594 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:09,751 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:16:35,873 INFO [train.py:873] (0/4) Epoch 1, batch 7200, loss[loss=0.2572, simple_loss=0.2446, pruned_loss=0.1349, over 14219.00 frames. ], tot_loss[loss=0.2648, simple_loss=0.2468, pruned_loss=0.1414, over 1943307.98 frames. ], batch size: 69, lr: 3.78e-02, grad_scale: 8.0 2022-12-07 06:16:42,298 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.199e+02 3.143e+02 4.183e+02 5.672e+02 1.523e+03, threshold=8.365e+02, percent-clipped=9.0 2022-12-07 06:16:56,273 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:16:59,954 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.98 vs. limit=5.0 2022-12-07 06:17:04,791 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:17:32,261 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:17:41,195 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:17:50,899 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:17:52,435 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.28 vs. limit=5.0 2022-12-07 06:18:04,543 INFO [train.py:873] (0/4) Epoch 1, batch 7300, loss[loss=0.2759, simple_loss=0.2478, pruned_loss=0.152, over 6908.00 frames. ], tot_loss[loss=0.2617, simple_loss=0.2446, pruned_loss=0.1394, over 1931130.49 frames. ], batch size: 100, lr: 3.76e-02, grad_scale: 8.0 2022-12-07 06:18:06,983 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:18:10,452 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 3.489e+02 4.612e+02 5.728e+02 1.039e+03, threshold=9.225e+02, percent-clipped=2.0 2022-12-07 06:18:32,885 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:18:35,741 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:18:59,619 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-07 06:19:33,639 INFO [train.py:873] (0/4) Epoch 1, batch 7400, loss[loss=0.2549, simple_loss=0.2338, pruned_loss=0.138, over 6933.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.2448, pruned_loss=0.1399, over 1954706.34 frames. ], batch size: 100, lr: 3.74e-02, grad_scale: 8.0 2022-12-07 06:19:36,419 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:19:36,489 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8419, 0.7443, 0.6560, 0.8649, 1.0227, 0.6067, 0.9264, 0.5189], device='cuda:0'), covar=tensor([0.0528, 0.0464, 0.0248, 0.0437, 0.0201, 0.0413, 0.0303, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0024, 0.0025, 0.0021, 0.0022, 0.0027, 0.0019, 0.0022], device='cuda:0'), out_proj_covar=tensor([3.2951e-05, 3.9323e-05, 3.7564e-05, 3.5045e-05, 3.1158e-05, 4.2042e-05, 3.0081e-05, 3.1430e-05], device='cuda:0') 2022-12-07 06:19:39,315 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:19:39,942 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 3.232e+02 4.380e+02 5.803e+02 1.577e+03, threshold=8.760e+02, percent-clipped=3.0 2022-12-07 06:19:55,822 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0242, 1.6365, 3.1681, 2.4719, 3.1945, 1.8142, 2.9963, 2.9060], device='cuda:0'), covar=tensor([0.0328, 0.4902, 0.0325, 0.6272, 0.0233, 0.3407, 0.0475, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0254, 0.0135, 0.0362, 0.0111, 0.0283, 0.0179, 0.0126], device='cuda:0'), out_proj_covar=tensor([1.3431e-04, 2.2881e-04, 1.1597e-04, 3.0682e-04, 9.9443e-05, 2.4044e-04, 1.4912e-04, 1.0511e-04], device='cuda:0') 2022-12-07 06:20:31,815 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 06:20:32,157 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:20:36,125 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2022-12-07 06:20:38,754 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3539, 5.2345, 4.8478, 5.2927, 5.1504, 5.1702, 5.3025, 5.3371], device='cuda:0'), covar=tensor([0.0328, 0.0300, 0.0515, 0.0208, 0.0265, 0.0209, 0.0498, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0092, 0.0116, 0.0083, 0.0096, 0.0098, 0.0109, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:20:50,534 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0022, 1.6475, 3.8289, 2.3037, 3.8575, 3.5427, 2.5625, 4.1766], device='cuda:0'), covar=tensor([0.0158, 0.2384, 0.0242, 0.1548, 0.0184, 0.0257, 0.0528, 0.0136], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0107, 0.0059, 0.0107, 0.0078, 0.0064, 0.0059, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:20:57,875 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 06:21:00,783 INFO [train.py:873] (0/4) Epoch 1, batch 7500, loss[loss=0.242, simple_loss=0.2391, pruned_loss=0.1225, over 14192.00 frames. ], tot_loss[loss=0.2603, simple_loss=0.2436, pruned_loss=0.1386, over 1898531.37 frames. ], batch size: 69, lr: 3.72e-02, grad_scale: 8.0 2022-12-07 06:21:06,494 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 3.575e+02 4.286e+02 5.078e+02 8.997e+02, threshold=8.573e+02, percent-clipped=1.0 2022-12-07 06:21:16,656 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:21:25,278 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6704, 1.5129, 2.7916, 1.8798, 2.7910, 2.5920, 1.6645, 3.0140], device='cuda:0'), covar=tensor([0.0234, 0.2304, 0.0376, 0.1576, 0.0264, 0.0395, 0.0870, 0.0225], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0107, 0.0060, 0.0108, 0.0078, 0.0064, 0.0059, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:21:26,129 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3804, 2.3701, 2.8427, 2.6615, 2.5395, 2.1656, 2.4761, 2.0492], device='cuda:0'), covar=tensor([0.0193, 0.0347, 0.0149, 0.0331, 0.0292, 0.0898, 0.0097, 0.0870], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0079, 0.0055, 0.0077, 0.0079, 0.0121, 0.0050, 0.0122], device='cuda:0'), out_proj_covar=tensor([6.7616e-05, 8.2885e-05, 5.9781e-05, 8.3039e-05, 8.5649e-05, 1.2466e-04, 4.8913e-05, 1.2385e-04], device='cuda:0') 2022-12-07 06:21:29,232 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:21:47,005 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-1.pt 2022-12-07 06:22:29,583 INFO [train.py:873] (0/4) Epoch 2, batch 0, loss[loss=0.3922, simple_loss=0.3272, pruned_loss=0.2286, over 9511.00 frames. ], tot_loss[loss=0.3922, simple_loss=0.3272, pruned_loss=0.2286, over 9511.00 frames. ], batch size: 100, lr: 3.64e-02, grad_scale: 8.0 2022-12-07 06:22:29,583 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 06:22:32,932 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0696, 2.2871, 4.5769, 4.0233, 4.1537, 4.3052, 3.7966, 4.7983], device='cuda:0'), covar=tensor([0.2136, 0.1670, 0.0168, 0.0122, 0.0206, 0.0201, 0.0143, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0102, 0.0050, 0.0065, 0.0054, 0.0056, 0.0050, 0.0049], device='cuda:0'), out_proj_covar=tensor([1.6264e-04, 1.5739e-04, 8.4816e-05, 1.1975e-04, 8.7741e-05, 9.4766e-05, 9.0431e-05, 7.8409e-05], device='cuda:0') 2022-12-07 06:22:36,878 INFO [train.py:905] (0/4) Epoch 2, validation: loss=0.201, simple_loss=0.225, pruned_loss=0.08852, over 857387.00 frames. 2022-12-07 06:22:36,879 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 06:22:38,722 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:22:54,100 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:23:16,974 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.396e+01 2.991e+02 4.594e+02 6.425e+02 1.765e+03, threshold=9.187e+02, percent-clipped=13.0 2022-12-07 06:23:18,163 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 06:23:21,528 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:23:31,701 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:23:37,414 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:24:05,654 INFO [train.py:873] (0/4) Epoch 2, batch 100, loss[loss=0.2417, simple_loss=0.2148, pruned_loss=0.1343, over 3803.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.2483, pruned_loss=0.1407, over 883068.26 frames. ], batch size: 100, lr: 3.62e-02, grad_scale: 8.0 2022-12-07 06:24:19,415 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2022-12-07 06:24:20,815 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7379, 3.2329, 3.6950, 3.2227, 3.6312, 3.5245, 1.7661, 3.6448], device='cuda:0'), covar=tensor([0.0331, 0.0586, 0.0567, 0.0569, 0.0489, 0.0366, 0.3420, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0070, 0.0073, 0.0063, 0.0095, 0.0067, 0.0116, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:24:24,556 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:24:41,567 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 06:24:44,979 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 3.684e+02 4.699e+02 5.885e+02 1.470e+03, threshold=9.397e+02, percent-clipped=3.0 2022-12-07 06:24:54,309 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.13 vs. limit=5.0 2022-12-07 06:25:16,245 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:25:23,879 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:25:32,820 INFO [train.py:873] (0/4) Epoch 2, batch 200, loss[loss=0.2775, simple_loss=0.2485, pruned_loss=0.1533, over 6988.00 frames. ], tot_loss[loss=0.2597, simple_loss=0.2441, pruned_loss=0.1376, over 1346298.72 frames. ], batch size: 100, lr: 3.61e-02, grad_scale: 8.0 2022-12-07 06:25:33,231 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:25:57,937 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.16 vs. limit=5.0 2022-12-07 06:26:07,039 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:08,733 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:12,299 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 3.123e+02 4.049e+02 5.757e+02 1.056e+03, threshold=8.099e+02, percent-clipped=3.0 2022-12-07 06:26:22,342 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:26:38,182 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0619, 1.9468, 4.0791, 2.8531, 3.6721, 2.2573, 4.0254, 3.7484], device='cuda:0'), covar=tensor([0.0378, 0.5775, 0.0246, 0.8161, 0.0128, 0.3438, 0.0486, 0.0170], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0264, 0.0138, 0.0359, 0.0112, 0.0280, 0.0184, 0.0122], device='cuda:0'), out_proj_covar=tensor([1.3993e-04, 2.4127e-04, 1.1977e-04, 3.0868e-04, 9.7619e-05, 2.4029e-04, 1.5687e-04, 1.0451e-04], device='cuda:0') 2022-12-07 06:26:59,710 INFO [train.py:873] (0/4) Epoch 2, batch 300, loss[loss=0.2826, simple_loss=0.2544, pruned_loss=0.1554, over 6908.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.2437, pruned_loss=0.1369, over 1683549.37 frames. ], batch size: 100, lr: 3.59e-02, grad_scale: 8.0 2022-12-07 06:26:59,901 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:26:59,928 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6810, 3.6698, 2.5631, 3.7939, 3.4810, 3.7196, 3.5163, 2.4826], device='cuda:0'), covar=tensor([0.0220, 0.0338, 0.2488, 0.0258, 0.0154, 0.0342, 0.0413, 0.2842], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0149, 0.0241, 0.0114, 0.0104, 0.0113, 0.0145, 0.0272], device='cuda:0'), out_proj_covar=tensor([8.2091e-05, 1.0039e-04, 1.5521e-04, 7.2929e-05, 7.4045e-05, 7.7346e-05, 1.0059e-04, 1.7230e-04], device='cuda:0') 2022-12-07 06:27:03,184 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:27:26,070 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8717, 1.4851, 2.1545, 1.9131, 2.1805, 1.5391, 1.9076, 2.2317], device='cuda:0'), covar=tensor([0.0792, 0.2880, 0.0201, 0.2650, 0.0192, 0.1699, 0.0790, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0271, 0.0136, 0.0361, 0.0114, 0.0288, 0.0184, 0.0124], device='cuda:0'), out_proj_covar=tensor([1.4135e-04, 2.4562e-04, 1.1857e-04, 3.0970e-04, 9.9186e-05, 2.4681e-04, 1.5705e-04, 1.0542e-04], device='cuda:0') 2022-12-07 06:27:38,763 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 3.418e+02 4.416e+02 5.390e+02 1.169e+03, threshold=8.833e+02, percent-clipped=6.0 2022-12-07 06:27:58,594 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 06:27:58,937 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:28:10,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7294, 2.6546, 2.5276, 2.5035, 2.5336, 2.6205, 2.7179, 2.7111], device='cuda:0'), covar=tensor([0.0404, 0.0461, 0.0701, 0.0463, 0.0438, 0.0412, 0.0574, 0.0585], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0096, 0.0119, 0.0085, 0.0101, 0.0101, 0.0114, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:28:26,418 INFO [train.py:873] (0/4) Epoch 2, batch 400, loss[loss=0.24, simple_loss=0.215, pruned_loss=0.1326, over 10042.00 frames. ], tot_loss[loss=0.255, simple_loss=0.2412, pruned_loss=0.1344, over 1813932.42 frames. ], batch size: 12, lr: 3.58e-02, grad_scale: 8.0 2022-12-07 06:28:40,960 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:28:41,808 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:29:05,270 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.294e+01 3.140e+02 4.082e+02 5.332e+02 1.723e+03, threshold=8.164e+02, percent-clipped=4.0 2022-12-07 06:29:09,340 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2022-12-07 06:29:53,230 INFO [train.py:873] (0/4) Epoch 2, batch 500, loss[loss=0.2111, simple_loss=0.1884, pruned_loss=0.1169, over 2548.00 frames. ], tot_loss[loss=0.2555, simple_loss=0.2406, pruned_loss=0.1352, over 1824054.12 frames. ], batch size: 100, lr: 3.56e-02, grad_scale: 8.0 2022-12-07 06:29:53,388 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:25,034 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:32,478 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.590e+01 3.381e+02 4.581e+02 6.129e+02 1.327e+03, threshold=9.162e+02, percent-clipped=13.0 2022-12-07 06:30:35,075 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:30:52,912 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 06:30:55,635 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2022-12-07 06:31:15,367 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:31:19,467 INFO [train.py:873] (0/4) Epoch 2, batch 600, loss[loss=0.2561, simple_loss=0.2437, pruned_loss=0.1343, over 14306.00 frames. ], tot_loss[loss=0.2538, simple_loss=0.2396, pruned_loss=0.134, over 1891141.85 frames. ], batch size: 46, lr: 3.54e-02, grad_scale: 8.0 2022-12-07 06:31:28,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6267, 1.5101, 3.3544, 1.9055, 3.5203, 3.2439, 2.3166, 3.6781], device='cuda:0'), covar=tensor([0.0156, 0.2530, 0.0336, 0.1972, 0.0229, 0.0309, 0.0759, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0107, 0.0064, 0.0116, 0.0086, 0.0069, 0.0065, 0.0067], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:31:30,323 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0013, 1.4827, 1.6717, 0.7838, 1.7831, 1.1068, 1.5180, 1.6822], device='cuda:0'), covar=tensor([0.0236, 0.2925, 0.0575, 0.1274, 0.0581, 0.0932, 0.0798, 0.0545], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0119, 0.0056, 0.0080, 0.0053, 0.0054, 0.0052, 0.0050], device='cuda:0'), out_proj_covar=tensor([7.0609e-05, 1.8314e-04, 9.1919e-05, 1.2748e-04, 9.6101e-05, 9.5597e-05, 9.5816e-05, 8.3153e-05], device='cuda:0') 2022-12-07 06:31:42,711 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7050, 0.7153, 0.7761, 0.8880, 0.8914, 0.6691, 0.7690, 0.8638], device='cuda:0'), covar=tensor([0.0432, 0.0631, 0.0388, 0.0469, 0.0360, 0.0369, 0.0491, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0022, 0.0023, 0.0019, 0.0019, 0.0022, 0.0018, 0.0019], device='cuda:0'), out_proj_covar=tensor([3.3215e-05, 3.9472e-05, 3.7836e-05, 3.5550e-05, 2.9885e-05, 3.6712e-05, 3.1258e-05, 3.0168e-05], device='cuda:0') 2022-12-07 06:31:59,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 3.066e+02 3.782e+02 5.192e+02 1.492e+03, threshold=7.564e+02, percent-clipped=6.0 2022-12-07 06:32:02,202 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4462, 2.4800, 3.4490, 2.8612, 3.6221, 3.6724, 3.3507, 3.0866], device='cuda:0'), covar=tensor([0.0145, 0.1691, 0.0106, 0.0835, 0.0155, 0.0335, 0.0611, 0.1762], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0223, 0.0090, 0.0146, 0.0093, 0.0121, 0.0105, 0.0263], device='cuda:0'), out_proj_covar=tensor([9.8208e-05, 1.8288e-04, 6.9462e-05, 1.1865e-04, 7.5788e-05, 9.5294e-05, 9.8967e-05, 2.0886e-04], device='cuda:0') 2022-12-07 06:32:06,964 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=18.95 vs. limit=5.0 2022-12-07 06:32:16,721 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 06:32:38,100 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7191, 1.3520, 2.0390, 1.8468, 1.8871, 1.8229, 1.5050, 1.9931], device='cuda:0'), covar=tensor([0.0433, 0.0748, 0.0135, 0.0195, 0.0162, 0.0140, 0.0313, 0.0089], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0106, 0.0052, 0.0070, 0.0059, 0.0059, 0.0052, 0.0051], device='cuda:0'), out_proj_covar=tensor([1.7515e-04, 1.7111e-04, 8.9350e-05, 1.3339e-04, 9.9601e-05, 1.0456e-04, 9.7966e-05, 8.7786e-05], device='cuda:0') 2022-12-07 06:32:47,258 INFO [train.py:873] (0/4) Epoch 2, batch 700, loss[loss=0.2771, simple_loss=0.2556, pruned_loss=0.1493, over 9467.00 frames. ], tot_loss[loss=0.2563, simple_loss=0.241, pruned_loss=0.1358, over 1894468.45 frames. ], batch size: 100, lr: 3.53e-02, grad_scale: 8.0 2022-12-07 06:32:49,132 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:33:01,903 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:33:26,031 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 3.101e+02 3.944e+02 5.331e+02 1.080e+03, threshold=7.888e+02, percent-clipped=5.0 2022-12-07 06:33:41,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:33:43,072 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:34:13,678 INFO [train.py:873] (0/4) Epoch 2, batch 800, loss[loss=0.1997, simple_loss=0.1736, pruned_loss=0.1129, over 2704.00 frames. ], tot_loss[loss=0.2547, simple_loss=0.2397, pruned_loss=0.1349, over 1844166.52 frames. ], batch size: 100, lr: 3.51e-02, grad_scale: 16.0 2022-12-07 06:34:45,232 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:34:53,071 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 2.975e+02 4.226e+02 5.564e+02 1.451e+03, threshold=8.453e+02, percent-clipped=7.0 2022-12-07 06:35:12,061 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 06:35:26,084 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:35:30,171 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2883, 2.1920, 2.5080, 2.8073, 2.4517, 1.9613, 2.4612, 2.0450], device='cuda:0'), covar=tensor([0.0161, 0.0240, 0.0176, 0.0185, 0.0159, 0.0665, 0.0090, 0.0590], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0092, 0.0065, 0.0090, 0.0092, 0.0141, 0.0060, 0.0137], device='cuda:0'), out_proj_covar=tensor([8.4235e-05, 1.0081e-04, 7.3429e-05, 1.0316e-04, 1.0519e-04, 1.5841e-04, 6.3234e-05, 1.4756e-04], device='cuda:0') 2022-12-07 06:35:36,567 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:35:40,661 INFO [train.py:873] (0/4) Epoch 2, batch 900, loss[loss=0.2531, simple_loss=0.2465, pruned_loss=0.1299, over 14264.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.2403, pruned_loss=0.135, over 1846148.25 frames. ], batch size: 25, lr: 3.50e-02, grad_scale: 16.0 2022-12-07 06:35:42,031 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2022-12-07 06:35:46,205 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.51 vs. limit=2.0 2022-12-07 06:36:02,287 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6253, 1.3091, 1.9553, 1.7208, 1.9459, 1.7175, 1.5132, 1.9436], device='cuda:0'), covar=tensor([0.0484, 0.0819, 0.0157, 0.0254, 0.0166, 0.0152, 0.0262, 0.0100], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0109, 0.0054, 0.0070, 0.0060, 0.0059, 0.0051, 0.0050], device='cuda:0'), out_proj_covar=tensor([1.7732e-04, 1.7926e-04, 9.7086e-05, 1.3580e-04, 1.0434e-04, 1.0620e-04, 9.7114e-05, 8.6578e-05], device='cuda:0') 2022-12-07 06:36:03,705 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3318, 5.2771, 5.0581, 5.5745, 5.3702, 4.3434, 5.6220, 5.4353], device='cuda:0'), covar=tensor([0.0763, 0.0492, 0.0664, 0.0626, 0.0436, 0.0488, 0.0614, 0.1176], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0068, 0.0091, 0.0087, 0.0089, 0.0060, 0.0081, 0.0085], device='cuda:0'), out_proj_covar=tensor([1.3190e-04, 1.0929e-04, 1.4284e-04, 1.3629e-04, 1.3914e-04, 9.5007e-05, 1.3190e-04, 1.3489e-04], device='cuda:0') 2022-12-07 06:36:09,567 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6380, 1.5892, 3.8067, 2.5757, 3.4194, 1.8477, 3.1600, 3.4367], device='cuda:0'), covar=tensor([0.0249, 0.6435, 0.0355, 0.9312, 0.0197, 0.4347, 0.0819, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0269, 0.0140, 0.0376, 0.0118, 0.0303, 0.0189, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0001, 0.0003, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:36:17,477 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:36:19,007 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 3.358e+02 4.277e+02 5.886e+02 1.045e+03, threshold=8.554e+02, percent-clipped=3.0 2022-12-07 06:36:44,551 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 06:37:03,607 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 06:37:06,426 INFO [train.py:873] (0/4) Epoch 2, batch 1000, loss[loss=0.2509, simple_loss=0.2103, pruned_loss=0.1458, over 3843.00 frames. ], tot_loss[loss=0.2554, simple_loss=0.2409, pruned_loss=0.135, over 1865856.98 frames. ], batch size: 100, lr: 3.48e-02, grad_scale: 16.0 2022-12-07 06:37:37,839 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7795, 2.0472, 4.6998, 2.5254, 4.4721, 4.5035, 4.2128, 5.0934], device='cuda:0'), covar=tensor([0.0146, 0.2039, 0.0176, 0.1694, 0.0177, 0.0203, 0.0228, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0110, 0.0065, 0.0119, 0.0084, 0.0068, 0.0067, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:37:46,390 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 3.331e+02 4.272e+02 5.638e+02 1.026e+03, threshold=8.545e+02, percent-clipped=1.0 2022-12-07 06:37:57,596 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:38:34,122 INFO [train.py:873] (0/4) Epoch 2, batch 1100, loss[loss=0.2309, simple_loss=0.1988, pruned_loss=0.1315, over 2602.00 frames. ], tot_loss[loss=0.2542, simple_loss=0.2403, pruned_loss=0.1341, over 1914099.30 frames. ], batch size: 100, lr: 3.47e-02, grad_scale: 8.0 2022-12-07 06:39:13,893 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 3.265e+02 4.282e+02 5.954e+02 1.002e+03, threshold=8.564e+02, percent-clipped=5.0 2022-12-07 06:39:34,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 06:39:58,094 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:40:00,476 INFO [train.py:873] (0/4) Epoch 2, batch 1200, loss[loss=0.254, simple_loss=0.2371, pruned_loss=0.1355, over 6926.00 frames. ], tot_loss[loss=0.2534, simple_loss=0.24, pruned_loss=0.1335, over 1921364.94 frames. ], batch size: 100, lr: 3.45e-02, grad_scale: 8.0 2022-12-07 06:40:10,649 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8011, 1.1911, 1.2169, 1.0581, 1.2869, 1.4896, 1.4715, 1.2777], device='cuda:0'), covar=tensor([0.2830, 0.1397, 0.2392, 0.1107, 0.0696, 0.0527, 0.0815, 0.0597], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0051, 0.0039, 0.0040, 0.0051, 0.0038, 0.0047, 0.0052], device='cuda:0'), out_proj_covar=tensor([2.0592e-04, 1.0723e-04, 8.9810e-05, 8.6493e-05, 9.7814e-05, 8.0963e-05, 1.0021e-04, 9.9472e-05], device='cuda:0') 2022-12-07 06:40:33,221 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.22 vs. limit=2.0 2022-12-07 06:40:40,233 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.965e+02 3.180e+02 4.401e+02 5.751e+02 1.287e+03, threshold=8.803e+02, percent-clipped=4.0 2022-12-07 06:40:51,385 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:41:00,665 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9354, 1.9176, 4.4743, 4.2982, 4.5100, 4.1984, 4.0446, 4.7106], device='cuda:0'), covar=tensor([0.1952, 0.1814, 0.0219, 0.0145, 0.0105, 0.0252, 0.0192, 0.0157], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0111, 0.0058, 0.0077, 0.0063, 0.0063, 0.0055, 0.0053], device='cuda:0'), out_proj_covar=tensor([1.8935e-04, 1.8651e-04, 1.0711e-04, 1.5107e-04, 1.1199e-04, 1.1638e-04, 1.0759e-04, 9.2565e-05], device='cuda:0') 2022-12-07 06:41:13,354 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8196, 1.7628, 4.1260, 2.2747, 4.1941, 4.3515, 4.0871, 4.9051], device='cuda:0'), covar=tensor([0.0159, 0.2641, 0.0413, 0.2107, 0.0244, 0.0264, 0.0238, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0112, 0.0066, 0.0120, 0.0084, 0.0070, 0.0067, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:41:25,277 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 06:41:27,152 INFO [train.py:873] (0/4) Epoch 2, batch 1300, loss[loss=0.2478, simple_loss=0.24, pruned_loss=0.1279, over 14336.00 frames. ], tot_loss[loss=0.2529, simple_loss=0.2399, pruned_loss=0.1329, over 1960205.96 frames. ], batch size: 55, lr: 3.44e-02, grad_scale: 8.0 2022-12-07 06:41:29,820 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9870, 1.1472, 2.0696, 1.4372, 1.8893, 1.9916, 1.3847, 2.0116], device='cuda:0'), covar=tensor([0.0131, 0.1049, 0.0151, 0.1112, 0.0196, 0.0194, 0.0515, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0112, 0.0066, 0.0120, 0.0085, 0.0071, 0.0066, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 06:41:35,709 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6891, 1.5818, 1.1700, 0.8276, 1.3158, 0.7878, 1.3733, 1.6261], device='cuda:0'), covar=tensor([0.0247, 0.1869, 0.0665, 0.1399, 0.0746, 0.0505, 0.0737, 0.0378], device='cuda:0'), in_proj_covar=tensor([0.0047, 0.0121, 0.0055, 0.0083, 0.0052, 0.0049, 0.0052, 0.0048], device='cuda:0'), out_proj_covar=tensor([7.7882e-05, 1.9094e-04, 9.4711e-05, 1.3883e-04, 9.8249e-05, 9.1520e-05, 1.0067e-04, 8.5714e-05], device='cuda:0') 2022-12-07 06:42:01,412 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.70 vs. limit=5.0 2022-12-07 06:42:06,753 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.973e+02 3.684e+02 4.915e+02 1.042e+03, threshold=7.368e+02, percent-clipped=2.0 2022-12-07 06:42:17,232 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:42:22,293 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:42:28,277 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0637, 1.7825, 1.0197, 0.8620, 1.1940, 1.2759, 1.6029, 1.8109], device='cuda:0'), covar=tensor([0.0229, 0.1903, 0.0728, 0.1176, 0.0728, 0.0542, 0.0814, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0117, 0.0053, 0.0080, 0.0051, 0.0048, 0.0050, 0.0045], device='cuda:0'), out_proj_covar=tensor([7.4769e-05, 1.8572e-04, 9.1737e-05, 1.3489e-04, 9.5923e-05, 8.8653e-05, 9.7408e-05, 8.0852e-05], device='cuda:0') 2022-12-07 06:42:30,138 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2102, 1.7657, 2.2069, 2.2411, 2.6152, 2.1589, 2.0533, 2.0288], device='cuda:0'), covar=tensor([0.0177, 0.1253, 0.0081, 0.0436, 0.0142, 0.0250, 0.0215, 0.1062], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0239, 0.0097, 0.0164, 0.0107, 0.0132, 0.0115, 0.0279], device='cuda:0'), out_proj_covar=tensor([1.0722e-04, 2.0017e-04, 7.7758e-05, 1.3483e-04, 9.1553e-05, 1.0528e-04, 1.1433e-04, 2.2571e-04], device='cuda:0') 2022-12-07 06:42:53,349 INFO [train.py:873] (0/4) Epoch 2, batch 1400, loss[loss=0.3074, simple_loss=0.2625, pruned_loss=0.1761, over 7746.00 frames. ], tot_loss[loss=0.2522, simple_loss=0.2393, pruned_loss=0.1326, over 1947973.97 frames. ], batch size: 100, lr: 3.42e-02, grad_scale: 8.0 2022-12-07 06:42:54,667 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2022-12-07 06:42:58,568 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:13,857 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:15,597 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:27,223 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 06:43:33,021 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 3.417e+02 4.711e+02 5.630e+02 1.118e+03, threshold=9.423e+02, percent-clipped=7.0 2022-12-07 06:43:36,656 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:43:57,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 06:44:06,571 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:44:12,438 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8140, 4.4661, 5.0434, 4.2293, 4.7137, 4.9782, 2.2060, 4.8006], device='cuda:0'), covar=tensor([0.0202, 0.0279, 0.0296, 0.0375, 0.0247, 0.0119, 0.2898, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0081, 0.0084, 0.0066, 0.0110, 0.0074, 0.0125, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:44:20,922 INFO [train.py:873] (0/4) Epoch 2, batch 1500, loss[loss=0.2609, simple_loss=0.2369, pruned_loss=0.1425, over 8602.00 frames. ], tot_loss[loss=0.2515, simple_loss=0.2386, pruned_loss=0.1322, over 1923323.18 frames. ], batch size: 100, lr: 3.41e-02, grad_scale: 8.0 2022-12-07 06:44:28,141 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 06:44:30,374 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:44:38,189 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2893, 1.0227, 0.9592, 1.1185, 1.0573, 1.1517, 1.0327, 0.9468], device='cuda:0'), covar=tensor([0.1960, 0.0855, 0.0665, 0.0422, 0.0605, 0.0529, 0.1115, 0.0692], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0053, 0.0041, 0.0041, 0.0053, 0.0042, 0.0053, 0.0055], device='cuda:0'), out_proj_covar=tensor([2.2962e-04, 1.1469e-04, 9.8044e-05, 9.0650e-05, 1.0401e-04, 9.0680e-05, 1.1476e-04, 1.0819e-04], device='cuda:0') 2022-12-07 06:44:59,855 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 3.213e+02 4.122e+02 5.427e+02 9.769e+02, threshold=8.244e+02, percent-clipped=1.0 2022-12-07 06:45:05,744 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:45:15,991 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0263, 3.2949, 3.7132, 3.6089, 3.3421, 2.9912, 3.7598, 2.6929], device='cuda:0'), covar=tensor([0.0271, 0.0268, 0.0204, 0.0259, 0.0264, 0.0727, 0.0061, 0.0730], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0098, 0.0074, 0.0097, 0.0095, 0.0146, 0.0060, 0.0143], device='cuda:0'), out_proj_covar=tensor([9.2269e-05, 1.1517e-04, 8.8627e-05, 1.1637e-04, 1.1184e-04, 1.7134e-04, 6.5986e-05, 1.6012e-04], device='cuda:0') 2022-12-07 06:45:46,184 INFO [train.py:873] (0/4) Epoch 2, batch 1600, loss[loss=0.1983, simple_loss=0.1812, pruned_loss=0.1078, over 3865.00 frames. ], tot_loss[loss=0.2512, simple_loss=0.2388, pruned_loss=0.1318, over 1955887.00 frames. ], batch size: 100, lr: 3.39e-02, grad_scale: 8.0 2022-12-07 06:46:26,068 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 3.267e+02 4.651e+02 6.782e+02 1.677e+03, threshold=9.302e+02, percent-clipped=10.0 2022-12-07 06:47:13,306 INFO [train.py:873] (0/4) Epoch 2, batch 1700, loss[loss=0.2838, simple_loss=0.2604, pruned_loss=0.1535, over 4985.00 frames. ], tot_loss[loss=0.2492, simple_loss=0.2376, pruned_loss=0.1304, over 1954705.00 frames. ], batch size: 100, lr: 3.38e-02, grad_scale: 8.0 2022-12-07 06:47:15,350 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5424, 1.1878, 1.9058, 1.7707, 1.9592, 1.8149, 1.4269, 2.0361], device='cuda:0'), covar=tensor([0.0404, 0.0807, 0.0141, 0.0205, 0.0122, 0.0110, 0.0289, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0116, 0.0059, 0.0078, 0.0064, 0.0066, 0.0055, 0.0054], device='cuda:0'), out_proj_covar=tensor([1.9837e-04, 1.9906e-04, 1.1087e-04, 1.5466e-04, 1.1569e-04, 1.2387e-04, 1.1033e-04, 9.7110e-05], device='cuda:0') 2022-12-07 06:47:19,762 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=3.15 vs. limit=2.0 2022-12-07 06:47:30,373 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:47:53,363 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.995e+02 4.223e+02 5.322e+02 8.026e+02, threshold=8.446e+02, percent-clipped=0.0 2022-12-07 06:48:21,582 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:48:32,159 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7421, 3.3240, 3.2750, 3.6614, 3.7655, 3.6714, 3.7120, 3.2640], device='cuda:0'), covar=tensor([0.0274, 0.0909, 0.0331, 0.0477, 0.0397, 0.0402, 0.0543, 0.0487], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0118, 0.0076, 0.0077, 0.0084, 0.0085, 0.0113, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:48:39,296 INFO [train.py:873] (0/4) Epoch 2, batch 1800, loss[loss=0.1806, simple_loss=0.1558, pruned_loss=0.1028, over 2631.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.2366, pruned_loss=0.1295, over 1977851.93 frames. ], batch size: 100, lr: 3.37e-02, grad_scale: 8.0 2022-12-07 06:48:44,780 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:49:19,721 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.949e+02 3.801e+02 5.110e+02 1.024e+03, threshold=7.602e+02, percent-clipped=3.0 2022-12-07 06:49:24,087 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:49:25,837 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:49:34,663 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 06:49:48,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=3.27 vs. limit=2.0 2022-12-07 06:49:50,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=4.86 vs. limit=2.0 2022-12-07 06:50:07,242 INFO [train.py:873] (0/4) Epoch 2, batch 1900, loss[loss=0.2807, simple_loss=0.231, pruned_loss=0.1652, over 2619.00 frames. ], tot_loss[loss=0.2501, simple_loss=0.2378, pruned_loss=0.1312, over 1943552.17 frames. ], batch size: 100, lr: 3.35e-02, grad_scale: 8.0 2022-12-07 06:50:08,453 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:50:17,826 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:50:46,648 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.987e+02 3.780e+02 5.318e+02 1.239e+03, threshold=7.560e+02, percent-clipped=8.0 2022-12-07 06:50:52,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5982, 3.0613, 3.9601, 2.2184, 2.9021, 2.4858, 0.9151, 2.8176], device='cuda:0'), covar=tensor([0.0779, 0.0305, 0.0378, 0.0516, 0.0291, 0.0552, 0.1983, 0.0532], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0056, 0.0052, 0.0056, 0.0054, 0.0050, 0.0098, 0.0060], device='cuda:0'), out_proj_covar=tensor([7.6739e-05, 7.6968e-05, 7.2869e-05, 7.2612e-05, 6.8789e-05, 6.8367e-05, 1.3522e-04, 8.2466e-05], device='cuda:0') 2022-12-07 06:51:33,029 INFO [train.py:873] (0/4) Epoch 2, batch 2000, loss[loss=0.2541, simple_loss=0.2441, pruned_loss=0.132, over 14280.00 frames. ], tot_loss[loss=0.251, simple_loss=0.2378, pruned_loss=0.1321, over 1895448.82 frames. ], batch size: 80, lr: 3.34e-02, grad_scale: 8.0 2022-12-07 06:51:43,848 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 06:51:50,735 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:51:51,707 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:06,324 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3896, 5.3531, 5.0230, 5.6820, 5.4151, 4.4671, 5.8005, 5.7436], device='cuda:0'), covar=tensor([0.0671, 0.0446, 0.0732, 0.0656, 0.0620, 0.0372, 0.0488, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0070, 0.0097, 0.0089, 0.0092, 0.0060, 0.0084, 0.0089], device='cuda:0'), out_proj_covar=tensor([1.3902e-04, 1.1780e-04, 1.5592e-04, 1.4311e-04, 1.5174e-04, 9.8130e-05, 1.4063e-04, 1.4542e-04], device='cuda:0') 2022-12-07 06:52:12,819 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 3.149e+02 4.158e+02 5.178e+02 1.006e+03, threshold=8.316e+02, percent-clipped=10.0 2022-12-07 06:52:15,371 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5946, 4.2552, 4.1914, 4.3963, 4.1670, 4.3624, 4.4723, 4.5395], device='cuda:0'), covar=tensor([0.0458, 0.0399, 0.0583, 0.0396, 0.0338, 0.0312, 0.0848, 0.0606], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0105, 0.0128, 0.0109, 0.0114, 0.0121, 0.0131, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:52:27,404 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8415, 1.5958, 2.7039, 1.4424, 1.9220, 2.0635, 0.9208, 1.8505], device='cuda:0'), covar=tensor([0.0694, 0.0757, 0.0339, 0.0638, 0.0446, 0.0475, 0.1918, 0.0618], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0058, 0.0053, 0.0057, 0.0055, 0.0050, 0.0098, 0.0062], device='cuda:0'), out_proj_covar=tensor([7.8709e-05, 8.1179e-05, 7.4390e-05, 7.5039e-05, 7.0364e-05, 6.8494e-05, 1.3582e-04, 8.4942e-05], device='cuda:0') 2022-12-07 06:52:32,379 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:41,417 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:44,067 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:54,890 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:52:59,448 INFO [train.py:873] (0/4) Epoch 2, batch 2100, loss[loss=0.2549, simple_loss=0.2428, pruned_loss=0.1335, over 11188.00 frames. ], tot_loss[loss=0.248, simple_loss=0.2361, pruned_loss=0.1299, over 1922422.61 frames. ], batch size: 100, lr: 3.32e-02, grad_scale: 8.0 2022-12-07 06:53:05,042 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:22,841 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:26,336 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7586, 1.3774, 1.9644, 1.5924, 1.9711, 1.3481, 1.7998, 1.9523], device='cuda:0'), covar=tensor([0.0427, 0.1423, 0.0194, 0.0988, 0.0120, 0.0931, 0.0455, 0.0129], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0292, 0.0148, 0.0398, 0.0121, 0.0310, 0.0208, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0004, 0.0001, 0.0003, 0.0002, 0.0001], device='cuda:0') 2022-12-07 06:53:33,505 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4597, 3.8246, 4.4065, 3.8224, 4.3061, 4.4592, 2.0826, 4.2268], device='cuda:0'), covar=tensor([0.0200, 0.0413, 0.0661, 0.0385, 0.0387, 0.0205, 0.3149, 0.0323], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0081, 0.0079, 0.0062, 0.0105, 0.0071, 0.0121, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:53:39,633 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 3.245e+02 4.428e+02 6.131e+02 1.111e+03, threshold=8.856e+02, percent-clipped=8.0 2022-12-07 06:53:46,125 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:47,975 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:53:53,720 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7953, 0.7170, 0.1574, 0.6653, 0.6767, 0.1786, 0.3723, 0.5992], device='cuda:0'), covar=tensor([0.0062, 0.0046, 0.0049, 0.0263, 0.0053, 0.0059, 0.0155, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0023, 0.0024, 0.0020, 0.0019, 0.0024, 0.0017, 0.0020], device='cuda:0'), out_proj_covar=tensor([3.9460e-05, 4.6061e-05, 4.2072e-05, 4.1010e-05, 3.5259e-05, 4.3934e-05, 3.4875e-05, 3.8389e-05], device='cuda:0') 2022-12-07 06:54:09,248 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:54:13,646 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1486, 1.1183, 0.9763, 1.0178, 0.7992, 0.9093, 0.5675, 0.9186], device='cuda:0'), covar=tensor([0.0150, 0.0331, 0.0405, 0.0235, 0.0457, 0.0299, 0.0255, 0.0334], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0017, 0.0017, 0.0017, 0.0017, 0.0017, 0.0016, 0.0017], device='cuda:0'), out_proj_covar=tensor([2.2456e-05, 2.6729e-05, 3.0362e-05, 2.5813e-05, 2.9524e-05, 2.7008e-05, 3.1700e-05, 2.8109e-05], device='cuda:0') 2022-12-07 06:54:26,278 INFO [train.py:873] (0/4) Epoch 2, batch 2200, loss[loss=0.2673, simple_loss=0.2477, pruned_loss=0.1435, over 14152.00 frames. ], tot_loss[loss=0.248, simple_loss=0.2358, pruned_loss=0.1301, over 1875356.25 frames. ], batch size: 37, lr: 3.31e-02, grad_scale: 8.0 2022-12-07 06:54:32,302 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:54:36,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 06:54:49,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 06:55:01,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0266, 1.9480, 2.0683, 1.9638, 2.0734, 1.8546, 1.1386, 1.9019], device='cuda:0'), covar=tensor([0.0261, 0.0247, 0.0267, 0.0184, 0.0203, 0.0305, 0.1235, 0.0312], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0084, 0.0085, 0.0065, 0.0108, 0.0073, 0.0125, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 06:55:01,520 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:55:05,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.644e+01 3.042e+02 4.179e+02 5.699e+02 1.039e+03, threshold=8.358e+02, percent-clipped=2.0 2022-12-07 06:55:53,424 INFO [train.py:873] (0/4) Epoch 2, batch 2300, loss[loss=0.2368, simple_loss=0.2341, pruned_loss=0.1197, over 14287.00 frames. ], tot_loss[loss=0.2468, simple_loss=0.2356, pruned_loss=0.129, over 1953911.68 frames. ], batch size: 44, lr: 3.30e-02, grad_scale: 8.0 2022-12-07 06:56:03,712 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 06:56:32,991 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2022-12-07 06:56:33,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 3.202e+02 3.964e+02 5.237e+02 8.345e+02, threshold=7.929e+02, percent-clipped=0.0 2022-12-07 06:56:45,122 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6198, 1.0938, 0.8251, 0.9225, 1.2710, 1.2410, 1.0155, 1.0965], device='cuda:0'), covar=tensor([0.0237, 0.0682, 0.0909, 0.0591, 0.0407, 0.0356, 0.0159, 0.0392], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0015, 0.0017, 0.0016, 0.0017, 0.0016, 0.0016], device='cuda:0'), out_proj_covar=tensor([2.2395e-05, 2.5580e-05, 2.8611e-05, 2.4782e-05, 2.7734e-05, 2.5622e-05, 3.1904e-05, 2.5980e-05], device='cuda:0') 2022-12-07 06:57:00,536 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:57:20,360 INFO [train.py:873] (0/4) Epoch 2, batch 2400, loss[loss=0.2347, simple_loss=0.229, pruned_loss=0.1202, over 14275.00 frames. ], tot_loss[loss=0.246, simple_loss=0.2353, pruned_loss=0.1284, over 1981939.89 frames. ], batch size: 63, lr: 3.28e-02, grad_scale: 8.0 2022-12-07 06:57:52,690 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-10000.pt 2022-12-07 06:58:03,556 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 3.087e+02 4.289e+02 5.805e+02 1.264e+03, threshold=8.579e+02, percent-clipped=4.0 2022-12-07 06:58:04,115 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 06:58:07,913 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:58:16,476 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 06:58:17,972 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.28 vs. limit=2.0 2022-12-07 06:58:30,047 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:58:50,355 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:58:50,992 INFO [train.py:873] (0/4) Epoch 2, batch 2500, loss[loss=0.2406, simple_loss=0.2341, pruned_loss=0.1236, over 14698.00 frames. ], tot_loss[loss=0.2469, simple_loss=0.2355, pruned_loss=0.1291, over 1944623.66 frames. ], batch size: 33, lr: 3.27e-02, grad_scale: 8.0 2022-12-07 06:58:57,330 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:21,258 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:22,244 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 06:59:30,093 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 3.039e+02 4.037e+02 4.970e+02 1.749e+03, threshold=8.074e+02, percent-clipped=6.0 2022-12-07 06:59:37,902 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 06:59:42,142 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 06:59:42,878 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8088, 4.6317, 4.3098, 5.0043, 4.7217, 3.9088, 5.1300, 4.9219], device='cuda:0'), covar=tensor([0.0750, 0.0544, 0.0758, 0.0651, 0.0569, 0.0656, 0.0560, 0.1023], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0070, 0.0094, 0.0091, 0.0097, 0.0063, 0.0086, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:00:05,351 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.45 vs. limit=2.0 2022-12-07 07:00:16,956 INFO [train.py:873] (0/4) Epoch 2, batch 2600, loss[loss=0.2177, simple_loss=0.2203, pruned_loss=0.1075, over 14239.00 frames. ], tot_loss[loss=0.2483, simple_loss=0.236, pruned_loss=0.1303, over 1914835.16 frames. ], batch size: 35, lr: 3.26e-02, grad_scale: 8.0 2022-12-07 07:00:18,023 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7969, 1.7514, 1.5544, 0.9614, 1.5764, 1.2324, 1.7588, 1.7176], device='cuda:0'), covar=tensor([0.0296, 0.2254, 0.0576, 0.1136, 0.0656, 0.0574, 0.0658, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0046, 0.0118, 0.0057, 0.0085, 0.0052, 0.0052, 0.0049, 0.0048], device='cuda:0'), out_proj_covar=tensor([8.5718e-05, 1.9971e-04, 1.0539e-04, 1.5532e-04, 1.0595e-04, 1.0075e-04, 1.0217e-04, 9.2769e-05], device='cuda:0') 2022-12-07 07:00:51,441 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8243, 2.1158, 3.1430, 3.1762, 3.3052, 2.1772, 2.7574, 2.0738], device='cuda:0'), covar=tensor([0.0142, 0.0307, 0.0220, 0.0183, 0.0088, 0.0589, 0.0054, 0.0618], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0109, 0.0088, 0.0108, 0.0099, 0.0159, 0.0065, 0.0154], device='cuda:0'), out_proj_covar=tensor([1.0648e-04, 1.3454e-04, 1.1596e-04, 1.3554e-04, 1.2175e-04, 1.9546e-04, 7.7176e-05, 1.7958e-04], device='cuda:0') 2022-12-07 07:00:56,780 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 3.080e+02 4.222e+02 5.388e+02 1.025e+03, threshold=8.444e+02, percent-clipped=5.0 2022-12-07 07:01:24,264 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:01:26,688 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0886, 2.9135, 2.8153, 3.1843, 2.8711, 2.3692, 3.0968, 3.1093], device='cuda:0'), covar=tensor([0.0816, 0.0686, 0.0857, 0.0608, 0.0780, 0.0727, 0.0771, 0.0704], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0068, 0.0094, 0.0086, 0.0093, 0.0063, 0.0083, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:01:44,393 INFO [train.py:873] (0/4) Epoch 2, batch 2700, loss[loss=0.2542, simple_loss=0.2447, pruned_loss=0.1319, over 14384.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.236, pruned_loss=0.1298, over 1892282.43 frames. ], batch size: 55, lr: 3.24e-02, grad_scale: 8.0 2022-12-07 07:02:06,310 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:02:20,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8394, 3.4559, 3.5197, 3.7997, 3.7474, 3.6926, 3.8684, 3.4497], device='cuda:0'), covar=tensor([0.0373, 0.0909, 0.0334, 0.0409, 0.0601, 0.0613, 0.0510, 0.0471], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0131, 0.0081, 0.0081, 0.0091, 0.0089, 0.0121, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:02:24,379 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 3.215e+02 3.934e+02 5.144e+02 9.049e+02, threshold=7.869e+02, percent-clipped=4.0 2022-12-07 07:02:29,015 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:09,552 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:10,326 INFO [train.py:873] (0/4) Epoch 2, batch 2800, loss[loss=0.2695, simple_loss=0.2593, pruned_loss=0.1399, over 14065.00 frames. ], tot_loss[loss=0.2461, simple_loss=0.2352, pruned_loss=0.1285, over 1956644.17 frames. ], batch size: 29, lr: 3.23e-02, grad_scale: 8.0 2022-12-07 07:03:20,355 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2022-12-07 07:03:38,848 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:03:42,167 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:50,604 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 3.431e+02 4.642e+02 5.590e+02 1.030e+03, threshold=9.284e+02, percent-clipped=5.0 2022-12-07 07:03:50,841 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:03:58,239 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:04:21,825 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 07:04:23,705 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:04:29,819 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:04:37,270 INFO [train.py:873] (0/4) Epoch 2, batch 2900, loss[loss=0.2301, simple_loss=0.2284, pruned_loss=0.1159, over 11281.00 frames. ], tot_loss[loss=0.2432, simple_loss=0.2334, pruned_loss=0.1265, over 1990683.61 frames. ], batch size: 14, lr: 3.22e-02, grad_scale: 8.0 2022-12-07 07:04:43,775 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:05:01,637 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.76 vs. limit=5.0 2022-12-07 07:05:02,118 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8351, 2.6761, 2.0936, 3.1022, 2.8425, 3.0364, 2.4834, 2.0165], device='cuda:0'), covar=tensor([0.0262, 0.0374, 0.2214, 0.0158, 0.0256, 0.0181, 0.0492, 0.2368], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0177, 0.0285, 0.0129, 0.0127, 0.0126, 0.0171, 0.0307], device='cuda:0'), out_proj_covar=tensor([1.0202e-04, 1.2396e-04, 1.9182e-04, 8.7499e-05, 9.4424e-05, 9.2520e-05, 1.2421e-04, 2.0460e-04], device='cuda:0') 2022-12-07 07:05:07,940 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5536, 1.1328, 1.2374, 1.1601, 1.0553, 1.3499, 1.2812, 1.1932], device='cuda:0'), covar=tensor([0.2401, 0.1105, 0.0937, 0.0498, 0.0632, 0.0422, 0.1004, 0.0650], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0055, 0.0043, 0.0045, 0.0056, 0.0042, 0.0052, 0.0059], device='cuda:0'), out_proj_covar=tensor([2.6536e-04, 1.2856e-04, 1.1224e-04, 1.0522e-04, 1.1968e-04, 9.9192e-05, 1.2656e-04, 1.2579e-04], device='cuda:0') 2022-12-07 07:05:09,215 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2022-12-07 07:05:18,327 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.981e+02 3.992e+02 4.851e+02 1.210e+03, threshold=7.983e+02, percent-clipped=1.0 2022-12-07 07:05:23,864 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 07:05:35,204 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2022-12-07 07:06:05,151 INFO [train.py:873] (0/4) Epoch 2, batch 3000, loss[loss=0.259, simple_loss=0.2503, pruned_loss=0.1339, over 14213.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.2348, pruned_loss=0.128, over 2025783.71 frames. ], batch size: 94, lr: 3.21e-02, grad_scale: 8.0 2022-12-07 07:06:05,152 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 07:06:13,212 INFO [train.py:905] (0/4) Epoch 2, validation: loss=0.1433, simple_loss=0.1828, pruned_loss=0.05186, over 857387.00 frames. 2022-12-07 07:06:13,213 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 07:06:33,155 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5948, 3.0697, 3.1337, 3.5588, 3.5256, 3.5170, 3.5520, 3.0704], device='cuda:0'), covar=tensor([0.0275, 0.1247, 0.0394, 0.0395, 0.0614, 0.0413, 0.0604, 0.0562], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0133, 0.0082, 0.0081, 0.0088, 0.0091, 0.0121, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:06:39,531 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 07:06:44,892 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0920, 1.6931, 3.7368, 1.9189, 3.8226, 3.7226, 2.7720, 4.1989], device='cuda:0'), covar=tensor([0.0171, 0.2540, 0.0359, 0.2150, 0.0235, 0.0297, 0.0577, 0.0172], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0129, 0.0081, 0.0137, 0.0096, 0.0085, 0.0077, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:06:54,389 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 3.162e+02 4.298e+02 5.900e+02 1.176e+03, threshold=8.596e+02, percent-clipped=5.0 2022-12-07 07:06:56,391 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.71 vs. limit=5.0 2022-12-07 07:07:08,661 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 07:07:12,047 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:07:14,665 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 07:07:41,333 INFO [train.py:873] (0/4) Epoch 2, batch 3100, loss[loss=0.2452, simple_loss=0.2179, pruned_loss=0.1363, over 3897.00 frames. ], tot_loss[loss=0.2436, simple_loss=0.2337, pruned_loss=0.1267, over 1990359.22 frames. ], batch size: 100, lr: 3.19e-02, grad_scale: 8.0 2022-12-07 07:07:58,895 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.27 vs. limit=2.0 2022-12-07 07:08:06,629 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 07:08:09,618 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:08:22,248 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 3.394e+02 4.427e+02 5.401e+02 1.921e+03, threshold=8.855e+02, percent-clipped=7.0 2022-12-07 07:08:29,208 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:08:51,473 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:08:52,384 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:08:57,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3719, 1.2184, 1.2608, 1.1000, 0.8406, 0.9368, 1.1900, 0.9808], device='cuda:0'), covar=tensor([0.0197, 0.1133, 0.0293, 0.0679, 0.0588, 0.0526, 0.0349, 0.0390], device='cuda:0'), in_proj_covar=tensor([0.0048, 0.0126, 0.0059, 0.0086, 0.0052, 0.0054, 0.0051, 0.0051], device='cuda:0'), out_proj_covar=tensor([9.4708e-05, 2.1808e-04, 1.1268e-04, 1.6099e-04, 1.1063e-04, 1.0947e-04, 1.0997e-04, 1.0105e-04], device='cuda:0') 2022-12-07 07:09:03,824 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.86 vs. limit=5.0 2022-12-07 07:09:08,152 INFO [train.py:873] (0/4) Epoch 2, batch 3200, loss[loss=0.2394, simple_loss=0.2236, pruned_loss=0.1276, over 6969.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.2334, pruned_loss=0.1272, over 1898472.01 frames. ], batch size: 100, lr: 3.18e-02, grad_scale: 8.0 2022-12-07 07:09:10,183 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:10,989 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:09:18,369 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:27,771 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8183, 1.1455, 2.4811, 2.4612, 2.5334, 2.4287, 1.3628, 2.5916], device='cuda:0'), covar=tensor([0.1230, 0.1806, 0.0226, 0.0289, 0.0220, 0.0195, 0.0632, 0.0168], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0118, 0.0062, 0.0080, 0.0069, 0.0070, 0.0059, 0.0058], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:09:29,813 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-07 07:09:44,989 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:09:49,654 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.759e+01 3.135e+02 4.113e+02 5.521e+02 8.692e+02, threshold=8.227e+02, percent-clipped=0.0 2022-12-07 07:09:49,791 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:10:11,425 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:10:21,944 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.05 vs. limit=2.0 2022-12-07 07:10:33,996 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6223, 0.8271, 0.9931, 0.7731, 0.7387, 0.8371, 1.3623, 0.6079], device='cuda:0'), covar=tensor([0.0265, 0.0229, 0.0211, 0.0262, 0.0276, 0.0218, 0.0124, 0.0365], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0015, 0.0015, 0.0017, 0.0015, 0.0016, 0.0015, 0.0017], device='cuda:0'), out_proj_covar=tensor([2.6156e-05, 2.6558e-05, 2.9041e-05, 2.7391e-05, 2.8698e-05, 2.6037e-05, 3.0746e-05, 2.9663e-05], device='cuda:0') 2022-12-07 07:10:35,550 INFO [train.py:873] (0/4) Epoch 2, batch 3300, loss[loss=0.1942, simple_loss=0.1759, pruned_loss=0.1062, over 2578.00 frames. ], tot_loss[loss=0.241, simple_loss=0.2316, pruned_loss=0.1252, over 1936371.83 frames. ], batch size: 100, lr: 3.17e-02, grad_scale: 8.0 2022-12-07 07:10:53,768 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:11:15,640 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.971e+02 4.122e+02 5.420e+02 1.001e+03, threshold=8.244e+02, percent-clipped=4.0 2022-12-07 07:11:26,415 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8123, 3.0639, 3.4535, 3.6938, 3.4711, 3.6309, 3.7178, 3.3274], device='cuda:0'), covar=tensor([0.0769, 0.1952, 0.0808, 0.1051, 0.1197, 0.0930, 0.1097, 0.1153], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0139, 0.0085, 0.0084, 0.0090, 0.0093, 0.0127, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:11:46,519 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 07:12:00,256 INFO [train.py:873] (0/4) Epoch 2, batch 3400, loss[loss=0.2708, simple_loss=0.2405, pruned_loss=0.1506, over 4955.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.2316, pruned_loss=0.1258, over 1868667.43 frames. ], batch size: 100, lr: 3.16e-02, grad_scale: 8.0 2022-12-07 07:12:42,275 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.877e+02 3.778e+02 5.417e+02 1.188e+03, threshold=7.556e+02, percent-clipped=7.0 2022-12-07 07:12:52,252 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0952, 1.7167, 2.2778, 2.1537, 2.3892, 2.0468, 2.0084, 2.0822], device='cuda:0'), covar=tensor([0.0146, 0.0744, 0.0064, 0.0281, 0.0086, 0.0195, 0.0141, 0.0724], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0284, 0.0112, 0.0202, 0.0137, 0.0167, 0.0155, 0.0319], device='cuda:0'), out_proj_covar=tensor([1.3721e-04, 2.4644e-04, 9.4503e-05, 1.7011e-04, 1.2691e-04, 1.4847e-04, 1.5563e-04, 2.7075e-04], device='cuda:0') 2022-12-07 07:13:25,158 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 07:13:28,830 INFO [train.py:873] (0/4) Epoch 2, batch 3500, loss[loss=0.206, simple_loss=0.2106, pruned_loss=0.1007, over 14303.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.2302, pruned_loss=0.1241, over 1858320.30 frames. ], batch size: 39, lr: 3.15e-02, grad_scale: 8.0 2022-12-07 07:13:30,329 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:13:57,959 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3184, 3.2422, 3.4449, 2.9645, 3.3615, 2.9976, 1.4609, 3.2042], device='cuda:0'), covar=tensor([0.0253, 0.0377, 0.0405, 0.0376, 0.0293, 0.0550, 0.2997, 0.0283], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0089, 0.0091, 0.0072, 0.0116, 0.0080, 0.0131, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 07:14:00,463 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:08,559 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 3.159e+02 4.112e+02 5.380e+02 8.623e+02, threshold=8.224e+02, percent-clipped=2.0 2022-12-07 07:14:08,700 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:14:11,089 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:26,031 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:28,611 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1054, 2.8067, 2.7589, 3.1412, 3.1570, 3.1112, 3.2164, 2.7920], device='cuda:0'), covar=tensor([0.0418, 0.1153, 0.0442, 0.0482, 0.0595, 0.0491, 0.0625, 0.0617], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0140, 0.0088, 0.0084, 0.0093, 0.0092, 0.0129, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:14:50,087 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:14:54,227 INFO [train.py:873] (0/4) Epoch 2, batch 3600, loss[loss=0.2187, simple_loss=0.2194, pruned_loss=0.109, over 14266.00 frames. ], tot_loss[loss=0.2395, simple_loss=0.2307, pruned_loss=0.1241, over 1898209.26 frames. ], batch size: 25, lr: 3.13e-02, grad_scale: 8.0 2022-12-07 07:15:04,186 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:15:35,173 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.817e+02 3.568e+02 4.695e+02 1.140e+03, threshold=7.135e+02, percent-clipped=4.0 2022-12-07 07:15:43,437 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0947, 1.4974, 3.0807, 1.6248, 3.0734, 2.9744, 2.1145, 3.2960], device='cuda:0'), covar=tensor([0.0301, 0.2478, 0.0371, 0.2237, 0.0475, 0.0453, 0.1126, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0125, 0.0078, 0.0134, 0.0101, 0.0087, 0.0078, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:15:55,224 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 07:15:56,896 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:16:02,122 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 07:16:21,363 INFO [train.py:873] (0/4) Epoch 2, batch 3700, loss[loss=0.1915, simple_loss=0.1715, pruned_loss=0.1057, over 2611.00 frames. ], tot_loss[loss=0.2411, simple_loss=0.2317, pruned_loss=0.1252, over 1900077.36 frames. ], batch size: 100, lr: 3.12e-02, grad_scale: 8.0 2022-12-07 07:16:55,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.79 vs. limit=5.0 2022-12-07 07:17:02,054 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 3.008e+02 3.803e+02 5.016e+02 1.141e+03, threshold=7.605e+02, percent-clipped=8.0 2022-12-07 07:17:15,426 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8251, 1.6549, 2.2092, 2.3535, 1.6958, 1.7421, 2.2213, 1.8949], device='cuda:0'), covar=tensor([0.0077, 0.0099, 0.0091, 0.0052, 0.0075, 0.0190, 0.0050, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0124, 0.0104, 0.0116, 0.0107, 0.0167, 0.0070, 0.0159], device='cuda:0'), out_proj_covar=tensor([1.1852e-04, 1.5930e-04, 1.4263e-04, 1.5230e-04, 1.3920e-04, 2.2020e-04, 8.8905e-05, 1.9401e-04], device='cuda:0') 2022-12-07 07:17:24,508 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1981, 1.4739, 1.7361, 1.5072, 1.0257, 1.6635, 1.8824, 1.7033], device='cuda:0'), covar=tensor([0.2774, 0.1106, 0.0972, 0.1677, 0.1140, 0.0599, 0.0761, 0.1007], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0056, 0.0044, 0.0046, 0.0059, 0.0044, 0.0056, 0.0063], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:17:47,318 INFO [train.py:873] (0/4) Epoch 2, batch 3800, loss[loss=0.2166, simple_loss=0.1812, pruned_loss=0.126, over 1257.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.232, pruned_loss=0.1252, over 1946566.05 frames. ], batch size: 100, lr: 3.11e-02, grad_scale: 8.0 2022-12-07 07:18:19,980 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:18:28,499 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 3.176e+02 4.502e+02 5.921e+02 1.352e+03, threshold=9.005e+02, percent-clipped=11.0 2022-12-07 07:18:46,156 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:02,213 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:15,324 INFO [train.py:873] (0/4) Epoch 2, batch 3900, loss[loss=0.2414, simple_loss=0.198, pruned_loss=0.1424, over 2648.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.2296, pruned_loss=0.1224, over 1969254.10 frames. ], batch size: 100, lr: 3.10e-02, grad_scale: 8.0 2022-12-07 07:19:27,728 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:19:38,784 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 07:19:55,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.890e+02 4.018e+02 5.176e+02 1.364e+03, threshold=8.037e+02, percent-clipped=4.0 2022-12-07 07:19:55,582 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8762, 1.6330, 4.0341, 3.8271, 3.9155, 4.1236, 3.4840, 4.1923], device='cuda:0'), covar=tensor([0.1571, 0.1624, 0.0158, 0.0136, 0.0127, 0.0104, 0.0225, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0126, 0.0065, 0.0084, 0.0073, 0.0072, 0.0062, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:20:02,370 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0978, 1.1341, 1.2303, 1.2008, 1.1419, 1.2382, 0.8535, 1.0386], device='cuda:0'), covar=tensor([0.1600, 0.0839, 0.0501, 0.0327, 0.0478, 0.0200, 0.0773, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0058, 0.0043, 0.0047, 0.0058, 0.0046, 0.0057, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:20:12,865 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:20:14,686 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:20:18,352 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3099, 2.0557, 2.7442, 2.8355, 2.4086, 1.9891, 2.7209, 2.0555], device='cuda:0'), covar=tensor([0.0108, 0.0158, 0.0141, 0.0136, 0.0079, 0.0359, 0.0041, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0123, 0.0104, 0.0118, 0.0108, 0.0165, 0.0072, 0.0159], device='cuda:0'), out_proj_covar=tensor([1.2022e-04, 1.5914e-04, 1.4405e-04, 1.5594e-04, 1.4164e-04, 2.1965e-04, 9.0381e-05, 1.9683e-04], device='cuda:0') 2022-12-07 07:20:22,681 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:20:40,966 INFO [train.py:873] (0/4) Epoch 2, batch 4000, loss[loss=0.2232, simple_loss=0.2278, pruned_loss=0.1092, over 14261.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.2309, pruned_loss=0.1231, over 2007077.54 frames. ], batch size: 39, lr: 3.09e-02, grad_scale: 8.0 2022-12-07 07:21:03,667 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:21:07,122 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:21:22,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 3.113e+02 4.062e+02 5.460e+02 1.002e+03, threshold=8.123e+02, percent-clipped=4.0 2022-12-07 07:21:31,208 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4561, 1.2058, 1.2617, 1.3045, 0.9690, 1.2269, 0.9624, 1.2197], device='cuda:0'), covar=tensor([0.2207, 0.0830, 0.0885, 0.0690, 0.0730, 0.0384, 0.1180, 0.0790], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0055, 0.0041, 0.0045, 0.0056, 0.0042, 0.0054, 0.0059], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:21:44,088 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:08,541 INFO [train.py:873] (0/4) Epoch 2, batch 4100, loss[loss=0.2731, simple_loss=0.256, pruned_loss=0.1451, over 12737.00 frames. ], tot_loss[loss=0.2387, simple_loss=0.2309, pruned_loss=0.1233, over 1998978.65 frames. ], batch size: 100, lr: 3.08e-02, grad_scale: 8.0 2022-12-07 07:22:33,942 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8287, 0.6849, 1.0481, 0.8548, 0.9137, 0.7112, 1.2713, 0.4913], device='cuda:0'), covar=tensor([0.0520, 0.0538, 0.0174, 0.0395, 0.0200, 0.0210, 0.0106, 0.0440], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0021, 0.0020, 0.0019, 0.0021, 0.0016, 0.0019], device='cuda:0'), out_proj_covar=tensor([4.2968e-05, 4.5329e-05, 4.3724e-05, 4.5213e-05, 3.9712e-05, 4.4632e-05, 3.6416e-05, 4.1026e-05], device='cuda:0') 2022-12-07 07:22:35,838 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8471, 1.3537, 2.4584, 2.3116, 2.4832, 2.2866, 1.7939, 2.4575], device='cuda:0'), covar=tensor([0.0624, 0.0950, 0.0099, 0.0207, 0.0120, 0.0094, 0.0284, 0.0082], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0123, 0.0065, 0.0084, 0.0071, 0.0072, 0.0060, 0.0059], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:22:37,455 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:39,921 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:22:49,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.791e+01 3.003e+02 4.031e+02 5.512e+02 8.434e+02, threshold=8.062e+02, percent-clipped=2.0 2022-12-07 07:23:32,854 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:23:35,180 INFO [train.py:873] (0/4) Epoch 2, batch 4200, loss[loss=0.247, simple_loss=0.2263, pruned_loss=0.1339, over 9503.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.2313, pruned_loss=0.124, over 1992819.09 frames. ], batch size: 100, lr: 3.07e-02, grad_scale: 8.0 2022-12-07 07:23:44,033 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2022-12-07 07:24:16,691 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 3.300e+02 4.114e+02 5.198e+02 1.405e+03, threshold=8.228e+02, percent-clipped=5.0 2022-12-07 07:24:22,286 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:24:34,056 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:24:41,898 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 07:24:47,721 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:02,801 INFO [train.py:873] (0/4) Epoch 2, batch 4300, loss[loss=0.2232, simple_loss=0.2216, pruned_loss=0.1124, over 14410.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.2313, pruned_loss=0.1237, over 1980937.04 frames. ], batch size: 73, lr: 3.06e-02, grad_scale: 8.0 2022-12-07 07:25:10,322 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 07:25:15,145 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:15,834 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:24,272 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:29,232 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9902, 0.9222, 0.8750, 1.1455, 1.0866, 0.6987, 1.0948, 0.8464], device='cuda:0'), covar=tensor([0.0530, 0.0303, 0.0162, 0.0328, 0.0440, 0.0395, 0.0226, 0.0294], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0021, 0.0019, 0.0020, 0.0023, 0.0017, 0.0019], device='cuda:0'), out_proj_covar=tensor([4.3604e-05, 4.5388e-05, 4.2836e-05, 4.4105e-05, 4.2218e-05, 4.9334e-05, 3.8517e-05, 4.2611e-05], device='cuda:0') 2022-12-07 07:25:41,728 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:25:44,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 3.023e+02 3.927e+02 4.940e+02 9.062e+02, threshold=7.854e+02, percent-clipped=2.0 2022-12-07 07:25:52,674 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:17,441 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:21,674 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1607, 5.1287, 4.9007, 5.4629, 5.2810, 4.3690, 5.4626, 5.4486], device='cuda:0'), covar=tensor([0.0648, 0.0396, 0.0598, 0.0449, 0.0443, 0.0446, 0.0520, 0.0493], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0072, 0.0094, 0.0089, 0.0098, 0.0066, 0.0086, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:26:28,088 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 07:26:28,227 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 07:26:30,061 INFO [train.py:873] (0/4) Epoch 2, batch 4400, loss[loss=0.2325, simple_loss=0.2342, pruned_loss=0.1155, over 14281.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.2301, pruned_loss=0.1228, over 1987863.47 frames. ], batch size: 44, lr: 3.04e-02, grad_scale: 8.0 2022-12-07 07:26:46,386 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:47,867 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7497, 1.6894, 3.8584, 3.5468, 3.5416, 3.5023, 3.0784, 3.9736], device='cuda:0'), covar=tensor([0.1575, 0.1603, 0.0109, 0.0148, 0.0166, 0.0170, 0.0257, 0.0084], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0126, 0.0067, 0.0088, 0.0075, 0.0077, 0.0062, 0.0064], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:26:48,654 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:26:54,396 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:10,436 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:11,348 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.977e+02 4.090e+02 5.341e+02 9.805e+02, threshold=8.180e+02, percent-clipped=2.0 2022-12-07 07:27:12,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 07:27:41,839 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:50,937 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:27:57,855 INFO [train.py:873] (0/4) Epoch 2, batch 4500, loss[loss=0.2468, simple_loss=0.1984, pruned_loss=0.1476, over 1315.00 frames. ], tot_loss[loss=0.2361, simple_loss=0.2295, pruned_loss=0.1214, over 1991325.33 frames. ], batch size: 100, lr: 3.03e-02, grad_scale: 8.0 2022-12-07 07:28:01,523 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 07:28:08,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:28:38,201 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 3.534e+02 4.689e+02 6.101e+02 2.207e+03, threshold=9.377e+02, percent-clipped=9.0 2022-12-07 07:29:01,429 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:23,549 INFO [train.py:873] (0/4) Epoch 2, batch 4600, loss[loss=0.2211, simple_loss=0.2145, pruned_loss=0.1138, over 5953.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.2308, pruned_loss=0.1231, over 1956144.92 frames. ], batch size: 100, lr: 3.02e-02, grad_scale: 8.0 2022-12-07 07:29:31,599 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:33,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.52 vs. limit=5.0 2022-12-07 07:29:45,286 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:29:52,379 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9429, 3.0416, 4.2524, 4.1664, 3.9009, 2.7647, 3.9567, 3.1394], device='cuda:0'), covar=tensor([0.0065, 0.0174, 0.0129, 0.0108, 0.0065, 0.0385, 0.0030, 0.0284], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0129, 0.0118, 0.0124, 0.0112, 0.0172, 0.0076, 0.0163], device='cuda:0'), out_proj_covar=tensor([1.2900e-04, 1.7179e-04, 1.6367e-04, 1.6579e-04, 1.4957e-04, 2.3437e-04, 9.6677e-05, 2.0277e-04], device='cuda:0') 2022-12-07 07:29:57,598 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:30:04,544 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.710e+02 3.814e+02 4.807e+02 1.336e+03, threshold=7.627e+02, percent-clipped=4.0 2022-12-07 07:30:05,470 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7673, 1.2571, 2.9002, 2.6893, 2.8028, 2.6143, 2.1169, 2.8653], device='cuda:0'), covar=tensor([0.1112, 0.1334, 0.0105, 0.0187, 0.0116, 0.0123, 0.0240, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0125, 0.0065, 0.0085, 0.0074, 0.0076, 0.0062, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:30:21,460 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 07:30:27,134 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:30:33,810 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8943, 1.7499, 2.5329, 1.5637, 1.7858, 1.8503, 1.0783, 2.0446], device='cuda:0'), covar=tensor([0.0737, 0.0795, 0.0358, 0.0932, 0.0623, 0.0856, 0.2111, 0.0508], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0058, 0.0054, 0.0061, 0.0057, 0.0054, 0.0107, 0.0063], device='cuda:0'), out_proj_covar=tensor([9.4441e-05, 9.4147e-05, 8.9450e-05, 9.5038e-05, 8.7712e-05, 9.0248e-05, 1.6248e-04, 1.0212e-04], device='cuda:0') 2022-12-07 07:30:50,578 INFO [train.py:873] (0/4) Epoch 2, batch 4700, loss[loss=0.1778, simple_loss=0.1831, pruned_loss=0.08623, over 13600.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.2281, pruned_loss=0.1205, over 1935005.52 frames. ], batch size: 17, lr: 3.01e-02, grad_scale: 8.0 2022-12-07 07:31:01,590 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:14,400 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:25,451 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:30,461 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.849e+02 3.785e+02 5.259e+02 1.272e+03, threshold=7.569e+02, percent-clipped=8.0 2022-12-07 07:31:55,996 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:31:56,809 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:02,101 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7757, 3.1290, 5.0230, 3.5170, 4.4703, 4.6635, 4.3130, 4.0937], device='cuda:0'), covar=tensor([0.0059, 0.1062, 0.0022, 0.0410, 0.0102, 0.0110, 0.0655, 0.0733], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0293, 0.0117, 0.0222, 0.0152, 0.0178, 0.0173, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 07:32:09,856 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:13,348 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5970, 1.1766, 1.4412, 0.9013, 1.0105, 1.1490, 1.2815, 1.1059], device='cuda:0'), covar=tensor([0.0173, 0.0852, 0.0339, 0.0491, 0.0527, 0.0492, 0.0260, 0.0436], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0144, 0.0072, 0.0095, 0.0064, 0.0061, 0.0058, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:32:16,668 INFO [train.py:873] (0/4) Epoch 2, batch 4800, loss[loss=0.3003, simple_loss=0.2661, pruned_loss=0.1672, over 8603.00 frames. ], tot_loss[loss=0.2343, simple_loss=0.2281, pruned_loss=0.1203, over 1997375.33 frames. ], batch size: 100, lr: 3.00e-02, grad_scale: 8.0 2022-12-07 07:32:51,083 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:32:57,419 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 3.011e+02 4.103e+02 5.225e+02 9.734e+02, threshold=8.206e+02, percent-clipped=3.0 2022-12-07 07:33:15,509 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:33:27,428 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:33:42,230 INFO [train.py:873] (0/4) Epoch 2, batch 4900, loss[loss=0.2418, simple_loss=0.2357, pruned_loss=0.124, over 14181.00 frames. ], tot_loss[loss=0.2351, simple_loss=0.2287, pruned_loss=0.1208, over 1999821.94 frames. ], batch size: 57, lr: 2.99e-02, grad_scale: 8.0 2022-12-07 07:33:49,641 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:15,558 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:19,795 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:21,912 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 3.534e+02 4.435e+02 5.863e+02 9.969e+02, threshold=8.870e+02, percent-clipped=5.0 2022-12-07 07:34:30,725 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:34:44,253 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 07:34:55,805 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2022-12-07 07:34:56,798 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:00,974 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2022-12-07 07:35:08,166 INFO [train.py:873] (0/4) Epoch 2, batch 5000, loss[loss=0.2206, simple_loss=0.2274, pruned_loss=0.1069, over 14317.00 frames. ], tot_loss[loss=0.2341, simple_loss=0.2282, pruned_loss=0.12, over 2010118.57 frames. ], batch size: 28, lr: 2.98e-02, grad_scale: 8.0 2022-12-07 07:35:18,412 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.53 vs. limit=5.0 2022-12-07 07:35:19,714 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:44,176 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:35:46,716 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6338, 2.3931, 1.7576, 1.1935, 2.2152, 2.3346, 2.0675, 2.3027], device='cuda:0'), covar=tensor([0.0218, 0.2993, 0.1240, 0.1562, 0.0700, 0.0373, 0.1439, 0.0451], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0142, 0.0075, 0.0097, 0.0065, 0.0062, 0.0055, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:35:49,096 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 3.073e+02 4.044e+02 5.211e+02 8.552e+02, threshold=8.087e+02, percent-clipped=0.0 2022-12-07 07:36:01,960 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:15,418 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:24,788 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8674, 4.3311, 4.4681, 5.0154, 4.5286, 3.7068, 5.0066, 4.9363], device='cuda:0'), covar=tensor([0.0687, 0.0567, 0.0612, 0.0468, 0.0634, 0.0626, 0.0543, 0.0632], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0075, 0.0095, 0.0090, 0.0102, 0.0066, 0.0089, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:36:25,603 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:36:35,312 INFO [train.py:873] (0/4) Epoch 2, batch 5100, loss[loss=0.2, simple_loss=0.2127, pruned_loss=0.09365, over 14138.00 frames. ], tot_loss[loss=0.2348, simple_loss=0.2284, pruned_loss=0.1206, over 1959690.68 frames. ], batch size: 29, lr: 2.97e-02, grad_scale: 8.0 2022-12-07 07:36:56,657 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:37:14,570 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-07 07:37:16,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 3.113e+02 4.447e+02 5.356e+02 1.001e+03, threshold=8.894e+02, percent-clipped=4.0 2022-12-07 07:37:22,955 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7779, 1.3067, 1.4282, 1.7124, 1.1837, 1.5764, 1.4139, 1.1064], device='cuda:0'), covar=tensor([0.4461, 0.1742, 0.2297, 0.0869, 0.1343, 0.0946, 0.1013, 0.1832], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0062, 0.0050, 0.0047, 0.0062, 0.0045, 0.0056, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:37:29,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2223, 4.5968, 4.6369, 5.2155, 4.8079, 4.5550, 5.1389, 4.4766], device='cuda:0'), covar=tensor([0.0200, 0.0938, 0.0251, 0.0265, 0.0594, 0.0363, 0.0417, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0150, 0.0095, 0.0086, 0.0097, 0.0094, 0.0134, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:37:34,895 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:37:37,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0885, 4.9251, 4.5600, 5.2400, 4.9100, 3.9378, 5.3341, 5.1410], device='cuda:0'), covar=tensor([0.0802, 0.0506, 0.0689, 0.0565, 0.0626, 0.0611, 0.0616, 0.0779], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0077, 0.0098, 0.0093, 0.0104, 0.0069, 0.0092, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:37:47,292 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6893, 1.4908, 1.9942, 1.6924, 2.0565, 1.4760, 1.7105, 2.0181], device='cuda:0'), covar=tensor([0.0607, 0.1143, 0.0138, 0.0821, 0.0089, 0.0638, 0.0806, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0306, 0.0159, 0.0414, 0.0141, 0.0320, 0.0250, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:38:01,733 INFO [train.py:873] (0/4) Epoch 2, batch 5200, loss[loss=0.254, simple_loss=0.2315, pruned_loss=0.1382, over 4979.00 frames. ], tot_loss[loss=0.2349, simple_loss=0.2289, pruned_loss=0.1205, over 1950990.33 frames. ], batch size: 100, lr: 2.96e-02, grad_scale: 8.0 2022-12-07 07:38:01,876 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:16,889 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:27,627 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2514, 1.3560, 1.8127, 0.9831, 1.2670, 1.5212, 0.9800, 1.2767], device='cuda:0'), covar=tensor([0.0867, 0.0643, 0.0229, 0.1367, 0.0624, 0.0480, 0.1601, 0.0640], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0057, 0.0054, 0.0062, 0.0061, 0.0053, 0.0106, 0.0068], device='cuda:0'), out_proj_covar=tensor([9.8413e-05, 9.5492e-05, 9.1161e-05, 1.0158e-04, 9.6365e-05, 9.0742e-05, 1.6644e-04, 1.1312e-04], device='cuda:0') 2022-12-07 07:38:31,537 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 07:38:35,772 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:38:43,164 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 3.055e+02 3.810e+02 5.193e+02 9.032e+02, threshold=7.619e+02, percent-clipped=1.0 2022-12-07 07:38:54,899 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 07:39:05,502 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7593, 4.2449, 4.8178, 3.9684, 4.5593, 4.8556, 1.8647, 4.5035], device='cuda:0'), covar=tensor([0.0123, 0.0222, 0.0267, 0.0380, 0.0237, 0.0079, 0.2763, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0099, 0.0095, 0.0076, 0.0131, 0.0087, 0.0137, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 07:39:28,088 INFO [train.py:873] (0/4) Epoch 2, batch 5300, loss[loss=0.2189, simple_loss=0.2249, pruned_loss=0.1065, over 13994.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.2278, pruned_loss=0.1194, over 1973396.37 frames. ], batch size: 23, lr: 2.95e-02, grad_scale: 8.0 2022-12-07 07:39:31,848 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.10 vs. limit=2.0 2022-12-07 07:40:09,055 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.807e+01 2.959e+02 3.875e+02 4.793e+02 1.192e+03, threshold=7.749e+02, percent-clipped=2.0 2022-12-07 07:40:53,785 INFO [train.py:873] (0/4) Epoch 2, batch 5400, loss[loss=0.2584, simple_loss=0.2422, pruned_loss=0.1373, over 9478.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.228, pruned_loss=0.1196, over 1981770.13 frames. ], batch size: 100, lr: 2.94e-02, grad_scale: 8.0 2022-12-07 07:41:03,548 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8474, 1.3385, 2.4727, 2.1746, 2.4810, 2.3612, 1.5345, 2.5104], device='cuda:0'), covar=tensor([0.0560, 0.0847, 0.0084, 0.0234, 0.0119, 0.0109, 0.0358, 0.0091], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0130, 0.0069, 0.0093, 0.0078, 0.0081, 0.0066, 0.0067], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 07:41:35,176 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 3.363e+02 4.433e+02 5.944e+02 1.364e+03, threshold=8.866e+02, percent-clipped=4.0 2022-12-07 07:42:03,934 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2022-12-07 07:42:21,077 INFO [train.py:873] (0/4) Epoch 2, batch 5500, loss[loss=0.2542, simple_loss=0.2404, pruned_loss=0.134, over 14249.00 frames. ], tot_loss[loss=0.2336, simple_loss=0.2279, pruned_loss=0.1197, over 1972162.26 frames. ], batch size: 69, lr: 2.93e-02, grad_scale: 8.0 2022-12-07 07:42:29,050 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 07:42:54,587 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:42:57,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2022-12-07 07:43:01,929 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 3.268e+02 4.131e+02 4.754e+02 9.036e+02, threshold=8.263e+02, percent-clipped=2.0 2022-12-07 07:43:08,797 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:43:35,172 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:43:39,854 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.18 vs. limit=5.0 2022-12-07 07:43:42,886 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0628, 2.9483, 2.8121, 3.2355, 2.7797, 2.3917, 3.0988, 3.2013], device='cuda:0'), covar=tensor([0.0755, 0.0648, 0.0762, 0.0596, 0.0755, 0.0839, 0.0668, 0.0616], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0075, 0.0094, 0.0089, 0.0101, 0.0066, 0.0090, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:43:46,159 INFO [train.py:873] (0/4) Epoch 2, batch 5600, loss[loss=0.215, simple_loss=0.2218, pruned_loss=0.1041, over 14532.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.2274, pruned_loss=0.1201, over 1909400.27 frames. ], batch size: 43, lr: 2.92e-02, grad_scale: 8.0 2022-12-07 07:44:28,273 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 3.284e+02 4.567e+02 6.071e+02 1.505e+03, threshold=9.133e+02, percent-clipped=13.0 2022-12-07 07:44:36,879 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 07:45:14,446 INFO [train.py:873] (0/4) Epoch 2, batch 5700, loss[loss=0.2398, simple_loss=0.2298, pruned_loss=0.125, over 11977.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.2277, pruned_loss=0.1206, over 1944296.42 frames. ], batch size: 100, lr: 2.91e-02, grad_scale: 8.0 2022-12-07 07:45:55,408 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.840e+02 3.959e+02 4.964e+02 1.183e+03, threshold=7.919e+02, percent-clipped=6.0 2022-12-07 07:46:40,287 INFO [train.py:873] (0/4) Epoch 2, batch 5800, loss[loss=0.239, simple_loss=0.2323, pruned_loss=0.1228, over 14266.00 frames. ], tot_loss[loss=0.2348, simple_loss=0.2282, pruned_loss=0.1207, over 1929818.21 frames. ], batch size: 39, lr: 2.90e-02, grad_scale: 8.0 2022-12-07 07:46:47,387 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5500, 1.2249, 1.2611, 0.9937, 1.1718, 1.2868, 1.3124, 1.2239], device='cuda:0'), covar=tensor([0.0317, 0.1409, 0.0485, 0.1120, 0.0610, 0.0436, 0.0344, 0.0556], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0156, 0.0077, 0.0104, 0.0072, 0.0065, 0.0061, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:46:56,610 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3204, 2.0603, 1.8487, 1.3421, 2.1169, 2.0788, 2.1914, 1.8389], device='cuda:0'), covar=tensor([0.0562, 0.2859, 0.0706, 0.1544, 0.0773, 0.0375, 0.0727, 0.0729], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0158, 0.0078, 0.0104, 0.0072, 0.0066, 0.0062, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:47:21,675 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 3.070e+02 4.104e+02 5.303e+02 9.108e+02, threshold=8.208e+02, percent-clipped=3.0 2022-12-07 07:47:29,063 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:48:07,860 INFO [train.py:873] (0/4) Epoch 2, batch 5900, loss[loss=0.2557, simple_loss=0.2477, pruned_loss=0.1319, over 14412.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.2276, pruned_loss=0.1204, over 1922813.14 frames. ], batch size: 53, lr: 2.89e-02, grad_scale: 8.0 2022-12-07 07:48:11,321 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:48:23,269 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9237, 3.5018, 3.4891, 3.8880, 3.7562, 3.5813, 3.8437, 3.3850], device='cuda:0'), covar=tensor([0.0311, 0.0834, 0.0312, 0.0376, 0.0587, 0.0698, 0.0587, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0149, 0.0096, 0.0091, 0.0096, 0.0092, 0.0142, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:48:49,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 3.527e+02 4.364e+02 5.819e+02 1.292e+03, threshold=8.728e+02, percent-clipped=7.0 2022-12-07 07:49:25,557 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1273, 3.8186, 3.6833, 3.7155, 3.9215, 3.8801, 4.1428, 4.0526], device='cuda:0'), covar=tensor([0.0647, 0.0678, 0.1028, 0.1370, 0.0460, 0.0403, 0.0571, 0.0705], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0148, 0.0186, 0.0212, 0.0154, 0.0170, 0.0181, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 07:49:28,990 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7882, 2.1894, 3.0797, 2.5539, 3.0125, 2.5975, 2.8607, 2.3907], device='cuda:0'), covar=tensor([0.0103, 0.0693, 0.0059, 0.0335, 0.0110, 0.0150, 0.0192, 0.0690], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0313, 0.0131, 0.0238, 0.0170, 0.0194, 0.0186, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 07:49:33,857 INFO [train.py:873] (0/4) Epoch 2, batch 6000, loss[loss=0.2458, simple_loss=0.2435, pruned_loss=0.124, over 14585.00 frames. ], tot_loss[loss=0.233, simple_loss=0.2269, pruned_loss=0.1195, over 1935940.05 frames. ], batch size: 22, lr: 2.88e-02, grad_scale: 8.0 2022-12-07 07:49:33,858 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 07:49:42,177 INFO [train.py:905] (0/4) Epoch 2, validation: loss=0.139, simple_loss=0.1792, pruned_loss=0.04941, over 857387.00 frames. 2022-12-07 07:49:42,178 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 07:50:24,256 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 3.403e+02 4.162e+02 5.697e+02 1.072e+03, threshold=8.324e+02, percent-clipped=3.0 2022-12-07 07:50:39,735 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4935, 3.9916, 4.1809, 4.6284, 4.2549, 3.4278, 4.6303, 4.5023], device='cuda:0'), covar=tensor([0.0618, 0.0509, 0.0482, 0.0515, 0.0458, 0.0606, 0.0459, 0.0558], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0073, 0.0094, 0.0088, 0.0100, 0.0063, 0.0088, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:50:48,887 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3799, 1.6274, 1.7101, 1.7220, 1.4667, 1.7371, 1.7904, 1.0387], device='cuda:0'), covar=tensor([0.3505, 0.2021, 0.1953, 0.1318, 0.0996, 0.1220, 0.1223, 0.2199], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0063, 0.0049, 0.0053, 0.0064, 0.0049, 0.0064, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:50:58,709 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9777, 1.6376, 3.1903, 2.2270, 3.1241, 1.8429, 2.2697, 3.1921], device='cuda:0'), covar=tensor([0.0448, 0.4020, 0.0296, 0.4974, 0.0153, 0.3007, 0.0853, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0320, 0.0163, 0.0412, 0.0147, 0.0327, 0.0268, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0001, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 07:51:09,281 INFO [train.py:873] (0/4) Epoch 2, batch 6100, loss[loss=0.2098, simple_loss=0.2142, pruned_loss=0.1027, over 14027.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.2271, pruned_loss=0.1189, over 2013338.57 frames. ], batch size: 29, lr: 2.87e-02, grad_scale: 8.0 2022-12-07 07:51:22,286 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0552, 1.7298, 2.2461, 2.4205, 1.9967, 1.7525, 2.2053, 1.9305], device='cuda:0'), covar=tensor([0.0049, 0.0083, 0.0058, 0.0037, 0.0048, 0.0121, 0.0035, 0.0073], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0140, 0.0140, 0.0137, 0.0120, 0.0184, 0.0082, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:51:25,168 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 07:51:50,708 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 3.186e+02 4.052e+02 4.953e+02 1.165e+03, threshold=8.105e+02, percent-clipped=5.0 2022-12-07 07:51:55,075 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 07:52:13,614 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:52:35,524 INFO [train.py:873] (0/4) Epoch 2, batch 6200, loss[loss=0.2409, simple_loss=0.1977, pruned_loss=0.1421, over 1200.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.2267, pruned_loss=0.1191, over 1938302.72 frames. ], batch size: 100, lr: 2.86e-02, grad_scale: 8.0 2022-12-07 07:52:49,412 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.41 vs. limit=5.0 2022-12-07 07:53:05,514 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 07:53:17,360 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.778e+02 3.775e+02 5.188e+02 1.088e+03, threshold=7.550e+02, percent-clipped=5.0 2022-12-07 07:53:17,892 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 07:54:02,988 INFO [train.py:873] (0/4) Epoch 2, batch 6300, loss[loss=0.2519, simple_loss=0.2146, pruned_loss=0.1446, over 3884.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.2254, pruned_loss=0.1178, over 1910021.08 frames. ], batch size: 100, lr: 2.86e-02, grad_scale: 8.0 2022-12-07 07:54:04,437 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 07:54:23,488 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0439, 4.7150, 5.1674, 4.1450, 4.7936, 5.1284, 2.3584, 4.6818], device='cuda:0'), covar=tensor([0.0096, 0.0179, 0.0206, 0.0228, 0.0206, 0.0234, 0.2373, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0105, 0.0099, 0.0080, 0.0135, 0.0091, 0.0141, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 07:54:31,412 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8373, 1.5717, 3.5868, 1.7411, 3.6361, 3.6102, 2.6945, 4.0456], device='cuda:0'), covar=tensor([0.0159, 0.2544, 0.0301, 0.2086, 0.0267, 0.0296, 0.0551, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0137, 0.0089, 0.0145, 0.0109, 0.0095, 0.0083, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:54:33,921 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2660, 1.5725, 2.4991, 1.4371, 2.3634, 2.4069, 1.7676, 2.4921], device='cuda:0'), covar=tensor([0.0170, 0.1427, 0.0192, 0.1384, 0.0216, 0.0244, 0.0529, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0137, 0.0089, 0.0146, 0.0109, 0.0095, 0.0083, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:54:43,892 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.964e+01 3.256e+02 4.500e+02 5.561e+02 1.186e+03, threshold=8.999e+02, percent-clipped=4.0 2022-12-07 07:55:29,725 INFO [train.py:873] (0/4) Epoch 2, batch 6400, loss[loss=0.2146, simple_loss=0.2216, pruned_loss=0.1038, over 14297.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.2251, pruned_loss=0.1176, over 1944811.60 frames. ], batch size: 63, lr: 2.85e-02, grad_scale: 8.0 2022-12-07 07:56:01,313 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2022-12-07 07:56:11,802 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 3.193e+02 4.162e+02 5.343e+02 1.659e+03, threshold=8.325e+02, percent-clipped=5.0 2022-12-07 07:56:15,291 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0404, 2.8684, 3.0249, 2.8731, 2.8056, 2.7850, 1.2422, 2.7423], device='cuda:0'), covar=tensor([0.0185, 0.0289, 0.0318, 0.0255, 0.0339, 0.0440, 0.2436, 0.0283], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0103, 0.0099, 0.0079, 0.0133, 0.0088, 0.0138, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 07:56:46,644 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2743, 2.3239, 3.6649, 3.5941, 3.3260, 2.2534, 3.3108, 2.7253], device='cuda:0'), covar=tensor([0.0059, 0.0154, 0.0122, 0.0078, 0.0051, 0.0315, 0.0025, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0142, 0.0144, 0.0134, 0.0122, 0.0184, 0.0085, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:0') 2022-12-07 07:56:57,034 INFO [train.py:873] (0/4) Epoch 2, batch 6500, loss[loss=0.2134, simple_loss=0.1806, pruned_loss=0.1231, over 2670.00 frames. ], tot_loss[loss=0.231, simple_loss=0.2261, pruned_loss=0.118, over 1983912.81 frames. ], batch size: 100, lr: 2.84e-02, grad_scale: 8.0 2022-12-07 07:57:03,781 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1218, 1.3802, 3.0442, 1.5258, 3.0475, 2.9583, 2.4843, 3.1988], device='cuda:0'), covar=tensor([0.0257, 0.2799, 0.0373, 0.2409, 0.0446, 0.0449, 0.0743, 0.0302], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0139, 0.0092, 0.0150, 0.0112, 0.0098, 0.0087, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:57:04,732 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:57:10,476 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2022-12-07 07:57:23,629 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:57:23,709 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7071, 2.5830, 1.9838, 1.5602, 2.2433, 2.1553, 2.6870, 2.2184], device='cuda:0'), covar=tensor([0.0286, 0.1836, 0.0605, 0.1506, 0.1216, 0.0325, 0.0535, 0.0557], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0161, 0.0079, 0.0107, 0.0070, 0.0068, 0.0059, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:57:31,983 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:57:38,491 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 3.036e+02 3.911e+02 5.054e+02 1.167e+03, threshold=7.822e+02, percent-clipped=3.0 2022-12-07 07:57:46,726 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:57:58,287 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:58:15,458 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 07:58:23,887 INFO [train.py:873] (0/4) Epoch 2, batch 6600, loss[loss=0.2474, simple_loss=0.2332, pruned_loss=0.1307, over 11947.00 frames. ], tot_loss[loss=0.23, simple_loss=0.2251, pruned_loss=0.1175, over 1982367.61 frames. ], batch size: 100, lr: 2.83e-02, grad_scale: 8.0 2022-12-07 07:58:24,898 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 07:58:39,620 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 07:58:52,442 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6299, 1.0000, 1.0244, 0.9316, 0.8067, 1.0987, 0.9027, 0.7278], device='cuda:0'), covar=tensor([0.2046, 0.0888, 0.0459, 0.0401, 0.0643, 0.0481, 0.1630, 0.0961], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0060, 0.0048, 0.0054, 0.0061, 0.0046, 0.0064, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 07:59:06,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.900e+02 3.866e+02 4.857e+02 9.306e+02, threshold=7.732e+02, percent-clipped=4.0 2022-12-07 07:59:20,770 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 07:59:27,058 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.44 vs. limit=2.0 2022-12-07 07:59:51,415 INFO [train.py:873] (0/4) Epoch 2, batch 6700, loss[loss=0.2275, simple_loss=0.2243, pruned_loss=0.1154, over 14183.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.2271, pruned_loss=0.119, over 2040880.58 frames. ], batch size: 99, lr: 2.82e-02, grad_scale: 8.0 2022-12-07 08:00:32,373 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 3.050e+02 4.021e+02 5.480e+02 1.335e+03, threshold=8.042e+02, percent-clipped=5.0 2022-12-07 08:00:45,417 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 08:01:18,007 INFO [train.py:873] (0/4) Epoch 2, batch 6800, loss[loss=0.2281, simple_loss=0.2087, pruned_loss=0.1238, over 3927.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.226, pruned_loss=0.1176, over 2034482.11 frames. ], batch size: 100, lr: 2.81e-02, grad_scale: 8.0 2022-12-07 08:01:33,488 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7266, 4.2509, 4.2314, 4.7360, 4.5343, 4.1153, 4.7291, 4.2112], device='cuda:0'), covar=tensor([0.0290, 0.0700, 0.0267, 0.0314, 0.0533, 0.0456, 0.0406, 0.0431], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0155, 0.0100, 0.0096, 0.0101, 0.0095, 0.0147, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:01:41,962 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7029, 1.7148, 4.0678, 1.8781, 4.2384, 4.2752, 3.9114, 4.8710], device='cuda:0'), covar=tensor([0.0129, 0.2584, 0.0376, 0.2199, 0.0225, 0.0257, 0.0339, 0.0145], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0139, 0.0094, 0.0149, 0.0111, 0.0098, 0.0089, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:01:43,706 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:01:50,120 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9707, 1.5818, 1.6211, 1.3982, 1.4118, 1.7611, 1.3181, 0.9024], device='cuda:0'), covar=tensor([0.3361, 0.1646, 0.1842, 0.1604, 0.1021, 0.0584, 0.1365, 0.1937], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0064, 0.0051, 0.0055, 0.0063, 0.0048, 0.0065, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:01:59,490 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 3.284e+02 4.234e+02 5.689e+02 1.698e+03, threshold=8.469e+02, percent-clipped=9.0 2022-12-07 08:02:01,454 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:14,254 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:25,183 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:26,471 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6453, 2.9326, 1.9407, 1.7079, 2.7640, 2.6432, 2.5834, 2.0121], device='cuda:0'), covar=tensor([0.0365, 0.3433, 0.0929, 0.1895, 0.0585, 0.0370, 0.1776, 0.1106], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0179, 0.0086, 0.0118, 0.0074, 0.0072, 0.0067, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:02:41,188 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:44,541 INFO [train.py:873] (0/4) Epoch 2, batch 6900, loss[loss=0.223, simple_loss=0.2252, pruned_loss=0.1104, over 14281.00 frames. ], tot_loss[loss=0.2311, simple_loss=0.2257, pruned_loss=0.1183, over 1964128.47 frames. ], batch size: 60, lr: 2.80e-02, grad_scale: 8.0 2022-12-07 08:02:54,003 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:02:55,549 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:03:05,482 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:03:12,254 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6848, 2.4525, 2.6661, 2.5594, 2.5223, 2.4713, 1.3378, 2.4675], device='cuda:0'), covar=tensor([0.0246, 0.0424, 0.0395, 0.0287, 0.0312, 0.0543, 0.2329, 0.0301], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0104, 0.0099, 0.0081, 0.0134, 0.0092, 0.0141, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:03:24,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0989, 1.4586, 2.5750, 1.9479, 2.3710, 1.6830, 1.8962, 2.2305], device='cuda:0'), covar=tensor([0.0505, 0.2456, 0.0147, 0.2393, 0.0196, 0.1927, 0.0756, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0317, 0.0163, 0.0424, 0.0157, 0.0329, 0.0276, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:03:25,507 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 3.007e+02 4.167e+02 5.148e+02 1.275e+03, threshold=8.334e+02, percent-clipped=3.0 2022-12-07 08:03:38,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 08:03:57,903 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:04:11,092 INFO [train.py:873] (0/4) Epoch 2, batch 7000, loss[loss=0.2634, simple_loss=0.2377, pruned_loss=0.1446, over 4954.00 frames. ], tot_loss[loss=0.2313, simple_loss=0.226, pruned_loss=0.1182, over 2020775.73 frames. ], batch size: 100, lr: 2.79e-02, grad_scale: 8.0 2022-12-07 08:04:53,422 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 3.149e+02 3.940e+02 5.259e+02 1.451e+03, threshold=7.881e+02, percent-clipped=6.0 2022-12-07 08:04:59,594 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 08:05:37,945 INFO [train.py:873] (0/4) Epoch 2, batch 7100, loss[loss=0.2079, simple_loss=0.18, pruned_loss=0.1179, over 1306.00 frames. ], tot_loss[loss=0.2287, simple_loss=0.2247, pruned_loss=0.1163, over 1998491.37 frames. ], batch size: 100, lr: 2.79e-02, grad_scale: 16.0 2022-12-07 08:06:15,969 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3769, 2.4153, 3.4043, 2.8033, 3.2185, 3.1712, 3.1961, 2.7257], device='cuda:0'), covar=tensor([0.0080, 0.0867, 0.0060, 0.0434, 0.0143, 0.0154, 0.0363, 0.0629], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0320, 0.0142, 0.0254, 0.0196, 0.0203, 0.0218, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:06:19,126 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.591e+01 3.151e+02 3.961e+02 5.034e+02 1.065e+03, threshold=7.923e+02, percent-clipped=3.0 2022-12-07 08:06:34,368 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:06:44,618 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:06:50,599 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0702, 1.9993, 1.8404, 2.1244, 1.7974, 1.8155, 2.0677, 2.1259], device='cuda:0'), covar=tensor([0.0755, 0.0808, 0.0999, 0.0619, 0.1038, 0.0663, 0.0749, 0.0577], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0076, 0.0094, 0.0090, 0.0099, 0.0065, 0.0093, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:07:00,718 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:04,456 INFO [train.py:873] (0/4) Epoch 2, batch 7200, loss[loss=0.2111, simple_loss=0.2175, pruned_loss=0.1023, over 14293.00 frames. ], tot_loss[loss=0.2307, simple_loss=0.2257, pruned_loss=0.1179, over 2001170.42 frames. ], batch size: 46, lr: 2.78e-02, grad_scale: 16.0 2022-12-07 08:07:09,577 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:15,476 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:15,574 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:07:37,008 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:42,108 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:07:45,743 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.988e+02 3.907e+02 5.109e+02 1.521e+03, threshold=7.813e+02, percent-clipped=5.0 2022-12-07 08:07:56,616 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:08:10,139 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0323, 3.7530, 3.6054, 3.5227, 3.8130, 3.8702, 3.9726, 3.9534], device='cuda:0'), covar=tensor([0.0586, 0.0499, 0.1003, 0.1582, 0.0441, 0.0442, 0.0659, 0.0702], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0152, 0.0201, 0.0239, 0.0167, 0.0184, 0.0195, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:08:12,980 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:08:22,512 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2303, 3.8827, 3.7672, 3.7701, 4.0036, 4.0361, 4.2459, 4.1648], device='cuda:0'), covar=tensor([0.0533, 0.0508, 0.1121, 0.1512, 0.0406, 0.0448, 0.0519, 0.0712], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0151, 0.0203, 0.0240, 0.0167, 0.0185, 0.0194, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:08:28,207 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9869, 1.9807, 1.9280, 2.0132, 1.9209, 1.8691, 0.9437, 1.7402], device='cuda:0'), covar=tensor([0.0314, 0.0371, 0.0527, 0.0249, 0.0398, 0.0507, 0.1986, 0.0501], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0100, 0.0096, 0.0078, 0.0132, 0.0088, 0.0141, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:08:29,762 INFO [train.py:873] (0/4) Epoch 2, batch 7300, loss[loss=0.1971, simple_loss=0.1706, pruned_loss=0.1118, over 1246.00 frames. ], tot_loss[loss=0.2276, simple_loss=0.2234, pruned_loss=0.1159, over 1989607.82 frames. ], batch size: 100, lr: 2.77e-02, grad_scale: 16.0 2022-12-07 08:08:55,000 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9294, 1.7611, 1.5210, 2.0858, 1.9884, 2.0769, 1.7970, 1.7903], device='cuda:0'), covar=tensor([0.0115, 0.0296, 0.0475, 0.0061, 0.0168, 0.0061, 0.0206, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0249, 0.0342, 0.0157, 0.0173, 0.0175, 0.0228, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0001, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:09:10,699 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.806e+02 3.909e+02 5.179e+02 1.856e+03, threshold=7.818e+02, percent-clipped=6.0 2022-12-07 08:09:21,965 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.13 vs. limit=2.0 2022-12-07 08:09:28,686 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.08 vs. limit=5.0 2022-12-07 08:09:55,278 INFO [train.py:873] (0/4) Epoch 2, batch 7400, loss[loss=0.218, simple_loss=0.2242, pruned_loss=0.1059, over 14224.00 frames. ], tot_loss[loss=0.227, simple_loss=0.2231, pruned_loss=0.1154, over 1969176.05 frames. ], batch size: 35, lr: 2.76e-02, grad_scale: 16.0 2022-12-07 08:10:15,619 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 08:10:26,831 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-15000.pt 2022-12-07 08:10:39,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.247e+01 3.005e+02 3.840e+02 4.787e+02 1.067e+03, threshold=7.681e+02, percent-clipped=4.0 2022-12-07 08:10:50,401 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0851, 1.6026, 1.5573, 1.6171, 1.2909, 1.6516, 1.6513, 0.8960], device='cuda:0'), covar=tensor([0.4537, 0.1468, 0.2928, 0.1462, 0.1207, 0.0831, 0.1074, 0.2568], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0062, 0.0050, 0.0055, 0.0062, 0.0050, 0.0062, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:11:24,370 INFO [train.py:873] (0/4) Epoch 2, batch 7500, loss[loss=0.2017, simple_loss=0.1954, pruned_loss=0.1041, over 4948.00 frames. ], tot_loss[loss=0.226, simple_loss=0.2224, pruned_loss=0.1148, over 1890193.93 frames. ], batch size: 100, lr: 2.75e-02, grad_scale: 8.0 2022-12-07 08:11:29,586 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:11:51,183 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4478, 3.0106, 4.3624, 3.2170, 4.2015, 3.5382, 3.8042, 3.6136], device='cuda:0'), covar=tensor([0.0094, 0.0895, 0.0031, 0.0411, 0.0140, 0.0281, 0.0678, 0.0638], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0326, 0.0145, 0.0260, 0.0199, 0.0217, 0.0221, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:11:52,692 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:12:03,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.166e+01 2.953e+02 4.080e+02 5.440e+02 1.285e+03, threshold=8.159e+02, percent-clipped=6.0 2022-12-07 08:12:05,496 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:12:05,543 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6774, 1.0180, 2.0241, 1.9096, 2.0544, 1.9627, 1.3639, 2.0397], device='cuda:0'), covar=tensor([0.0341, 0.0794, 0.0099, 0.0206, 0.0104, 0.0124, 0.0305, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0136, 0.0071, 0.0099, 0.0083, 0.0088, 0.0068, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:12:06,647 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.97 vs. limit=5.0 2022-12-07 08:12:10,428 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-2.pt 2022-12-07 08:12:48,568 INFO [train.py:873] (0/4) Epoch 3, batch 0, loss[loss=0.2724, simple_loss=0.2392, pruned_loss=0.1528, over 5024.00 frames. ], tot_loss[loss=0.2724, simple_loss=0.2392, pruned_loss=0.1528, over 5024.00 frames. ], batch size: 100, lr: 2.61e-02, grad_scale: 8.0 2022-12-07 08:12:48,569 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 08:12:55,593 INFO [train.py:905] (0/4) Epoch 3, validation: loss=0.159, simple_loss=0.1997, pruned_loss=0.05911, over 857387.00 frames. 2022-12-07 08:12:55,594 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 08:13:12,537 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:13:47,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-07 08:13:53,451 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:13:57,123 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:14:02,117 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 08:14:11,335 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.526e+01 3.048e+02 3.683e+02 5.299e+02 1.167e+03, threshold=7.365e+02, percent-clipped=4.0 2022-12-07 08:14:23,257 INFO [train.py:873] (0/4) Epoch 3, batch 100, loss[loss=0.2211, simple_loss=0.2193, pruned_loss=0.1114, over 13528.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.226, pruned_loss=0.1156, over 897059.30 frames. ], batch size: 100, lr: 2.60e-02, grad_scale: 8.0 2022-12-07 08:14:49,283 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:15:01,261 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5480, 4.1180, 4.0385, 4.5981, 4.4566, 4.0033, 4.5909, 3.8663], device='cuda:0'), covar=tensor([0.0312, 0.0913, 0.0290, 0.0372, 0.0547, 0.0486, 0.0428, 0.0509], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0163, 0.0102, 0.0098, 0.0101, 0.0094, 0.0144, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:15:34,757 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2868, 2.3154, 1.8511, 2.5027, 2.2493, 2.4657, 2.1649, 1.9730], device='cuda:0'), covar=tensor([0.0148, 0.0177, 0.0827, 0.0090, 0.0122, 0.0074, 0.0270, 0.0607], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0257, 0.0354, 0.0159, 0.0181, 0.0177, 0.0241, 0.0358], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0001, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:15:38,765 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 3.082e+02 4.015e+02 5.606e+02 9.270e+02, threshold=8.031e+02, percent-clipped=6.0 2022-12-07 08:15:49,796 INFO [train.py:873] (0/4) Epoch 3, batch 200, loss[loss=0.2029, simple_loss=0.1731, pruned_loss=0.1164, over 2636.00 frames. ], tot_loss[loss=0.2276, simple_loss=0.224, pruned_loss=0.1156, over 1319348.75 frames. ], batch size: 100, lr: 2.59e-02, grad_scale: 8.0 2022-12-07 08:16:14,980 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6890, 1.4856, 1.2355, 1.5342, 1.3960, 1.7181, 1.3910, 1.8528], device='cuda:0'), covar=tensor([0.1873, 0.1883, 0.1152, 0.0519, 0.0766, 0.0311, 0.1108, 0.0121], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0012, 0.0012, 0.0012, 0.0015, 0.0013, 0.0015], device='cuda:0'), out_proj_covar=tensor([2.8975e-05, 2.9200e-05, 3.2643e-05, 2.8626e-05, 2.8550e-05, 3.3806e-05, 3.9252e-05, 3.4794e-05], device='cuda:0') 2022-12-07 08:16:17,589 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2530, 1.7830, 1.8280, 1.6620, 1.2383, 1.6328, 1.5927, 0.9682], device='cuda:0'), covar=tensor([0.2813, 0.1247, 0.1415, 0.1249, 0.1165, 0.0696, 0.1006, 0.2341], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0061, 0.0052, 0.0057, 0.0064, 0.0050, 0.0063, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:16:31,893 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4582, 2.4417, 3.4787, 2.5759, 3.3681, 3.0149, 3.2445, 2.6966], device='cuda:0'), covar=tensor([0.0082, 0.0818, 0.0055, 0.0521, 0.0136, 0.0130, 0.0401, 0.0912], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0323, 0.0151, 0.0259, 0.0206, 0.0211, 0.0229, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:16:50,885 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:17:04,620 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.006e+02 3.006e+02 3.776e+02 5.007e+02 1.251e+03, threshold=7.551e+02, percent-clipped=5.0 2022-12-07 08:17:14,494 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.06 vs. limit=5.0 2022-12-07 08:17:16,242 INFO [train.py:873] (0/4) Epoch 3, batch 300, loss[loss=0.2452, simple_loss=0.2402, pruned_loss=0.1251, over 14553.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.2219, pruned_loss=0.1134, over 1641975.57 frames. ], batch size: 43, lr: 2.59e-02, grad_scale: 8.0 2022-12-07 08:17:32,187 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:17:39,006 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5820, 2.4800, 4.7304, 3.1045, 4.4950, 1.9546, 3.6700, 4.5117], device='cuda:0'), covar=tensor([0.0188, 0.3422, 0.0135, 0.5972, 0.0085, 0.3012, 0.0679, 0.0073], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0331, 0.0173, 0.0430, 0.0156, 0.0335, 0.0289, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:18:31,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.693e+02 3.512e+02 4.941e+02 9.484e+02, threshold=7.024e+02, percent-clipped=2.0 2022-12-07 08:18:35,319 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1730, 2.8070, 2.8606, 3.1868, 3.1465, 3.1955, 3.2262, 2.6709], device='cuda:0'), covar=tensor([0.0369, 0.1194, 0.0422, 0.0424, 0.0596, 0.0353, 0.0490, 0.0689], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0168, 0.0103, 0.0101, 0.0103, 0.0099, 0.0150, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:18:40,096 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 08:18:41,945 INFO [train.py:873] (0/4) Epoch 3, batch 400, loss[loss=0.199, simple_loss=0.182, pruned_loss=0.108, over 2533.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2206, pruned_loss=0.1123, over 1737321.21 frames. ], batch size: 100, lr: 2.58e-02, grad_scale: 8.0 2022-12-07 08:18:52,165 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-07 08:19:04,004 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:19:07,576 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5373, 2.8776, 3.4773, 1.8807, 2.4627, 2.5842, 1.2579, 2.4296], device='cuda:0'), covar=tensor([0.1098, 0.0335, 0.0699, 0.1780, 0.0829, 0.0935, 0.3076, 0.1163], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0063, 0.0057, 0.0069, 0.0074, 0.0059, 0.0119, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:19:56,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 3.036e+02 3.993e+02 5.394e+02 1.334e+03, threshold=7.985e+02, percent-clipped=7.0 2022-12-07 08:20:07,944 INFO [train.py:873] (0/4) Epoch 3, batch 500, loss[loss=0.2333, simple_loss=0.217, pruned_loss=0.1248, over 3899.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.2194, pruned_loss=0.1112, over 1823096.24 frames. ], batch size: 100, lr: 2.57e-02, grad_scale: 8.0 2022-12-07 08:21:21,549 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8434, 4.3843, 5.1094, 4.5544, 4.6578, 5.0664, 2.0059, 4.5763], device='cuda:0'), covar=tensor([0.0144, 0.0249, 0.0233, 0.0184, 0.0254, 0.0095, 0.2785, 0.0176], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0107, 0.0100, 0.0083, 0.0138, 0.0096, 0.0142, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:21:22,225 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.935e+02 3.905e+02 4.916e+02 1.140e+03, threshold=7.811e+02, percent-clipped=3.0 2022-12-07 08:21:32,702 INFO [train.py:873] (0/4) Epoch 3, batch 600, loss[loss=0.248, simple_loss=0.2353, pruned_loss=0.1304, over 4987.00 frames. ], tot_loss[loss=0.223, simple_loss=0.2207, pruned_loss=0.1126, over 1873483.58 frames. ], batch size: 100, lr: 2.56e-02, grad_scale: 8.0 2022-12-07 08:21:44,172 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=4.22 vs. limit=2.0 2022-12-07 08:22:18,770 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3308, 3.1045, 3.3915, 2.9905, 3.1362, 2.9174, 1.2936, 3.1237], device='cuda:0'), covar=tensor([0.0189, 0.0306, 0.0347, 0.0353, 0.0347, 0.0577, 0.2979, 0.0239], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0105, 0.0100, 0.0083, 0.0138, 0.0095, 0.0142, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:22:33,866 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4223, 2.9922, 4.4689, 3.1102, 3.9414, 4.0257, 3.8349, 3.7296], device='cuda:0'), covar=tensor([0.0083, 0.1112, 0.0051, 0.0498, 0.0216, 0.0188, 0.0636, 0.0723], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0332, 0.0156, 0.0269, 0.0212, 0.0221, 0.0232, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:22:39,215 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5500, 1.4589, 1.9164, 2.0557, 1.3445, 1.5235, 1.8164, 1.6978], device='cuda:0'), covar=tensor([0.0029, 0.0036, 0.0025, 0.0015, 0.0045, 0.0065, 0.0018, 0.0025], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0146, 0.0164, 0.0150, 0.0127, 0.0191, 0.0092, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:0') 2022-12-07 08:22:47,078 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 3.132e+02 3.850e+02 5.057e+02 1.369e+03, threshold=7.700e+02, percent-clipped=5.0 2022-12-07 08:22:58,910 INFO [train.py:873] (0/4) Epoch 3, batch 700, loss[loss=0.1978, simple_loss=0.2066, pruned_loss=0.09448, over 14266.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.2204, pruned_loss=0.1129, over 1828726.87 frames. ], batch size: 80, lr: 2.56e-02, grad_scale: 8.0 2022-12-07 08:23:05,790 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6170, 4.2722, 4.9677, 3.6696, 4.4937, 4.9651, 1.8672, 4.5437], device='cuda:0'), covar=tensor([0.0148, 0.0264, 0.0272, 0.0426, 0.0286, 0.0085, 0.2949, 0.0175], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0106, 0.0100, 0.0083, 0.0138, 0.0094, 0.0142, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:23:16,715 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-07 08:23:20,522 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:23:30,347 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.42 vs. limit=5.0 2022-12-07 08:23:33,294 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-07 08:23:37,139 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8190, 2.7416, 2.8769, 2.7208, 2.7189, 2.6777, 1.2450, 2.6377], device='cuda:0'), covar=tensor([0.0222, 0.0324, 0.0381, 0.0341, 0.0281, 0.0410, 0.2438, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0106, 0.0099, 0.0083, 0.0138, 0.0095, 0.0142, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:23:47,992 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:23:50,388 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3216, 1.4850, 2.4563, 1.3625, 2.4184, 2.4576, 1.6447, 2.4707], device='cuda:0'), covar=tensor([0.0230, 0.1578, 0.0212, 0.1472, 0.0232, 0.0276, 0.0743, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0140, 0.0097, 0.0149, 0.0116, 0.0103, 0.0092, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:24:01,232 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:24:09,662 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9970, 1.9368, 1.7912, 2.1123, 1.5864, 1.8540, 2.0069, 2.1185], device='cuda:0'), covar=tensor([0.0985, 0.0858, 0.1122, 0.0739, 0.1400, 0.0640, 0.0883, 0.0750], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0078, 0.0097, 0.0091, 0.0105, 0.0067, 0.0096, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:24:12,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 3.132e+02 4.108e+02 5.472e+02 1.235e+03, threshold=8.216e+02, percent-clipped=5.0 2022-12-07 08:24:23,970 INFO [train.py:873] (0/4) Epoch 3, batch 800, loss[loss=0.2599, simple_loss=0.2349, pruned_loss=0.1425, over 9503.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.2195, pruned_loss=0.1132, over 1809335.47 frames. ], batch size: 100, lr: 2.55e-02, grad_scale: 8.0 2022-12-07 08:24:40,900 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:25:30,385 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:25:39,380 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.828e+02 4.017e+02 4.886e+02 9.739e+02, threshold=8.033e+02, percent-clipped=2.0 2022-12-07 08:25:51,105 INFO [train.py:873] (0/4) Epoch 3, batch 900, loss[loss=0.2463, simple_loss=0.2063, pruned_loss=0.1432, over 2574.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.2189, pruned_loss=0.1122, over 1853549.15 frames. ], batch size: 100, lr: 2.54e-02, grad_scale: 8.0 2022-12-07 08:26:22,591 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:26:42,844 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:26:52,601 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1549, 2.7943, 2.9699, 3.1648, 3.1099, 3.1947, 3.2217, 2.7659], device='cuda:0'), covar=tensor([0.0375, 0.1149, 0.0373, 0.0480, 0.0579, 0.0352, 0.0554, 0.0596], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0165, 0.0101, 0.0100, 0.0099, 0.0093, 0.0150, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:27:05,929 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=5.45 vs. limit=2.0 2022-12-07 08:27:06,256 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 2.979e+02 3.878e+02 4.942e+02 1.281e+03, threshold=7.755e+02, percent-clipped=6.0 2022-12-07 08:27:16,834 INFO [train.py:873] (0/4) Epoch 3, batch 1000, loss[loss=0.2194, simple_loss=0.2204, pruned_loss=0.1091, over 14154.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.219, pruned_loss=0.1117, over 1878080.20 frames. ], batch size: 99, lr: 2.54e-02, grad_scale: 8.0 2022-12-07 08:27:35,226 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:27:38,014 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:28:29,870 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:28:31,353 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 3.033e+02 4.056e+02 5.147e+02 1.062e+03, threshold=8.113e+02, percent-clipped=5.0 2022-12-07 08:28:31,607 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0844, 0.6653, 0.7597, 1.1850, 1.0723, 0.4679, 1.2069, 1.0023], device='cuda:0'), covar=tensor([0.0400, 0.0275, 0.0189, 0.0194, 0.0350, 0.0197, 0.0220, 0.0340], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0017, 0.0018, 0.0015, 0.0017, 0.0021, 0.0015, 0.0015], device='cuda:0'), out_proj_covar=tensor([4.4102e-05, 4.4295e-05, 4.5009e-05, 4.3797e-05, 4.3312e-05, 5.2498e-05, 4.4475e-05, 3.8429e-05], device='cuda:0') 2022-12-07 08:28:42,880 INFO [train.py:873] (0/4) Epoch 3, batch 1100, loss[loss=0.2528, simple_loss=0.2034, pruned_loss=0.1512, over 1196.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.219, pruned_loss=0.1118, over 1883669.42 frames. ], batch size: 100, lr: 2.53e-02, grad_scale: 8.0 2022-12-07 08:28:45,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5994, 3.9495, 2.9248, 4.8554, 3.8240, 4.2541, 4.0829, 3.4104], device='cuda:0'), covar=tensor([0.0124, 0.0368, 0.2208, 0.0079, 0.0171, 0.0517, 0.0367, 0.1813], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0265, 0.0348, 0.0158, 0.0183, 0.0180, 0.0236, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:28:55,093 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:29:42,287 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 08:29:52,130 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7796, 4.2686, 4.2777, 4.7422, 4.5716, 4.2498, 4.7711, 4.1651], device='cuda:0'), covar=tensor([0.0262, 0.0899, 0.0264, 0.0371, 0.0589, 0.0460, 0.0493, 0.0434], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0168, 0.0103, 0.0101, 0.0101, 0.0093, 0.0153, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:29:58,310 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.546e+02 3.661e+02 5.045e+02 9.524e+02, threshold=7.322e+02, percent-clipped=3.0 2022-12-07 08:30:09,551 INFO [train.py:873] (0/4) Epoch 3, batch 1200, loss[loss=0.2033, simple_loss=0.2119, pruned_loss=0.09735, over 14294.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2184, pruned_loss=0.1105, over 1963149.05 frames. ], batch size: 63, lr: 2.52e-02, grad_scale: 8.0 2022-12-07 08:30:37,126 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:30:48,990 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:31:23,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 3.060e+02 3.691e+02 4.545e+02 9.377e+02, threshold=7.381e+02, percent-clipped=3.0 2022-12-07 08:31:35,382 INFO [train.py:873] (0/4) Epoch 3, batch 1300, loss[loss=0.2637, simple_loss=0.2414, pruned_loss=0.143, over 8584.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.2177, pruned_loss=0.11, over 1935856.76 frames. ], batch size: 100, lr: 2.51e-02, grad_scale: 8.0 2022-12-07 08:31:41,475 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:31:49,047 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:31:57,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2348, 2.7576, 4.1054, 3.0635, 4.0823, 3.8673, 3.6274, 3.3323], device='cuda:0'), covar=tensor([0.0068, 0.0963, 0.0061, 0.0514, 0.0154, 0.0142, 0.0597, 0.0703], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0334, 0.0166, 0.0269, 0.0213, 0.0220, 0.0230, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:32:37,040 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2022-12-07 08:32:43,982 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:32:48,642 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 08:32:49,717 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.808e+02 3.737e+02 4.916e+02 1.118e+03, threshold=7.474e+02, percent-clipped=8.0 2022-12-07 08:32:52,723 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0329, 1.8718, 1.7317, 0.9460, 1.6311, 1.7129, 2.0606, 1.7391], device='cuda:0'), covar=tensor([0.0634, 0.4226, 0.1562, 0.3629, 0.1149, 0.0955, 0.0817, 0.1334], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0196, 0.0094, 0.0121, 0.0075, 0.0080, 0.0067, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:33:01,053 INFO [train.py:873] (0/4) Epoch 3, batch 1400, loss[loss=0.2188, simple_loss=0.2194, pruned_loss=0.109, over 14227.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.2184, pruned_loss=0.1105, over 1930536.11 frames. ], batch size: 94, lr: 2.51e-02, grad_scale: 8.0 2022-12-07 08:33:12,676 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:33:54,100 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:34:14,491 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-07 08:34:15,419 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.141e+01 3.095e+02 4.057e+02 5.380e+02 9.685e+02, threshold=8.113e+02, percent-clipped=8.0 2022-12-07 08:34:27,198 INFO [train.py:873] (0/4) Epoch 3, batch 1500, loss[loss=0.2342, simple_loss=0.2334, pruned_loss=0.1175, over 14094.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2176, pruned_loss=0.1104, over 1895420.38 frames. ], batch size: 29, lr: 2.50e-02, grad_scale: 8.0 2022-12-07 08:34:35,166 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2959, 0.7472, 0.5888, 1.5246, 1.0426, 0.4870, 1.4484, 1.3047], device='cuda:0'), covar=tensor([0.0275, 0.0380, 0.0109, 0.0118, 0.0363, 0.0173, 0.0133, 0.0162], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0019, 0.0018, 0.0015, 0.0018, 0.0021, 0.0016, 0.0015], device='cuda:0'), out_proj_covar=tensor([4.5073e-05, 4.8516e-05, 4.6013e-05, 4.4911e-05, 4.6051e-05, 5.3824e-05, 4.5597e-05, 3.8674e-05], device='cuda:0') 2022-12-07 08:34:54,273 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:35:35,803 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:35:41,765 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.784e+02 3.654e+02 4.523e+02 7.286e+02, threshold=7.308e+02, percent-clipped=0.0 2022-12-07 08:35:52,814 INFO [train.py:873] (0/4) Epoch 3, batch 1600, loss[loss=0.2477, simple_loss=0.2419, pruned_loss=0.1268, over 14276.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.2185, pruned_loss=0.1107, over 1942061.70 frames. ], batch size: 76, lr: 2.49e-02, grad_scale: 8.0 2022-12-07 08:35:54,557 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:36:06,501 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:36:48,048 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:01,956 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:07,697 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 3.316e+02 4.284e+02 5.393e+02 9.437e+02, threshold=8.568e+02, percent-clipped=8.0 2022-12-07 08:37:19,028 INFO [train.py:873] (0/4) Epoch 3, batch 1700, loss[loss=0.1998, simple_loss=0.2098, pruned_loss=0.0949, over 14414.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.2183, pruned_loss=0.1111, over 1887406.89 frames. ], batch size: 41, lr: 2.49e-02, grad_scale: 8.0 2022-12-07 08:37:42,665 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:37:55,916 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:38:28,227 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:38:33,243 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.887e+02 3.816e+02 4.798e+02 1.128e+03, threshold=7.633e+02, percent-clipped=3.0 2022-12-07 08:38:44,614 INFO [train.py:873] (0/4) Epoch 3, batch 1800, loss[loss=0.2073, simple_loss=0.2154, pruned_loss=0.09957, over 14380.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2176, pruned_loss=0.1096, over 1946681.23 frames. ], batch size: 73, lr: 2.48e-02, grad_scale: 8.0 2022-12-07 08:38:48,270 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:39:02,733 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1586, 1.1240, 1.4702, 0.9312, 1.2644, 1.3924, 0.9265, 1.3735], device='cuda:0'), covar=tensor([0.0833, 0.1128, 0.0290, 0.1266, 0.0739, 0.0289, 0.1072, 0.0462], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0068, 0.0062, 0.0072, 0.0077, 0.0063, 0.0123, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:39:11,518 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7246, 2.9002, 2.0667, 3.2336, 2.6728, 3.2224, 2.8124, 2.2373], device='cuda:0'), covar=tensor([0.0163, 0.0443, 0.2082, 0.0142, 0.0252, 0.0148, 0.0546, 0.2454], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0272, 0.0354, 0.0167, 0.0195, 0.0190, 0.0248, 0.0352], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:39:14,719 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1711, 2.0971, 1.6474, 1.5978, 2.0890, 2.2180, 2.1145, 2.1478], device='cuda:0'), covar=tensor([0.1230, 0.1399, 0.2987, 0.3692, 0.1281, 0.1205, 0.2355, 0.1581], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0174, 0.0228, 0.0284, 0.0183, 0.0203, 0.0223, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:39:21,128 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:39:51,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7272, 1.3340, 3.4571, 1.4309, 3.5912, 3.5508, 2.5786, 3.8075], device='cuda:0'), covar=tensor([0.0142, 0.2428, 0.0300, 0.2013, 0.0274, 0.0234, 0.0559, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0145, 0.0102, 0.0152, 0.0118, 0.0107, 0.0096, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:39:59,563 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 3.016e+02 3.995e+02 5.163e+02 9.262e+02, threshold=7.990e+02, percent-clipped=4.0 2022-12-07 08:40:11,093 INFO [train.py:873] (0/4) Epoch 3, batch 1900, loss[loss=0.201, simple_loss=0.2116, pruned_loss=0.09518, over 14675.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2175, pruned_loss=0.1094, over 1980787.67 frames. ], batch size: 33, lr: 2.47e-02, grad_scale: 8.0 2022-12-07 08:40:12,990 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:40:13,418 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 08:40:25,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 08:40:42,312 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 08:40:53,613 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:41:25,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.833e+02 3.865e+02 5.052e+02 1.376e+03, threshold=7.730e+02, percent-clipped=3.0 2022-12-07 08:41:36,266 INFO [train.py:873] (0/4) Epoch 3, batch 2000, loss[loss=0.2123, simple_loss=0.1847, pruned_loss=0.12, over 1200.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.2179, pruned_loss=0.1095, over 1981426.74 frames. ], batch size: 100, lr: 2.47e-02, grad_scale: 16.0 2022-12-07 08:41:46,471 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8481, 1.2486, 3.3629, 3.3712, 3.6305, 3.4033, 3.0028, 3.5510], device='cuda:0'), covar=tensor([0.1320, 0.1719, 0.0194, 0.0198, 0.0146, 0.0178, 0.0256, 0.0170], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0145, 0.0076, 0.0105, 0.0089, 0.0095, 0.0072, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:42:30,537 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:42:49,868 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 3.144e+02 4.016e+02 5.229e+02 1.105e+03, threshold=8.031e+02, percent-clipped=5.0 2022-12-07 08:43:00,458 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:43:01,530 INFO [train.py:873] (0/4) Epoch 3, batch 2100, loss[loss=0.2146, simple_loss=0.1811, pruned_loss=0.124, over 1266.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2183, pruned_loss=0.1102, over 1976998.94 frames. ], batch size: 100, lr: 2.46e-02, grad_scale: 16.0 2022-12-07 08:43:22,651 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:43:32,873 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 08:43:39,874 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2861, 3.0659, 2.8940, 2.8959, 3.1457, 3.1655, 3.2719, 3.2547], device='cuda:0'), covar=tensor([0.0688, 0.0596, 0.1411, 0.2164, 0.0644, 0.0545, 0.0769, 0.0835], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0170, 0.0228, 0.0288, 0.0188, 0.0208, 0.0221, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:44:16,384 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.883e+02 3.756e+02 4.821e+02 7.319e+02, threshold=7.512e+02, percent-clipped=0.0 2022-12-07 08:44:27,586 INFO [train.py:873] (0/4) Epoch 3, batch 2200, loss[loss=0.1916, simple_loss=0.1959, pruned_loss=0.09367, over 5997.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.2175, pruned_loss=0.1099, over 1918755.82 frames. ], batch size: 100, lr: 2.45e-02, grad_scale: 16.0 2022-12-07 08:44:34,670 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.29 vs. limit=5.0 2022-12-07 08:44:59,076 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:45:17,017 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:45:42,085 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 3.126e+02 3.760e+02 4.600e+02 9.821e+02, threshold=7.519e+02, percent-clipped=3.0 2022-12-07 08:45:51,656 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:45:53,185 INFO [train.py:873] (0/4) Epoch 3, batch 2300, loss[loss=0.2405, simple_loss=0.2256, pruned_loss=0.1277, over 3860.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.2173, pruned_loss=0.1096, over 1959878.57 frames. ], batch size: 100, lr: 2.45e-02, grad_scale: 16.0 2022-12-07 08:46:09,727 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:47:07,374 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.910e+02 3.962e+02 5.251e+02 8.178e+02, threshold=7.923e+02, percent-clipped=2.0 2022-12-07 08:47:18,003 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:47:18,682 INFO [train.py:873] (0/4) Epoch 3, batch 2400, loss[loss=0.1858, simple_loss=0.1978, pruned_loss=0.08692, over 13810.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.2176, pruned_loss=0.1101, over 1966770.25 frames. ], batch size: 23, lr: 2.44e-02, grad_scale: 16.0 2022-12-07 08:47:35,590 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:47:35,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.88 vs. limit=2.0 2022-12-07 08:47:50,045 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:47:58,493 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:48:24,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2332, 1.9526, 2.5569, 2.5897, 2.0883, 1.7964, 2.7475, 2.2643], device='cuda:0'), covar=tensor([0.0046, 0.0087, 0.0073, 0.0041, 0.0054, 0.0160, 0.0026, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0151, 0.0180, 0.0160, 0.0135, 0.0198, 0.0094, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002], device='cuda:0') 2022-12-07 08:48:31,664 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:48:34,155 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.796e+02 3.862e+02 5.213e+02 1.653e+03, threshold=7.725e+02, percent-clipped=7.0 2022-12-07 08:48:44,388 INFO [train.py:873] (0/4) Epoch 3, batch 2500, loss[loss=0.1967, simple_loss=0.1775, pruned_loss=0.108, over 1282.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2164, pruned_loss=0.108, over 1991055.01 frames. ], batch size: 100, lr: 2.43e-02, grad_scale: 8.0 2022-12-07 08:49:44,752 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6904, 0.6004, 0.3578, 0.7910, 0.6638, 0.2544, 0.6268, 0.5134], device='cuda:0'), covar=tensor([0.0034, 0.0040, 0.0020, 0.0064, 0.0034, 0.0023, 0.0078, 0.0057], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0020, 0.0020, 0.0017, 0.0018, 0.0021, 0.0017, 0.0016], device='cuda:0'), out_proj_covar=tensor([4.7549e-05, 5.2871e-05, 5.0957e-05, 5.1421e-05, 4.9041e-05, 5.4135e-05, 4.9083e-05, 4.0992e-05], device='cuda:0') 2022-12-07 08:49:49,066 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:49:59,744 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.968e+02 3.839e+02 5.013e+02 8.291e+02, threshold=7.678e+02, percent-clipped=4.0 2022-12-07 08:50:04,227 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:50:10,237 INFO [train.py:873] (0/4) Epoch 3, batch 2600, loss[loss=0.2463, simple_loss=0.2068, pruned_loss=0.1428, over 1297.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.2161, pruned_loss=0.1078, over 1984653.26 frames. ], batch size: 100, lr: 2.43e-02, grad_scale: 8.0 2022-12-07 08:50:14,758 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:50:22,582 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 08:50:32,509 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8988, 1.8783, 1.5834, 1.9862, 1.8476, 1.9940, 1.7452, 1.6460], device='cuda:0'), covar=tensor([0.0108, 0.0272, 0.0505, 0.0123, 0.0138, 0.0076, 0.0316, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0278, 0.0354, 0.0171, 0.0205, 0.0195, 0.0258, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:50:41,068 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:50:51,509 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9088, 3.5462, 4.4822, 2.6752, 2.8504, 3.1538, 1.1355, 3.1119], device='cuda:0'), covar=tensor([0.1504, 0.0660, 0.0626, 0.1503, 0.0947, 0.0873, 0.3702, 0.1449], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0070, 0.0061, 0.0071, 0.0080, 0.0064, 0.0127, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:51:06,981 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:51:20,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7865, 1.9918, 1.0430, 1.8164, 1.4995, 1.2918, 1.3641, 0.9538], device='cuda:0'), covar=tensor([0.0400, 0.0514, 0.0562, 0.0485, 0.0343, 0.0116, 0.0185, 0.0324], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0012, 0.0013, 0.0013, 0.0015, 0.0014, 0.0015], device='cuda:0'), out_proj_covar=tensor([3.3957e-05, 3.3226e-05, 3.6064e-05, 3.3910e-05, 3.3770e-05, 3.5199e-05, 4.4630e-05, 4.2189e-05], device='cuda:0') 2022-12-07 08:51:25,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.813e+02 3.769e+02 4.969e+02 1.379e+03, threshold=7.537e+02, percent-clipped=10.0 2022-12-07 08:51:36,295 INFO [train.py:873] (0/4) Epoch 3, batch 2700, loss[loss=0.21, simple_loss=0.2017, pruned_loss=0.1091, over 5035.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2155, pruned_loss=0.1076, over 2003143.52 frames. ], batch size: 100, lr: 2.42e-02, grad_scale: 8.0 2022-12-07 08:51:53,219 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:52:11,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 08:52:18,047 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2022-12-07 08:52:23,238 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 08:52:34,435 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:52:37,480 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.15 vs. limit=2.0 2022-12-07 08:52:51,605 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 3.286e+02 4.021e+02 4.924e+02 9.078e+02, threshold=8.042e+02, percent-clipped=2.0 2022-12-07 08:52:54,882 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.89 vs. limit=5.0 2022-12-07 08:53:01,816 INFO [train.py:873] (0/4) Epoch 3, batch 2800, loss[loss=0.2153, simple_loss=0.1918, pruned_loss=0.1194, over 2640.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2163, pruned_loss=0.1087, over 1945279.22 frames. ], batch size: 100, lr: 2.41e-02, grad_scale: 8.0 2022-12-07 08:53:04,630 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5420, 2.3706, 3.2790, 2.5841, 3.3620, 3.2623, 3.0916, 2.6336], device='cuda:0'), covar=tensor([0.0088, 0.1003, 0.0076, 0.0486, 0.0148, 0.0127, 0.0382, 0.0871], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0331, 0.0171, 0.0275, 0.0221, 0.0218, 0.0234, 0.0358], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 08:53:18,378 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7751, 0.7364, 0.5439, 0.8677, 0.7333, 0.3874, 0.8589, 0.9452], device='cuda:0'), covar=tensor([0.0201, 0.0148, 0.0067, 0.0133, 0.0091, 0.0123, 0.0171, 0.0136], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0019, 0.0020, 0.0016, 0.0018, 0.0021, 0.0016, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.6768e-05, 5.2801e-05, 5.1132e-05, 5.0084e-05, 4.7417e-05, 5.3638e-05, 4.6944e-05, 4.3980e-05], device='cuda:0') 2022-12-07 08:53:31,814 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:03,123 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9324, 0.7950, 0.8737, 0.9781, 0.8474, 0.6368, 1.1209, 0.8537], device='cuda:0'), covar=tensor([0.0225, 0.0311, 0.0233, 0.0426, 0.0152, 0.0163, 0.0228, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0020, 0.0021, 0.0017, 0.0018, 0.0021, 0.0016, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.7395e-05, 5.3622e-05, 5.3070e-05, 5.2460e-05, 4.8258e-05, 5.4505e-05, 4.7417e-05, 4.4597e-05], device='cuda:0') 2022-12-07 08:54:08,741 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 08:54:16,684 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.972e+02 3.622e+02 4.500e+02 9.762e+02, threshold=7.243e+02, percent-clipped=2.0 2022-12-07 08:54:21,472 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:23,940 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:54:24,811 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9776, 1.8746, 1.6019, 1.5766, 1.9111, 1.8018, 1.9601, 1.8500], device='cuda:0'), covar=tensor([0.1033, 0.0962, 0.1936, 0.2706, 0.0736, 0.1073, 0.1249, 0.1340], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0180, 0.0243, 0.0311, 0.0198, 0.0220, 0.0228, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:54:27,360 INFO [train.py:873] (0/4) Epoch 3, batch 2900, loss[loss=0.2, simple_loss=0.2091, pruned_loss=0.09542, over 14575.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2159, pruned_loss=0.1076, over 2002792.87 frames. ], batch size: 22, lr: 2.41e-02, grad_scale: 8.0 2022-12-07 08:54:39,647 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 08:54:53,279 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:01,988 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:05,762 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9625, 2.7033, 2.5160, 2.5954, 2.7848, 2.7423, 2.9055, 2.8769], device='cuda:0'), covar=tensor([0.0824, 0.0758, 0.1688, 0.2147, 0.0696, 0.0733, 0.0929, 0.0882], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0177, 0.0243, 0.0305, 0.0196, 0.0218, 0.0230, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 08:55:19,244 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:55:20,052 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 08:55:26,697 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 08:55:42,573 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 3.066e+02 4.043e+02 5.337e+02 1.703e+03, threshold=8.087e+02, percent-clipped=7.0 2022-12-07 08:55:52,803 INFO [train.py:873] (0/4) Epoch 3, batch 3000, loss[loss=0.2347, simple_loss=0.2263, pruned_loss=0.1215, over 14542.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2155, pruned_loss=0.1075, over 1951259.70 frames. ], batch size: 34, lr: 2.40e-02, grad_scale: 8.0 2022-12-07 08:55:52,804 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 08:55:59,625 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4222, 2.4506, 4.7365, 4.2867, 4.4017, 4.6750, 4.6637, 4.8746], device='cuda:0'), covar=tensor([0.1286, 0.1254, 0.0076, 0.0098, 0.0111, 0.0103, 0.0038, 0.0059], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0141, 0.0078, 0.0108, 0.0088, 0.0097, 0.0072, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 08:56:03,704 INFO [train.py:905] (0/4) Epoch 3, validation: loss=0.1337, simple_loss=0.176, pruned_loss=0.04573, over 857387.00 frames. 2022-12-07 08:56:03,704 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 08:56:40,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.93 vs. limit=2.0 2022-12-07 08:56:45,798 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:56:46,894 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:56:56,929 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7398, 1.6142, 2.0423, 1.7180, 2.0788, 1.5727, 1.6226, 1.8154], device='cuda:0'), covar=tensor([0.0467, 0.0763, 0.0079, 0.0722, 0.0090, 0.0554, 0.0755, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0323, 0.0181, 0.0417, 0.0165, 0.0323, 0.0298, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0004, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:57:06,479 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3080, 2.2102, 4.0469, 2.8935, 4.2940, 1.9167, 2.9298, 3.9728], device='cuda:0'), covar=tensor([0.0247, 0.4638, 0.0331, 0.8672, 0.0119, 0.3574, 0.1155, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0329, 0.0183, 0.0421, 0.0167, 0.0328, 0.0302, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0004, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 08:57:19,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 3.113e+02 4.054e+02 5.125e+02 9.563e+02, threshold=8.108e+02, percent-clipped=4.0 2022-12-07 08:57:30,105 INFO [train.py:873] (0/4) Epoch 3, batch 3100, loss[loss=0.2128, simple_loss=0.1749, pruned_loss=0.1254, over 1231.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.215, pruned_loss=0.1074, over 1949296.89 frames. ], batch size: 100, lr: 2.40e-02, grad_scale: 8.0 2022-12-07 08:57:34,832 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 08:57:38,630 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:57:39,574 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:58:03,672 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1214, 1.2323, 1.4972, 0.8578, 1.0040, 1.3093, 0.7548, 1.2554], device='cuda:0'), covar=tensor([0.0779, 0.0718, 0.0288, 0.1100, 0.0748, 0.0315, 0.1108, 0.0377], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0067, 0.0061, 0.0072, 0.0080, 0.0060, 0.0121, 0.0067], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:58:45,517 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.872e+02 3.739e+02 4.989e+02 1.887e+03, threshold=7.477e+02, percent-clipped=4.0 2022-12-07 08:58:48,256 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:58:55,686 INFO [train.py:873] (0/4) Epoch 3, batch 3200, loss[loss=0.2342, simple_loss=0.2318, pruned_loss=0.1183, over 14405.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.2154, pruned_loss=0.108, over 1925262.43 frames. ], batch size: 53, lr: 2.39e-02, grad_scale: 8.0 2022-12-07 08:59:02,986 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4509, 1.4080, 1.9059, 1.1026, 1.3586, 1.7023, 0.9389, 1.6556], device='cuda:0'), covar=tensor([0.0787, 0.1172, 0.0253, 0.1438, 0.1214, 0.0468, 0.2251, 0.0420], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0070, 0.0063, 0.0074, 0.0083, 0.0062, 0.0126, 0.0067], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:0') 2022-12-07 08:59:22,828 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 08:59:48,407 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:03,903 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:11,699 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 3.073e+02 3.846e+02 5.259e+02 1.192e+03, threshold=7.691e+02, percent-clipped=7.0 2022-12-07 09:00:22,198 INFO [train.py:873] (0/4) Epoch 3, batch 3300, loss[loss=0.223, simple_loss=0.2124, pruned_loss=0.1168, over 6950.00 frames. ], tot_loss[loss=0.2156, simple_loss=0.2154, pruned_loss=0.1079, over 1945208.05 frames. ], batch size: 100, lr: 2.38e-02, grad_scale: 8.0 2022-12-07 09:00:29,914 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:00:49,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8697, 1.4992, 1.7764, 1.2295, 1.3292, 1.4376, 1.3573, 1.5686], device='cuda:0'), covar=tensor([0.0211, 0.0859, 0.0392, 0.0811, 0.0463, 0.0910, 0.0299, 0.0520], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0205, 0.0097, 0.0126, 0.0083, 0.0079, 0.0070, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:01:13,873 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4386, 1.7139, 3.9928, 1.9098, 4.1685, 4.1887, 3.5128, 4.7376], device='cuda:0'), covar=tensor([0.0159, 0.2609, 0.0281, 0.2065, 0.0242, 0.0247, 0.0336, 0.0093], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0144, 0.0103, 0.0154, 0.0119, 0.0110, 0.0097, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:01:18,091 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:01:33,647 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:01:37,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 3.014e+02 3.926e+02 4.952e+02 1.093e+03, threshold=7.852e+02, percent-clipped=5.0 2022-12-07 09:01:44,732 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7074, 5.0092, 5.2884, 5.6354, 5.4098, 4.6623, 5.7097, 4.5934], device='cuda:0'), covar=tensor([0.0315, 0.1246, 0.0254, 0.0455, 0.0574, 0.0327, 0.0441, 0.0514], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0171, 0.0114, 0.0107, 0.0103, 0.0098, 0.0161, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:01:48,039 INFO [train.py:873] (0/4) Epoch 3, batch 3400, loss[loss=0.2387, simple_loss=0.2209, pruned_loss=0.1282, over 5994.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2156, pruned_loss=0.1082, over 1939998.91 frames. ], batch size: 100, lr: 2.38e-02, grad_scale: 8.0 2022-12-07 09:01:52,000 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:01:53,114 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:02:10,143 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:02:25,552 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:02:28,297 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2021, 3.4449, 5.0321, 3.7023, 4.7144, 5.0348, 4.2032, 4.3692], device='cuda:0'), covar=tensor([0.0049, 0.0866, 0.0059, 0.0418, 0.0148, 0.0070, 0.0544, 0.0466], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0331, 0.0184, 0.0272, 0.0225, 0.0217, 0.0242, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:02:34,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 09:03:03,541 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 3.062e+02 4.109e+02 5.827e+02 9.337e+02, threshold=8.217e+02, percent-clipped=6.0 2022-12-07 09:03:06,412 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:03:13,686 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.49 vs. limit=5.0 2022-12-07 09:03:13,884 INFO [train.py:873] (0/4) Epoch 3, batch 3500, loss[loss=0.2024, simple_loss=0.1749, pruned_loss=0.115, over 2674.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2142, pruned_loss=0.1065, over 1951380.28 frames. ], batch size: 100, lr: 2.37e-02, grad_scale: 8.0 2022-12-07 09:03:46,486 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:03:54,840 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:04:26,703 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3672, 4.9071, 4.8272, 5.4553, 4.9998, 4.1791, 5.6559, 5.5589], device='cuda:0'), covar=tensor([0.0668, 0.0521, 0.0682, 0.0613, 0.0694, 0.0426, 0.0549, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0077, 0.0100, 0.0092, 0.0102, 0.0068, 0.0100, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:04:28,294 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.925e+02 3.819e+02 4.873e+02 8.102e+02, threshold=7.639e+02, percent-clipped=0.0 2022-12-07 09:04:33,881 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 09:04:38,566 INFO [train.py:873] (0/4) Epoch 3, batch 3600, loss[loss=0.2263, simple_loss=0.2122, pruned_loss=0.1202, over 5988.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2135, pruned_loss=0.1047, over 1985087.59 frames. ], batch size: 100, lr: 2.37e-02, grad_scale: 8.0 2022-12-07 09:04:45,908 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.06 vs. limit=5.0 2022-12-07 09:04:47,565 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:04:52,160 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8431, 0.7942, 0.8521, 0.7101, 0.7813, 0.8275, 1.5114, 1.4773], device='cuda:0'), covar=tensor([0.1103, 0.0320, 0.0145, 0.0750, 0.0585, 0.0202, 0.0169, 0.0203], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0019, 0.0020, 0.0017, 0.0017, 0.0022, 0.0016, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.7570e-05, 5.4076e-05, 5.3650e-05, 5.2405e-05, 4.8181e-05, 5.7203e-05, 4.7720e-05, 4.4476e-05], device='cuda:0') 2022-12-07 09:05:16,140 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1521, 1.5730, 3.9790, 3.8515, 3.8462, 3.9432, 3.1770, 4.0708], device='cuda:0'), covar=tensor([0.1037, 0.1334, 0.0083, 0.0097, 0.0091, 0.0083, 0.0221, 0.0069], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0143, 0.0079, 0.0110, 0.0092, 0.0099, 0.0076, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:05:16,973 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8217, 1.7831, 3.0709, 2.3182, 2.8810, 1.7792, 2.3593, 2.8228], device='cuda:0'), covar=tensor([0.0389, 0.3244, 0.0248, 0.4345, 0.0214, 0.2890, 0.0942, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0317, 0.0177, 0.0411, 0.0168, 0.0328, 0.0293, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0004, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:05:54,917 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 3.054e+02 3.647e+02 5.491e+02 1.580e+03, threshold=7.295e+02, percent-clipped=4.0 2022-12-07 09:06:02,762 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 09:06:05,676 INFO [train.py:873] (0/4) Epoch 3, batch 3700, loss[loss=0.1902, simple_loss=0.2029, pruned_loss=0.08875, over 14265.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2134, pruned_loss=0.1043, over 2006570.06 frames. ], batch size: 57, lr: 2.36e-02, grad_scale: 8.0 2022-12-07 09:06:09,869 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:10,715 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:23,432 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:39,729 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 09:06:51,533 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:06:52,320 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:07:21,106 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.825e+02 3.727e+02 4.870e+02 9.642e+02, threshold=7.453e+02, percent-clipped=3.0 2022-12-07 09:07:31,080 INFO [train.py:873] (0/4) Epoch 3, batch 3800, loss[loss=0.1944, simple_loss=0.2012, pruned_loss=0.09377, over 13544.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2137, pruned_loss=0.1054, over 1978796.52 frames. ], batch size: 100, lr: 2.35e-02, grad_scale: 8.0 2022-12-07 09:07:35,773 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:07:59,068 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8903, 1.2452, 2.0590, 1.3782, 2.0435, 2.0514, 1.5402, 2.0151], device='cuda:0'), covar=tensor([0.0194, 0.1104, 0.0182, 0.1134, 0.0223, 0.0309, 0.0525, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0147, 0.0107, 0.0158, 0.0123, 0.0114, 0.0100, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:08:09,329 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7831, 5.0594, 5.2415, 5.7075, 5.5227, 4.5815, 5.6641, 4.6751], device='cuda:0'), covar=tensor([0.0271, 0.0920, 0.0231, 0.0336, 0.0515, 0.0324, 0.0443, 0.0418], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0175, 0.0118, 0.0109, 0.0106, 0.0100, 0.0168, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:08:28,731 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:08:47,291 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.617e+02 3.029e+02 3.748e+02 4.757e+02 1.336e+03, threshold=7.495e+02, percent-clipped=2.0 2022-12-07 09:08:58,125 INFO [train.py:873] (0/4) Epoch 3, batch 3900, loss[loss=0.1984, simple_loss=0.2127, pruned_loss=0.09206, over 14576.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2129, pruned_loss=0.1047, over 1934666.82 frames. ], batch size: 34, lr: 2.35e-02, grad_scale: 8.0 2022-12-07 09:09:02,369 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:09:35,783 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7350, 5.1848, 4.9598, 5.7141, 5.2755, 4.6524, 5.6146, 4.7694], device='cuda:0'), covar=tensor([0.0323, 0.1142, 0.0267, 0.0357, 0.0718, 0.0321, 0.0459, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0173, 0.0118, 0.0111, 0.0107, 0.0098, 0.0166, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:09:47,659 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9756, 1.2576, 3.0444, 2.9319, 3.0539, 3.0414, 2.2647, 3.0708], device='cuda:0'), covar=tensor([0.0795, 0.1177, 0.0097, 0.0146, 0.0119, 0.0080, 0.0300, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0146, 0.0080, 0.0113, 0.0092, 0.0100, 0.0077, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:10:13,854 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.994e+02 3.505e+02 4.615e+02 1.018e+03, threshold=7.009e+02, percent-clipped=3.0 2022-12-07 09:10:19,727 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9548, 1.3512, 3.4257, 3.1673, 3.3776, 3.4086, 2.9144, 3.5130], device='cuda:0'), covar=tensor([0.0985, 0.1335, 0.0078, 0.0167, 0.0117, 0.0100, 0.0184, 0.0075], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0146, 0.0080, 0.0115, 0.0093, 0.0100, 0.0078, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:10:23,709 INFO [train.py:873] (0/4) Epoch 3, batch 4000, loss[loss=0.1857, simple_loss=0.1648, pruned_loss=0.1033, over 1193.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2129, pruned_loss=0.1042, over 1946124.96 frames. ], batch size: 100, lr: 2.34e-02, grad_scale: 8.0 2022-12-07 09:10:27,204 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:10:27,449 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=20.22 vs. limit=5.0 2022-12-07 09:10:42,416 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:10:57,668 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 09:11:20,473 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:11:23,547 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:11:32,924 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7646, 1.4813, 1.6641, 1.1308, 1.3285, 1.5209, 1.5412, 1.4645], device='cuda:0'), covar=tensor([0.0247, 0.0999, 0.0343, 0.0717, 0.0497, 0.0551, 0.0276, 0.0682], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0208, 0.0097, 0.0127, 0.0086, 0.0083, 0.0069, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0005, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:11:39,096 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 09:11:39,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.771e+02 3.795e+02 5.006e+02 9.210e+02, threshold=7.589e+02, percent-clipped=4.0 2022-12-07 09:11:50,339 INFO [train.py:873] (0/4) Epoch 3, batch 4100, loss[loss=0.2118, simple_loss=0.2141, pruned_loss=0.1048, over 14248.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2131, pruned_loss=0.1048, over 1918790.14 frames. ], batch size: 80, lr: 2.34e-02, grad_scale: 8.0 2022-12-07 09:12:42,490 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:12:53,447 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1840, 2.2780, 1.8086, 2.4595, 2.1106, 2.4028, 1.9936, 1.8629], device='cuda:0'), covar=tensor([0.0256, 0.0467, 0.1355, 0.0118, 0.0313, 0.0414, 0.0805, 0.1435], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0282, 0.0351, 0.0171, 0.0219, 0.0208, 0.0256, 0.0345], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:13:05,132 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 2.865e+02 3.600e+02 4.882e+02 1.320e+03, threshold=7.199e+02, percent-clipped=6.0 2022-12-07 09:13:15,262 INFO [train.py:873] (0/4) Epoch 3, batch 4200, loss[loss=0.2021, simple_loss=0.2018, pruned_loss=0.1012, over 9503.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.213, pruned_loss=0.1048, over 1948264.45 frames. ], batch size: 100, lr: 2.33e-02, grad_scale: 8.0 2022-12-07 09:13:19,781 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:13:37,763 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 09:14:01,565 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:14:09,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2022-12-07 09:14:11,081 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7506, 2.5277, 3.1791, 2.1241, 2.4531, 2.5749, 0.9885, 3.1198], device='cuda:0'), covar=tensor([0.2621, 0.1196, 0.0946, 0.2904, 0.2015, 0.1338, 0.6301, 0.1466], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0063, 0.0059, 0.0071, 0.0085, 0.0063, 0.0126, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:14:15,283 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:14:18,932 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 09:14:31,171 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 3.180e+02 3.832e+02 4.985e+02 8.220e+02, threshold=7.663e+02, percent-clipped=2.0 2022-12-07 09:14:34,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 09:14:36,162 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-07 09:14:42,547 INFO [train.py:873] (0/4) Epoch 3, batch 4300, loss[loss=0.2274, simple_loss=0.2235, pruned_loss=0.1157, over 11178.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2136, pruned_loss=0.1052, over 1976475.61 frames. ], batch size: 100, lr: 2.33e-02, grad_scale: 8.0 2022-12-07 09:15:08,471 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:15:09,602 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.15 vs. limit=5.0 2022-12-07 09:15:34,002 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:15:56,157 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2022-12-07 09:15:57,829 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.698e+02 3.428e+02 4.598e+02 7.579e+02, threshold=6.855e+02, percent-clipped=0.0 2022-12-07 09:16:08,043 INFO [train.py:873] (0/4) Epoch 3, batch 4400, loss[loss=0.2457, simple_loss=0.2219, pruned_loss=0.1347, over 3864.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2136, pruned_loss=0.1054, over 1985114.33 frames. ], batch size: 100, lr: 2.32e-02, grad_scale: 8.0 2022-12-07 09:16:17,837 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4079, 1.2349, 1.0499, 1.0319, 1.0094, 0.4999, 1.2176, 0.8263], device='cuda:0'), covar=tensor([0.0423, 0.0423, 0.0731, 0.0611, 0.0529, 0.0222, 0.0151, 0.0402], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0012, 0.0011, 0.0013, 0.0013, 0.0013, 0.0015], device='cuda:0'), out_proj_covar=tensor([3.4768e-05, 3.4626e-05, 3.7520e-05, 3.3847e-05, 3.6391e-05, 3.4688e-05, 4.3750e-05, 4.5797e-05], device='cuda:0') 2022-12-07 09:16:56,796 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2712, 3.8822, 4.4280, 3.5388, 4.2205, 4.3058, 1.6211, 4.0982], device='cuda:0'), covar=tensor([0.0135, 0.0268, 0.0253, 0.0371, 0.0226, 0.0157, 0.2859, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0117, 0.0110, 0.0092, 0.0150, 0.0102, 0.0151, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:17:00,306 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:23,968 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 2.902e+02 3.784e+02 4.919e+02 1.720e+03, threshold=7.568e+02, percent-clipped=8.0 2022-12-07 09:17:28,349 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:34,595 INFO [train.py:873] (0/4) Epoch 3, batch 4500, loss[loss=0.2081, simple_loss=0.2156, pruned_loss=0.1003, over 14269.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2122, pruned_loss=0.1038, over 1998544.41 frames. ], batch size: 46, lr: 2.31e-02, grad_scale: 8.0 2022-12-07 09:17:38,210 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4696, 3.0046, 2.0399, 3.5997, 3.4749, 3.6654, 2.5213, 2.4170], device='cuda:0'), covar=tensor([0.0272, 0.0679, 0.2948, 0.0218, 0.0250, 0.0344, 0.1156, 0.2701], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0283, 0.0344, 0.0177, 0.0223, 0.0206, 0.0261, 0.0342], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:17:42,278 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:17:44,332 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 09:18:14,747 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 09:18:20,799 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:18:50,393 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.725e+01 3.027e+02 3.828e+02 4.881e+02 1.178e+03, threshold=7.656e+02, percent-clipped=6.0 2022-12-07 09:18:59,478 INFO [train.py:873] (0/4) Epoch 3, batch 4600, loss[loss=0.2264, simple_loss=0.2262, pruned_loss=0.1133, over 14229.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2129, pruned_loss=0.1048, over 1945522.90 frames. ], batch size: 94, lr: 2.31e-02, grad_scale: 8.0 2022-12-07 09:19:21,759 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:19:48,361 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9837, 1.4261, 3.5762, 1.4551, 3.7383, 3.7665, 2.6805, 4.0792], device='cuda:0'), covar=tensor([0.0132, 0.2451, 0.0307, 0.2311, 0.0281, 0.0275, 0.0577, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0140, 0.0105, 0.0155, 0.0121, 0.0110, 0.0097, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:19:52,013 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 09:20:16,183 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 3.281e+02 3.991e+02 5.326e+02 1.100e+03, threshold=7.982e+02, percent-clipped=6.0 2022-12-07 09:20:26,813 INFO [train.py:873] (0/4) Epoch 3, batch 4700, loss[loss=0.2316, simple_loss=0.2029, pruned_loss=0.1301, over 2611.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2124, pruned_loss=0.1042, over 1972499.02 frames. ], batch size: 100, lr: 2.30e-02, grad_scale: 8.0 2022-12-07 09:20:33,475 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 09:21:43,694 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 3.110e+02 3.871e+02 4.666e+02 7.831e+02, threshold=7.741e+02, percent-clipped=0.0 2022-12-07 09:21:47,258 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7399, 4.2932, 4.2441, 4.6457, 4.4278, 4.2285, 4.7499, 4.0326], device='cuda:0'), covar=tensor([0.0328, 0.0849, 0.0312, 0.0421, 0.0648, 0.0479, 0.0447, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0169, 0.0115, 0.0106, 0.0103, 0.0096, 0.0163, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:21:52,841 INFO [train.py:873] (0/4) Epoch 3, batch 4800, loss[loss=0.2168, simple_loss=0.1874, pruned_loss=0.1231, over 2636.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2119, pruned_loss=0.1041, over 1932903.81 frames. ], batch size: 100, lr: 2.30e-02, grad_scale: 8.0 2022-12-07 09:22:35,281 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:22:37,829 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1920, 1.6759, 2.3905, 1.9871, 2.5060, 2.1705, 2.1440, 2.0259], device='cuda:0'), covar=tensor([0.0103, 0.0622, 0.0080, 0.0306, 0.0100, 0.0146, 0.0201, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0342, 0.0199, 0.0289, 0.0249, 0.0230, 0.0259, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:0') 2022-12-07 09:22:57,562 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-20000.pt 2022-12-07 09:23:13,337 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.781e+02 3.610e+02 5.418e+02 1.055e+03, threshold=7.220e+02, percent-clipped=5.0 2022-12-07 09:23:19,285 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:23:23,443 INFO [train.py:873] (0/4) Epoch 3, batch 4900, loss[loss=0.233, simple_loss=0.1974, pruned_loss=0.1343, over 2561.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2122, pruned_loss=0.1042, over 1965148.77 frames. ], batch size: 100, lr: 2.29e-02, grad_scale: 8.0 2022-12-07 09:23:40,516 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 09:23:44,385 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:04,200 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6487, 1.9104, 3.7661, 2.5844, 3.6445, 1.9313, 2.6278, 3.4177], device='cuda:0'), covar=tensor([0.0359, 0.5053, 0.0395, 0.7564, 0.0168, 0.3609, 0.1386, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0312, 0.0175, 0.0408, 0.0171, 0.0318, 0.0289, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0004, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:24:10,891 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:17,490 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:25,503 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:37,942 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:24:37,997 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2464, 1.9335, 2.5176, 2.4750, 1.9491, 1.7336, 2.7036, 1.9110], device='cuda:0'), covar=tensor([0.0097, 0.0200, 0.0159, 0.0136, 0.0154, 0.0356, 0.0042, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0158, 0.0201, 0.0168, 0.0143, 0.0205, 0.0110, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:24:39,472 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 3.078e+02 3.886e+02 5.087e+02 8.123e+02, threshold=7.771e+02, percent-clipped=4.0 2022-12-07 09:24:48,394 INFO [train.py:873] (0/4) Epoch 3, batch 5000, loss[loss=0.1559, simple_loss=0.1841, pruned_loss=0.06385, over 14271.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2132, pruned_loss=0.1044, over 1998073.35 frames. ], batch size: 31, lr: 2.29e-02, grad_scale: 8.0 2022-12-07 09:25:10,561 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:25:23,379 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 09:25:29,514 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:25:42,255 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9371, 2.8329, 2.5165, 2.5764, 2.8585, 2.6610, 2.8897, 2.8430], device='cuda:0'), covar=tensor([0.0720, 0.0682, 0.1505, 0.2225, 0.0634, 0.0787, 0.1264, 0.0850], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0188, 0.0266, 0.0336, 0.0207, 0.0239, 0.0256, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:25:47,065 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2862, 2.9089, 2.9515, 3.3493, 3.0434, 2.5988, 3.3099, 3.2569], device='cuda:0'), covar=tensor([0.0797, 0.0871, 0.0784, 0.0614, 0.0886, 0.0777, 0.0820, 0.0823], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0082, 0.0101, 0.0097, 0.0109, 0.0076, 0.0106, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:25:56,519 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:26:03,514 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.93 vs. limit=5.0 2022-12-07 09:26:04,801 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 2.928e+02 3.762e+02 4.834e+02 9.442e+02, threshold=7.523e+02, percent-clipped=4.0 2022-12-07 09:26:14,941 INFO [train.py:873] (0/4) Epoch 3, batch 5100, loss[loss=0.1717, simple_loss=0.1985, pruned_loss=0.07249, over 14282.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2125, pruned_loss=0.1044, over 1915965.17 frames. ], batch size: 44, lr: 2.28e-02, grad_scale: 8.0 2022-12-07 09:26:49,090 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:26:56,637 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:27:19,213 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 09:27:30,744 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.933e+02 3.750e+02 4.548e+02 8.499e+02, threshold=7.500e+02, percent-clipped=1.0 2022-12-07 09:27:37,604 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:27:39,973 INFO [train.py:873] (0/4) Epoch 3, batch 5200, loss[loss=0.2263, simple_loss=0.2128, pruned_loss=0.1199, over 6958.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2135, pruned_loss=0.1063, over 1899413.75 frames. ], batch size: 100, lr: 2.28e-02, grad_scale: 8.0 2022-12-07 09:28:16,876 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-07 09:28:23,148 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:28:34,148 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:28:35,673 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7597, 1.8859, 4.2376, 2.0154, 4.3030, 4.5720, 4.3218, 5.0755], device='cuda:0'), covar=tensor([0.0133, 0.2812, 0.0279, 0.2248, 0.0283, 0.0217, 0.0197, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0154, 0.0112, 0.0164, 0.0129, 0.0117, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:28:56,776 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.824e+02 3.945e+02 4.935e+02 1.895e+03, threshold=7.890e+02, percent-clipped=6.0 2022-12-07 09:28:59,470 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:06,135 INFO [train.py:873] (0/4) Epoch 3, batch 5300, loss[loss=0.2355, simple_loss=0.2296, pruned_loss=0.1207, over 3834.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2133, pruned_loss=0.1058, over 1902689.85 frames. ], batch size: 100, lr: 2.27e-02, grad_scale: 4.0 2022-12-07 09:29:23,351 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:26,906 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:43,751 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:29:52,890 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:30:23,816 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 2.763e+02 3.648e+02 4.928e+02 1.065e+03, threshold=7.297e+02, percent-clipped=2.0 2022-12-07 09:30:32,154 INFO [train.py:873] (0/4) Epoch 3, batch 5400, loss[loss=0.2131, simple_loss=0.2016, pruned_loss=0.1123, over 5975.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2135, pruned_loss=0.1054, over 1928596.43 frames. ], batch size: 100, lr: 2.27e-02, grad_scale: 4.0 2022-12-07 09:31:00,924 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 09:31:02,929 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:31:27,545 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:31:50,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.948e+02 3.775e+02 4.848e+02 6.830e+02, threshold=7.550e+02, percent-clipped=0.0 2022-12-07 09:31:59,000 INFO [train.py:873] (0/4) Epoch 3, batch 5500, loss[loss=0.1967, simple_loss=0.2, pruned_loss=0.09671, over 14512.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2118, pruned_loss=0.1035, over 1912202.42 frames. ], batch size: 51, lr: 2.26e-02, grad_scale: 4.0 2022-12-07 09:32:19,536 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:32:29,430 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7405, 2.6483, 2.4653, 1.4527, 2.1377, 2.4786, 2.7634, 2.0206], device='cuda:0'), covar=tensor([0.0284, 0.2029, 0.0801, 0.2673, 0.0727, 0.0275, 0.0395, 0.1344], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0219, 0.0098, 0.0128, 0.0088, 0.0083, 0.0070, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0005, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:32:32,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9641, 0.9349, 0.7614, 0.8392, 0.7102, 0.5029, 0.9554, 1.0213], device='cuda:0'), covar=tensor([0.0110, 0.0122, 0.0124, 0.0266, 0.0129, 0.0156, 0.0215, 0.0103], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0018, 0.0019, 0.0016, 0.0017, 0.0022, 0.0017, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.9413e-05, 5.4057e-05, 5.1234e-05, 5.1240e-05, 4.9610e-05, 6.0903e-05, 5.3451e-05, 4.9153e-05], device='cuda:0') 2022-12-07 09:32:42,426 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:14,659 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2247, 4.0483, 3.6903, 3.7469, 3.8968, 4.0489, 4.2091, 4.1367], device='cuda:0'), covar=tensor([0.0572, 0.0564, 0.1226, 0.1966, 0.0568, 0.0503, 0.0817, 0.0736], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0188, 0.0264, 0.0337, 0.0214, 0.0242, 0.0256, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:33:15,378 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.742e+02 3.691e+02 4.963e+02 1.196e+03, threshold=7.382e+02, percent-clipped=6.0 2022-12-07 09:33:22,776 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:23,569 INFO [train.py:873] (0/4) Epoch 3, batch 5600, loss[loss=0.1995, simple_loss=0.2023, pruned_loss=0.09835, over 13937.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2131, pruned_loss=0.1051, over 1899682.65 frames. ], batch size: 19, lr: 2.26e-02, grad_scale: 8.0 2022-12-07 09:33:40,096 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:41,246 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:33:52,928 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 09:34:00,939 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:06,483 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:21,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 09:34:24,033 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:43,760 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 3.037e+02 4.375e+02 5.410e+02 1.130e+03, threshold=8.749e+02, percent-clipped=7.0 2022-12-07 09:34:44,723 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:34:46,542 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6839, 4.5866, 4.7427, 4.2173, 4.6033, 5.1448, 1.6340, 4.4036], device='cuda:0'), covar=tensor([0.0187, 0.0237, 0.0516, 0.0406, 0.0348, 0.0108, 0.3669, 0.0292], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0113, 0.0105, 0.0091, 0.0146, 0.0099, 0.0141, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:34:49,898 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 09:34:53,316 INFO [train.py:873] (0/4) Epoch 3, batch 5700, loss[loss=0.1951, simple_loss=0.2087, pruned_loss=0.09078, over 14627.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2122, pruned_loss=0.1036, over 1950208.14 frames. ], batch size: 22, lr: 2.25e-02, grad_scale: 8.0 2022-12-07 09:34:56,229 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 09:35:24,861 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:35:27,962 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 09:36:06,585 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:36:12,609 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.426e+01 2.770e+02 3.515e+02 4.337e+02 1.206e+03, threshold=7.030e+02, percent-clipped=1.0 2022-12-07 09:36:21,396 INFO [train.py:873] (0/4) Epoch 3, batch 5800, loss[loss=0.2249, simple_loss=0.2157, pruned_loss=0.1171, over 11962.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2123, pruned_loss=0.1035, over 1983786.98 frames. ], batch size: 100, lr: 2.25e-02, grad_scale: 8.0 2022-12-07 09:36:23,850 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7502, 2.7439, 4.8248, 3.2327, 4.7441, 2.0054, 3.3494, 4.6178], device='cuda:0'), covar=tensor([0.0325, 0.4409, 0.0364, 1.0516, 0.0400, 0.3830, 0.1472, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0306, 0.0178, 0.0410, 0.0178, 0.0319, 0.0286, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:36:24,685 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1593, 2.1139, 1.6360, 1.6426, 1.4742, 1.7047, 1.8688, 0.6806], device='cuda:0'), covar=tensor([0.5227, 0.1430, 0.2813, 0.1941, 0.1900, 0.0952, 0.1529, 0.4765], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0064, 0.0054, 0.0059, 0.0069, 0.0059, 0.0076, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:36:37,677 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:36:59,273 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 09:37:16,837 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0789, 0.8900, 1.0064, 1.1216, 0.8717, 0.4814, 1.0759, 0.7551], device='cuda:0'), covar=tensor([0.0108, 0.0315, 0.0106, 0.0099, 0.0227, 0.0246, 0.0085, 0.0416], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0010, 0.0011, 0.0014, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([3.6810e-05, 3.7026e-05, 3.9456e-05, 3.4515e-05, 3.6164e-05, 3.9702e-05, 4.7616e-05, 5.1018e-05], device='cuda:0') 2022-12-07 09:37:39,551 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.882e+02 3.467e+02 4.356e+02 1.149e+03, threshold=6.934e+02, percent-clipped=1.0 2022-12-07 09:37:40,247 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 09:37:47,926 INFO [train.py:873] (0/4) Epoch 3, batch 5900, loss[loss=0.205, simple_loss=0.1744, pruned_loss=0.1179, over 1258.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2112, pruned_loss=0.1029, over 1994430.43 frames. ], batch size: 100, lr: 2.24e-02, grad_scale: 8.0 2022-12-07 09:38:04,840 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:11,858 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:32,099 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:38:36,845 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-07 09:38:42,932 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 09:38:47,683 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:07,996 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:09,614 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 3.189e+02 4.370e+02 6.428e+02 1.112e+03, threshold=8.740e+02, percent-clipped=21.0 2022-12-07 09:39:12,486 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:16,010 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:39:18,490 INFO [train.py:873] (0/4) Epoch 3, batch 6000, loss[loss=0.1928, simple_loss=0.2052, pruned_loss=0.09026, over 14080.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2117, pruned_loss=0.1027, over 2001134.12 frames. ], batch size: 22, lr: 2.24e-02, grad_scale: 8.0 2022-12-07 09:39:18,491 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 09:39:27,745 INFO [train.py:905] (0/4) Epoch 3, validation: loss=0.1295, simple_loss=0.172, pruned_loss=0.04354, over 857387.00 frames. 2022-12-07 09:39:27,746 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 09:39:49,371 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3880, 2.3779, 1.8894, 2.5108, 2.2203, 2.4569, 2.0401, 1.8730], device='cuda:0'), covar=tensor([0.0176, 0.0362, 0.1258, 0.0111, 0.0283, 0.0124, 0.0637, 0.1100], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0275, 0.0328, 0.0170, 0.0220, 0.0199, 0.0259, 0.0327], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2022-12-07 09:40:07,011 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2022-12-07 09:40:16,140 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:40:35,790 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5314, 1.7947, 2.7755, 2.7759, 2.5852, 1.9996, 2.6861, 2.1527], device='cuda:0'), covar=tensor([0.0061, 0.0178, 0.0134, 0.0095, 0.0066, 0.0247, 0.0035, 0.0211], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0160, 0.0201, 0.0174, 0.0144, 0.0207, 0.0111, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:40:47,318 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.723e+02 3.536e+02 4.487e+02 9.124e+02, threshold=7.073e+02, percent-clipped=3.0 2022-12-07 09:40:56,101 INFO [train.py:873] (0/4) Epoch 3, batch 6100, loss[loss=0.223, simple_loss=0.1976, pruned_loss=0.1242, over 3867.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2111, pruned_loss=0.102, over 2026629.06 frames. ], batch size: 100, lr: 2.23e-02, grad_scale: 8.0 2022-12-07 09:41:12,162 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:41:15,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-07 09:41:54,623 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:11,679 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:14,076 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 3.459e+02 4.303e+02 5.209e+02 1.109e+03, threshold=8.606e+02, percent-clipped=4.0 2022-12-07 09:42:22,732 INFO [train.py:873] (0/4) Epoch 3, batch 6200, loss[loss=0.2454, simple_loss=0.2117, pruned_loss=0.1396, over 1300.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2101, pruned_loss=0.1018, over 1983175.29 frames. ], batch size: 100, lr: 2.23e-02, grad_scale: 8.0 2022-12-07 09:42:25,288 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:42:26,160 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3071, 3.0689, 3.7838, 2.3690, 2.3711, 3.0123, 1.3137, 2.5882], device='cuda:0'), covar=tensor([0.1471, 0.1116, 0.0677, 0.2360, 0.1533, 0.0936, 0.5663, 0.2301], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0069, 0.0068, 0.0076, 0.0089, 0.0062, 0.0134, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:43:02,291 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.29 vs. limit=2.0 2022-12-07 09:43:04,565 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:12,626 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5295, 3.4820, 3.1278, 2.3145, 2.9584, 3.2287, 3.4700, 2.2968], device='cuda:0'), covar=tensor([0.0365, 0.2073, 0.0793, 0.2206, 0.0962, 0.0347, 0.0734, 0.1706], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0225, 0.0103, 0.0134, 0.0091, 0.0087, 0.0079, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:43:18,562 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:25,113 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8199, 2.7026, 2.5015, 1.3747, 2.4236, 2.4402, 2.8894, 2.0976], device='cuda:0'), covar=tensor([0.0646, 0.2476, 0.0991, 0.3093, 0.1009, 0.0652, 0.0530, 0.1776], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0228, 0.0104, 0.0135, 0.0091, 0.0087, 0.0078, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:43:34,701 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:43:41,035 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.777e+02 3.357e+02 4.403e+02 6.655e+02, threshold=6.715e+02, percent-clipped=0.0 2022-12-07 09:43:49,959 INFO [train.py:873] (0/4) Epoch 3, batch 6300, loss[loss=0.2119, simple_loss=0.2127, pruned_loss=0.1056, over 14235.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2085, pruned_loss=0.1003, over 1987491.65 frames. ], batch size: 94, lr: 2.22e-02, grad_scale: 8.0 2022-12-07 09:44:18,800 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8468, 1.3472, 1.9608, 1.2215, 2.0644, 2.0133, 1.5480, 2.0582], device='cuda:0'), covar=tensor([0.0174, 0.0995, 0.0170, 0.1124, 0.0181, 0.0270, 0.0447, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0149, 0.0111, 0.0158, 0.0130, 0.0118, 0.0101, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:44:32,696 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:44:42,737 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0880, 4.6344, 4.4699, 5.0648, 4.7686, 4.2621, 5.0678, 4.2907], device='cuda:0'), covar=tensor([0.0251, 0.0909, 0.0242, 0.0342, 0.0634, 0.0427, 0.0409, 0.0454], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0183, 0.0119, 0.0111, 0.0115, 0.0101, 0.0171, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:44:48,406 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6496, 1.2433, 1.2263, 1.2228, 1.0193, 1.1257, 1.0051, 0.7749], device='cuda:0'), covar=tensor([0.1883, 0.0532, 0.0314, 0.0323, 0.0712, 0.0206, 0.0920, 0.0931], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0061, 0.0049, 0.0056, 0.0067, 0.0056, 0.0075, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:45:07,700 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.996e+02 3.929e+02 5.404e+02 9.451e+02, threshold=7.857e+02, percent-clipped=12.0 2022-12-07 09:45:16,313 INFO [train.py:873] (0/4) Epoch 3, batch 6400, loss[loss=0.1884, simple_loss=0.1677, pruned_loss=0.1046, over 2578.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2085, pruned_loss=0.09982, over 1967992.54 frames. ], batch size: 100, lr: 2.22e-02, grad_scale: 8.0 2022-12-07 09:46:13,391 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2701, 2.8906, 3.8561, 2.8960, 4.0143, 3.8602, 3.5530, 3.0773], device='cuda:0'), covar=tensor([0.0083, 0.1022, 0.0139, 0.0672, 0.0152, 0.0198, 0.0754, 0.0853], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0340, 0.0225, 0.0291, 0.0254, 0.0240, 0.0267, 0.0357], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 09:46:16,505 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:46:32,405 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:46:34,034 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.688e+02 3.501e+02 4.661e+02 7.945e+02, threshold=7.001e+02, percent-clipped=1.0 2022-12-07 09:46:43,049 INFO [train.py:873] (0/4) Epoch 3, batch 6500, loss[loss=0.1557, simple_loss=0.1796, pruned_loss=0.06594, over 13873.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2097, pruned_loss=0.1009, over 1999965.02 frames. ], batch size: 20, lr: 2.21e-02, grad_scale: 8.0 2022-12-07 09:47:09,498 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:17,607 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 09:47:19,621 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:22,184 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7419, 2.4680, 2.6170, 2.7615, 2.7324, 2.7229, 2.8873, 2.4046], device='cuda:0'), covar=tensor([0.0619, 0.1651, 0.0514, 0.0637, 0.0935, 0.0649, 0.0691, 0.0752], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0188, 0.0125, 0.0115, 0.0121, 0.0106, 0.0175, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:47:24,786 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:33,140 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:47:54,430 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:00,153 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 3.136e+02 4.193e+02 5.448e+02 1.293e+03, threshold=8.387e+02, percent-clipped=8.0 2022-12-07 09:48:08,228 INFO [train.py:873] (0/4) Epoch 3, batch 6600, loss[loss=0.2261, simple_loss=0.2185, pruned_loss=0.1169, over 11947.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2091, pruned_loss=0.1011, over 2003370.50 frames. ], batch size: 100, lr: 2.21e-02, grad_scale: 8.0 2022-12-07 09:48:17,901 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8751, 1.5568, 2.0316, 1.7919, 2.1738, 1.8117, 1.8433, 1.9174], device='cuda:0'), covar=tensor([0.0063, 0.0313, 0.0047, 0.0080, 0.0047, 0.0068, 0.0040, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0344, 0.0232, 0.0297, 0.0257, 0.0240, 0.0267, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 09:48:34,764 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:40,812 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:48:51,193 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:05,084 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 09:49:13,083 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-07 09:49:27,547 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 09:49:27,684 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.849e+02 3.705e+02 5.166e+02 8.548e+02, threshold=7.409e+02, percent-clipped=1.0 2022-12-07 09:49:34,262 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:36,091 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:49:36,765 INFO [train.py:873] (0/4) Epoch 3, batch 6700, loss[loss=0.2001, simple_loss=0.2024, pruned_loss=0.09887, over 14313.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.209, pruned_loss=0.1012, over 1933450.95 frames. ], batch size: 46, lr: 2.20e-02, grad_scale: 8.0 2022-12-07 09:49:57,642 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:50:13,205 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7709, 2.6506, 2.3241, 2.4221, 2.6406, 2.6420, 2.7761, 2.6616], device='cuda:0'), covar=tensor([0.0711, 0.0758, 0.1662, 0.2252, 0.0791, 0.0666, 0.0885, 0.0957], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0195, 0.0272, 0.0359, 0.0222, 0.0252, 0.0266, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:50:31,560 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 09:50:47,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 09:50:49,124 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1301, 1.5799, 2.3801, 1.9750, 2.3663, 1.6451, 1.7834, 2.1141], device='cuda:0'), covar=tensor([0.0950, 0.2046, 0.0234, 0.2654, 0.0136, 0.1556, 0.1190, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0306, 0.0182, 0.0409, 0.0176, 0.0319, 0.0279, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:50:51,647 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:50:54,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.679e+02 3.700e+02 5.034e+02 8.925e+02, threshold=7.401e+02, percent-clipped=4.0 2022-12-07 09:51:03,452 INFO [train.py:873] (0/4) Epoch 3, batch 6800, loss[loss=0.213, simple_loss=0.2219, pruned_loss=0.102, over 14484.00 frames. ], tot_loss[loss=0.207, simple_loss=0.21, pruned_loss=0.102, over 1917948.97 frames. ], batch size: 49, lr: 2.20e-02, grad_scale: 8.0 2022-12-07 09:51:26,005 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:40,271 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:41,082 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:41,966 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:51:54,850 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:19,491 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8006, 1.9209, 2.4096, 1.6092, 1.6997, 2.1208, 1.1808, 2.0041], device='cuda:0'), covar=tensor([0.1432, 0.1351, 0.0952, 0.2117, 0.2012, 0.0765, 0.4960, 0.0996], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0066, 0.0069, 0.0074, 0.0087, 0.0060, 0.0135, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:52:21,335 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.962e+02 3.808e+02 5.028e+02 7.344e+02, threshold=7.616e+02, percent-clipped=0.0 2022-12-07 09:52:22,230 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:29,726 INFO [train.py:873] (0/4) Epoch 3, batch 6900, loss[loss=0.1928, simple_loss=0.1835, pruned_loss=0.1011, over 3881.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2085, pruned_loss=0.1, over 1951920.17 frames. ], batch size: 100, lr: 2.19e-02, grad_scale: 8.0 2022-12-07 09:52:35,053 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:52:35,738 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:53:16,764 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5089, 5.0472, 5.1364, 5.6585, 5.4446, 4.7738, 5.8056, 5.5952], device='cuda:0'), covar=tensor([0.0550, 0.0514, 0.0527, 0.0509, 0.0414, 0.0322, 0.0418, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0080, 0.0098, 0.0096, 0.0107, 0.0074, 0.0103, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 09:53:46,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.783e+02 2.990e+02 4.011e+02 5.296e+02 1.371e+03, threshold=8.022e+02, percent-clipped=6.0 2022-12-07 09:53:50,267 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:53:55,140 INFO [train.py:873] (0/4) Epoch 3, batch 7000, loss[loss=0.2174, simple_loss=0.1989, pruned_loss=0.1179, over 3888.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2092, pruned_loss=0.101, over 1960150.21 frames. ], batch size: 100, lr: 2.19e-02, grad_scale: 8.0 2022-12-07 09:53:59,785 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:54:06,809 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4903, 4.1770, 4.1129, 4.4849, 4.3660, 3.9460, 4.5167, 3.7478], device='cuda:0'), covar=tensor([0.0339, 0.0765, 0.0283, 0.0359, 0.0579, 0.0774, 0.0435, 0.0493], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0189, 0.0123, 0.0113, 0.0121, 0.0103, 0.0179, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:54:39,488 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:54:52,523 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:05,119 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:12,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.681e+02 3.418e+02 4.673e+02 9.486e+02, threshold=6.836e+02, percent-clipped=4.0 2022-12-07 09:55:13,080 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:14,942 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7537, 2.6068, 2.6983, 2.8617, 2.7844, 2.8847, 2.9595, 2.4539], device='cuda:0'), covar=tensor([0.0534, 0.1126, 0.0480, 0.0534, 0.0690, 0.0412, 0.0593, 0.0652], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0187, 0.0123, 0.0115, 0.0122, 0.0101, 0.0176, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:55:21,727 INFO [train.py:873] (0/4) Epoch 3, batch 7100, loss[loss=0.1945, simple_loss=0.1993, pruned_loss=0.0949, over 14147.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.209, pruned_loss=0.1004, over 1960210.76 frames. ], batch size: 29, lr: 2.18e-02, grad_scale: 8.0 2022-12-07 09:55:32,446 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:32,467 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:55:44,091 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:00,008 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:05,978 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:16,577 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6876, 4.4962, 4.1235, 4.2529, 4.2190, 4.5245, 4.6822, 4.6286], device='cuda:0'), covar=tensor([0.0637, 0.0467, 0.1284, 0.1906, 0.0689, 0.0458, 0.0636, 0.0776], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0198, 0.0278, 0.0362, 0.0223, 0.0258, 0.0263, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:56:24,911 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:25,595 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:38,993 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.756e+01 2.795e+02 3.480e+02 4.240e+02 8.610e+02, threshold=6.960e+02, percent-clipped=4.0 2022-12-07 09:56:40,766 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:47,594 INFO [train.py:873] (0/4) Epoch 3, batch 7200, loss[loss=0.2331, simple_loss=0.194, pruned_loss=0.1361, over 1163.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2096, pruned_loss=0.101, over 1985818.69 frames. ], batch size: 100, lr: 2.18e-02, grad_scale: 8.0 2022-12-07 09:56:48,442 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:56:48,510 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6931, 2.6709, 3.6255, 2.4462, 2.5512, 2.8015, 1.2777, 3.0622], device='cuda:0'), covar=tensor([0.0985, 0.1031, 0.0865, 0.1463, 0.1151, 0.1134, 0.3982, 0.1037], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0069, 0.0071, 0.0076, 0.0087, 0.0062, 0.0136, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:57:39,283 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:57:41,904 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8387, 3.0416, 2.3060, 2.2566, 2.8085, 2.9044, 3.0318, 2.8692], device='cuda:0'), covar=tensor([0.1352, 0.0825, 0.2707, 0.3900, 0.1381, 0.1105, 0.1384, 0.1397], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0196, 0.0274, 0.0361, 0.0221, 0.0258, 0.0263, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 09:57:43,122 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2022-12-07 09:57:59,127 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0311, 3.1958, 4.1000, 2.6883, 2.5432, 3.4374, 1.7465, 3.1644], device='cuda:0'), covar=tensor([0.1515, 0.0778, 0.0429, 0.1619, 0.1242, 0.0725, 0.3928, 0.1965], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0068, 0.0071, 0.0078, 0.0089, 0.0061, 0.0138, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 09:58:05,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.951e+02 3.875e+02 4.878e+02 1.117e+03, threshold=7.750e+02, percent-clipped=7.0 2022-12-07 09:58:09,199 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:13,856 INFO [train.py:873] (0/4) Epoch 3, batch 7300, loss[loss=0.2185, simple_loss=0.1933, pruned_loss=0.1219, over 3918.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2083, pruned_loss=0.09972, over 1943333.37 frames. ], batch size: 100, lr: 2.17e-02, grad_scale: 16.0 2022-12-07 09:58:31,604 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:38,278 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2022-12-07 09:58:38,529 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2022-12-07 09:58:49,758 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:58:49,893 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7571, 0.9173, 0.6062, 0.9668, 1.0362, 0.5292, 1.1432, 1.0550], device='cuda:0'), covar=tensor([0.0681, 0.0373, 0.0172, 0.0243, 0.0364, 0.0227, 0.0135, 0.0355], device='cuda:0'), in_proj_covar=tensor([0.0016, 0.0019, 0.0019, 0.0016, 0.0016, 0.0023, 0.0017, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.5021e-05, 5.8313e-05, 5.5734e-05, 5.4686e-05, 5.0590e-05, 6.4504e-05, 5.8582e-05, 5.3775e-05], device='cuda:0') 2022-12-07 09:59:06,146 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:59:23,381 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 09:59:31,052 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.878e+02 3.644e+02 4.636e+02 1.010e+03, threshold=7.288e+02, percent-clipped=2.0 2022-12-07 09:59:38,849 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6464, 2.0381, 2.8854, 2.2272, 2.8558, 2.6845, 2.6039, 2.4046], device='cuda:0'), covar=tensor([0.0127, 0.1104, 0.0196, 0.0557, 0.0180, 0.0174, 0.0275, 0.0912], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0358, 0.0251, 0.0308, 0.0274, 0.0252, 0.0276, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 09:59:39,444 INFO [train.py:873] (0/4) Epoch 3, batch 7400, loss[loss=0.2357, simple_loss=0.2192, pruned_loss=0.1261, over 5996.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2083, pruned_loss=0.1001, over 1986962.26 frames. ], batch size: 100, lr: 2.17e-02, grad_scale: 16.0 2022-12-07 09:59:45,720 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:05,236 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:19,770 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:38,809 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:00:58,627 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.799e+02 3.655e+02 4.825e+02 8.691e+02, threshold=7.309e+02, percent-clipped=2.0 2022-12-07 10:01:06,422 INFO [train.py:873] (0/4) Epoch 3, batch 7500, loss[loss=0.198, simple_loss=0.2107, pruned_loss=0.09267, over 14632.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2087, pruned_loss=0.1009, over 1986476.68 frames. ], batch size: 22, lr: 2.16e-02, grad_scale: 8.0 2022-12-07 10:01:07,383 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:01:18,023 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:01:37,745 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6785, 1.7809, 1.7498, 1.7738, 1.5472, 1.6567, 1.6667, 0.8708], device='cuda:0'), covar=tensor([0.2361, 0.1244, 0.0886, 0.0629, 0.1213, 0.0592, 0.1540, 0.3335], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0060, 0.0051, 0.0054, 0.0070, 0.0054, 0.0076, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:01:44,222 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:01:51,351 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-3.pt 2022-12-07 10:02:33,690 INFO [train.py:873] (0/4) Epoch 4, batch 0, loss[loss=0.2368, simple_loss=0.2339, pruned_loss=0.1199, over 14250.00 frames. ], tot_loss[loss=0.2368, simple_loss=0.2339, pruned_loss=0.1199, over 14250.00 frames. ], batch size: 94, lr: 2.02e-02, grad_scale: 8.0 2022-12-07 10:02:33,691 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 10:02:40,698 INFO [train.py:905] (0/4) Epoch 4, validation: loss=0.1426, simple_loss=0.185, pruned_loss=0.0501, over 857387.00 frames. 2022-12-07 10:02:40,698 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 10:02:52,906 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 10:03:04,316 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1421, 1.2752, 1.4868, 1.0737, 1.0801, 0.9847, 1.4065, 1.0369], device='cuda:0'), covar=tensor([0.0725, 0.0746, 0.0473, 0.0952, 0.0669, 0.0251, 0.0426, 0.0810], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0010, 0.0011, 0.0010, 0.0011, 0.0013, 0.0013, 0.0015], device='cuda:0'), out_proj_covar=tensor([3.8388e-05, 3.5750e-05, 3.9494e-05, 3.6777e-05, 3.8267e-05, 4.1742e-05, 5.1955e-05, 5.2440e-05], device='cuda:0') 2022-12-07 10:03:06,590 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.563e+01 2.095e+02 3.755e+02 5.061e+02 1.365e+03, threshold=7.510e+02, percent-clipped=11.0 2022-12-07 10:03:28,537 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:05,559 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 10:04:07,828 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:08,914 INFO [train.py:873] (0/4) Epoch 4, batch 100, loss[loss=0.1617, simple_loss=0.1634, pruned_loss=0.07996, over 3901.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2081, pruned_loss=0.0979, over 858147.45 frames. ], batch size: 100, lr: 2.02e-02, grad_scale: 8.0 2022-12-07 10:04:14,986 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:33,903 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.273e+02 2.903e+02 3.464e+02 4.472e+02 7.664e+02, threshold=6.927e+02, percent-clipped=2.0 2022-12-07 10:04:41,801 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:48,111 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:49,741 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:04:54,197 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0643, 2.2031, 4.1550, 4.1266, 4.4768, 2.5804, 4.3438, 3.1858], device='cuda:0'), covar=tensor([0.0059, 0.0205, 0.0184, 0.0073, 0.0039, 0.0311, 0.0025, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0162, 0.0212, 0.0175, 0.0149, 0.0207, 0.0114, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:05:01,765 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 10:05:08,146 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:08,446 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.39 vs. limit=2.0 2022-12-07 10:05:22,193 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:29,639 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:35,011 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:05:35,728 INFO [train.py:873] (0/4) Epoch 4, batch 200, loss[loss=0.2007, simple_loss=0.2122, pruned_loss=0.09461, over 14138.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2075, pruned_loss=0.09771, over 1338859.15 frames. ], batch size: 84, lr: 2.01e-02, grad_scale: 8.0 2022-12-07 10:05:38,453 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0424, 1.6216, 2.4544, 1.8823, 2.2365, 1.7029, 1.9077, 2.0960], device='cuda:0'), covar=tensor([0.0823, 0.2011, 0.0219, 0.1846, 0.0124, 0.1790, 0.1220, 0.0483], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0299, 0.0174, 0.0401, 0.0173, 0.0318, 0.0293, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0002, 0.0004, 0.0004, 0.0002], device='cuda:0') 2022-12-07 10:05:40,923 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:00,925 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.534e+02 3.313e+02 4.252e+02 7.669e+02, threshold=6.626e+02, percent-clipped=3.0 2022-12-07 10:06:03,412 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:22,005 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:06:44,737 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6667, 1.4082, 1.7727, 2.0949, 1.3259, 1.7387, 1.8524, 1.8474], device='cuda:0'), covar=tensor([0.0026, 0.0039, 0.0022, 0.0013, 0.0037, 0.0052, 0.0019, 0.0018], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0162, 0.0215, 0.0177, 0.0150, 0.0205, 0.0115, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:07:01,607 INFO [train.py:873] (0/4) Epoch 4, batch 300, loss[loss=0.2038, simple_loss=0.208, pruned_loss=0.09983, over 14031.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2074, pruned_loss=0.09778, over 1635551.30 frames. ], batch size: 22, lr: 2.01e-02, grad_scale: 8.0 2022-12-07 10:07:08,464 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:07:26,196 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.872e+01 2.724e+02 3.752e+02 4.773e+02 9.337e+02, threshold=7.504e+02, percent-clipped=9.0 2022-12-07 10:07:46,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 10:07:48,223 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:08:09,989 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.63 vs. limit=5.0 2022-12-07 10:08:16,624 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7629, 1.8010, 1.7794, 1.9055, 1.5182, 1.7611, 1.7776, 0.8869], device='cuda:0'), covar=tensor([0.2716, 0.1004, 0.1305, 0.0590, 0.1203, 0.0624, 0.1430, 0.3235], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0062, 0.0054, 0.0053, 0.0071, 0.0056, 0.0079, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:08:27,271 INFO [train.py:873] (0/4) Epoch 4, batch 400, loss[loss=0.2468, simple_loss=0.2305, pruned_loss=0.1316, over 11207.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2069, pruned_loss=0.09738, over 1812901.23 frames. ], batch size: 100, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:08:29,391 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:08:46,014 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:08:52,656 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.809e+02 3.789e+02 4.581e+02 1.009e+03, threshold=7.577e+02, percent-clipped=4.0 2022-12-07 10:09:05,869 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.48 vs. limit=5.0 2022-12-07 10:09:09,789 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5607, 1.8540, 3.7468, 3.6740, 3.6860, 2.2241, 3.5596, 2.6797], device='cuda:0'), covar=tensor([0.0071, 0.0223, 0.0214, 0.0111, 0.0059, 0.0332, 0.0043, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0160, 0.0211, 0.0177, 0.0149, 0.0204, 0.0114, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:09:22,132 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:23,963 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 10:09:27,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.55 vs. limit=2.0 2022-12-07 10:09:38,571 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:48,757 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:09:54,463 INFO [train.py:873] (0/4) Epoch 4, batch 500, loss[loss=0.1696, simple_loss=0.1895, pruned_loss=0.07488, over 14651.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2069, pruned_loss=0.09769, over 1887423.05 frames. ], batch size: 33, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:10:19,852 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.033e+01 2.776e+02 3.736e+02 4.463e+02 9.726e+02, threshold=7.473e+02, percent-clipped=4.0 2022-12-07 10:10:20,887 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:10:31,649 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:13,490 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:20,516 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.94 vs. limit=5.0 2022-12-07 10:11:20,902 INFO [train.py:873] (0/4) Epoch 4, batch 600, loss[loss=0.1357, simple_loss=0.128, pruned_loss=0.07169, over 1240.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2068, pruned_loss=0.09777, over 1951157.93 frames. ], batch size: 100, lr: 2.00e-02, grad_scale: 8.0 2022-12-07 10:11:23,683 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:27,802 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:11:45,413 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.720e+02 3.786e+02 4.640e+02 1.060e+03, threshold=7.571e+02, percent-clipped=5.0 2022-12-07 10:12:08,381 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:12:46,507 INFO [train.py:873] (0/4) Epoch 4, batch 700, loss[loss=0.2054, simple_loss=0.1734, pruned_loss=0.1187, over 1217.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2063, pruned_loss=0.0976, over 1929503.44 frames. ], batch size: 100, lr: 1.99e-02, grad_scale: 8.0 2022-12-07 10:13:11,266 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.855e+02 3.544e+02 4.274e+02 7.481e+02, threshold=7.089e+02, percent-clipped=0.0 2022-12-07 10:13:40,589 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:13:52,681 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:07,436 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:09,179 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5553, 2.4571, 2.1856, 1.3222, 2.0027, 2.1163, 2.5591, 1.9402], device='cuda:0'), covar=tensor([0.0442, 0.2293, 0.1107, 0.2612, 0.1062, 0.0537, 0.0756, 0.1721], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0227, 0.0106, 0.0129, 0.0097, 0.0089, 0.0080, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:14:12,389 INFO [train.py:873] (0/4) Epoch 4, batch 800, loss[loss=0.2128, simple_loss=0.2172, pruned_loss=0.1041, over 13954.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2065, pruned_loss=0.09838, over 1921190.57 frames. ], batch size: 23, lr: 1.99e-02, grad_scale: 8.0 2022-12-07 10:14:13,311 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:22,039 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:14:37,484 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 3.074e+02 3.598e+02 4.736e+02 1.088e+03, threshold=7.196e+02, percent-clipped=3.0 2022-12-07 10:14:48,550 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:05,919 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 10:15:26,577 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:36,429 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:15:38,302 INFO [train.py:873] (0/4) Epoch 4, batch 900, loss[loss=0.1798, simple_loss=0.1757, pruned_loss=0.09193, over 3879.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2061, pruned_loss=0.09778, over 1964736.48 frames. ], batch size: 100, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:16:03,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.303e+02 2.960e+02 3.909e+02 5.159e+02 1.247e+03, threshold=7.817e+02, percent-clipped=4.0 2022-12-07 10:16:11,816 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8881, 1.6371, 2.1167, 1.8178, 2.1756, 1.8435, 1.8687, 1.9946], device='cuda:0'), covar=tensor([0.0096, 0.0561, 0.0061, 0.0115, 0.0059, 0.0104, 0.0074, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0351, 0.0258, 0.0305, 0.0286, 0.0246, 0.0281, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:16:57,104 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2022-12-07 10:17:03,223 INFO [train.py:873] (0/4) Epoch 4, batch 1000, loss[loss=0.2131, simple_loss=0.2211, pruned_loss=0.1026, over 14189.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.206, pruned_loss=0.09792, over 1940747.40 frames. ], batch size: 84, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:17:17,887 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8944, 1.5061, 3.0691, 1.5253, 3.0746, 3.0349, 2.1086, 3.1852], device='cuda:0'), covar=tensor([0.0161, 0.1967, 0.0209, 0.1713, 0.0231, 0.0278, 0.0611, 0.0135], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0148, 0.0115, 0.0160, 0.0130, 0.0127, 0.0109, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:17:28,303 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 2.746e+02 3.575e+02 4.376e+02 8.279e+02, threshold=7.149e+02, percent-clipped=2.0 2022-12-07 10:18:09,542 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:18:20,752 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2022-12-07 10:18:28,811 INFO [train.py:873] (0/4) Epoch 4, batch 1100, loss[loss=0.1957, simple_loss=0.184, pruned_loss=0.1037, over 3841.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2057, pruned_loss=0.09863, over 1876024.48 frames. ], batch size: 100, lr: 1.98e-02, grad_scale: 8.0 2022-12-07 10:18:50,542 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:18:53,787 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.927e+02 4.016e+02 5.307e+02 1.051e+03, threshold=8.031e+02, percent-clipped=8.0 2022-12-07 10:19:18,359 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:19:43,596 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:19:53,907 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:19:55,484 INFO [train.py:873] (0/4) Epoch 4, batch 1200, loss[loss=0.2162, simple_loss=0.2135, pruned_loss=0.1095, over 14174.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2059, pruned_loss=0.09812, over 1876633.89 frames. ], batch size: 99, lr: 1.97e-02, grad_scale: 8.0 2022-12-07 10:20:11,316 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=15.76 vs. limit=5.0 2022-12-07 10:20:21,446 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.900e+02 3.609e+02 4.734e+02 8.045e+02, threshold=7.219e+02, percent-clipped=1.0 2022-12-07 10:20:24,876 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:20:34,945 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:21:07,434 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7624, 3.4873, 3.4790, 3.9074, 3.3361, 3.0599, 3.8846, 3.8290], device='cuda:0'), covar=tensor([0.0750, 0.0656, 0.0655, 0.0523, 0.0976, 0.0627, 0.0580, 0.0643], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0083, 0.0104, 0.0101, 0.0110, 0.0076, 0.0111, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:21:20,670 INFO [train.py:873] (0/4) Epoch 4, batch 1300, loss[loss=0.2067, simple_loss=0.1753, pruned_loss=0.1191, over 1323.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2054, pruned_loss=0.0979, over 1860930.13 frames. ], batch size: 100, lr: 1.97e-02, grad_scale: 4.0 2022-12-07 10:21:28,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 10:21:42,261 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 10:21:46,583 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.737e+02 3.421e+02 4.208e+02 7.387e+02, threshold=6.842e+02, percent-clipped=1.0 2022-12-07 10:21:48,798 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0315, 2.2247, 3.1996, 2.4891, 3.0969, 2.8862, 2.8210, 2.4516], device='cuda:0'), covar=tensor([0.0126, 0.1315, 0.0203, 0.0792, 0.0274, 0.0241, 0.0676, 0.1095], device='cuda:0'), in_proj_covar=tensor([0.0233, 0.0344, 0.0256, 0.0302, 0.0287, 0.0243, 0.0277, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:22:04,350 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:22:35,910 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8266, 1.2368, 1.1687, 1.1097, 0.9541, 1.2034, 1.1862, 0.8897], device='cuda:0'), covar=tensor([0.2568, 0.0892, 0.0432, 0.0435, 0.0891, 0.0270, 0.0929, 0.0902], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0059, 0.0053, 0.0052, 0.0069, 0.0055, 0.0078, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:22:46,181 INFO [train.py:873] (0/4) Epoch 4, batch 1400, loss[loss=0.256, simple_loss=0.2156, pruned_loss=0.1482, over 1259.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2053, pruned_loss=0.09676, over 1912265.69 frames. ], batch size: 100, lr: 1.96e-02, grad_scale: 4.0 2022-12-07 10:22:55,448 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:23:12,193 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.769e+02 3.319e+02 4.343e+02 7.993e+02, threshold=6.638e+02, percent-clipped=1.0 2022-12-07 10:23:23,818 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 10:23:35,089 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 10:24:11,898 INFO [train.py:873] (0/4) Epoch 4, batch 1500, loss[loss=0.1591, simple_loss=0.1835, pruned_loss=0.06736, over 13924.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2048, pruned_loss=0.09708, over 1937859.27 frames. ], batch size: 26, lr: 1.96e-02, grad_scale: 4.0 2022-12-07 10:24:16,761 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:24:38,553 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.634e+02 3.340e+02 4.470e+02 8.988e+02, threshold=6.680e+02, percent-clipped=4.0 2022-12-07 10:24:44,564 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 10:24:45,602 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1425, 1.3526, 3.2415, 1.4729, 2.9900, 3.2408, 2.2746, 3.3524], device='cuda:0'), covar=tensor([0.0177, 0.2313, 0.0213, 0.1923, 0.0633, 0.0274, 0.0750, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0148, 0.0118, 0.0160, 0.0130, 0.0126, 0.0110, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:25:16,775 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.15 vs. limit=5.0 2022-12-07 10:25:19,004 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2022-12-07 10:25:39,219 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.02 vs. limit=5.0 2022-12-07 10:25:39,350 INFO [train.py:873] (0/4) Epoch 4, batch 1600, loss[loss=0.2765, simple_loss=0.2218, pruned_loss=0.1656, over 1341.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.204, pruned_loss=0.09642, over 1921490.29 frames. ], batch size: 100, lr: 1.96e-02, grad_scale: 8.0 2022-12-07 10:26:05,382 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.687e+02 3.404e+02 4.412e+02 1.604e+03, threshold=6.807e+02, percent-clipped=2.0 2022-12-07 10:26:21,307 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1514, 2.8798, 2.9258, 3.2053, 3.0822, 3.1732, 3.2585, 2.6919], device='cuda:0'), covar=tensor([0.0455, 0.1016, 0.0428, 0.0444, 0.0734, 0.0347, 0.0557, 0.0594], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0187, 0.0125, 0.0116, 0.0123, 0.0103, 0.0180, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:27:05,336 INFO [train.py:873] (0/4) Epoch 4, batch 1700, loss[loss=0.2232, simple_loss=0.2149, pruned_loss=0.1157, over 8648.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2038, pruned_loss=0.0954, over 2004442.37 frames. ], batch size: 100, lr: 1.95e-02, grad_scale: 8.0 2022-12-07 10:27:10,500 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:27:31,011 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.793e+02 3.554e+02 4.451e+02 1.467e+03, threshold=7.108e+02, percent-clipped=5.0 2022-12-07 10:28:02,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7707, 3.5078, 3.3933, 3.8267, 3.7578, 3.4910, 3.8169, 3.3183], device='cuda:0'), covar=tensor([0.0433, 0.0799, 0.0383, 0.0354, 0.0524, 0.0753, 0.0498, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0187, 0.0125, 0.0115, 0.0124, 0.0102, 0.0180, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:28:31,821 INFO [train.py:873] (0/4) Epoch 4, batch 1800, loss[loss=0.1849, simple_loss=0.1984, pruned_loss=0.08568, over 14293.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2049, pruned_loss=0.09683, over 1989421.97 frames. ], batch size: 39, lr: 1.95e-02, grad_scale: 8.0 2022-12-07 10:28:52,426 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4009, 2.2364, 2.9656, 2.0141, 2.0953, 2.3761, 1.1764, 2.5279], device='cuda:0'), covar=tensor([0.1641, 0.1062, 0.1076, 0.2294, 0.2199, 0.1367, 0.5696, 0.1237], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0073, 0.0066, 0.0080, 0.0093, 0.0064, 0.0138, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:28:57,764 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.645e+02 3.516e+02 4.741e+02 1.240e+03, threshold=7.031e+02, percent-clipped=7.0 2022-12-07 10:29:23,682 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:29:43,025 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:29:49,406 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-07 10:29:58,458 INFO [train.py:873] (0/4) Epoch 4, batch 1900, loss[loss=0.2005, simple_loss=0.1893, pruned_loss=0.1059, over 4943.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2046, pruned_loss=0.09691, over 1952859.54 frames. ], batch size: 100, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:30:04,700 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5772, 2.2685, 3.2556, 2.1490, 2.1682, 2.3126, 1.3229, 2.7698], device='cuda:0'), covar=tensor([0.1407, 0.0988, 0.0503, 0.1812, 0.1785, 0.1820, 0.4324, 0.0894], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0069, 0.0065, 0.0080, 0.0093, 0.0063, 0.0136, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:30:17,380 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:30:24,686 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.734e+02 3.498e+02 4.101e+02 8.612e+02, threshold=6.995e+02, percent-clipped=1.0 2022-12-07 10:30:36,031 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:31:19,151 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8274, 3.3161, 2.9031, 2.2101, 2.8282, 2.9243, 3.2864, 2.2171], device='cuda:0'), covar=tensor([0.0383, 0.3949, 0.1182, 0.2981, 0.1370, 0.0512, 0.1585, 0.1924], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0232, 0.0110, 0.0136, 0.0101, 0.0094, 0.0082, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:31:24,623 INFO [train.py:873] (0/4) Epoch 4, batch 2000, loss[loss=0.2522, simple_loss=0.2139, pruned_loss=0.1452, over 1286.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2055, pruned_loss=0.09775, over 1938597.20 frames. ], batch size: 100, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:31:29,955 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:31:50,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.930e+02 3.788e+02 4.800e+02 1.678e+03, threshold=7.576e+02, percent-clipped=5.0 2022-12-07 10:32:02,158 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1383, 2.9922, 2.3651, 2.6104, 1.7453, 2.8140, 2.8846, 1.1373], device='cuda:0'), covar=tensor([0.2831, 0.0872, 0.3037, 0.1131, 0.1261, 0.0404, 0.1482, 0.3305], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0059, 0.0052, 0.0053, 0.0068, 0.0055, 0.0078, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:32:06,372 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2730, 1.0152, 1.0129, 1.1994, 1.0547, 0.6097, 1.1096, 0.7516], device='cuda:0'), covar=tensor([0.0321, 0.0290, 0.0513, 0.0198, 0.0455, 0.0275, 0.0174, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0011, 0.0011, 0.0009, 0.0012, 0.0014, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.2565e-05, 4.1314e-05, 4.5458e-05, 3.7054e-05, 4.3729e-05, 5.0261e-05, 5.2727e-05, 6.0853e-05], device='cuda:0') 2022-12-07 10:32:11,142 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:32:14,276 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6844, 2.4781, 3.5193, 2.7100, 3.5061, 3.3308, 3.1893, 2.8675], device='cuda:0'), covar=tensor([0.0137, 0.1514, 0.0246, 0.0894, 0.0296, 0.0272, 0.0962, 0.1159], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0365, 0.0268, 0.0310, 0.0291, 0.0256, 0.0291, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:32:19,854 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-07 10:32:32,582 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:32:35,039 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9758, 1.5760, 3.7334, 3.5256, 3.6209, 3.7201, 3.0892, 3.7945], device='cuda:0'), covar=tensor([0.1005, 0.1114, 0.0065, 0.0129, 0.0115, 0.0077, 0.0150, 0.0077], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0147, 0.0086, 0.0121, 0.0104, 0.0107, 0.0078, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:32:51,148 INFO [train.py:873] (0/4) Epoch 4, batch 2100, loss[loss=0.1699, simple_loss=0.1829, pruned_loss=0.0785, over 13891.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2045, pruned_loss=0.09604, over 1982085.44 frames. ], batch size: 23, lr: 1.94e-02, grad_scale: 8.0 2022-12-07 10:33:17,442 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.608e+02 3.300e+02 4.069e+02 1.028e+03, threshold=6.599e+02, percent-clipped=1.0 2022-12-07 10:33:25,487 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:33:44,868 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:18,569 INFO [train.py:873] (0/4) Epoch 4, batch 2200, loss[loss=0.1978, simple_loss=0.2064, pruned_loss=0.09464, over 14391.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2037, pruned_loss=0.09528, over 1961861.07 frames. ], batch size: 41, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:34:31,860 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:37,033 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:43,259 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1233, 1.8572, 2.0646, 1.3038, 1.8114, 1.9211, 2.0723, 1.9003], device='cuda:0'), covar=tensor([0.0372, 0.1464, 0.0845, 0.2096, 0.0754, 0.0588, 0.0411, 0.1172], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0220, 0.0106, 0.0127, 0.0097, 0.0088, 0.0076, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:34:43,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.748e+02 3.566e+02 4.781e+02 9.343e+02, threshold=7.131e+02, percent-clipped=9.0 2022-12-07 10:34:50,881 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:34:54,352 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0682, 0.8090, 1.0461, 1.0260, 0.9647, 0.6376, 1.1098, 1.0419], device='cuda:0'), covar=tensor([0.0487, 0.0435, 0.0288, 0.0623, 0.0293, 0.0268, 0.0524, 0.0424], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0018, 0.0019, 0.0016, 0.0016, 0.0022, 0.0019, 0.0018], device='cuda:0'), out_proj_covar=tensor([5.4682e-05, 5.9442e-05, 5.8502e-05, 5.7045e-05, 5.4030e-05, 6.8223e-05, 6.4242e-05, 5.8101e-05], device='cuda:0') 2022-12-07 10:34:59,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6013, 0.6238, 0.7003, 0.6719, 0.6473, 0.3507, 0.4141, 0.5609], device='cuda:0'), covar=tensor([0.0126, 0.0102, 0.0180, 0.0073, 0.0236, 0.0360, 0.0166, 0.0339], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0011, 0.0011, 0.0010, 0.0011, 0.0014, 0.0013, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.3343e-05, 4.1820e-05, 4.5172e-05, 3.8004e-05, 4.3005e-05, 5.0987e-05, 5.6184e-05, 6.2665e-05], device='cuda:0') 2022-12-07 10:35:44,684 INFO [train.py:873] (0/4) Epoch 4, batch 2300, loss[loss=0.1844, simple_loss=0.1816, pruned_loss=0.09356, over 3892.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2033, pruned_loss=0.09454, over 2006175.46 frames. ], batch size: 100, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:35:51,861 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:35:56,368 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-25000.pt 2022-12-07 10:36:14,179 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.734e+02 3.348e+02 4.422e+02 1.416e+03, threshold=6.696e+02, percent-clipped=4.0 2022-12-07 10:36:38,910 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7379, 0.7234, 0.7630, 0.7551, 0.6398, 0.3123, 0.7253, 0.6437], device='cuda:0'), covar=tensor([0.0134, 0.0060, 0.0107, 0.0064, 0.0176, 0.0207, 0.0092, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0011, 0.0011, 0.0009, 0.0011, 0.0014, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([4.3679e-05, 4.0567e-05, 4.4552e-05, 3.6554e-05, 4.2279e-05, 5.1055e-05, 5.4855e-05, 6.1132e-05], device='cuda:0') 2022-12-07 10:36:40,982 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 10:36:48,417 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:36:49,465 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 10:37:11,675 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 10:37:15,452 INFO [train.py:873] (0/4) Epoch 4, batch 2400, loss[loss=0.1796, simple_loss=0.1617, pruned_loss=0.09879, over 2647.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.204, pruned_loss=0.09539, over 2018137.80 frames. ], batch size: 100, lr: 1.93e-02, grad_scale: 8.0 2022-12-07 10:37:41,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.617e+02 3.623e+02 4.580e+02 1.018e+03, threshold=7.246e+02, percent-clipped=1.0 2022-12-07 10:37:44,892 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:38:41,250 INFO [train.py:873] (0/4) Epoch 4, batch 2500, loss[loss=0.2064, simple_loss=0.214, pruned_loss=0.09939, over 14420.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2043, pruned_loss=0.09575, over 1981658.83 frames. ], batch size: 73, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:38:42,328 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9445, 2.8965, 2.0772, 3.2391, 2.8955, 3.1503, 2.5660, 2.2437], device='cuda:0'), covar=tensor([0.0262, 0.0822, 0.3257, 0.0203, 0.0400, 0.0365, 0.0808, 0.3035], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0285, 0.0328, 0.0187, 0.0231, 0.0224, 0.0261, 0.0318], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:38:55,807 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:38:56,513 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:07,387 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.934e+02 3.822e+02 4.724e+02 8.736e+02, threshold=7.644e+02, percent-clipped=3.0 2022-12-07 10:39:14,162 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:15,438 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 10:39:37,039 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:39:49,581 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9860, 0.8411, 0.4135, 1.0994, 0.8760, 0.2292, 0.7480, 0.9182], device='cuda:0'), covar=tensor([0.0114, 0.0156, 0.0066, 0.0080, 0.0086, 0.0048, 0.0259, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0019, 0.0020, 0.0016, 0.0016, 0.0022, 0.0019, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.6649e-05, 6.1364e-05, 6.0042e-05, 5.7094e-05, 5.3800e-05, 6.9801e-05, 6.5422e-05, 5.6262e-05], device='cuda:0') 2022-12-07 10:39:56,056 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:40:07,854 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0268, 3.7743, 3.7857, 4.1644, 4.0489, 3.6031, 4.2069, 3.5326], device='cuda:0'), covar=tensor([0.0534, 0.1131, 0.0325, 0.0395, 0.0633, 0.0936, 0.0474, 0.0521], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0195, 0.0128, 0.0121, 0.0128, 0.0104, 0.0189, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:40:08,587 INFO [train.py:873] (0/4) Epoch 4, batch 2600, loss[loss=0.2143, simple_loss=0.2178, pruned_loss=0.1053, over 11181.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2034, pruned_loss=0.09527, over 1963415.64 frames. ], batch size: 100, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:40:34,420 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.970e+01 2.618e+02 3.461e+02 4.480e+02 1.117e+03, threshold=6.922e+02, percent-clipped=3.0 2022-12-07 10:40:58,611 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7395, 4.4872, 4.9369, 3.9936, 4.8229, 5.1820, 1.5199, 4.5363], device='cuda:0'), covar=tensor([0.0144, 0.0244, 0.0383, 0.0317, 0.0183, 0.0100, 0.3276, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0123, 0.0111, 0.0097, 0.0154, 0.0109, 0.0143, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:41:03,025 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:41:33,622 INFO [train.py:873] (0/4) Epoch 4, batch 2700, loss[loss=0.1907, simple_loss=0.2008, pruned_loss=0.09026, over 14270.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2041, pruned_loss=0.09599, over 1928389.98 frames. ], batch size: 76, lr: 1.92e-02, grad_scale: 8.0 2022-12-07 10:41:59,633 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.867e+02 3.536e+02 4.473e+02 9.787e+02, threshold=7.071e+02, percent-clipped=5.0 2022-12-07 10:42:01,633 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3517, 2.1414, 4.5083, 3.0194, 4.2380, 2.2661, 3.1575, 4.0943], device='cuda:0'), covar=tensor([0.0284, 0.5107, 0.0171, 0.9122, 0.0263, 0.3686, 0.1160, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0286, 0.0169, 0.0384, 0.0176, 0.0298, 0.0266, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:42:03,224 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:42:09,235 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:42:44,230 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:43:00,507 INFO [train.py:873] (0/4) Epoch 4, batch 2800, loss[loss=0.1975, simple_loss=0.1771, pruned_loss=0.1089, over 2640.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2033, pruned_loss=0.09493, over 1928747.59 frames. ], batch size: 100, lr: 1.91e-02, grad_scale: 8.0 2022-12-07 10:43:02,427 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:43:14,999 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:43:25,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.496e+02 3.123e+02 4.474e+02 9.387e+02, threshold=6.246e+02, percent-clipped=3.0 2022-12-07 10:43:55,768 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:44:26,291 INFO [train.py:873] (0/4) Epoch 4, batch 2900, loss[loss=0.1912, simple_loss=0.202, pruned_loss=0.09026, over 14649.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2024, pruned_loss=0.09371, over 1937907.23 frames. ], batch size: 33, lr: 1.91e-02, grad_scale: 8.0 2022-12-07 10:44:52,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 2.734e+02 3.612e+02 4.564e+02 7.787e+02, threshold=7.225e+02, percent-clipped=1.0 2022-12-07 10:45:01,541 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5133, 1.5826, 2.8091, 2.1263, 2.7376, 1.7899, 2.0656, 2.5344], device='cuda:0'), covar=tensor([0.0595, 0.4943, 0.0296, 0.7159, 0.0337, 0.4062, 0.1870, 0.0283], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0283, 0.0170, 0.0378, 0.0174, 0.0292, 0.0269, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:45:22,127 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:45:48,370 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1619, 2.6696, 4.0335, 2.9463, 3.8822, 3.8003, 3.7212, 3.3551], device='cuda:0'), covar=tensor([0.0178, 0.1562, 0.0263, 0.0883, 0.0363, 0.0252, 0.1241, 0.1256], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0360, 0.0277, 0.0315, 0.0301, 0.0253, 0.0298, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:45:53,686 INFO [train.py:873] (0/4) Epoch 4, batch 3000, loss[loss=0.2082, simple_loss=0.2172, pruned_loss=0.09957, over 14280.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2036, pruned_loss=0.0946, over 2066380.86 frames. ], batch size: 66, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:45:53,687 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 10:46:03,274 INFO [train.py:905] (0/4) Epoch 4, validation: loss=0.1268, simple_loss=0.1698, pruned_loss=0.0419, over 857387.00 frames. 2022-12-07 10:46:03,274 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 10:46:14,146 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:46:29,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.570e+02 3.689e+02 4.619e+02 9.776e+02, threshold=7.378e+02, percent-clipped=3.0 2022-12-07 10:46:34,248 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3619, 2.0973, 4.3097, 3.0396, 4.2442, 2.1911, 3.0737, 3.9439], device='cuda:0'), covar=tensor([0.0221, 0.4894, 0.0263, 0.8657, 0.0163, 0.3739, 0.1281, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0281, 0.0174, 0.0380, 0.0176, 0.0294, 0.0272, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:46:35,849 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8730, 0.8316, 0.6811, 0.8163, 0.7285, 0.1634, 0.8700, 0.8230], device='cuda:0'), covar=tensor([0.0123, 0.0104, 0.0123, 0.0166, 0.0064, 0.0043, 0.0312, 0.0222], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0018, 0.0020, 0.0016, 0.0016, 0.0022, 0.0018, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.5599e-05, 5.9864e-05, 5.9584e-05, 5.7754e-05, 5.4295e-05, 7.0680e-05, 6.4950e-05, 5.7295e-05], device='cuda:0') 2022-12-07 10:46:51,610 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3139, 2.1323, 2.0863, 1.1822, 2.0590, 2.3655, 2.3174, 1.9439], device='cuda:0'), covar=tensor([0.0640, 0.2419, 0.1518, 0.4215, 0.1180, 0.0545, 0.0695, 0.1909], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0226, 0.0108, 0.0133, 0.0101, 0.0090, 0.0080, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:47:04,015 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6681, 1.8331, 3.8743, 2.6439, 3.7754, 1.7782, 2.8232, 3.2646], device='cuda:0'), covar=tensor([0.0330, 0.4575, 0.0276, 0.7350, 0.0206, 0.4101, 0.1194, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0280, 0.0174, 0.0383, 0.0175, 0.0294, 0.0272, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:47:28,458 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:47:30,953 INFO [train.py:873] (0/4) Epoch 4, batch 3100, loss[loss=0.1873, simple_loss=0.189, pruned_loss=0.09284, over 6896.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2027, pruned_loss=0.09385, over 2031152.36 frames. ], batch size: 100, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:47:37,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2770, 4.0628, 4.5122, 3.9113, 4.1770, 4.4696, 1.4773, 3.9967], device='cuda:0'), covar=tensor([0.0139, 0.0228, 0.0271, 0.0318, 0.0256, 0.0140, 0.2856, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0125, 0.0115, 0.0098, 0.0152, 0.0106, 0.0144, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:47:42,338 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0476, 2.0658, 1.9590, 2.2044, 1.8282, 1.9226, 2.0696, 2.0963], device='cuda:0'), covar=tensor([0.1150, 0.0943, 0.0942, 0.0818, 0.1285, 0.0832, 0.1037, 0.0940], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0087, 0.0104, 0.0101, 0.0113, 0.0079, 0.0113, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:47:42,404 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6137, 0.5958, 0.6317, 0.4852, 0.4638, 0.3313, 0.4985, 0.6851], device='cuda:0'), covar=tensor([0.0268, 0.0163, 0.0214, 0.0110, 0.0516, 0.0407, 0.0119, 0.0568], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0012, 0.0015, 0.0012, 0.0018], device='cuda:0'), out_proj_covar=tensor([4.5459e-05, 4.3994e-05, 4.7842e-05, 4.0033e-05, 4.6998e-05, 5.4793e-05, 5.5105e-05, 6.6919e-05], device='cuda:0') 2022-12-07 10:47:48,444 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8321, 3.0046, 4.6207, 3.3437, 4.5106, 4.2582, 3.8805, 3.8460], device='cuda:0'), covar=tensor([0.0090, 0.1671, 0.0224, 0.0837, 0.0199, 0.0232, 0.1008, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0358, 0.0277, 0.0312, 0.0296, 0.0255, 0.0299, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:47:57,005 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.578e+02 3.545e+02 4.492e+02 7.913e+02, threshold=7.089e+02, percent-clipped=1.0 2022-12-07 10:48:24,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.41 vs. limit=2.0 2022-12-07 10:48:33,320 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:48:57,815 INFO [train.py:873] (0/4) Epoch 4, batch 3200, loss[loss=0.2057, simple_loss=0.2168, pruned_loss=0.09731, over 14279.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2027, pruned_loss=0.09375, over 2052780.80 frames. ], batch size: 66, lr: 1.90e-02, grad_scale: 8.0 2022-12-07 10:49:14,673 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8144, 2.6345, 2.6144, 2.9066, 2.4588, 2.4236, 2.8371, 2.8284], device='cuda:0'), covar=tensor([0.0673, 0.0740, 0.0729, 0.0531, 0.0852, 0.0627, 0.0731, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0104, 0.0086, 0.0101, 0.0099, 0.0111, 0.0077, 0.0110, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:49:21,756 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0700, 0.8505, 1.2353, 1.0956, 1.1490, 0.8162, 1.1352, 1.1508], device='cuda:0'), covar=tensor([0.1044, 0.0365, 0.0223, 0.1213, 0.0343, 0.0226, 0.0501, 0.0397], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0018, 0.0020, 0.0017, 0.0017, 0.0023, 0.0018, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.7632e-05, 6.1887e-05, 6.1644e-05, 6.0713e-05, 5.6262e-05, 7.3995e-05, 6.5739e-05, 5.7172e-05], device='cuda:0') 2022-12-07 10:49:24,095 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.801e+02 3.510e+02 4.555e+02 1.491e+03, threshold=7.021e+02, percent-clipped=7.0 2022-12-07 10:49:25,923 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:49:33,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 10:50:15,246 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5720, 2.6184, 1.7825, 2.7765, 2.5788, 2.6570, 2.0729, 1.9860], device='cuda:0'), covar=tensor([0.0453, 0.1229, 0.4938, 0.0388, 0.0801, 0.0554, 0.2039, 0.5563], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0300, 0.0319, 0.0189, 0.0236, 0.0230, 0.0260, 0.0321], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:50:25,177 INFO [train.py:873] (0/4) Epoch 4, batch 3300, loss[loss=0.1674, simple_loss=0.1824, pruned_loss=0.07623, over 13977.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.202, pruned_loss=0.09345, over 1982391.45 frames. ], batch size: 19, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:50:51,957 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.499e+02 3.235e+02 4.436e+02 1.031e+03, threshold=6.470e+02, percent-clipped=3.0 2022-12-07 10:51:16,335 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3472, 2.1768, 2.3510, 2.0005, 1.6030, 2.3777, 2.2770, 1.0261], device='cuda:0'), covar=tensor([0.2778, 0.1171, 0.1075, 0.1013, 0.1099, 0.0485, 0.1263, 0.3772], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0060, 0.0049, 0.0052, 0.0068, 0.0054, 0.0077, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:51:44,149 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.62 vs. limit=2.0 2022-12-07 10:51:48,809 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:51:51,017 INFO [train.py:873] (0/4) Epoch 4, batch 3400, loss[loss=0.199, simple_loss=0.1703, pruned_loss=0.1139, over 1285.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2017, pruned_loss=0.09291, over 1934543.49 frames. ], batch size: 100, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:52:04,563 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:52:06,276 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4425, 3.3070, 3.0585, 3.5409, 3.0064, 2.6324, 3.4813, 3.4671], device='cuda:0'), covar=tensor([0.0833, 0.0681, 0.0799, 0.0808, 0.0917, 0.0818, 0.0791, 0.0921], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0085, 0.0101, 0.0101, 0.0111, 0.0076, 0.0112, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2022-12-07 10:52:18,305 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.979e+02 3.585e+02 4.923e+02 1.213e+03, threshold=7.170e+02, percent-clipped=7.0 2022-12-07 10:52:30,052 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 10:52:57,455 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:53:01,602 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:53:18,122 INFO [train.py:873] (0/4) Epoch 4, batch 3500, loss[loss=0.2265, simple_loss=0.1905, pruned_loss=0.1313, over 1228.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2023, pruned_loss=0.09331, over 1990786.46 frames. ], batch size: 100, lr: 1.89e-02, grad_scale: 8.0 2022-12-07 10:53:41,619 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 10:53:44,755 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.647e+02 3.463e+02 4.115e+02 9.483e+02, threshold=6.926e+02, percent-clipped=4.0 2022-12-07 10:53:54,557 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:54:11,995 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:54:27,179 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2022-12-07 10:54:44,199 INFO [train.py:873] (0/4) Epoch 4, batch 3600, loss[loss=0.1823, simple_loss=0.2, pruned_loss=0.08237, over 14430.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2018, pruned_loss=0.09267, over 1969586.34 frames. ], batch size: 53, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:55:05,332 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:55:10,969 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.617e+02 3.267e+02 4.043e+02 9.670e+02, threshold=6.534e+02, percent-clipped=3.0 2022-12-07 10:55:39,249 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4410, 3.3142, 2.6813, 1.8771, 3.0150, 3.0939, 3.3877, 2.5058], device='cuda:0'), covar=tensor([0.0492, 0.2184, 0.1270, 0.2810, 0.0888, 0.0379, 0.0677, 0.1784], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0220, 0.0108, 0.0129, 0.0099, 0.0092, 0.0080, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 10:56:10,704 INFO [train.py:873] (0/4) Epoch 4, batch 3700, loss[loss=0.1852, simple_loss=0.1979, pruned_loss=0.08624, over 14280.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2013, pruned_loss=0.09294, over 1943100.84 frames. ], batch size: 66, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:56:17,428 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2655, 1.8400, 2.3949, 2.4869, 2.2727, 1.7258, 2.5175, 2.0770], device='cuda:0'), covar=tensor([0.0068, 0.0152, 0.0083, 0.0065, 0.0059, 0.0277, 0.0053, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0173, 0.0238, 0.0201, 0.0160, 0.0223, 0.0135, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 10:56:33,759 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:56:37,277 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.639e+02 3.495e+02 4.493e+02 9.568e+02, threshold=6.990e+02, percent-clipped=8.0 2022-12-07 10:56:45,552 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4723, 1.6386, 4.2673, 1.6408, 4.2043, 4.4032, 3.8746, 4.7848], device='cuda:0'), covar=tensor([0.0122, 0.2383, 0.0220, 0.2080, 0.0208, 0.0223, 0.0282, 0.0090], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0151, 0.0118, 0.0159, 0.0133, 0.0128, 0.0111, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 10:56:49,082 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7459, 2.2638, 3.7510, 3.9947, 3.9228, 2.3020, 3.6986, 3.1972], device='cuda:0'), covar=tensor([0.0073, 0.0255, 0.0252, 0.0120, 0.0051, 0.0384, 0.0069, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0173, 0.0240, 0.0202, 0.0160, 0.0222, 0.0136, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 10:57:11,830 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:57:26,627 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:57:29,157 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0503, 0.7277, 0.9440, 0.9229, 1.1033, 0.3628, 1.2322, 1.0651], device='cuda:0'), covar=tensor([0.0721, 0.0789, 0.0282, 0.0680, 0.0273, 0.0307, 0.0459, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0016, 0.0018, 0.0020, 0.0016, 0.0017, 0.0022, 0.0018, 0.0018], device='cuda:0'), out_proj_covar=tensor([5.8371e-05, 6.1518e-05, 6.0318e-05, 5.7980e-05, 5.5569e-05, 7.3247e-05, 6.5308e-05, 6.0321e-05], device='cuda:0') 2022-12-07 10:57:36,082 INFO [train.py:873] (0/4) Epoch 4, batch 3800, loss[loss=0.1646, simple_loss=0.1896, pruned_loss=0.06981, over 14097.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2011, pruned_loss=0.09324, over 1900806.73 frames. ], batch size: 29, lr: 1.88e-02, grad_scale: 8.0 2022-12-07 10:57:36,193 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([6.0337, 5.5119, 5.3832, 5.9745, 5.7236, 4.8830, 5.9417, 4.7735], device='cuda:0'), covar=tensor([0.0227, 0.0869, 0.0204, 0.0282, 0.0498, 0.0282, 0.0358, 0.0393], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0203, 0.0131, 0.0125, 0.0134, 0.0110, 0.0196, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 10:57:58,007 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 10:58:00,280 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:58:03,348 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.752e+02 3.266e+02 4.284e+02 8.876e+02, threshold=6.533e+02, percent-clipped=2.0 2022-12-07 10:58:08,533 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:58:41,450 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 10:59:03,303 INFO [train.py:873] (0/4) Epoch 4, batch 3900, loss[loss=0.1521, simple_loss=0.1778, pruned_loss=0.06316, over 14581.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2019, pruned_loss=0.09385, over 1927602.38 frames. ], batch size: 43, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 10:59:19,334 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 10:59:23,167 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2022-12-07 10:59:29,337 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.794e+02 3.682e+02 4.681e+02 9.431e+02, threshold=7.364e+02, percent-clipped=3.0 2022-12-07 11:00:28,848 INFO [train.py:873] (0/4) Epoch 4, batch 4000, loss[loss=0.1562, simple_loss=0.1781, pruned_loss=0.06711, over 13929.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.202, pruned_loss=0.0935, over 1956704.60 frames. ], batch size: 26, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 11:00:55,537 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.732e+02 2.728e+02 3.637e+02 4.905e+02 8.607e+02, threshold=7.274e+02, percent-clipped=2.0 2022-12-07 11:01:15,254 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 11:01:29,193 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:01:39,500 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:01:53,837 INFO [train.py:873] (0/4) Epoch 4, batch 4100, loss[loss=0.1955, simple_loss=0.1984, pruned_loss=0.09632, over 14394.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2022, pruned_loss=0.09392, over 1951720.96 frames. ], batch size: 53, lr: 1.87e-02, grad_scale: 8.0 2022-12-07 11:02:10,431 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:02:16,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.07 vs. limit=2.0 2022-12-07 11:02:20,838 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.498e+02 3.224e+02 4.280e+02 1.572e+03, threshold=6.447e+02, percent-clipped=4.0 2022-12-07 11:02:26,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:02:41,369 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2022-12-07 11:02:54,544 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.29 vs. limit=5.0 2022-12-07 11:03:07,343 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:03:19,882 INFO [train.py:873] (0/4) Epoch 4, batch 4200, loss[loss=0.1969, simple_loss=0.1883, pruned_loss=0.1028, over 2648.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2027, pruned_loss=0.09435, over 1945192.00 frames. ], batch size: 100, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:03:34,670 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:03:36,782 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:03:46,350 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.657e+02 3.822e+02 4.895e+02 8.860e+02, threshold=7.643e+02, percent-clipped=9.0 2022-12-07 11:04:10,199 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5826, 4.2748, 4.1850, 4.6428, 4.4242, 4.0066, 4.6602, 3.8822], device='cuda:0'), covar=tensor([0.0347, 0.0979, 0.0291, 0.0330, 0.0688, 0.0643, 0.0404, 0.0464], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0200, 0.0130, 0.0123, 0.0132, 0.0109, 0.0191, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:04:13,588 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:04:16,891 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:04:26,273 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:04:45,369 INFO [train.py:873] (0/4) Epoch 4, batch 4300, loss[loss=0.1859, simple_loss=0.204, pruned_loss=0.08391, over 14270.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.203, pruned_loss=0.09428, over 1990950.43 frames. ], batch size: 57, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:05:05,717 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:05:11,530 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.870e+02 3.578e+02 4.388e+02 9.558e+02, threshold=7.157e+02, percent-clipped=3.0 2022-12-07 11:05:45,024 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3472, 1.8540, 2.3891, 2.0271, 2.5245, 2.2460, 2.2017, 2.1907], device='cuda:0'), covar=tensor([0.0194, 0.0984, 0.0284, 0.0626, 0.0166, 0.0246, 0.0239, 0.0659], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0337, 0.0278, 0.0303, 0.0298, 0.0250, 0.0292, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:05:57,097 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:06:11,320 INFO [train.py:873] (0/4) Epoch 4, batch 4400, loss[loss=0.1866, simple_loss=0.1982, pruned_loss=0.08749, over 14025.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2021, pruned_loss=0.09387, over 1988846.81 frames. ], batch size: 19, lr: 1.86e-02, grad_scale: 8.0 2022-12-07 11:06:13,313 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0914, 1.7757, 3.2436, 2.3735, 3.1570, 1.7176, 2.4039, 2.9657], device='cuda:0'), covar=tensor([0.0575, 0.4936, 0.0368, 0.7506, 0.0376, 0.4379, 0.1482, 0.0295], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0281, 0.0175, 0.0379, 0.0177, 0.0296, 0.0260, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 11:06:21,664 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 11:06:38,400 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.559e+02 3.461e+02 4.606e+02 7.213e+02, threshold=6.922e+02, percent-clipped=1.0 2022-12-07 11:06:38,490 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:06:42,063 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1612, 1.9945, 3.3242, 2.1577, 3.4060, 2.9163, 3.0539, 2.5845], device='cuda:0'), covar=tensor([0.0300, 0.2969, 0.0435, 0.1851, 0.0485, 0.0696, 0.1175, 0.2332], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0336, 0.0285, 0.0303, 0.0302, 0.0251, 0.0292, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:06:55,032 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-07 11:06:57,827 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7890, 1.5550, 2.0819, 1.7739, 2.0563, 1.4202, 1.7124, 1.8698], device='cuda:0'), covar=tensor([0.0729, 0.1647, 0.0178, 0.1649, 0.0196, 0.0789, 0.0747, 0.0280], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0275, 0.0170, 0.0372, 0.0174, 0.0290, 0.0254, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0005, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 11:07:04,502 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:07:37,301 INFO [train.py:873] (0/4) Epoch 4, batch 4500, loss[loss=0.2, simple_loss=0.1854, pruned_loss=0.1073, over 3854.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2009, pruned_loss=0.09246, over 1978071.66 frames. ], batch size: 100, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:07:57,246 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:08:04,791 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.722e+01 2.647e+02 3.341e+02 4.493e+02 1.287e+03, threshold=6.683e+02, percent-clipped=3.0 2022-12-07 11:08:26,618 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2532, 0.5727, 0.8214, 0.9184, 1.2240, 0.7659, 1.2358, 1.0987], device='cuda:0'), covar=tensor([0.1233, 0.0462, 0.0666, 0.1432, 0.1201, 0.0399, 0.0538, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0016, 0.0017, 0.0021, 0.0017, 0.0017, 0.0023, 0.0017, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.8848e-05, 6.0032e-05, 6.4092e-05, 6.0488e-05, 5.7142e-05, 7.5792e-05, 6.3825e-05, 5.9710e-05], device='cuda:0') 2022-12-07 11:08:28,740 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2022-12-07 11:08:40,935 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:08:47,816 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5964, 1.6659, 1.6039, 1.5446, 1.2379, 1.5850, 1.4250, 1.0458], device='cuda:0'), covar=tensor([0.2197, 0.1138, 0.0705, 0.0716, 0.1341, 0.0662, 0.1914, 0.3475], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0057, 0.0049, 0.0051, 0.0068, 0.0052, 0.0079, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:09:04,080 INFO [train.py:873] (0/4) Epoch 4, batch 4600, loss[loss=0.2072, simple_loss=0.2087, pruned_loss=0.1029, over 11182.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2016, pruned_loss=0.09375, over 1958016.93 frames. ], batch size: 100, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:09:20,162 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:09:25,725 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9369, 1.6592, 1.9900, 1.2118, 1.7590, 1.8348, 1.9160, 1.7413], device='cuda:0'), covar=tensor([0.0474, 0.1097, 0.0808, 0.2079, 0.0841, 0.0562, 0.0395, 0.1380], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0221, 0.0109, 0.0130, 0.0096, 0.0095, 0.0078, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:09:30,612 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 3.048e+02 4.012e+02 4.961e+02 8.817e+02, threshold=8.024e+02, percent-clipped=5.0 2022-12-07 11:09:47,403 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0487, 2.0774, 1.9232, 2.1863, 1.7695, 1.8763, 2.1107, 2.1251], device='cuda:0'), covar=tensor([0.1002, 0.0841, 0.0862, 0.0719, 0.1144, 0.0715, 0.0857, 0.0837], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0089, 0.0105, 0.0106, 0.0115, 0.0083, 0.0115, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:10:06,789 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1434, 3.0400, 3.2703, 2.9956, 3.0671, 2.6627, 1.1928, 2.8754], device='cuda:0'), covar=tensor([0.0234, 0.0303, 0.0419, 0.0436, 0.0352, 0.0797, 0.2927, 0.0357], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0126, 0.0117, 0.0101, 0.0157, 0.0110, 0.0149, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 11:10:09,181 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1587, 4.2163, 3.5678, 3.4270, 4.0162, 4.1623, 4.3076, 4.1610], device='cuda:0'), covar=tensor([0.1440, 0.0748, 0.2951, 0.4905, 0.1213, 0.1262, 0.1481, 0.1561], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0219, 0.0320, 0.0408, 0.0244, 0.0294, 0.0299, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:10:10,005 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1637, 3.9233, 3.7192, 3.7842, 4.0126, 3.9454, 4.2097, 4.1192], device='cuda:0'), covar=tensor([0.0933, 0.0628, 0.1800, 0.2570, 0.0727, 0.0893, 0.0843, 0.0834], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0220, 0.0320, 0.0408, 0.0244, 0.0294, 0.0299, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:10:29,693 INFO [train.py:873] (0/4) Epoch 4, batch 4700, loss[loss=0.2204, simple_loss=0.214, pruned_loss=0.1134, over 5025.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2017, pruned_loss=0.09397, over 1921921.10 frames. ], batch size: 100, lr: 1.85e-02, grad_scale: 8.0 2022-12-07 11:10:56,345 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.544e+02 3.410e+02 4.417e+02 8.558e+02, threshold=6.819e+02, percent-clipped=1.0 2022-12-07 11:10:57,998 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.56 vs. limit=2.0 2022-12-07 11:11:01,978 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0830, 2.6390, 4.9101, 3.1888, 4.8216, 2.1751, 3.4131, 4.4911], device='cuda:0'), covar=tensor([0.0192, 0.4637, 0.0224, 0.9878, 0.0168, 0.3919, 0.1300, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0279, 0.0171, 0.0376, 0.0174, 0.0293, 0.0263, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:11:48,045 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2022-12-07 11:11:56,834 INFO [train.py:873] (0/4) Epoch 4, batch 4800, loss[loss=0.2638, simple_loss=0.2411, pruned_loss=0.1432, over 8603.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2012, pruned_loss=0.0932, over 1943293.44 frames. ], batch size: 100, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:12:11,703 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:12:24,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 3.057e+02 3.716e+02 4.446e+02 1.010e+03, threshold=7.432e+02, percent-clipped=2.0 2022-12-07 11:12:51,228 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.19 vs. limit=5.0 2022-12-07 11:12:53,324 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7795, 3.3220, 3.3365, 3.8044, 3.3617, 2.9492, 3.7801, 3.7520], device='cuda:0'), covar=tensor([0.0643, 0.0758, 0.0776, 0.0592, 0.0820, 0.0772, 0.0685, 0.0687], device='cuda:0'), in_proj_covar=tensor([0.0106, 0.0090, 0.0106, 0.0106, 0.0113, 0.0083, 0.0115, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:12:58,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8397, 2.3817, 4.8937, 3.3623, 4.6010, 2.2487, 3.5222, 4.3830], device='cuda:0'), covar=tensor([0.0262, 0.4604, 0.0235, 0.8344, 0.0220, 0.3625, 0.1143, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0276, 0.0172, 0.0379, 0.0174, 0.0292, 0.0265, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:13:00,406 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:13:23,628 INFO [train.py:873] (0/4) Epoch 4, batch 4900, loss[loss=0.164, simple_loss=0.1869, pruned_loss=0.07054, over 14279.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2015, pruned_loss=0.09349, over 1966219.69 frames. ], batch size: 31, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:13:40,692 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:13:42,439 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:13:45,417 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 11:13:51,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.749e+02 3.689e+02 4.632e+02 8.661e+02, threshold=7.378e+02, percent-clipped=1.0 2022-12-07 11:14:21,724 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:14:50,476 INFO [train.py:873] (0/4) Epoch 4, batch 5000, loss[loss=0.1735, simple_loss=0.1957, pruned_loss=0.07565, over 14317.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2016, pruned_loss=0.09339, over 1928197.08 frames. ], batch size: 28, lr: 1.84e-02, grad_scale: 8.0 2022-12-07 11:15:14,309 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:15:18,161 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.479e+02 3.081e+02 3.772e+02 7.331e+02, threshold=6.161e+02, percent-clipped=0.0 2022-12-07 11:15:32,952 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-07 11:15:42,684 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.65 vs. limit=5.0 2022-12-07 11:16:02,931 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1463, 1.6025, 4.4218, 4.2280, 4.1927, 4.4988, 4.2580, 4.6019], device='cuda:0'), covar=tensor([0.1053, 0.1219, 0.0080, 0.0111, 0.0099, 0.0069, 0.0093, 0.0073], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0154, 0.0096, 0.0132, 0.0111, 0.0116, 0.0083, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:16:06,370 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:16:16,332 INFO [train.py:873] (0/4) Epoch 4, batch 5100, loss[loss=0.1893, simple_loss=0.1961, pruned_loss=0.0912, over 14241.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2006, pruned_loss=0.09239, over 1950807.55 frames. ], batch size: 46, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:16:31,614 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:16:43,937 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.535e+02 3.343e+02 4.225e+02 1.167e+03, threshold=6.685e+02, percent-clipped=6.0 2022-12-07 11:17:12,648 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:17:30,353 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0486, 1.8257, 3.1942, 2.2409, 2.9353, 1.7475, 2.4991, 2.6978], device='cuda:0'), covar=tensor([0.0600, 0.5153, 0.0467, 0.7800, 0.0334, 0.3952, 0.1424, 0.0487], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0279, 0.0173, 0.0379, 0.0174, 0.0292, 0.0266, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:17:43,049 INFO [train.py:873] (0/4) Epoch 4, batch 5200, loss[loss=0.1941, simple_loss=0.1992, pruned_loss=0.09453, over 9470.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2007, pruned_loss=0.09209, over 1951476.39 frames. ], batch size: 100, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:17:48,362 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 11:17:49,720 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4077, 2.9049, 3.8275, 2.9301, 2.8774, 2.6723, 1.5203, 3.1567], device='cuda:0'), covar=tensor([0.1997, 0.1353, 0.0801, 0.1389, 0.1413, 0.2258, 0.6298, 0.1258], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0074, 0.0068, 0.0083, 0.0096, 0.0065, 0.0138, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:18:10,409 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.745e+02 3.569e+02 4.514e+02 6.730e+02, threshold=7.138e+02, percent-clipped=2.0 2022-12-07 11:18:40,494 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5420, 2.4246, 2.1938, 2.2027, 2.3717, 2.4214, 2.4964, 2.4844], device='cuda:0'), covar=tensor([0.0931, 0.0792, 0.2181, 0.2555, 0.1077, 0.1069, 0.1220, 0.0886], device='cuda:0'), in_proj_covar=tensor([0.0262, 0.0216, 0.0333, 0.0416, 0.0247, 0.0292, 0.0305, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:19:08,657 INFO [train.py:873] (0/4) Epoch 4, batch 5300, loss[loss=0.192, simple_loss=0.1708, pruned_loss=0.1066, over 2579.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2008, pruned_loss=0.09206, over 1939921.12 frames. ], batch size: 100, lr: 1.83e-02, grad_scale: 8.0 2022-12-07 11:19:31,016 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2022-12-07 11:19:36,271 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.605e+02 3.441e+02 4.511e+02 8.751e+02, threshold=6.883e+02, percent-clipped=1.0 2022-12-07 11:20:20,554 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:20:35,859 INFO [train.py:873] (0/4) Epoch 4, batch 5400, loss[loss=0.187, simple_loss=0.1938, pruned_loss=0.0901, over 11157.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.1989, pruned_loss=0.08996, over 1921563.14 frames. ], batch size: 100, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:21:03,785 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.371e+02 3.121e+02 4.059e+02 8.001e+02, threshold=6.242e+02, percent-clipped=5.0 2022-12-07 11:22:02,364 INFO [train.py:873] (0/4) Epoch 4, batch 5500, loss[loss=0.1803, simple_loss=0.1868, pruned_loss=0.08687, over 6975.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.1991, pruned_loss=0.0908, over 1885681.57 frames. ], batch size: 100, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:22:30,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.632e+02 3.347e+02 4.419e+02 1.188e+03, threshold=6.694e+02, percent-clipped=11.0 2022-12-07 11:22:39,453 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:23:01,981 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:23:29,471 INFO [train.py:873] (0/4) Epoch 4, batch 5600, loss[loss=0.1609, simple_loss=0.1861, pruned_loss=0.06786, over 14219.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2004, pruned_loss=0.0917, over 1905826.93 frames. ], batch size: 35, lr: 1.82e-02, grad_scale: 8.0 2022-12-07 11:23:29,640 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:23:32,018 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:23:54,692 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:23:55,378 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2571, 3.0418, 2.7626, 2.8944, 3.1714, 3.1152, 3.1885, 3.1750], device='cuda:0'), covar=tensor([0.0775, 0.0867, 0.1925, 0.2861, 0.0760, 0.0819, 0.1159, 0.0917], device='cuda:0'), in_proj_covar=tensor([0.0261, 0.0228, 0.0333, 0.0423, 0.0245, 0.0297, 0.0313, 0.0250], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:23:57,191 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.230e+02 2.669e+02 3.555e+02 4.575e+02 8.799e+02, threshold=7.110e+02, percent-clipped=5.0 2022-12-07 11:24:10,844 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.89 vs. limit=5.0 2022-12-07 11:24:21,853 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:24:41,046 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:24:48,613 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2359, 1.8282, 2.3337, 2.4990, 2.1332, 1.8327, 2.5358, 2.0480], device='cuda:0'), covar=tensor([0.0069, 0.0153, 0.0092, 0.0062, 0.0087, 0.0213, 0.0050, 0.0142], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0175, 0.0252, 0.0201, 0.0163, 0.0226, 0.0141, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 11:24:55,321 INFO [train.py:873] (0/4) Epoch 4, batch 5700, loss[loss=0.1685, simple_loss=0.1847, pruned_loss=0.07611, over 14212.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2003, pruned_loss=0.09261, over 1864444.50 frames. ], batch size: 89, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:25:15,392 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0721, 3.8141, 3.6274, 3.6911, 3.9856, 4.0121, 4.1327, 4.0862], device='cuda:0'), covar=tensor([0.0793, 0.0628, 0.1699, 0.2695, 0.0643, 0.0667, 0.0932, 0.0746], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0218, 0.0320, 0.0412, 0.0239, 0.0289, 0.0303, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:25:22,827 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:25:23,563 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 2.835e+02 3.398e+02 4.455e+02 8.619e+02, threshold=6.796e+02, percent-clipped=2.0 2022-12-07 11:25:35,528 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 11:26:22,858 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:26:23,531 INFO [train.py:873] (0/4) Epoch 4, batch 5800, loss[loss=0.2163, simple_loss=0.1803, pruned_loss=0.1261, over 2618.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.1996, pruned_loss=0.0919, over 1872882.33 frames. ], batch size: 100, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:26:27,977 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:26:52,116 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.705e+02 3.370e+02 4.199e+02 7.683e+02, threshold=6.740e+02, percent-clipped=3.0 2022-12-07 11:27:16,504 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:21,763 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:49,613 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:27:51,186 INFO [train.py:873] (0/4) Epoch 4, batch 5900, loss[loss=0.2034, simple_loss=0.2124, pruned_loss=0.09721, over 14265.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2, pruned_loss=0.09181, over 1955833.57 frames. ], batch size: 76, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:28:03,860 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2022-12-07 11:28:13,002 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:28:19,509 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.929e+02 3.624e+02 4.382e+02 1.301e+03, threshold=7.248e+02, percent-clipped=4.0 2022-12-07 11:28:40,742 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:28:58,913 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 11:29:19,380 INFO [train.py:873] (0/4) Epoch 4, batch 6000, loss[loss=0.1869, simple_loss=0.2055, pruned_loss=0.08412, over 14296.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.1997, pruned_loss=0.09154, over 1944569.94 frames. ], batch size: 25, lr: 1.81e-02, grad_scale: 8.0 2022-12-07 11:29:19,381 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 11:29:27,927 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5562, 4.9038, 5.0618, 5.3324, 5.0731, 4.5972, 5.3927, 4.6842], device='cuda:0'), covar=tensor([0.0265, 0.0626, 0.0314, 0.0474, 0.0671, 0.0177, 0.0378, 0.0504], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0204, 0.0135, 0.0130, 0.0134, 0.0108, 0.0199, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:29:30,471 INFO [train.py:905] (0/4) Epoch 4, validation: loss=0.1258, simple_loss=0.1688, pruned_loss=0.04138, over 857387.00 frames. 2022-12-07 11:29:30,471 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 11:29:58,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.688e+02 3.368e+02 4.685e+02 1.249e+03, threshold=6.736e+02, percent-clipped=4.0 2022-12-07 11:30:23,447 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.28 vs. limit=5.0 2022-12-07 11:30:33,892 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.65 vs. limit=5.0 2022-12-07 11:30:58,675 INFO [train.py:873] (0/4) Epoch 4, batch 6100, loss[loss=0.2429, simple_loss=0.2331, pruned_loss=0.1264, over 9533.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.1994, pruned_loss=0.09157, over 1873155.72 frames. ], batch size: 100, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:31:06,926 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 11:31:25,483 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2022-12-07 11:31:27,398 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.104e+02 2.717e+02 3.239e+02 4.102e+02 7.823e+02, threshold=6.479e+02, percent-clipped=2.0 2022-12-07 11:31:47,325 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:31:52,838 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:32:04,629 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:32:07,407 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2022-12-07 11:32:25,386 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:32:27,170 INFO [train.py:873] (0/4) Epoch 4, batch 6200, loss[loss=0.1868, simple_loss=0.2028, pruned_loss=0.08534, over 14268.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2005, pruned_loss=0.09256, over 1939844.25 frames. ], batch size: 37, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:32:48,582 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:32:55,224 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.759e+02 3.767e+02 5.172e+02 1.797e+03, threshold=7.535e+02, percent-clipped=14.0 2022-12-07 11:32:57,908 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:07,485 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:10,726 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 11:33:16,618 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:33:21,071 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5998, 3.4406, 3.1663, 3.2289, 3.5416, 3.4676, 3.6486, 3.5104], device='cuda:0'), covar=tensor([0.0845, 0.0578, 0.1773, 0.2396, 0.0718, 0.0773, 0.0980, 0.1010], device='cuda:0'), in_proj_covar=tensor([0.0258, 0.0218, 0.0327, 0.0414, 0.0248, 0.0295, 0.0303, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:33:30,100 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 11:33:55,196 INFO [train.py:873] (0/4) Epoch 4, batch 6300, loss[loss=0.1896, simple_loss=0.1687, pruned_loss=0.1053, over 1275.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2007, pruned_loss=0.09289, over 1884815.91 frames. ], batch size: 100, lr: 1.80e-02, grad_scale: 8.0 2022-12-07 11:33:58,972 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:34:02,012 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.95 vs. limit=5.0 2022-12-07 11:34:04,681 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 11:34:08,940 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:34:12,344 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1948, 1.2817, 1.5005, 1.0496, 1.0464, 1.3020, 1.0576, 1.2438], device='cuda:0'), covar=tensor([0.1352, 0.2529, 0.0552, 0.1907, 0.2017, 0.0604, 0.2261, 0.0752], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0075, 0.0070, 0.0083, 0.0101, 0.0063, 0.0139, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:34:23,887 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.322e+02 3.000e+02 3.617e+02 7.085e+02, threshold=6.000e+02, percent-clipped=0.0 2022-12-07 11:35:03,252 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:35:24,087 INFO [train.py:873] (0/4) Epoch 4, batch 6400, loss[loss=0.1707, simple_loss=0.1887, pruned_loss=0.07632, over 13530.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.1998, pruned_loss=0.09134, over 1886141.75 frames. ], batch size: 100, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:35:50,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7483, 2.6244, 2.5659, 1.3156, 2.3303, 2.4568, 2.8213, 2.1871], device='cuda:0'), covar=tensor([0.0665, 0.2490, 0.1017, 0.3084, 0.0826, 0.0561, 0.0673, 0.1651], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0219, 0.0110, 0.0129, 0.0095, 0.0095, 0.0083, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:35:52,467 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.620e+02 3.469e+02 4.373e+02 1.002e+03, threshold=6.937e+02, percent-clipped=9.0 2022-12-07 11:36:12,744 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:36:17,939 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:36:52,305 INFO [train.py:873] (0/4) Epoch 4, batch 6500, loss[loss=0.1733, simple_loss=0.1916, pruned_loss=0.07755, over 13941.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.1994, pruned_loss=0.09164, over 1887083.60 frames. ], batch size: 23, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:36:54,882 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:00,022 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:18,800 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:20,462 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.481e+02 2.712e+02 3.572e+02 4.670e+02 1.139e+03, threshold=7.145e+02, percent-clipped=7.0 2022-12-07 11:37:35,553 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:37:48,095 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6762, 2.4758, 3.3079, 2.6334, 3.3551, 3.3144, 3.1607, 2.7299], device='cuda:0'), covar=tensor([0.0300, 0.2112, 0.0610, 0.1534, 0.0457, 0.0500, 0.1458, 0.1795], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0351, 0.0331, 0.0316, 0.0320, 0.0275, 0.0321, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:38:20,129 INFO [train.py:873] (0/4) Epoch 4, batch 6600, loss[loss=0.1628, simple_loss=0.1905, pruned_loss=0.06754, over 14235.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.1989, pruned_loss=0.09177, over 1882730.95 frames. ], batch size: 35, lr: 1.79e-02, grad_scale: 8.0 2022-12-07 11:38:29,390 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:38:38,239 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0586, 3.1395, 2.6799, 1.8627, 2.0947, 3.1000, 3.0087, 1.1696], device='cuda:0'), covar=tensor([0.4124, 0.0767, 0.3705, 0.3736, 0.1548, 0.0777, 0.1760, 0.5086], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0056, 0.0050, 0.0050, 0.0069, 0.0052, 0.0078, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:38:48,061 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.617e+02 3.366e+02 4.262e+02 9.069e+02, threshold=6.731e+02, percent-clipped=3.0 2022-12-07 11:39:22,441 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:39:47,259 INFO [train.py:873] (0/4) Epoch 4, batch 6700, loss[loss=0.1842, simple_loss=0.1957, pruned_loss=0.08634, over 14279.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.1995, pruned_loss=0.09131, over 1942238.48 frames. ], batch size: 60, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:40:14,795 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.696e+02 3.507e+02 4.703e+02 1.037e+03, threshold=7.015e+02, percent-clipped=5.0 2022-12-07 11:40:35,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2022-12-07 11:40:51,694 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-07 11:40:52,932 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7457, 1.9151, 2.3890, 2.0584, 2.6850, 2.4474, 2.3384, 2.0653], device='cuda:0'), covar=tensor([0.0295, 0.1819, 0.0462, 0.1142, 0.0315, 0.0565, 0.0504, 0.1406], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0341, 0.0327, 0.0308, 0.0312, 0.0266, 0.0315, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:41:13,497 INFO [train.py:873] (0/4) Epoch 4, batch 6800, loss[loss=0.2114, simple_loss=0.2174, pruned_loss=0.1028, over 14526.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.1995, pruned_loss=0.0911, over 2011277.35 frames. ], batch size: 51, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:41:25,840 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4991, 1.2166, 1.9946, 1.8318, 1.9940, 1.9806, 1.4715, 1.9546], device='cuda:0'), covar=tensor([0.0377, 0.0813, 0.0123, 0.0232, 0.0186, 0.0117, 0.0266, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0150, 0.0093, 0.0126, 0.0107, 0.0113, 0.0083, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:41:39,861 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:41:42,142 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.836e+02 3.618e+02 4.474e+02 6.792e+02, threshold=7.235e+02, percent-clipped=0.0 2022-12-07 11:42:04,287 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4503, 3.3021, 3.1636, 3.6028, 3.0501, 2.9469, 3.5425, 3.5035], device='cuda:0'), covar=tensor([0.0703, 0.0662, 0.0734, 0.0549, 0.0805, 0.0580, 0.0710, 0.0637], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0088, 0.0106, 0.0107, 0.0114, 0.0082, 0.0117, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:42:09,558 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:42:21,758 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:42:41,252 INFO [train.py:873] (0/4) Epoch 4, batch 6900, loss[loss=0.1969, simple_loss=0.2047, pruned_loss=0.09459, over 14281.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.1987, pruned_loss=0.09083, over 1960616.74 frames. ], batch size: 60, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:42:45,769 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:01,270 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-07 11:43:02,754 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:09,418 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.737e+02 3.624e+02 4.962e+02 1.279e+03, threshold=7.248e+02, percent-clipped=6.0 2022-12-07 11:43:42,677 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:43:45,825 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:44:07,054 INFO [train.py:873] (0/4) Epoch 4, batch 7000, loss[loss=0.2243, simple_loss=0.2173, pruned_loss=0.1157, over 8569.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.1985, pruned_loss=0.09008, over 1959388.79 frames. ], batch size: 100, lr: 1.78e-02, grad_scale: 8.0 2022-12-07 11:44:23,766 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:44:35,728 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.834e+02 3.387e+02 4.469e+02 7.718e+02, threshold=6.775e+02, percent-clipped=1.0 2022-12-07 11:44:38,324 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:45:15,597 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.85 vs. limit=5.0 2022-12-07 11:45:29,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1756, 2.9803, 3.9681, 2.7573, 2.5700, 2.6800, 1.6415, 3.0609], device='cuda:0'), covar=tensor([0.0967, 0.0673, 0.0715, 0.2918, 0.2152, 0.1023, 0.5134, 0.1264], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0073, 0.0072, 0.0083, 0.0099, 0.0063, 0.0135, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:45:33,332 INFO [train.py:873] (0/4) Epoch 4, batch 7100, loss[loss=0.1834, simple_loss=0.1986, pruned_loss=0.08409, over 14264.00 frames. ], tot_loss[loss=0.19, simple_loss=0.1991, pruned_loss=0.09045, over 1984428.72 frames. ], batch size: 57, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:45:34,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7080, 1.3854, 3.7370, 1.5871, 3.7370, 3.8085, 2.9728, 4.1125], device='cuda:0'), covar=tensor([0.0353, 0.3740, 0.0511, 0.2882, 0.0466, 0.0418, 0.0643, 0.0240], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0159, 0.0128, 0.0168, 0.0146, 0.0137, 0.0117, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 11:45:46,036 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1789, 2.9025, 2.1987, 3.4389, 3.0710, 3.1908, 2.5090, 2.2233], device='cuda:0'), covar=tensor([0.0411, 0.1141, 0.3397, 0.0396, 0.0689, 0.0855, 0.1377, 0.4180], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0299, 0.0311, 0.0189, 0.0233, 0.0235, 0.0253, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 11:45:46,784 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:46:02,683 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.583e+02 3.436e+02 4.712e+02 1.166e+03, threshold=6.872e+02, percent-clipped=4.0 2022-12-07 11:46:23,603 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9058, 0.8517, 0.9849, 0.9272, 0.5964, 0.7655, 0.8263, 0.6124], device='cuda:0'), covar=tensor([0.0343, 0.0273, 0.0230, 0.0182, 0.0426, 0.0454, 0.0209, 0.0538], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0010, 0.0011, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([5.1279e-05, 5.2728e-05, 5.2920e-05, 4.8581e-05, 5.1771e-05, 6.5512e-05, 5.9615e-05, 6.8150e-05], device='cuda:0') 2022-12-07 11:46:33,728 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 11:46:40,651 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:46:58,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1084, 2.3005, 5.1067, 4.4069, 4.6521, 5.0616, 4.8787, 5.1297], device='cuda:0'), covar=tensor([0.1089, 0.1011, 0.0050, 0.0116, 0.0084, 0.0059, 0.0056, 0.0070], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0154, 0.0096, 0.0132, 0.0111, 0.0117, 0.0086, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:47:01,594 INFO [train.py:873] (0/4) Epoch 4, batch 7200, loss[loss=0.2543, simple_loss=0.2321, pruned_loss=0.1382, over 10352.00 frames. ], tot_loss[loss=0.19, simple_loss=0.1984, pruned_loss=0.09075, over 1864321.64 frames. ], batch size: 100, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:47:06,078 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:19,224 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:30,074 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7197, 2.4800, 4.6372, 2.9489, 4.4319, 1.8525, 3.5269, 4.1642], device='cuda:0'), covar=tensor([0.0221, 0.5026, 0.0558, 1.1478, 0.0226, 0.4480, 0.1232, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0279, 0.0168, 0.0369, 0.0176, 0.0283, 0.0259, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:47:30,663 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.626e+02 3.344e+02 4.272e+02 1.259e+03, threshold=6.687e+02, percent-clipped=4.0 2022-12-07 11:47:47,975 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:47:49,157 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.91 vs. limit=5.0 2022-12-07 11:48:03,906 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6125, 1.7185, 2.9597, 2.1903, 2.7679, 1.7403, 2.3717, 2.6450], device='cuda:0'), covar=tensor([0.0577, 0.4639, 0.0432, 0.6988, 0.0365, 0.3544, 0.1299, 0.0340], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0276, 0.0167, 0.0370, 0.0175, 0.0280, 0.0257, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 11:48:28,905 INFO [train.py:873] (0/4) Epoch 4, batch 7300, loss[loss=0.196, simple_loss=0.1948, pruned_loss=0.09858, over 4969.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.1968, pruned_loss=0.08912, over 1910312.56 frames. ], batch size: 100, lr: 1.77e-02, grad_scale: 8.0 2022-12-07 11:48:38,732 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6518, 0.5841, 0.6758, 0.4914, 0.5177, 0.4788, 0.4172, 0.6017], device='cuda:0'), covar=tensor([0.0206, 0.0155, 0.0168, 0.0150, 0.0268, 0.0392, 0.0338, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0012, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([5.2160e-05, 5.4617e-05, 5.4061e-05, 4.8919e-05, 5.3641e-05, 6.7442e-05, 6.1856e-05, 7.0353e-05], device='cuda:0') 2022-12-07 11:48:40,731 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-30000.pt 2022-12-07 11:48:51,930 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 11:48:59,471 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 11:49:01,143 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.428e+02 2.508e+02 3.092e+02 3.972e+02 1.080e+03, threshold=6.184e+02, percent-clipped=3.0 2022-12-07 11:49:32,345 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6559, 3.8452, 4.3935, 4.6884, 4.5184, 3.9582, 4.7201, 3.8895], device='cuda:0'), covar=tensor([0.0711, 0.2346, 0.0780, 0.0980, 0.0958, 0.0747, 0.1093, 0.1008], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0208, 0.0139, 0.0130, 0.0136, 0.0110, 0.0197, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:49:38,914 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3762, 2.1503, 3.4881, 3.6210, 3.6062, 2.1437, 3.6631, 2.8198], device='cuda:0'), covar=tensor([0.0087, 0.0252, 0.0212, 0.0100, 0.0076, 0.0400, 0.0048, 0.0276], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0176, 0.0249, 0.0204, 0.0161, 0.0221, 0.0143, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 11:49:59,920 INFO [train.py:873] (0/4) Epoch 4, batch 7400, loss[loss=0.2097, simple_loss=0.2191, pruned_loss=0.1002, over 14466.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.1975, pruned_loss=0.08995, over 1872548.51 frames. ], batch size: 51, lr: 1.76e-02, grad_scale: 8.0 2022-12-07 11:50:04,629 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-07 11:50:18,399 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-07 11:50:29,607 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.725e+02 3.785e+02 4.899e+02 8.566e+02, threshold=7.571e+02, percent-clipped=11.0 2022-12-07 11:50:33,052 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9803, 1.7017, 4.2461, 4.0325, 4.0718, 4.2441, 3.7431, 4.2672], device='cuda:0'), covar=tensor([0.1063, 0.1126, 0.0060, 0.0101, 0.0108, 0.0071, 0.0148, 0.0079], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0151, 0.0093, 0.0128, 0.0110, 0.0115, 0.0084, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 11:50:33,862 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:02,854 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:03,756 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4331, 5.0196, 4.9600, 5.4212, 5.1844, 4.4415, 5.4197, 4.4454], device='cuda:0'), covar=tensor([0.0282, 0.0840, 0.0231, 0.0363, 0.0532, 0.0339, 0.0353, 0.0424], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0205, 0.0138, 0.0130, 0.0135, 0.0109, 0.0195, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:51:27,333 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:28,005 INFO [train.py:873] (0/4) Epoch 4, batch 7500, loss[loss=0.1901, simple_loss=0.1948, pruned_loss=0.09271, over 11948.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.1975, pruned_loss=0.08986, over 1853112.53 frames. ], batch size: 100, lr: 1.76e-02, grad_scale: 8.0 2022-12-07 11:51:45,994 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:47,599 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:51:53,396 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2256, 3.9508, 3.7679, 4.2276, 4.1371, 3.8117, 4.3006, 3.5543], device='cuda:0'), covar=tensor([0.0366, 0.0963, 0.0366, 0.0466, 0.0665, 0.0805, 0.0521, 0.0538], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0211, 0.0141, 0.0132, 0.0137, 0.0111, 0.0199, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 11:51:56,539 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.609e+02 3.309e+02 4.020e+02 7.663e+02, threshold=6.618e+02, percent-clipped=1.0 2022-12-07 11:52:14,528 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-4.pt 2022-12-07 11:52:55,164 INFO [train.py:873] (0/4) Epoch 5, batch 0, loss[loss=0.231, simple_loss=0.2273, pruned_loss=0.1174, over 14250.00 frames. ], tot_loss[loss=0.231, simple_loss=0.2273, pruned_loss=0.1174, over 14250.00 frames. ], batch size: 94, lr: 1.64e-02, grad_scale: 8.0 2022-12-07 11:52:55,165 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 11:53:02,316 INFO [train.py:905] (0/4) Epoch 5, validation: loss=0.1366, simple_loss=0.1807, pruned_loss=0.04626, over 857387.00 frames. 2022-12-07 11:53:02,317 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 11:53:02,756 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 11:53:07,512 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:53:21,378 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:54:03,604 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 11:54:05,438 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.465e+01 2.449e+02 3.602e+02 5.226e+02 1.817e+03, threshold=7.204e+02, percent-clipped=11.0 2022-12-07 11:54:10,508 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 11:54:31,491 INFO [train.py:873] (0/4) Epoch 5, batch 100, loss[loss=0.2024, simple_loss=0.2094, pruned_loss=0.09774, over 10324.00 frames. ], tot_loss[loss=0.185, simple_loss=0.1962, pruned_loss=0.08688, over 810745.39 frames. ], batch size: 100, lr: 1.64e-02, grad_scale: 8.0 2022-12-07 11:54:35,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8771, 1.2254, 1.0044, 1.3159, 1.1127, 1.1606, 1.1753, 1.0209], device='cuda:0'), covar=tensor([0.1481, 0.1299, 0.0734, 0.1172, 0.1545, 0.0449, 0.0352, 0.1555], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0012, 0.0015, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.3862e-05, 5.5178e-05, 5.3600e-05, 4.9783e-05, 5.4871e-05, 6.7512e-05, 6.0660e-05, 7.3316e-05], device='cuda:0') 2022-12-07 11:54:40,806 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:54:45,805 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:55:09,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9557, 2.0687, 2.7668, 2.2403, 2.8923, 2.7150, 2.7304, 2.4257], device='cuda:0'), covar=tensor([0.0273, 0.1648, 0.0355, 0.1179, 0.0321, 0.0359, 0.0638, 0.1445], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0348, 0.0335, 0.0318, 0.0330, 0.0269, 0.0324, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:55:17,131 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.12 vs. limit=2.0 2022-12-07 11:55:32,798 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.778e+02 3.417e+02 4.208e+02 9.792e+02, threshold=6.833e+02, percent-clipped=3.0 2022-12-07 11:55:33,894 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:55:58,007 INFO [train.py:873] (0/4) Epoch 5, batch 200, loss[loss=0.197, simple_loss=0.1875, pruned_loss=0.1033, over 6027.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.1955, pruned_loss=0.08698, over 1248313.90 frames. ], batch size: 100, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:56:06,132 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:56:26,446 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:56:32,607 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1628, 2.0225, 2.0905, 1.2215, 1.8528, 2.2016, 2.1545, 1.8072], device='cuda:0'), covar=tensor([0.0495, 0.1075, 0.1057, 0.2494, 0.0963, 0.0579, 0.0488, 0.1695], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0221, 0.0116, 0.0129, 0.0098, 0.0097, 0.0084, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 11:56:40,647 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.48 vs. limit=5.0 2022-12-07 11:56:48,235 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:57:00,757 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.564e+02 3.377e+02 4.486e+02 9.729e+02, threshold=6.754e+02, percent-clipped=3.0 2022-12-07 11:57:25,910 INFO [train.py:873] (0/4) Epoch 5, batch 300, loss[loss=0.197, simple_loss=0.2057, pruned_loss=0.09419, over 14227.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.1952, pruned_loss=0.08666, over 1542529.19 frames. ], batch size: 60, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:57:39,898 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:58:27,950 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.626e+02 3.332e+02 4.486e+02 9.530e+02, threshold=6.664e+02, percent-clipped=3.0 2022-12-07 11:58:53,706 INFO [train.py:873] (0/4) Epoch 5, batch 400, loss[loss=0.2029, simple_loss=0.2129, pruned_loss=0.09645, over 12746.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.1955, pruned_loss=0.08771, over 1687595.60 frames. ], batch size: 100, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 11:58:55,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7377, 0.6011, 0.4643, 0.8177, 0.6936, 0.1793, 0.6189, 0.8111], device='cuda:0'), covar=tensor([0.0089, 0.0188, 0.0081, 0.0126, 0.0113, 0.0060, 0.0359, 0.0085], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0019, 0.0017, 0.0019, 0.0025, 0.0019, 0.0019], device='cuda:0'), out_proj_covar=tensor([6.8196e-05, 7.0587e-05, 6.5650e-05, 6.4423e-05, 6.8227e-05, 8.8415e-05, 7.4227e-05, 6.7659e-05], device='cuda:0') 2022-12-07 11:58:58,988 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:59:35,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 11:59:52,610 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 11:59:52,714 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 11:59:56,041 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.557e+02 3.010e+02 3.908e+02 7.694e+02, threshold=6.020e+02, percent-clipped=1.0 2022-12-07 12:00:21,154 INFO [train.py:873] (0/4) Epoch 5, batch 500, loss[loss=0.189, simple_loss=0.1951, pruned_loss=0.09149, over 14381.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.1963, pruned_loss=0.08842, over 1784603.49 frames. ], batch size: 55, lr: 1.63e-02, grad_scale: 8.0 2022-12-07 12:00:45,798 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 12:00:49,056 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:01:23,932 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.567e+02 3.338e+02 3.998e+02 8.568e+02, threshold=6.676e+02, percent-clipped=3.0 2022-12-07 12:01:31,368 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:01:48,856 INFO [train.py:873] (0/4) Epoch 5, batch 600, loss[loss=0.2209, simple_loss=0.1913, pruned_loss=0.1252, over 2621.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.1969, pruned_loss=0.08846, over 1856798.30 frames. ], batch size: 100, lr: 1.62e-02, grad_scale: 4.0 2022-12-07 12:02:02,939 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:36,346 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:45,015 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:02:51,899 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.564e+02 3.109e+02 3.998e+02 8.746e+02, threshold=6.217e+02, percent-clipped=4.0 2022-12-07 12:02:55,498 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:03,267 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:16,248 INFO [train.py:873] (0/4) Epoch 5, batch 700, loss[loss=0.1883, simple_loss=0.2063, pruned_loss=0.08522, over 14302.00 frames. ], tot_loss[loss=0.185, simple_loss=0.1958, pruned_loss=0.08708, over 1898095.77 frames. ], batch size: 31, lr: 1.62e-02, grad_scale: 4.0 2022-12-07 12:03:21,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 12:03:29,806 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:37,729 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 12:03:49,055 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:03:54,763 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5204, 1.1604, 1.9984, 1.9475, 1.9745, 2.0675, 1.5927, 1.9883], device='cuda:0'), covar=tensor([0.0389, 0.0675, 0.0084, 0.0178, 0.0174, 0.0104, 0.0210, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0152, 0.0098, 0.0134, 0.0113, 0.0117, 0.0087, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 12:03:56,552 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:04:08,956 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7357, 3.5445, 3.0435, 2.1475, 3.0597, 3.3260, 3.6809, 3.0342], device='cuda:0'), covar=tensor([0.0455, 0.2623, 0.1295, 0.2523, 0.0800, 0.0468, 0.0812, 0.1331], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0217, 0.0115, 0.0126, 0.0097, 0.0097, 0.0082, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:04:10,435 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:04:10,496 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:04:14,770 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:04:18,900 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.586e+02 3.335e+02 4.382e+02 1.879e+03, threshold=6.670e+02, percent-clipped=6.0 2022-12-07 12:04:43,833 INFO [train.py:873] (0/4) Epoch 5, batch 800, loss[loss=0.1774, simple_loss=0.1954, pruned_loss=0.07974, over 14233.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.195, pruned_loss=0.08616, over 1959365.61 frames. ], batch size: 80, lr: 1.62e-02, grad_scale: 8.0 2022-12-07 12:04:57,146 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:05:04,529 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:05:17,853 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2598, 3.0890, 2.7280, 1.8213, 2.8342, 3.2066, 3.0679, 2.4579], device='cuda:0'), covar=tensor([0.0597, 0.2177, 0.1456, 0.2830, 0.0821, 0.0360, 0.1643, 0.1744], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0219, 0.0115, 0.0127, 0.0097, 0.0098, 0.0083, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:05:41,441 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2087, 0.5057, 0.9502, 1.0465, 1.0335, 0.9573, 1.1905, 1.4551], device='cuda:0'), covar=tensor([0.0721, 0.0806, 0.0550, 0.0635, 0.1291, 0.0514, 0.0656, 0.0645], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0018, 0.0019, 0.0026, 0.0019, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.0400e-05, 7.4596e-05, 6.7759e-05, 6.7415e-05, 6.9402e-05, 9.3212e-05, 7.3761e-05, 6.8857e-05], device='cuda:0') 2022-12-07 12:05:47,141 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.770e+02 3.066e+02 3.847e+02 6.961e+02, threshold=6.133e+02, percent-clipped=2.0 2022-12-07 12:05:48,212 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9620, 1.5626, 3.6683, 3.3921, 3.6157, 3.7601, 3.0812, 3.7307], device='cuda:0'), covar=tensor([0.1046, 0.1172, 0.0079, 0.0154, 0.0127, 0.0071, 0.0178, 0.0075], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0151, 0.0097, 0.0134, 0.0112, 0.0115, 0.0086, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 12:06:11,660 INFO [train.py:873] (0/4) Epoch 5, batch 900, loss[loss=0.1934, simple_loss=0.1733, pruned_loss=0.1067, over 2638.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.1945, pruned_loss=0.0859, over 1963703.13 frames. ], batch size: 100, lr: 1.62e-02, grad_scale: 8.0 2022-12-07 12:06:11,759 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3300, 4.3650, 3.5047, 3.4760, 3.9996, 4.3613, 4.4912, 4.4276], device='cuda:0'), covar=tensor([0.1523, 0.0950, 0.3619, 0.4499, 0.1268, 0.1113, 0.1308, 0.1526], device='cuda:0'), in_proj_covar=tensor([0.0264, 0.0224, 0.0330, 0.0424, 0.0254, 0.0298, 0.0302, 0.0253], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:06:36,911 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7997, 2.6898, 1.8375, 2.9211, 2.6090, 2.6935, 2.3226, 2.1461], device='cuda:0'), covar=tensor([0.0365, 0.0983, 0.2997, 0.0265, 0.0570, 0.0473, 0.1324, 0.3231], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0307, 0.0319, 0.0188, 0.0249, 0.0245, 0.0267, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:07:14,863 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.339e+02 3.126e+02 3.777e+02 6.609e+02, threshold=6.252e+02, percent-clipped=2.0 2022-12-07 12:07:15,326 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2022-12-07 12:07:24,665 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3168, 1.9450, 2.7374, 1.7602, 1.7423, 2.2592, 1.2973, 2.1367], device='cuda:0'), covar=tensor([0.0921, 0.1469, 0.0477, 0.2656, 0.2519, 0.0932, 0.4703, 0.0882], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0074, 0.0069, 0.0084, 0.0100, 0.0064, 0.0137, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:07:39,740 INFO [train.py:873] (0/4) Epoch 5, batch 1000, loss[loss=0.198, simple_loss=0.2037, pruned_loss=0.09615, over 14154.00 frames. ], tot_loss[loss=0.184, simple_loss=0.1949, pruned_loss=0.08659, over 1924724.63 frames. ], batch size: 99, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:07:48,318 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:07:59,522 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 12:08:07,786 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:11,680 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 12:08:15,258 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:33,879 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:08:33,979 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3747, 2.2187, 3.2473, 2.6327, 3.3086, 3.1726, 2.9533, 2.5814], device='cuda:0'), covar=tensor([0.0308, 0.2528, 0.0475, 0.1396, 0.0489, 0.0461, 0.0875, 0.2042], device='cuda:0'), in_proj_covar=tensor([0.0269, 0.0344, 0.0342, 0.0316, 0.0333, 0.0274, 0.0324, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:08:36,926 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.17 vs. limit=5.0 2022-12-07 12:08:42,295 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.699e+02 3.594e+02 4.260e+02 1.045e+03, threshold=7.189e+02, percent-clipped=6.0 2022-12-07 12:08:56,527 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2022-12-07 12:09:06,865 INFO [train.py:873] (0/4) Epoch 5, batch 1100, loss[loss=0.2121, simple_loss=0.2085, pruned_loss=0.1079, over 11994.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.1952, pruned_loss=0.08713, over 1947082.62 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:09:15,211 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:09:22,294 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:09:25,686 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7252, 1.2219, 2.0379, 1.1724, 1.9776, 1.9919, 1.7496, 1.9396], device='cuda:0'), covar=tensor([0.0320, 0.1910, 0.0279, 0.1671, 0.0381, 0.0373, 0.0606, 0.0361], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0159, 0.0131, 0.0165, 0.0149, 0.0136, 0.0114, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:09:26,612 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0535, 0.5758, 1.0198, 1.0694, 1.0496, 0.7932, 0.8853, 1.1479], device='cuda:0'), covar=tensor([0.0589, 0.1150, 0.0475, 0.0924, 0.0585, 0.0414, 0.0594, 0.0404], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0020, 0.0019, 0.0019, 0.0026, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([6.9485e-05, 7.3722e-05, 6.8585e-05, 7.0804e-05, 7.0478e-05, 9.2604e-05, 7.6335e-05, 6.9932e-05], device='cuda:0') 2022-12-07 12:09:53,927 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.03 vs. limit=5.0 2022-12-07 12:09:58,008 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 12:09:59,479 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:10:03,940 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-07 12:10:09,481 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.558e+02 3.355e+02 4.456e+02 1.089e+03, threshold=6.709e+02, percent-clipped=2.0 2022-12-07 12:10:34,138 INFO [train.py:873] (0/4) Epoch 5, batch 1200, loss[loss=0.2821, simple_loss=0.2422, pruned_loss=0.161, over 7794.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.1949, pruned_loss=0.08691, over 1946871.59 frames. ], batch size: 100, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:10:52,689 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:11:10,727 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:11:37,433 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.031e+02 2.496e+02 3.332e+02 4.341e+02 8.542e+02, threshold=6.664e+02, percent-clipped=7.0 2022-12-07 12:12:02,328 INFO [train.py:873] (0/4) Epoch 5, batch 1300, loss[loss=0.2054, simple_loss=0.209, pruned_loss=0.1009, over 14173.00 frames. ], tot_loss[loss=0.184, simple_loss=0.1947, pruned_loss=0.08661, over 1964456.08 frames. ], batch size: 84, lr: 1.61e-02, grad_scale: 8.0 2022-12-07 12:12:05,381 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:12:08,242 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.04 vs. limit=5.0 2022-12-07 12:12:11,297 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:15,541 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.88 vs. limit=5.0 2022-12-07 12:12:30,385 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:38,832 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:12:54,064 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:05,507 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.035e+02 2.616e+02 3.313e+02 3.855e+02 7.867e+02, threshold=6.626e+02, percent-clipped=4.0 2022-12-07 12:13:13,017 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:20,929 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:13:23,893 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6388, 3.1889, 2.7831, 1.9904, 2.9916, 2.9287, 3.3401, 2.7130], device='cuda:0'), covar=tensor([0.0436, 0.3018, 0.1611, 0.3306, 0.0650, 0.0653, 0.1095, 0.1769], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0215, 0.0116, 0.0129, 0.0096, 0.0100, 0.0083, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:13:30,391 INFO [train.py:873] (0/4) Epoch 5, batch 1400, loss[loss=0.1408, simple_loss=0.1679, pruned_loss=0.05687, over 14469.00 frames. ], tot_loss[loss=0.183, simple_loss=0.1947, pruned_loss=0.08563, over 1983310.93 frames. ], batch size: 18, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:13:45,870 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:14:18,266 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:14:28,097 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:14:33,108 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.505e+02 2.718e+02 3.448e+02 4.143e+02 7.580e+02, threshold=6.896e+02, percent-clipped=4.0 2022-12-07 12:14:41,949 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8987, 2.4432, 3.9294, 4.0908, 4.0125, 2.6368, 4.1998, 3.0087], device='cuda:0'), covar=tensor([0.0122, 0.0313, 0.0248, 0.0118, 0.0088, 0.0496, 0.0071, 0.0359], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0186, 0.0268, 0.0211, 0.0169, 0.0234, 0.0153, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 12:14:44,416 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0711, 1.7273, 2.0810, 1.2789, 1.5066, 1.9164, 1.9826, 1.7930], device='cuda:0'), covar=tensor([0.0345, 0.0964, 0.0552, 0.1621, 0.0847, 0.0378, 0.0281, 0.1139], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0211, 0.0114, 0.0127, 0.0097, 0.0097, 0.0082, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:14:52,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=12.04 vs. limit=5.0 2022-12-07 12:14:57,257 INFO [train.py:873] (0/4) Epoch 5, batch 1500, loss[loss=0.2062, simple_loss=0.179, pruned_loss=0.1167, over 1241.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.193, pruned_loss=0.08431, over 1950526.28 frames. ], batch size: 100, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:15:11,627 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:15:11,741 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:15:31,505 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:00,993 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.523e+02 3.162e+02 3.978e+02 7.805e+02, threshold=6.325e+02, percent-clipped=2.0 2022-12-07 12:16:01,990 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9157, 2.8364, 3.0233, 2.9092, 2.9029, 2.8214, 1.2614, 2.6985], device='cuda:0'), covar=tensor([0.0243, 0.0299, 0.0422, 0.0286, 0.0293, 0.0498, 0.2780, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0131, 0.0122, 0.0105, 0.0161, 0.0112, 0.0148, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:16:02,938 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:24,805 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:16:26,443 INFO [train.py:873] (0/4) Epoch 5, batch 1600, loss[loss=0.1595, simple_loss=0.1539, pruned_loss=0.08259, over 3865.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.1939, pruned_loss=0.08584, over 1957121.50 frames. ], batch size: 100, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:16:26,689 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:52,794 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:16:58,402 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:25,421 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:30,354 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.483e+02 3.139e+02 4.013e+02 6.840e+02, threshold=6.277e+02, percent-clipped=1.0 2022-12-07 12:17:46,271 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:17:49,258 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 12:17:54,776 INFO [train.py:873] (0/4) Epoch 5, batch 1700, loss[loss=0.1899, simple_loss=0.166, pruned_loss=0.1069, over 1198.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.1938, pruned_loss=0.08573, over 1941846.27 frames. ], batch size: 100, lr: 1.60e-02, grad_scale: 8.0 2022-12-07 12:17:56,112 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:16,198 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5237, 2.2305, 2.4028, 1.2248, 2.0925, 2.3832, 2.6550, 2.0650], device='cuda:0'), covar=tensor([0.0588, 0.1515, 0.1089, 0.2831, 0.0866, 0.0621, 0.0497, 0.1633], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0212, 0.0116, 0.0129, 0.0099, 0.0102, 0.0084, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:18:17,872 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2867, 2.2859, 2.0158, 2.0361, 1.7709, 2.1457, 2.1320, 0.8639], device='cuda:0'), covar=tensor([0.2577, 0.0857, 0.1659, 0.0929, 0.1275, 0.0602, 0.1152, 0.4290], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0057, 0.0051, 0.0050, 0.0071, 0.0053, 0.0079, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:18:18,870 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:50,904 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:18:59,553 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.546e+02 3.116e+02 4.047e+02 1.395e+03, threshold=6.231e+02, percent-clipped=6.0 2022-12-07 12:19:25,249 INFO [train.py:873] (0/4) Epoch 5, batch 1800, loss[loss=0.1282, simple_loss=0.1507, pruned_loss=0.05288, over 10776.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.1943, pruned_loss=0.08569, over 1984522.54 frames. ], batch size: 13, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:19:34,913 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:19:39,344 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:22,869 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:26,301 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9503, 4.3620, 3.4194, 5.1140, 4.5281, 4.8482, 3.8624, 3.3199], device='cuda:0'), covar=tensor([0.0455, 0.0680, 0.3456, 0.0197, 0.0460, 0.1100, 0.1116, 0.3265], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0292, 0.0313, 0.0182, 0.0232, 0.0237, 0.0255, 0.0291], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:20:30,237 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.694e+02 3.195e+02 4.070e+02 6.631e+02, threshold=6.390e+02, percent-clipped=2.0 2022-12-07 12:20:51,057 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:20:53,875 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:20:55,473 INFO [train.py:873] (0/4) Epoch 5, batch 1900, loss[loss=0.1727, simple_loss=0.1685, pruned_loss=0.0884, over 4943.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.1943, pruned_loss=0.08605, over 2007929.29 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:20:56,881 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 12:21:23,367 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:21:38,179 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 12:21:39,084 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0021, 1.8974, 1.9987, 2.1110, 1.9496, 1.8872, 1.2407, 1.7694], device='cuda:0'), covar=tensor([0.0330, 0.0362, 0.0473, 0.0220, 0.0318, 0.0584, 0.1769, 0.0351], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0134, 0.0127, 0.0109, 0.0167, 0.0114, 0.0153, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:22:01,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.008e+02 2.535e+02 3.167e+02 4.268e+02 9.066e+02, threshold=6.334e+02, percent-clipped=2.0 2022-12-07 12:22:13,242 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:22:26,469 INFO [train.py:873] (0/4) Epoch 5, batch 2000, loss[loss=0.189, simple_loss=0.2001, pruned_loss=0.08899, over 10318.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.1944, pruned_loss=0.08602, over 1925183.14 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:22:29,200 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2095, 4.7178, 4.6748, 5.1955, 4.9881, 4.4221, 5.2070, 4.2933], device='cuda:0'), covar=tensor([0.0253, 0.0889, 0.0268, 0.0363, 0.0541, 0.0422, 0.0419, 0.0421], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0205, 0.0140, 0.0133, 0.0141, 0.0111, 0.0207, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:22:41,242 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1632, 3.0804, 2.7410, 2.8101, 3.1437, 3.0289, 3.2163, 3.1571], device='cuda:0'), covar=tensor([0.1311, 0.0734, 0.2264, 0.3347, 0.0832, 0.0975, 0.1269, 0.1081], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0223, 0.0346, 0.0440, 0.0255, 0.0304, 0.0317, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:22:45,446 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:23:16,433 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:23:30,260 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.510e+02 3.096e+02 3.855e+02 9.073e+02, threshold=6.191e+02, percent-clipped=1.0 2022-12-07 12:23:50,823 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4063, 4.8506, 4.9484, 5.3377, 5.1241, 4.4010, 5.3362, 4.5216], device='cuda:0'), covar=tensor([0.0234, 0.0916, 0.0238, 0.0407, 0.0610, 0.0353, 0.0489, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0204, 0.0139, 0.0132, 0.0139, 0.0110, 0.0205, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:23:55,266 INFO [train.py:873] (0/4) Epoch 5, batch 2100, loss[loss=0.2, simple_loss=0.1989, pruned_loss=0.1005, over 3827.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.194, pruned_loss=0.08539, over 1943232.33 frames. ], batch size: 100, lr: 1.59e-02, grad_scale: 8.0 2022-12-07 12:24:05,664 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:24:29,029 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7270, 4.0436, 3.1137, 5.0615, 4.5518, 4.8152, 3.9808, 3.4024], device='cuda:0'), covar=tensor([0.0397, 0.0925, 0.3805, 0.0601, 0.0456, 0.1140, 0.0889, 0.3147], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0297, 0.0310, 0.0182, 0.0233, 0.0234, 0.0254, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:24:50,029 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:25:01,678 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.695e+02 3.113e+02 4.251e+02 1.070e+03, threshold=6.225e+02, percent-clipped=3.0 2022-12-07 12:25:23,328 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:25:27,727 INFO [train.py:873] (0/4) Epoch 5, batch 2200, loss[loss=0.1898, simple_loss=0.1994, pruned_loss=0.09006, over 14265.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.1938, pruned_loss=0.08576, over 1887042.41 frames. ], batch size: 76, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:25:54,748 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:05,774 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:31,378 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.575e+02 3.323e+02 4.480e+02 8.616e+02, threshold=6.645e+02, percent-clipped=4.0 2022-12-07 12:26:36,975 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:43,110 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:26:47,017 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.19 vs. limit=5.0 2022-12-07 12:26:56,110 INFO [train.py:873] (0/4) Epoch 5, batch 2300, loss[loss=0.1727, simple_loss=0.1646, pruned_loss=0.09035, over 2622.00 frames. ], tot_loss[loss=0.181, simple_loss=0.1926, pruned_loss=0.08467, over 1897324.63 frames. ], batch size: 100, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:26:59,739 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1714, 1.2754, 1.4799, 0.9621, 0.8458, 1.3201, 1.0259, 1.1356], device='cuda:0'), covar=tensor([0.1406, 0.1728, 0.0506, 0.1921, 0.3141, 0.0852, 0.2250, 0.0947], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0071, 0.0069, 0.0079, 0.0096, 0.0062, 0.0132, 0.0069], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:27:00,646 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.23 vs. limit=2.0 2022-12-07 12:27:16,002 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:25,530 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:29,052 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8941, 1.4045, 3.1129, 1.3002, 3.0872, 3.0757, 2.3211, 3.2824], device='cuda:0'), covar=tensor([0.0238, 0.2627, 0.0326, 0.2364, 0.0589, 0.0359, 0.0762, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0157, 0.0131, 0.0169, 0.0148, 0.0139, 0.0118, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:27:48,236 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:27:54,312 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1998, 1.3121, 1.1794, 1.3493, 1.3021, 0.7470, 1.0266, 0.9298], device='cuda:0'), covar=tensor([0.0612, 0.0825, 0.0611, 0.0425, 0.0591, 0.0341, 0.0312, 0.0845], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0011, 0.0011, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([5.5229e-05, 5.5329e-05, 5.5554e-05, 4.9429e-05, 5.6789e-05, 7.2213e-05, 6.3337e-05, 7.4513e-05], device='cuda:0') 2022-12-07 12:27:59,293 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:28:00,942 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.763e+02 3.478e+02 4.434e+02 7.343e+02, threshold=6.957e+02, percent-clipped=3.0 2022-12-07 12:28:26,684 INFO [train.py:873] (0/4) Epoch 5, batch 2400, loss[loss=0.1634, simple_loss=0.1826, pruned_loss=0.07207, over 14393.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.1929, pruned_loss=0.08516, over 1864576.41 frames. ], batch size: 53, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:28:30,924 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:28:50,075 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8354, 1.7494, 3.0697, 2.1297, 2.9287, 1.7174, 2.3830, 2.7522], device='cuda:0'), covar=tensor([0.0630, 0.4716, 0.0298, 0.7047, 0.0438, 0.4445, 0.1426, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0264, 0.0168, 0.0357, 0.0173, 0.0275, 0.0250, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:29:31,402 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.127e+02 2.520e+02 3.446e+02 4.436e+02 8.213e+02, threshold=6.892e+02, percent-clipped=3.0 2022-12-07 12:29:34,236 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-07 12:29:56,012 INFO [train.py:873] (0/4) Epoch 5, batch 2500, loss[loss=0.1982, simple_loss=0.1875, pruned_loss=0.1044, over 3888.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.193, pruned_loss=0.08524, over 1868856.38 frames. ], batch size: 100, lr: 1.58e-02, grad_scale: 8.0 2022-12-07 12:30:23,960 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1955, 2.0507, 2.2579, 1.3692, 1.9894, 2.2075, 2.3554, 1.9458], device='cuda:0'), covar=tensor([0.0540, 0.1631, 0.0929, 0.2308, 0.0930, 0.0525, 0.0439, 0.1557], device='cuda:0'), in_proj_covar=tensor([0.0100, 0.0215, 0.0118, 0.0132, 0.0102, 0.0105, 0.0086, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:31:01,909 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.569e+02 3.218e+02 4.442e+02 9.627e+02, threshold=6.437e+02, percent-clipped=1.0 2022-12-07 12:31:27,893 INFO [train.py:873] (0/4) Epoch 5, batch 2600, loss[loss=0.2039, simple_loss=0.2079, pruned_loss=0.1, over 12752.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.1937, pruned_loss=0.08635, over 1861335.38 frames. ], batch size: 100, lr: 1.57e-02, grad_scale: 16.0 2022-12-07 12:32:11,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2022-12-07 12:32:18,878 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4206, 2.3545, 1.7929, 2.5425, 2.1690, 2.4120, 2.1360, 1.9693], device='cuda:0'), covar=tensor([0.0290, 0.0646, 0.2093, 0.0241, 0.0547, 0.0310, 0.1101, 0.1767], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0289, 0.0302, 0.0180, 0.0230, 0.0234, 0.0250, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:32:33,213 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.535e+02 3.293e+02 4.332e+02 8.259e+02, threshold=6.586e+02, percent-clipped=6.0 2022-12-07 12:32:56,510 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6294, 1.4765, 2.9799, 1.3882, 2.9370, 2.8822, 2.2093, 3.0426], device='cuda:0'), covar=tensor([0.0203, 0.2071, 0.0196, 0.1816, 0.0204, 0.0279, 0.0625, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0154, 0.0130, 0.0165, 0.0144, 0.0138, 0.0115, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 12:32:57,305 INFO [train.py:873] (0/4) Epoch 5, batch 2700, loss[loss=0.1867, simple_loss=0.1876, pruned_loss=0.09284, over 5035.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.1941, pruned_loss=0.08662, over 1836861.40 frames. ], batch size: 100, lr: 1.57e-02, grad_scale: 8.0 2022-12-07 12:33:13,532 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:34:03,195 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.537e+02 3.172e+02 3.779e+02 6.676e+02, threshold=6.344e+02, percent-clipped=1.0 2022-12-07 12:34:07,341 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:34:19,519 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8903, 0.7996, 0.7303, 0.7444, 1.0531, 0.5352, 0.6363, 0.8112], device='cuda:0'), covar=tensor([0.0657, 0.0628, 0.0368, 0.0444, 0.0329, 0.0387, 0.0617, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0019, 0.0018, 0.0019, 0.0025, 0.0019, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.0082e-05, 7.3159e-05, 6.7504e-05, 7.0954e-05, 7.2887e-05, 9.0284e-05, 7.6468e-05, 6.8421e-05], device='cuda:0') 2022-12-07 12:34:22,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-07 12:34:27,431 INFO [train.py:873] (0/4) Epoch 5, batch 2800, loss[loss=0.1458, simple_loss=0.1369, pruned_loss=0.07737, over 1261.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.1939, pruned_loss=0.08614, over 1833291.68 frames. ], batch size: 100, lr: 1.57e-02, grad_scale: 8.0 2022-12-07 12:34:35,203 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 12:34:45,092 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:34:54,540 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4205, 3.4563, 3.0595, 2.0310, 3.0539, 3.3763, 3.4059, 2.7709], device='cuda:0'), covar=tensor([0.0539, 0.2047, 0.1132, 0.2220, 0.0605, 0.0411, 0.1344, 0.1549], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0212, 0.0118, 0.0128, 0.0099, 0.0105, 0.0085, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:34:55,727 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9286, 1.4664, 1.2686, 1.4653, 1.2863, 1.4270, 1.0625, 1.0350], device='cuda:0'), covar=tensor([0.1939, 0.0577, 0.0704, 0.0333, 0.1143, 0.0457, 0.1811, 0.2385], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0056, 0.0053, 0.0051, 0.0074, 0.0053, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 12:35:19,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2647, 3.2381, 2.6915, 2.6518, 3.3076, 3.2140, 3.2821, 3.2583], device='cuda:0'), covar=tensor([0.1084, 0.0742, 0.2405, 0.3675, 0.0930, 0.0952, 0.1546, 0.1189], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0229, 0.0339, 0.0429, 0.0252, 0.0303, 0.0322, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:35:33,623 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.428e+02 3.052e+02 4.047e+02 7.917e+02, threshold=6.105e+02, percent-clipped=1.0 2022-12-07 12:35:40,388 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:35:56,746 INFO [train.py:873] (0/4) Epoch 5, batch 2900, loss[loss=0.1406, simple_loss=0.1681, pruned_loss=0.05662, over 13965.00 frames. ], tot_loss[loss=0.181, simple_loss=0.1928, pruned_loss=0.08464, over 1908976.70 frames. ], batch size: 19, lr: 1.57e-02, grad_scale: 4.0 2022-12-07 12:36:01,989 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:36:26,719 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2022-12-07 12:36:31,439 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6746, 0.7920, 0.7589, 0.4371, 0.5933, 0.4331, 0.6269, 0.5295], device='cuda:0'), covar=tensor([0.0101, 0.0122, 0.0148, 0.0101, 0.0216, 0.0448, 0.0150, 0.0419], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0011, 0.0011, 0.0010, 0.0012, 0.0015, 0.0011, 0.0017], device='cuda:0'), out_proj_covar=tensor([5.5960e-05, 5.5622e-05, 5.7826e-05, 5.1007e-05, 5.7822e-05, 7.5730e-05, 6.2642e-05, 7.7473e-05], device='cuda:0') 2022-12-07 12:36:48,831 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1207, 2.0136, 1.9441, 2.1566, 1.7395, 1.8828, 2.0848, 2.1224], device='cuda:0'), covar=tensor([0.0846, 0.0889, 0.0922, 0.0714, 0.1232, 0.0810, 0.0964, 0.0912], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0092, 0.0108, 0.0107, 0.0114, 0.0085, 0.0122, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 12:36:56,612 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:37:02,204 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 2.752e+02 3.414e+02 4.319e+02 7.862e+02, threshold=6.827e+02, percent-clipped=8.0 2022-12-07 12:37:26,084 INFO [train.py:873] (0/4) Epoch 5, batch 3000, loss[loss=0.155, simple_loss=0.1799, pruned_loss=0.06502, over 13994.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.1934, pruned_loss=0.08469, over 1958994.00 frames. ], batch size: 22, lr: 1.56e-02, grad_scale: 4.0 2022-12-07 12:37:26,084 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 12:37:38,399 INFO [train.py:905] (0/4) Epoch 5, validation: loss=0.1238, simple_loss=0.167, pruned_loss=0.04033, over 857387.00 frames. 2022-12-07 12:37:38,400 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 12:38:44,263 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:38:44,935 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.483e+02 3.391e+02 4.332e+02 8.491e+02, threshold=6.783e+02, percent-clipped=1.0 2022-12-07 12:39:08,696 INFO [train.py:873] (0/4) Epoch 5, batch 3100, loss[loss=0.1765, simple_loss=0.19, pruned_loss=0.08149, over 14244.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.1925, pruned_loss=0.0843, over 1943610.02 frames. ], batch size: 69, lr: 1.56e-02, grad_scale: 4.0 2022-12-07 12:40:15,025 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.236e+02 3.030e+02 3.906e+02 6.925e+02, threshold=6.060e+02, percent-clipped=1.0 2022-12-07 12:40:17,217 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:40:38,101 INFO [train.py:873] (0/4) Epoch 5, batch 3200, loss[loss=0.1838, simple_loss=0.2006, pruned_loss=0.08355, over 14536.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.1926, pruned_loss=0.08427, over 1938715.27 frames. ], batch size: 34, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:40:56,967 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3648, 2.3251, 3.0580, 2.5630, 3.0787, 3.0809, 3.0104, 2.6176], device='cuda:0'), covar=tensor([0.0331, 0.2190, 0.0563, 0.1410, 0.0556, 0.0469, 0.1018, 0.1935], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0345, 0.0352, 0.0322, 0.0349, 0.0277, 0.0323, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 12:40:59,559 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:41:14,225 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6364, 1.7575, 1.9992, 1.5052, 1.3608, 1.8267, 1.2954, 1.7351], device='cuda:0'), covar=tensor([0.0927, 0.1109, 0.0391, 0.1454, 0.1962, 0.0403, 0.2428, 0.0500], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0077, 0.0074, 0.0084, 0.0105, 0.0064, 0.0138, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 12:41:17,860 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1198, 1.8180, 2.1449, 2.3797, 1.8711, 1.8263, 2.2678, 2.1170], device='cuda:0'), covar=tensor([0.0063, 0.0145, 0.0073, 0.0060, 0.0096, 0.0228, 0.0069, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0189, 0.0272, 0.0219, 0.0175, 0.0236, 0.0160, 0.0227], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 12:41:34,531 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:41:44,543 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.667e+02 3.351e+02 4.253e+02 1.055e+03, threshold=6.703e+02, percent-clipped=3.0 2022-12-07 12:41:53,273 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:42:08,402 INFO [train.py:873] (0/4) Epoch 5, batch 3300, loss[loss=0.2116, simple_loss=0.2119, pruned_loss=0.1057, over 14127.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1916, pruned_loss=0.08297, over 2029754.86 frames. ], batch size: 99, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:42:28,554 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 12:43:13,044 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:43:13,764 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.043e+02 2.484e+02 3.236e+02 4.553e+02 1.226e+03, threshold=6.473e+02, percent-clipped=4.0 2022-12-07 12:43:14,082 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9293, 2.4352, 4.8625, 3.1947, 4.7568, 2.2388, 3.4085, 4.5811], device='cuda:0'), covar=tensor([0.0257, 0.5419, 0.0210, 1.0140, 0.0225, 0.4058, 0.1291, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0268, 0.0165, 0.0348, 0.0172, 0.0267, 0.0244, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:43:37,538 INFO [train.py:873] (0/4) Epoch 5, batch 3400, loss[loss=0.1639, simple_loss=0.1852, pruned_loss=0.07135, over 14272.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.1918, pruned_loss=0.08266, over 2039355.39 frames. ], batch size: 28, lr: 1.56e-02, grad_scale: 8.0 2022-12-07 12:43:54,618 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9354, 1.8354, 2.3252, 1.7677, 1.6113, 2.0329, 1.3076, 1.9926], device='cuda:0'), covar=tensor([0.0959, 0.2148, 0.0708, 0.1880, 0.2762, 0.1029, 0.4847, 0.0803], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0077, 0.0072, 0.0083, 0.0102, 0.0063, 0.0137, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 12:43:57,508 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:03,429 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:22,524 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9042, 1.4447, 3.9846, 3.8745, 3.8211, 4.1475, 3.6176, 4.1289], device='cuda:0'), covar=tensor([0.1092, 0.1372, 0.0120, 0.0122, 0.0141, 0.0094, 0.0122, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0100, 0.0138, 0.0117, 0.0120, 0.0088, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 12:44:43,756 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.478e+02 3.354e+02 4.337e+02 1.168e+03, threshold=6.709e+02, percent-clipped=6.0 2022-12-07 12:44:45,670 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:44:46,227 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-07 12:44:56,909 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 12:45:07,676 INFO [train.py:873] (0/4) Epoch 5, batch 3500, loss[loss=0.1635, simple_loss=0.1861, pruned_loss=0.07046, over 14221.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.1921, pruned_loss=0.08365, over 1994282.01 frames. ], batch size: 60, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:45:28,348 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:45:36,808 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6317, 2.4299, 2.5131, 1.5546, 2.2845, 2.5513, 2.7729, 2.1389], device='cuda:0'), covar=tensor([0.0660, 0.2117, 0.1001, 0.2480, 0.1003, 0.0500, 0.0643, 0.1732], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0215, 0.0119, 0.0133, 0.0103, 0.0105, 0.0088, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 12:46:02,393 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:13,096 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.895e+01 2.306e+02 2.839e+02 3.644e+02 6.341e+02, threshold=5.678e+02, percent-clipped=0.0 2022-12-07 12:46:18,212 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:46:26,791 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2022-12-07 12:46:36,554 INFO [train.py:873] (0/4) Epoch 5, batch 3600, loss[loss=0.1826, simple_loss=0.1564, pruned_loss=0.1045, over 2641.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1914, pruned_loss=0.08308, over 1995193.43 frames. ], batch size: 100, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:46:45,901 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:47:21,350 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0416, 2.9005, 2.8336, 3.1347, 2.7298, 2.5303, 3.0679, 3.0716], device='cuda:0'), covar=tensor([0.0847, 0.0818, 0.0854, 0.0739, 0.0988, 0.0931, 0.0846, 0.0852], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0095, 0.0110, 0.0110, 0.0118, 0.0088, 0.0125, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 12:47:44,854 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.078e+02 2.977e+02 3.494e+02 4.554e+02 8.959e+02, threshold=6.988e+02, percent-clipped=9.0 2022-12-07 12:47:59,697 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:48:09,174 INFO [train.py:873] (0/4) Epoch 5, batch 3700, loss[loss=0.2271, simple_loss=0.2254, pruned_loss=0.1144, over 12739.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.1919, pruned_loss=0.08334, over 1998559.01 frames. ], batch size: 100, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:48:48,177 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8041, 1.4603, 2.0371, 1.7061, 1.9669, 1.4450, 1.5445, 1.6776], device='cuda:0'), covar=tensor([0.0871, 0.2407, 0.0149, 0.1483, 0.0363, 0.1021, 0.0862, 0.0529], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0262, 0.0167, 0.0356, 0.0175, 0.0269, 0.0247, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 12:48:54,472 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9697, 0.9248, 0.9256, 0.9996, 1.2501, 0.5881, 0.7949, 0.9773], device='cuda:0'), covar=tensor([0.0584, 0.0618, 0.0246, 0.0421, 0.0263, 0.0403, 0.0536, 0.0464], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0018, 0.0017, 0.0019, 0.0025, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([6.9007e-05, 7.0844e-05, 6.6297e-05, 6.8973e-05, 7.3237e-05, 9.1204e-05, 7.8150e-05, 7.1870e-05], device='cuda:0') 2022-12-07 12:48:54,486 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:48:57,795 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6173, 5.0111, 4.9848, 5.4761, 5.2694, 4.5667, 5.4564, 4.4781], device='cuda:0'), covar=tensor([0.0230, 0.0829, 0.0219, 0.0352, 0.0594, 0.0364, 0.0446, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0219, 0.0144, 0.0138, 0.0144, 0.0114, 0.0216, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:49:15,045 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.468e+02 3.019e+02 3.918e+02 8.852e+02, threshold=6.038e+02, percent-clipped=4.0 2022-12-07 12:49:24,196 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:49:33,975 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:49:38,095 INFO [train.py:873] (0/4) Epoch 5, batch 3800, loss[loss=0.2143, simple_loss=0.1935, pruned_loss=0.1176, over 3860.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.1926, pruned_loss=0.08408, over 1910878.13 frames. ], batch size: 100, lr: 1.55e-02, grad_scale: 8.0 2022-12-07 12:50:22,834 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 12:50:29,199 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:50:45,937 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.542e+02 3.059e+02 3.773e+02 1.009e+03, threshold=6.118e+02, percent-clipped=4.0 2022-12-07 12:50:50,744 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:50:59,167 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8773, 2.6264, 2.6842, 2.8671, 2.8150, 2.7779, 2.9447, 2.4518], device='cuda:0'), covar=tensor([0.0475, 0.1208, 0.0464, 0.0507, 0.0662, 0.0446, 0.0665, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0214, 0.0140, 0.0133, 0.0139, 0.0112, 0.0209, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 12:51:05,917 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:51:09,664 INFO [train.py:873] (0/4) Epoch 5, batch 3900, loss[loss=0.1561, simple_loss=0.187, pruned_loss=0.06257, over 14576.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.1921, pruned_loss=0.08288, over 1973122.37 frames. ], batch size: 34, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:51:33,807 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:52:02,494 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:52:17,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.452e+02 3.217e+02 4.044e+02 1.041e+03, threshold=6.433e+02, percent-clipped=5.0 2022-12-07 12:52:41,188 INFO [train.py:873] (0/4) Epoch 5, batch 4000, loss[loss=0.1815, simple_loss=0.1771, pruned_loss=0.09299, over 6968.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.1919, pruned_loss=0.08343, over 1957918.13 frames. ], batch size: 100, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:53:22,811 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:53:48,676 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.578e+02 3.207e+02 3.995e+02 8.945e+02, threshold=6.415e+02, percent-clipped=4.0 2022-12-07 12:53:49,148 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2022-12-07 12:53:57,358 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:54:12,179 INFO [train.py:873] (0/4) Epoch 5, batch 4100, loss[loss=0.1868, simple_loss=0.202, pruned_loss=0.08577, over 12737.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.1914, pruned_loss=0.083, over 1939578.68 frames. ], batch size: 100, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:54:40,229 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:54:57,946 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:55:18,132 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.334e+02 3.145e+02 3.965e+02 8.915e+02, threshold=6.291e+02, percent-clipped=4.0 2022-12-07 12:55:41,761 INFO [train.py:873] (0/4) Epoch 5, batch 4200, loss[loss=0.189, simple_loss=0.2005, pruned_loss=0.08874, over 14291.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.1914, pruned_loss=0.08247, over 1974319.44 frames. ], batch size: 66, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:56:28,595 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:56:48,746 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.089e+02 2.566e+02 3.165e+02 3.922e+02 1.587e+03, threshold=6.330e+02, percent-clipped=5.0 2022-12-07 12:57:11,680 INFO [train.py:873] (0/4) Epoch 5, batch 4300, loss[loss=0.1698, simple_loss=0.1835, pruned_loss=0.078, over 14410.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.1909, pruned_loss=0.08215, over 1998093.54 frames. ], batch size: 73, lr: 1.54e-02, grad_scale: 8.0 2022-12-07 12:57:51,401 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:57:53,687 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:14,959 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3040, 0.8142, 0.9077, 0.9576, 1.1710, 0.4902, 1.0755, 1.0371], device='cuda:0'), covar=tensor([0.0485, 0.0828, 0.0464, 0.0826, 0.0473, 0.0414, 0.0305, 0.0559], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0019, 0.0018, 0.0018, 0.0020, 0.0025, 0.0019, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.0628e-05, 7.4518e-05, 6.8631e-05, 7.1660e-05, 7.7470e-05, 9.4132e-05, 7.8941e-05, 7.2364e-05], device='cuda:0') 2022-12-07 12:58:17,826 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:18,608 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.121e+02 2.611e+02 3.299e+02 4.173e+02 8.760e+02, threshold=6.598e+02, percent-clipped=4.0 2022-12-07 12:58:21,421 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:36,742 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:58:41,838 INFO [train.py:873] (0/4) Epoch 5, batch 4400, loss[loss=0.2016, simple_loss=0.1973, pruned_loss=0.103, over 3864.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.1915, pruned_loss=0.08314, over 2002795.63 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 12:58:46,338 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:03,460 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6202, 2.0454, 2.7341, 2.6217, 2.6083, 1.8825, 2.8264, 2.1135], device='cuda:0'), covar=tensor([0.0118, 0.0273, 0.0224, 0.0127, 0.0128, 0.0472, 0.0080, 0.0345], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0192, 0.0279, 0.0226, 0.0178, 0.0239, 0.0166, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 12:59:12,368 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 12:59:16,033 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:27,207 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 12:59:36,245 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 12:59:48,180 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.387e+02 3.073e+02 3.868e+02 1.059e+03, threshold=6.147e+02, percent-clipped=1.0 2022-12-07 13:00:02,145 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0546, 3.6110, 3.1983, 2.6076, 3.0529, 3.5488, 3.8544, 3.0908], device='cuda:0'), covar=tensor([0.0417, 0.3627, 0.1205, 0.2704, 0.1040, 0.0608, 0.1141, 0.1517], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0215, 0.0118, 0.0130, 0.0103, 0.0110, 0.0088, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:00:10,398 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:00:11,141 INFO [train.py:873] (0/4) Epoch 5, batch 4500, loss[loss=0.144, simple_loss=0.1738, pruned_loss=0.05713, over 14279.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.1905, pruned_loss=0.08197, over 1968984.37 frames. ], batch size: 44, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:00:24,431 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 13:00:31,281 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:00:44,604 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0963, 0.8209, 0.9978, 0.8970, 1.0735, 0.6816, 1.1955, 1.0100], device='cuda:0'), covar=tensor([0.0854, 0.1164, 0.0551, 0.0698, 0.0972, 0.0483, 0.0613, 0.1862], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0018, 0.0017, 0.0019, 0.0024, 0.0019, 0.0018], device='cuda:0'), out_proj_covar=tensor([6.9429e-05, 7.2472e-05, 6.7169e-05, 6.9628e-05, 7.5022e-05, 9.2041e-05, 7.7321e-05, 7.0869e-05], device='cuda:0') 2022-12-07 13:00:57,895 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:01:16,933 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.392e+02 3.325e+02 4.104e+02 6.523e+02, threshold=6.650e+02, percent-clipped=2.0 2022-12-07 13:01:40,486 INFO [train.py:873] (0/4) Epoch 5, batch 4600, loss[loss=0.1643, simple_loss=0.1812, pruned_loss=0.07363, over 14310.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.1907, pruned_loss=0.08234, over 1932704.85 frames. ], batch size: 63, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:01:40,575 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:02:19,354 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9521, 1.5776, 3.0690, 2.8928, 3.0246, 3.0736, 2.4735, 3.0952], device='cuda:0'), covar=tensor([0.0884, 0.0944, 0.0100, 0.0183, 0.0170, 0.0104, 0.0254, 0.0104], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0156, 0.0101, 0.0141, 0.0117, 0.0120, 0.0094, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:02:46,772 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 2.771e+02 3.403e+02 4.151e+02 6.915e+02, threshold=6.807e+02, percent-clipped=1.0 2022-12-07 13:03:10,235 INFO [train.py:873] (0/4) Epoch 5, batch 4700, loss[loss=0.198, simple_loss=0.1802, pruned_loss=0.1079, over 2611.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.191, pruned_loss=0.083, over 1890865.00 frames. ], batch size: 100, lr: 1.53e-02, grad_scale: 8.0 2022-12-07 13:03:10,316 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:03:35,701 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:03:39,282 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:03:55,972 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-35000.pt 2022-12-07 13:04:02,501 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1049, 1.5248, 1.5036, 1.4683, 1.4643, 1.5732, 1.1671, 0.9929], device='cuda:0'), covar=tensor([0.2573, 0.1423, 0.0561, 0.0906, 0.1074, 0.0857, 0.1888, 0.3566], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0059, 0.0052, 0.0051, 0.0070, 0.0054, 0.0081, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:04:21,152 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.441e+02 3.432e+02 4.388e+02 1.189e+03, threshold=6.863e+02, percent-clipped=8.0 2022-12-07 13:04:37,329 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6141, 5.3360, 5.0823, 5.5126, 5.2457, 5.1069, 5.6501, 5.4340], device='cuda:0'), covar=tensor([0.0696, 0.0683, 0.0625, 0.0674, 0.0617, 0.0386, 0.0678, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0097, 0.0110, 0.0109, 0.0115, 0.0088, 0.0126, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:04:45,219 INFO [train.py:873] (0/4) Epoch 5, batch 4800, loss[loss=0.1713, simple_loss=0.1933, pruned_loss=0.07464, over 14538.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.1912, pruned_loss=0.08319, over 1899266.97 frames. ], batch size: 34, lr: 1.52e-02, grad_scale: 8.0 2022-12-07 13:05:01,161 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:05:24,478 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:05:53,125 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.350e+02 3.192e+02 4.148e+02 1.074e+03, threshold=6.385e+02, percent-clipped=3.0 2022-12-07 13:06:16,617 INFO [train.py:873] (0/4) Epoch 5, batch 4900, loss[loss=0.2185, simple_loss=0.2116, pruned_loss=0.1127, over 14533.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.1913, pruned_loss=0.08342, over 1919076.86 frames. ], batch size: 49, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:06:20,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:07:23,191 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6914, 2.3123, 3.4202, 2.5002, 3.4093, 3.2186, 3.2433, 2.6859], device='cuda:0'), covar=tensor([0.0356, 0.3025, 0.1032, 0.2329, 0.0826, 0.0793, 0.1282, 0.2469], device='cuda:0'), in_proj_covar=tensor([0.0271, 0.0336, 0.0360, 0.0313, 0.0346, 0.0282, 0.0321, 0.0352], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 13:07:24,094 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.440e+01 2.537e+02 3.134e+02 4.134e+02 8.105e+02, threshold=6.269e+02, percent-clipped=2.0 2022-12-07 13:07:24,763 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 13:07:39,502 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2429, 1.8878, 3.3997, 2.2844, 3.1753, 1.8182, 2.4916, 3.1364], device='cuda:0'), covar=tensor([0.0605, 0.4601, 0.0250, 0.6600, 0.0438, 0.3563, 0.1279, 0.0332], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0263, 0.0165, 0.0354, 0.0175, 0.0268, 0.0243, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:07:47,532 INFO [train.py:873] (0/4) Epoch 5, batch 5000, loss[loss=0.1939, simple_loss=0.1949, pruned_loss=0.09642, over 14010.00 frames. ], tot_loss[loss=0.179, simple_loss=0.1917, pruned_loss=0.08315, over 1940803.18 frames. ], batch size: 22, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:07:47,647 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:13,618 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:08:17,372 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:30,438 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:08:53,564 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.768e+02 3.459e+02 4.682e+02 1.378e+03, threshold=6.919e+02, percent-clipped=12.0 2022-12-07 13:08:56,552 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:09:00,104 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:09:13,327 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:09:13,906 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 13:09:17,079 INFO [train.py:873] (0/4) Epoch 5, batch 5100, loss[loss=0.1789, simple_loss=0.1919, pruned_loss=0.08302, over 14220.00 frames. ], tot_loss[loss=0.18, simple_loss=0.1919, pruned_loss=0.08404, over 1864262.63 frames. ], batch size: 60, lr: 1.52e-02, grad_scale: 16.0 2022-12-07 13:09:24,878 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-07 13:09:32,885 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:09:39,882 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:10:08,787 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:10:13,831 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2022-12-07 13:10:15,184 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:10:23,150 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.613e+02 3.282e+02 4.367e+02 7.735e+02, threshold=6.564e+02, percent-clipped=1.0 2022-12-07 13:10:34,112 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:10:44,925 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:10:45,709 INFO [train.py:873] (0/4) Epoch 5, batch 5200, loss[loss=0.1697, simple_loss=0.151, pruned_loss=0.09418, over 1273.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.1933, pruned_loss=0.0856, over 1877407.54 frames. ], batch size: 100, lr: 1.52e-02, grad_scale: 8.0 2022-12-07 13:10:49,207 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8441, 1.7007, 4.5354, 4.2680, 4.0181, 4.5726, 4.1511, 4.5980], device='cuda:0'), covar=tensor([0.1112, 0.1138, 0.0075, 0.0131, 0.0133, 0.0075, 0.0132, 0.0087], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0151, 0.0099, 0.0139, 0.0115, 0.0117, 0.0092, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:10:56,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2022-12-07 13:11:49,887 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8481, 0.7193, 0.7345, 0.7976, 1.0073, 0.6459, 0.9411, 0.9231], device='cuda:0'), covar=tensor([0.0641, 0.0817, 0.0542, 0.0802, 0.0710, 0.0554, 0.0683, 0.0709], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0018, 0.0019, 0.0018, 0.0020, 0.0024, 0.0019, 0.0020], device='cuda:0'), out_proj_covar=tensor([7.2706e-05, 7.4669e-05, 7.0035e-05, 7.4196e-05, 7.7998e-05, 9.2973e-05, 8.0775e-05, 7.7774e-05], device='cuda:0') 2022-12-07 13:11:51,440 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.356e+02 3.494e+02 4.184e+02 1.202e+03, threshold=6.988e+02, percent-clipped=4.0 2022-12-07 13:11:57,279 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:12:13,733 INFO [train.py:873] (0/4) Epoch 5, batch 5300, loss[loss=0.1922, simple_loss=0.2096, pruned_loss=0.08739, over 13966.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.193, pruned_loss=0.08488, over 1872807.73 frames. ], batch size: 26, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:12:51,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:12:53,467 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5076, 3.0016, 2.2685, 3.6637, 3.2488, 3.3953, 2.7790, 2.3421], device='cuda:0'), covar=tensor([0.0438, 0.1617, 0.3942, 0.0280, 0.0829, 0.1341, 0.1444, 0.4502], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0294, 0.0290, 0.0183, 0.0246, 0.0245, 0.0252, 0.0276], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:13:20,928 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.787e+02 3.586e+02 4.482e+02 8.195e+02, threshold=7.172e+02, percent-clipped=2.0 2022-12-07 13:13:23,087 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1134, 2.0527, 2.0404, 2.2336, 1.8788, 1.9654, 2.1795, 2.2372], device='cuda:0'), covar=tensor([0.0889, 0.0959, 0.0916, 0.0759, 0.1112, 0.0791, 0.0907, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0095, 0.0107, 0.0108, 0.0113, 0.0085, 0.0122, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:13:35,955 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1139, 1.1846, 1.1300, 1.1446, 0.7081, 0.8440, 0.8863, 0.7023], device='cuda:0'), covar=tensor([0.0268, 0.0373, 0.0346, 0.0349, 0.0433, 0.0377, 0.0282, 0.0825], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0011, 0.0012, 0.0016, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([6.0644e-05, 6.4161e-05, 6.3069e-05, 5.8199e-05, 6.4785e-05, 8.8549e-05, 6.8137e-05, 8.5150e-05], device='cuda:0') 2022-12-07 13:13:43,969 INFO [train.py:873] (0/4) Epoch 5, batch 5400, loss[loss=0.1734, simple_loss=0.1868, pruned_loss=0.07994, over 11157.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.1914, pruned_loss=0.08372, over 1828992.13 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:13:45,070 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8534, 0.6725, 0.7373, 0.8736, 0.8907, 0.3725, 0.7899, 0.9634], device='cuda:0'), covar=tensor([0.0461, 0.0772, 0.0358, 0.0653, 0.0434, 0.0379, 0.0680, 0.0401], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0018, 0.0018, 0.0019, 0.0024, 0.0019, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.1183e-05, 7.3639e-05, 6.8434e-05, 7.3020e-05, 7.5919e-05, 9.0267e-05, 7.8802e-05, 7.5577e-05], device='cuda:0') 2022-12-07 13:14:22,883 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0202, 3.0631, 3.1571, 2.9964, 3.0390, 2.6384, 1.1738, 2.8472], device='cuda:0'), covar=tensor([0.0301, 0.0343, 0.0485, 0.0418, 0.0345, 0.1116, 0.3381, 0.0338], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0137, 0.0126, 0.0112, 0.0168, 0.0120, 0.0152, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:14:24,255 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 13:14:30,641 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:14:50,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 13:14:51,488 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.811e+02 3.690e+02 4.403e+02 1.133e+03, threshold=7.380e+02, percent-clipped=4.0 2022-12-07 13:14:52,066 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 13:14:52,933 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.50 vs. limit=5.0 2022-12-07 13:14:57,178 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:12,984 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:13,720 INFO [train.py:873] (0/4) Epoch 5, batch 5500, loss[loss=0.161, simple_loss=0.1632, pruned_loss=0.07943, over 4932.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.1901, pruned_loss=0.08209, over 1870068.72 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:15:20,855 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2022-12-07 13:15:33,591 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:15:55,919 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:09,277 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:20,153 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.649e+02 3.424e+02 4.152e+02 8.919e+02, threshold=6.848e+02, percent-clipped=1.0 2022-12-07 13:16:28,217 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:16:42,803 INFO [train.py:873] (0/4) Epoch 5, batch 5600, loss[loss=0.2064, simple_loss=0.2148, pruned_loss=0.09904, over 14264.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.1912, pruned_loss=0.0828, over 1890908.96 frames. ], batch size: 69, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:17:03,267 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:17:15,296 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:17:48,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.385e+02 3.015e+02 3.800e+02 5.921e+02, threshold=6.029e+02, percent-clipped=0.0 2022-12-07 13:18:10,021 INFO [train.py:873] (0/4) Epoch 5, batch 5700, loss[loss=0.1639, simple_loss=0.1758, pruned_loss=0.07601, over 6920.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.1909, pruned_loss=0.0827, over 1877894.38 frames. ], batch size: 100, lr: 1.51e-02, grad_scale: 8.0 2022-12-07 13:18:55,435 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:19:09,773 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 13:19:10,111 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7168, 3.3691, 3.4050, 3.7294, 3.5447, 3.7547, 3.7538, 3.1218], device='cuda:0'), covar=tensor([0.0380, 0.1273, 0.0389, 0.0469, 0.0737, 0.0301, 0.0534, 0.0575], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0221, 0.0147, 0.0137, 0.0144, 0.0116, 0.0219, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 13:19:15,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.574e+02 3.156e+02 4.047e+02 9.769e+02, threshold=6.311e+02, percent-clipped=9.0 2022-12-07 13:19:17,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 13:19:20,514 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:19:25,512 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2459, 1.3670, 1.3861, 1.3950, 1.1973, 0.8337, 1.0010, 0.8424], device='cuda:0'), covar=tensor([0.0773, 0.1438, 0.1005, 0.0703, 0.0784, 0.0383, 0.0597, 0.1494], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0012, 0.0011, 0.0013, 0.0017, 0.0013, 0.0018], device='cuda:0'), out_proj_covar=tensor([6.4678e-05, 6.7394e-05, 6.4346e-05, 5.9907e-05, 6.9316e-05, 9.2960e-05, 7.4802e-05, 8.8178e-05], device='cuda:0') 2022-12-07 13:19:28,752 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8263, 2.7045, 2.6431, 2.8986, 2.5013, 2.4778, 2.8765, 2.8676], device='cuda:0'), covar=tensor([0.0640, 0.0695, 0.0649, 0.0634, 0.0812, 0.0739, 0.0728, 0.0657], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0098, 0.0109, 0.0112, 0.0114, 0.0087, 0.0123, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:19:37,071 INFO [train.py:873] (0/4) Epoch 5, batch 5800, loss[loss=0.1733, simple_loss=0.1983, pruned_loss=0.07413, over 14285.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.191, pruned_loss=0.08225, over 1927042.76 frames. ], batch size: 31, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:19:37,136 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:20:02,747 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:20:13,705 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1209, 1.4487, 3.2504, 1.3147, 3.0343, 3.2068, 2.2722, 3.3798], device='cuda:0'), covar=tensor([0.0200, 0.2387, 0.0252, 0.2087, 0.0885, 0.0313, 0.0810, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0159, 0.0136, 0.0170, 0.0152, 0.0150, 0.0122, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:20:38,551 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1107, 1.3018, 1.5378, 1.3579, 1.2747, 0.9076, 1.4891, 1.1935], device='cuda:0'), covar=tensor([0.0800, 0.1734, 0.0899, 0.1548, 0.1275, 0.0478, 0.0295, 0.0903], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0013, 0.0016, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([6.1762e-05, 6.4030e-05, 6.1089e-05, 5.7592e-05, 6.7027e-05, 8.8361e-05, 6.9268e-05, 8.4836e-05], device='cuda:0') 2022-12-07 13:20:42,662 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.552e+02 3.407e+02 4.382e+02 1.034e+03, threshold=6.813e+02, percent-clipped=3.0 2022-12-07 13:20:45,482 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:04,916 INFO [train.py:873] (0/4) Epoch 5, batch 5900, loss[loss=0.1655, simple_loss=0.1795, pruned_loss=0.07571, over 5979.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.1899, pruned_loss=0.08121, over 1929339.60 frames. ], batch size: 100, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:21:20,649 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:32,809 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:36,992 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:21:39,605 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.23 vs. limit=2.0 2022-12-07 13:21:54,563 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 13:22:10,495 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.756e+02 3.283e+02 4.209e+02 7.399e+02, threshold=6.567e+02, percent-clipped=1.0 2022-12-07 13:22:19,810 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:22:27,217 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:22:33,228 INFO [train.py:873] (0/4) Epoch 5, batch 6000, loss[loss=0.1497, simple_loss=0.1782, pruned_loss=0.06056, over 14080.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.1899, pruned_loss=0.08145, over 1941593.81 frames. ], batch size: 22, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:22:33,228 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 13:22:45,290 INFO [train.py:905] (0/4) Epoch 5, validation: loss=0.1225, simple_loss=0.1661, pruned_loss=0.03949, over 857387.00 frames. 2022-12-07 13:22:45,291 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 13:23:51,459 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.659e+02 3.309e+02 3.881e+02 6.355e+02, threshold=6.617e+02, percent-clipped=0.0 2022-12-07 13:24:13,083 INFO [train.py:873] (0/4) Epoch 5, batch 6100, loss[loss=0.1412, simple_loss=0.1721, pruned_loss=0.05513, over 13925.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.1911, pruned_loss=0.08231, over 1961477.70 frames. ], batch size: 23, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:24:22,178 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1725, 4.0652, 4.2853, 3.8347, 4.1432, 4.5801, 1.5392, 3.9021], device='cuda:0'), covar=tensor([0.0265, 0.0390, 0.0614, 0.0473, 0.0407, 0.0182, 0.3957, 0.0383], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0136, 0.0126, 0.0112, 0.0169, 0.0118, 0.0150, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:24:44,961 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6409, 3.4701, 3.1620, 3.2253, 3.5415, 3.5027, 3.5817, 3.5390], device='cuda:0'), covar=tensor([0.0692, 0.0531, 0.2132, 0.2430, 0.0678, 0.0730, 0.1004, 0.0843], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0224, 0.0356, 0.0445, 0.0258, 0.0316, 0.0325, 0.0271], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:25:06,749 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2481, 1.3713, 1.5055, 1.0135, 0.9353, 1.3410, 0.8045, 1.1777], device='cuda:0'), covar=tensor([0.1142, 0.2078, 0.0586, 0.2564, 0.2711, 0.0560, 0.1705, 0.0727], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0076, 0.0071, 0.0082, 0.0103, 0.0065, 0.0132, 0.0071], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 13:25:19,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.518e+02 3.418e+02 4.191e+02 1.416e+03, threshold=6.837e+02, percent-clipped=13.0 2022-12-07 13:25:22,020 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:25:40,681 INFO [train.py:873] (0/4) Epoch 5, batch 6200, loss[loss=0.1599, simple_loss=0.1849, pruned_loss=0.06749, over 14410.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.191, pruned_loss=0.0823, over 1975266.09 frames. ], batch size: 31, lr: 1.50e-02, grad_scale: 8.0 2022-12-07 13:25:56,840 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:03,268 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2022-12-07 13:26:03,699 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:38,824 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:26:46,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.596e+02 3.351e+02 4.378e+02 1.450e+03, threshold=6.701e+02, percent-clipped=3.0 2022-12-07 13:26:57,908 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:27:08,587 INFO [train.py:873] (0/4) Epoch 5, batch 6300, loss[loss=0.109, simple_loss=0.1395, pruned_loss=0.03919, over 14405.00 frames. ], tot_loss[loss=0.178, simple_loss=0.1913, pruned_loss=0.08236, over 1971518.31 frames. ], batch size: 18, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:27:35,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 13:27:57,547 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:00,215 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:01,972 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0672, 2.5548, 4.9708, 3.2886, 4.7093, 2.3461, 3.4848, 4.5123], device='cuda:0'), covar=tensor([0.0266, 0.4915, 0.0273, 0.8717, 0.0230, 0.3665, 0.1316, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0262, 0.0171, 0.0352, 0.0179, 0.0267, 0.0245, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:28:04,055 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:14,423 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.701e+02 3.163e+02 3.734e+02 7.022e+02, threshold=6.325e+02, percent-clipped=2.0 2022-12-07 13:28:35,795 INFO [train.py:873] (0/4) Epoch 5, batch 6400, loss[loss=0.1888, simple_loss=0.1723, pruned_loss=0.1027, over 3870.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.1909, pruned_loss=0.08213, over 1950312.72 frames. ], batch size: 100, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:28:40,899 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 13:28:51,061 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:53,751 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:28:57,062 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:29:03,101 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2519, 5.0372, 4.7502, 5.3904, 4.8515, 4.6487, 5.3502, 5.2253], device='cuda:0'), covar=tensor([0.0730, 0.0456, 0.0632, 0.0488, 0.0594, 0.0378, 0.0511, 0.0599], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0096, 0.0110, 0.0113, 0.0115, 0.0087, 0.0124, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:29:11,704 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5876, 1.7641, 1.9617, 1.3680, 1.4118, 1.8142, 1.0823, 1.6160], device='cuda:0'), covar=tensor([0.1941, 0.1475, 0.0588, 0.2400, 0.2448, 0.0631, 0.3450, 0.0949], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0078, 0.0076, 0.0085, 0.0106, 0.0068, 0.0137, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:29:20,789 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:29:31,877 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9107, 0.6480, 0.8129, 0.7855, 0.6389, 0.6185, 0.5769, 0.4655], device='cuda:0'), covar=tensor([0.0403, 0.0461, 0.0509, 0.0326, 0.0635, 0.0604, 0.0382, 0.1173], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0013, 0.0016, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([6.2405e-05, 6.4754e-05, 6.1520e-05, 5.7224e-05, 6.6491e-05, 8.7931e-05, 7.2431e-05, 8.6963e-05], device='cuda:0') 2022-12-07 13:29:41,187 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.894e+02 3.789e+02 4.825e+02 7.842e+02, threshold=7.578e+02, percent-clipped=4.0 2022-12-07 13:29:58,644 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2225, 2.2550, 2.1644, 2.2436, 1.7234, 2.2660, 2.1249, 0.8856], device='cuda:0'), covar=tensor([0.2789, 0.0717, 0.0898, 0.0955, 0.1476, 0.0775, 0.1990, 0.4476], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0061, 0.0053, 0.0054, 0.0074, 0.0057, 0.0084, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:30:02,669 INFO [train.py:873] (0/4) Epoch 5, batch 6500, loss[loss=0.1546, simple_loss=0.176, pruned_loss=0.06656, over 13873.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.1924, pruned_loss=0.08343, over 1941026.06 frames. ], batch size: 20, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:30:13,310 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:30:17,429 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0224, 3.7108, 3.6712, 3.9910, 3.8756, 3.5675, 4.0792, 3.4022], device='cuda:0'), covar=tensor([0.0459, 0.1016, 0.0387, 0.0530, 0.0660, 0.1049, 0.0510, 0.0553], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0219, 0.0143, 0.0138, 0.0140, 0.0115, 0.0212, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 13:30:51,582 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:30:55,961 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.31 vs. limit=2.0 2022-12-07 13:31:07,675 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.577e+02 3.368e+02 4.224e+02 7.324e+02, threshold=6.736e+02, percent-clipped=0.0 2022-12-07 13:31:18,698 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:31:28,830 INFO [train.py:873] (0/4) Epoch 5, batch 6600, loss[loss=0.1745, simple_loss=0.1874, pruned_loss=0.08081, over 13542.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.1906, pruned_loss=0.0818, over 1955537.01 frames. ], batch size: 100, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:31:44,243 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:32:00,561 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:32:23,256 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:32:34,637 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.557e+01 2.430e+02 3.121e+02 4.365e+02 9.299e+02, threshold=6.241e+02, percent-clipped=3.0 2022-12-07 13:32:56,984 INFO [train.py:873] (0/4) Epoch 5, batch 6700, loss[loss=0.173, simple_loss=0.193, pruned_loss=0.07652, over 14155.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.1905, pruned_loss=0.08148, over 1989938.26 frames. ], batch size: 29, lr: 1.49e-02, grad_scale: 8.0 2022-12-07 13:33:04,795 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:07,462 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:10,032 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:13,470 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:16,919 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:45,506 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:33:59,092 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:34:03,149 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.484e+02 3.120e+02 4.102e+02 1.074e+03, threshold=6.239e+02, percent-clipped=7.0 2022-12-07 13:34:24,642 INFO [train.py:873] (0/4) Epoch 5, batch 6800, loss[loss=0.1957, simple_loss=0.2044, pruned_loss=0.09355, over 14526.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.1902, pruned_loss=0.08193, over 1969166.00 frames. ], batch size: 51, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:34:31,238 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:34:38,168 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8555, 0.9738, 0.9073, 0.9525, 0.9811, 0.5682, 1.0832, 0.9636], device='cuda:0'), covar=tensor([0.0828, 0.0764, 0.0509, 0.0873, 0.1071, 0.0558, 0.0449, 0.0885], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0018, 0.0018, 0.0020, 0.0025, 0.0019, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.6017e-05, 7.9403e-05, 7.1826e-05, 7.5691e-05, 8.1779e-05, 9.7430e-05, 8.2113e-05, 7.6311e-05], device='cuda:0') 2022-12-07 13:34:38,995 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:35:03,067 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0093, 2.8037, 2.5496, 2.6340, 2.9574, 2.8570, 2.9807, 2.9339], device='cuda:0'), covar=tensor([0.0853, 0.0746, 0.2053, 0.2772, 0.0784, 0.0892, 0.1100, 0.0909], device='cuda:0'), in_proj_covar=tensor([0.0303, 0.0234, 0.0371, 0.0464, 0.0275, 0.0329, 0.0341, 0.0288], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:35:20,168 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.42 vs. limit=2.0 2022-12-07 13:35:21,469 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7911, 0.8252, 0.6772, 0.9824, 0.9955, 0.4584, 0.8868, 0.8129], device='cuda:0'), covar=tensor([0.0779, 0.0486, 0.0428, 0.0445, 0.0316, 0.0439, 0.0391, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0019, 0.0018, 0.0018, 0.0020, 0.0026, 0.0019, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.6484e-05, 7.9786e-05, 7.1908e-05, 7.5724e-05, 8.1989e-05, 9.8086e-05, 8.2519e-05, 7.7294e-05], device='cuda:0') 2022-12-07 13:35:30,147 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.403e+02 2.984e+02 3.959e+02 1.091e+03, threshold=5.968e+02, percent-clipped=8.0 2022-12-07 13:35:52,585 INFO [train.py:873] (0/4) Epoch 5, batch 6900, loss[loss=0.1852, simple_loss=0.1968, pruned_loss=0.08678, over 14283.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.1901, pruned_loss=0.08218, over 1912264.81 frames. ], batch size: 69, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:36:02,903 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 13:36:02,931 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4858, 3.5051, 3.7570, 3.3042, 3.5383, 3.3890, 1.3127, 3.3574], device='cuda:0'), covar=tensor([0.0267, 0.0297, 0.0372, 0.0417, 0.0301, 0.0447, 0.3340, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0139, 0.0124, 0.0117, 0.0171, 0.0119, 0.0153, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:36:09,319 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8206, 0.9003, 0.9148, 0.9243, 0.7998, 0.6205, 0.8585, 0.5361], device='cuda:0'), covar=tensor([0.0304, 0.0338, 0.0280, 0.0106, 0.0478, 0.0723, 0.0320, 0.0635], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0011, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.0917e-05, 6.4084e-05, 6.1681e-05, 5.6052e-05, 6.5521e-05, 8.5246e-05, 7.0511e-05, 8.1648e-05], device='cuda:0') 2022-12-07 13:36:15,098 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 13:36:39,880 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7342, 5.4757, 5.1345, 5.9769, 5.3905, 4.6604, 5.8501, 5.7324], device='cuda:0'), covar=tensor([0.0645, 0.0500, 0.0558, 0.0354, 0.0665, 0.0417, 0.0531, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0095, 0.0110, 0.0113, 0.0117, 0.0089, 0.0126, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:36:58,808 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.434e+02 3.159e+02 4.022e+02 7.807e+02, threshold=6.317e+02, percent-clipped=6.0 2022-12-07 13:37:20,127 INFO [train.py:873] (0/4) Epoch 5, batch 7000, loss[loss=0.1201, simple_loss=0.1482, pruned_loss=0.04604, over 13584.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.1891, pruned_loss=0.08068, over 2000869.49 frames. ], batch size: 17, lr: 1.48e-02, grad_scale: 4.0 2022-12-07 13:37:31,311 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:33,840 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:36,245 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:37,162 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:37:37,201 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:12,649 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:14,991 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:16,760 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:18,495 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:26,221 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.611e+02 3.349e+02 4.594e+02 9.259e+02, threshold=6.698e+02, percent-clipped=6.0 2022-12-07 13:38:29,800 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:46,857 INFO [train.py:873] (0/4) Epoch 5, batch 7100, loss[loss=0.1664, simple_loss=0.1508, pruned_loss=0.091, over 1276.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.1892, pruned_loss=0.08089, over 2000284.94 frames. ], batch size: 100, lr: 1.48e-02, grad_scale: 4.0 2022-12-07 13:38:52,919 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 13:38:56,251 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:38:59,720 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:39:25,681 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2022-12-07 13:39:34,761 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:39:49,910 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 13:39:53,448 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.303e+02 3.351e+02 4.171e+02 6.616e+02, threshold=6.701e+02, percent-clipped=0.0 2022-12-07 13:39:53,734 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:40:14,166 INFO [train.py:873] (0/4) Epoch 5, batch 7200, loss[loss=0.1592, simple_loss=0.1807, pruned_loss=0.06884, over 14369.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.19, pruned_loss=0.08129, over 2026394.47 frames. ], batch size: 41, lr: 1.48e-02, grad_scale: 8.0 2022-12-07 13:40:25,498 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 13:41:05,525 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6300, 1.3745, 2.9569, 1.4997, 2.8262, 2.8554, 2.0628, 2.8915], device='cuda:0'), covar=tensor([0.0281, 0.2634, 0.0280, 0.1936, 0.0359, 0.0423, 0.0775, 0.0286], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0158, 0.0134, 0.0167, 0.0155, 0.0149, 0.0122, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:41:08,035 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 13:41:21,481 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.612e+02 3.336e+02 4.094e+02 6.871e+02, threshold=6.672e+02, percent-clipped=1.0 2022-12-07 13:41:21,661 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9102, 0.9083, 0.7461, 0.9645, 1.0187, 0.1862, 0.8816, 0.8734], device='cuda:0'), covar=tensor([0.0186, 0.0243, 0.0080, 0.0164, 0.0108, 0.0075, 0.0287, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0018, 0.0018, 0.0018, 0.0020, 0.0024, 0.0019, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.4007e-05, 7.7068e-05, 7.0311e-05, 7.5209e-05, 8.0017e-05, 9.4390e-05, 8.2852e-05, 7.6719e-05], device='cuda:0') 2022-12-07 13:41:24,416 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0036, 2.0796, 2.5641, 1.5885, 1.6877, 2.2120, 1.1476, 2.1902], device='cuda:0'), covar=tensor([0.1019, 0.1395, 0.0475, 0.2269, 0.2911, 0.0685, 0.5491, 0.0504], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0082, 0.0078, 0.0085, 0.0110, 0.0070, 0.0141, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:41:32,710 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0379, 3.1617, 4.6346, 3.5389, 4.5735, 4.5348, 4.1615, 3.9340], device='cuda:0'), covar=tensor([0.0287, 0.2845, 0.0743, 0.1512, 0.0593, 0.0562, 0.2171, 0.2012], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0337, 0.0368, 0.0313, 0.0351, 0.0288, 0.0341, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:41:42,810 INFO [train.py:873] (0/4) Epoch 5, batch 7300, loss[loss=0.1807, simple_loss=0.1968, pruned_loss=0.08229, over 14292.00 frames. ], tot_loss[loss=0.174, simple_loss=0.1884, pruned_loss=0.07986, over 2039538.60 frames. ], batch size: 76, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:41:58,555 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:36,056 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9788, 2.7247, 2.7835, 1.8698, 2.5165, 2.6568, 3.0469, 2.4853], device='cuda:0'), covar=tensor([0.0621, 0.1632, 0.1033, 0.2027, 0.1039, 0.0539, 0.0558, 0.1504], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0205, 0.0121, 0.0127, 0.0105, 0.0113, 0.0088, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 13:42:41,343 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:42,179 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:50,447 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:42:51,224 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.435e+02 3.251e+02 4.059e+02 7.905e+02, threshold=6.501e+02, percent-clipped=2.0 2022-12-07 13:42:57,316 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9120, 2.7527, 2.4421, 2.5008, 2.8699, 2.8022, 2.9233, 2.8765], device='cuda:0'), covar=tensor([0.1107, 0.0955, 0.2537, 0.3392, 0.0985, 0.1066, 0.1341, 0.1080], device='cuda:0'), in_proj_covar=tensor([0.0297, 0.0231, 0.0370, 0.0466, 0.0276, 0.0329, 0.0341, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:43:12,150 INFO [train.py:873] (0/4) Epoch 5, batch 7400, loss[loss=0.1671, simple_loss=0.1726, pruned_loss=0.08077, over 4986.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.1892, pruned_loss=0.08083, over 2009816.25 frames. ], batch size: 100, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:43:22,048 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:43:24,020 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:44:01,704 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6381, 3.5714, 3.7325, 3.2894, 3.6501, 3.5761, 1.5317, 3.4660], device='cuda:0'), covar=tensor([0.0203, 0.0259, 0.0403, 0.0545, 0.0268, 0.0322, 0.2703, 0.0240], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0136, 0.0123, 0.0114, 0.0168, 0.0117, 0.0148, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:44:05,011 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:44:15,037 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:44:17,758 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1479, 0.9495, 1.1257, 0.8876, 0.9178, 0.7467, 0.9509, 0.8146], device='cuda:0'), covar=tensor([0.0280, 0.0603, 0.0521, 0.0360, 0.0476, 0.0312, 0.0232, 0.0634], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0009, 0.0012, 0.0015, 0.0011, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.0768e-05, 6.4506e-05, 5.9434e-05, 5.5125e-05, 6.4624e-05, 8.2959e-05, 6.8533e-05, 8.2027e-05], device='cuda:0') 2022-12-07 13:44:19,313 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.420e+02 3.401e+02 4.272e+02 7.052e+02, threshold=6.801e+02, percent-clipped=3.0 2022-12-07 13:44:41,302 INFO [train.py:873] (0/4) Epoch 5, batch 7500, loss[loss=0.1818, simple_loss=0.1573, pruned_loss=0.1032, over 1211.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.19, pruned_loss=0.08154, over 1996432.97 frames. ], batch size: 100, lr: 1.47e-02, grad_scale: 8.0 2022-12-07 13:45:09,833 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2022-12-07 13:45:25,372 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2369, 1.3338, 1.5096, 0.8341, 0.8564, 1.4093, 0.8325, 1.1884], device='cuda:0'), covar=tensor([0.0764, 0.1747, 0.0455, 0.2543, 0.2404, 0.0519, 0.2024, 0.0611], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0079, 0.0075, 0.0083, 0.0107, 0.0068, 0.0134, 0.0070], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:45:29,211 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-5.pt 2022-12-07 13:46:12,626 INFO [train.py:873] (0/4) Epoch 6, batch 0, loss[loss=0.2104, simple_loss=0.2178, pruned_loss=0.1016, over 14248.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2178, pruned_loss=0.1016, over 14248.00 frames. ], batch size: 31, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:46:12,627 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 13:46:20,009 INFO [train.py:905] (0/4) Epoch 6, validation: loss=0.1313, simple_loss=0.1749, pruned_loss=0.04388, over 857387.00 frames. 2022-12-07 13:46:20,010 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 13:46:32,650 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.323e+01 1.772e+02 2.799e+02 3.695e+02 8.641e+02, threshold=5.598e+02, percent-clipped=1.0 2022-12-07 13:46:43,523 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 13:47:50,541 INFO [train.py:873] (0/4) Epoch 6, batch 100, loss[loss=0.1736, simple_loss=0.1894, pruned_loss=0.07889, over 13528.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.1896, pruned_loss=0.0795, over 856691.96 frames. ], batch size: 100, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:47:56,915 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2022-12-07 13:47:57,553 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2132, 2.2351, 4.3304, 2.7030, 4.0428, 2.1197, 3.0743, 3.8263], device='cuda:0'), covar=tensor([0.0635, 0.5655, 0.0356, 1.0745, 0.0504, 0.4312, 0.1468, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0256, 0.0163, 0.0345, 0.0176, 0.0263, 0.0240, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:48:01,911 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:48:02,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.495e+02 3.181e+02 4.062e+02 9.836e+02, threshold=6.363e+02, percent-clipped=3.0 2022-12-07 13:48:10,544 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1455, 3.6069, 2.7528, 4.3064, 4.0795, 4.1388, 3.4513, 2.9793], device='cuda:0'), covar=tensor([0.0512, 0.1298, 0.3715, 0.0274, 0.0655, 0.1117, 0.1124, 0.3775], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0301, 0.0291, 0.0184, 0.0251, 0.0253, 0.0253, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:48:44,529 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:48:50,778 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:49:16,192 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5547, 3.3876, 3.0039, 2.4356, 2.8670, 3.2085, 3.3329, 2.7151], device='cuda:0'), covar=tensor([0.0665, 0.2623, 0.1395, 0.2336, 0.1061, 0.0576, 0.1194, 0.1821], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0201, 0.0118, 0.0125, 0.0104, 0.0110, 0.0089, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 13:49:19,530 INFO [train.py:873] (0/4) Epoch 6, batch 200, loss[loss=0.1705, simple_loss=0.1878, pruned_loss=0.07657, over 10352.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.1881, pruned_loss=0.07836, over 1311702.62 frames. ], batch size: 100, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:49:22,294 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8590, 2.4960, 3.6499, 2.8206, 3.5219, 3.4916, 3.4185, 2.9081], device='cuda:0'), covar=tensor([0.0352, 0.2595, 0.0753, 0.1846, 0.0703, 0.0620, 0.1388, 0.2100], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0336, 0.0370, 0.0318, 0.0351, 0.0291, 0.0336, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 13:49:24,565 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4787, 2.4949, 5.3542, 4.8326, 4.8467, 5.3524, 4.9677, 5.4490], device='cuda:0'), covar=tensor([0.1049, 0.0986, 0.0040, 0.0095, 0.0078, 0.0070, 0.0079, 0.0051], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0154, 0.0103, 0.0142, 0.0117, 0.0121, 0.0092, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:49:27,113 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:49:31,828 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.443e+02 3.000e+02 3.883e+02 6.442e+02, threshold=6.000e+02, percent-clipped=2.0 2022-12-07 13:49:45,636 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:50:09,562 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:50:47,793 INFO [train.py:873] (0/4) Epoch 6, batch 300, loss[loss=0.1552, simple_loss=0.1744, pruned_loss=0.06799, over 14254.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.187, pruned_loss=0.07835, over 1579519.59 frames. ], batch size: 37, lr: 1.37e-02, grad_scale: 8.0 2022-12-07 13:50:50,503 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9727, 1.7116, 3.2282, 2.2407, 3.3058, 1.6808, 2.4877, 2.9632], device='cuda:0'), covar=tensor([0.0652, 0.5629, 0.0429, 0.7989, 0.0509, 0.4562, 0.1733, 0.0504], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0255, 0.0167, 0.0346, 0.0176, 0.0262, 0.0239, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 13:50:59,800 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.428e+02 3.008e+02 3.765e+02 6.349e+02, threshold=6.017e+02, percent-clipped=1.0 2022-12-07 13:51:23,797 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1409, 3.0988, 3.3057, 3.0585, 3.0896, 2.8517, 1.2784, 2.9042], device='cuda:0'), covar=tensor([0.0230, 0.0258, 0.0320, 0.0334, 0.0236, 0.0739, 0.2725, 0.0242], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0135, 0.0122, 0.0115, 0.0168, 0.0116, 0.0150, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:52:07,491 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2498, 1.2627, 2.6707, 1.4329, 2.5642, 2.5771, 2.0437, 2.6094], device='cuda:0'), covar=tensor([0.0434, 0.3171, 0.0365, 0.2478, 0.0484, 0.0570, 0.0852, 0.0434], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0158, 0.0136, 0.0169, 0.0153, 0.0151, 0.0123, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:52:16,128 INFO [train.py:873] (0/4) Epoch 6, batch 400, loss[loss=0.1548, simple_loss=0.1807, pruned_loss=0.06443, over 14117.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.1864, pruned_loss=0.07745, over 1765807.70 frames. ], batch size: 29, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:52:21,612 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6659, 1.5718, 4.4788, 2.1002, 4.1848, 4.5766, 4.1032, 4.9890], device='cuda:0'), covar=tensor([0.0143, 0.2738, 0.0239, 0.1955, 0.0287, 0.0271, 0.0286, 0.0101], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0157, 0.0135, 0.0168, 0.0153, 0.0150, 0.0123, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:52:28,766 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.719e+02 3.194e+02 3.868e+02 8.474e+02, threshold=6.387e+02, percent-clipped=5.0 2022-12-07 13:52:42,717 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-07 13:53:01,392 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8607, 3.7441, 4.0386, 3.6089, 3.8634, 3.9333, 1.4518, 3.6258], device='cuda:0'), covar=tensor([0.0192, 0.0295, 0.0357, 0.0413, 0.0286, 0.0345, 0.2798, 0.0272], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0136, 0.0123, 0.0116, 0.0168, 0.0115, 0.0150, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:53:05,029 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5244, 3.2580, 3.3074, 3.6753, 3.2117, 3.0178, 3.6462, 3.5656], device='cuda:0'), covar=tensor([0.0985, 0.0831, 0.0774, 0.0672, 0.1007, 0.0711, 0.0753, 0.0824], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0099, 0.0111, 0.0113, 0.0115, 0.0087, 0.0128, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:53:27,387 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9978, 1.9817, 1.9273, 2.0896, 1.7069, 1.9594, 2.0730, 2.0623], device='cuda:0'), covar=tensor([0.0892, 0.1036, 0.0907, 0.0774, 0.1369, 0.0837, 0.1068, 0.0782], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0098, 0.0111, 0.0113, 0.0114, 0.0086, 0.0127, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:53:45,478 INFO [train.py:873] (0/4) Epoch 6, batch 500, loss[loss=0.1713, simple_loss=0.1884, pruned_loss=0.07711, over 11171.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.1869, pruned_loss=0.07805, over 1847546.87 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:53:57,615 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.649e+02 3.367e+02 4.608e+02 8.885e+02, threshold=6.735e+02, percent-clipped=8.0 2022-12-07 13:54:06,488 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:54:28,800 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:55:12,570 INFO [train.py:873] (0/4) Epoch 6, batch 600, loss[loss=0.1988, simple_loss=0.2018, pruned_loss=0.09787, over 4987.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.1872, pruned_loss=0.07848, over 1848219.91 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:55:21,381 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:55:22,073 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3153, 3.0124, 3.0993, 3.4202, 2.9745, 2.8016, 3.3999, 3.3361], device='cuda:0'), covar=tensor([0.0753, 0.0823, 0.0757, 0.0641, 0.0925, 0.0708, 0.0664, 0.0805], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0097, 0.0111, 0.0113, 0.0113, 0.0086, 0.0126, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:55:24,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.417e+02 3.018e+02 3.905e+02 9.436e+02, threshold=6.036e+02, percent-clipped=5.0 2022-12-07 13:56:35,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7694, 2.6559, 2.5804, 2.8844, 2.3436, 2.4169, 2.8682, 2.8152], device='cuda:0'), covar=tensor([0.0853, 0.0822, 0.0766, 0.0655, 0.1147, 0.0804, 0.0866, 0.0796], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0099, 0.0112, 0.0116, 0.0117, 0.0089, 0.0130, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 13:56:41,307 INFO [train.py:873] (0/4) Epoch 6, batch 700, loss[loss=0.195, simple_loss=0.1932, pruned_loss=0.09836, over 7805.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.1868, pruned_loss=0.07817, over 1965807.64 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:56:46,369 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9412, 4.4748, 4.4086, 4.8409, 4.7071, 4.2749, 4.8071, 4.2200], device='cuda:0'), covar=tensor([0.0322, 0.0911, 0.0303, 0.0400, 0.0540, 0.0491, 0.0525, 0.0455], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0225, 0.0149, 0.0141, 0.0143, 0.0116, 0.0219, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 13:56:53,909 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.144e+02 2.377e+02 3.088e+02 3.950e+02 6.999e+02, threshold=6.177e+02, percent-clipped=2.0 2022-12-07 13:57:15,859 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5466, 1.3157, 2.7798, 1.4638, 2.7112, 2.7384, 2.0491, 2.8614], device='cuda:0'), covar=tensor([0.0225, 0.2140, 0.0265, 0.1712, 0.0333, 0.0382, 0.0826, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0160, 0.0140, 0.0172, 0.0155, 0.0153, 0.0125, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 13:58:10,602 INFO [train.py:873] (0/4) Epoch 6, batch 800, loss[loss=0.172, simple_loss=0.1951, pruned_loss=0.07448, over 13988.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.1865, pruned_loss=0.07809, over 1912502.42 frames. ], batch size: 22, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:58:23,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.424e+02 2.905e+02 3.924e+02 6.932e+02, threshold=5.809e+02, percent-clipped=2.0 2022-12-07 13:58:27,196 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 13:58:32,330 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:58:32,693 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 13:59:15,088 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:59:30,977 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 13:59:39,506 INFO [train.py:873] (0/4) Epoch 6, batch 900, loss[loss=0.1997, simple_loss=0.1997, pruned_loss=0.0999, over 10328.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.1872, pruned_loss=0.07892, over 1920867.21 frames. ], batch size: 100, lr: 1.36e-02, grad_scale: 8.0 2022-12-07 13:59:44,640 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 13:59:52,236 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.457e+02 3.272e+02 3.954e+02 1.451e+03, threshold=6.544e+02, percent-clipped=7.0 2022-12-07 14:00:40,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1266, 1.6743, 1.6203, 1.6599, 1.5081, 1.7460, 1.3328, 1.0862], device='cuda:0'), covar=tensor([0.2171, 0.0641, 0.0404, 0.0368, 0.0690, 0.0329, 0.1473, 0.1432], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0061, 0.0051, 0.0053, 0.0073, 0.0057, 0.0083, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:01:02,993 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1822, 3.0598, 2.7022, 2.8512, 3.1428, 3.0492, 3.1682, 3.1323], device='cuda:0'), covar=tensor([0.0911, 0.0740, 0.2334, 0.2867, 0.0805, 0.0898, 0.1143, 0.0918], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0234, 0.0373, 0.0461, 0.0277, 0.0336, 0.0345, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:01:08,897 INFO [train.py:873] (0/4) Epoch 6, batch 1000, loss[loss=0.1611, simple_loss=0.1888, pruned_loss=0.0667, over 14201.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.1867, pruned_loss=0.07832, over 1949128.94 frames. ], batch size: 46, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:01:18,856 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-07 14:01:19,406 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:01:21,140 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 2.592e+02 3.216e+02 4.248e+02 8.098e+02, threshold=6.432e+02, percent-clipped=2.0 2022-12-07 14:01:58,849 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-07 14:02:13,785 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:02:31,485 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 14:02:37,249 INFO [train.py:873] (0/4) Epoch 6, batch 1100, loss[loss=0.1588, simple_loss=0.1812, pruned_loss=0.06822, over 14620.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.1864, pruned_loss=0.07854, over 1901242.88 frames. ], batch size: 22, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:02:46,432 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2022-12-07 14:02:50,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.088e+02 2.398e+02 3.098e+02 3.888e+02 7.053e+02, threshold=6.197e+02, percent-clipped=1.0 2022-12-07 14:03:33,208 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4625, 1.0331, 1.3807, 0.8345, 1.1396, 1.3332, 1.1917, 1.1986], device='cuda:0'), covar=tensor([0.0296, 0.0832, 0.0489, 0.0470, 0.0883, 0.0734, 0.0296, 0.1052], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0205, 0.0118, 0.0128, 0.0109, 0.0114, 0.0088, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:04:07,241 INFO [train.py:873] (0/4) Epoch 6, batch 1200, loss[loss=0.1673, simple_loss=0.1839, pruned_loss=0.07532, over 14468.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.1862, pruned_loss=0.07831, over 1900065.89 frames. ], batch size: 51, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:04:11,625 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:04:14,663 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 14:04:19,444 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.136e+01 2.396e+02 3.010e+02 3.930e+02 7.519e+02, threshold=6.020e+02, percent-clipped=3.0 2022-12-07 14:04:19,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 14:04:24,524 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2022-12-07 14:04:35,079 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9096, 0.9423, 0.8092, 0.8922, 0.7694, 0.6078, 0.9061, 0.7644], device='cuda:0'), covar=tensor([0.0848, 0.1102, 0.0461, 0.1071, 0.0691, 0.0415, 0.0738, 0.0659], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0020, 0.0026, 0.0021, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.9599e-05, 8.2467e-05, 7.4747e-05, 7.9573e-05, 8.3349e-05, 1.0041e-04, 8.8227e-05, 7.9483e-05], device='cuda:0') 2022-12-07 14:04:54,187 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:05:35,512 INFO [train.py:873] (0/4) Epoch 6, batch 1300, loss[loss=0.2271, simple_loss=0.1919, pruned_loss=0.1311, over 1290.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.1857, pruned_loss=0.07799, over 1864442.25 frames. ], batch size: 100, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:05:48,564 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.179e+02 2.329e+02 2.773e+02 3.737e+02 7.179e+02, threshold=5.547e+02, percent-clipped=2.0 2022-12-07 14:06:36,437 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:06:41,007 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6300, 0.5426, 0.6827, 0.7108, 0.5927, 0.5459, 0.3981, 0.4616], device='cuda:0'), covar=tensor([0.0163, 0.0147, 0.0147, 0.0135, 0.0304, 0.0305, 0.0180, 0.0378], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.3672e-05, 6.8206e-05, 6.1242e-05, 5.9487e-05, 6.7547e-05, 8.8613e-05, 7.3654e-05, 8.5059e-05], device='cuda:0') 2022-12-07 14:06:41,852 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:07:05,236 INFO [train.py:873] (0/4) Epoch 6, batch 1400, loss[loss=0.1821, simple_loss=0.1591, pruned_loss=0.1026, over 1219.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.1863, pruned_loss=0.07781, over 1909067.83 frames. ], batch size: 100, lr: 1.35e-02, grad_scale: 8.0 2022-12-07 14:07:17,497 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.370e+02 2.964e+02 3.794e+02 7.798e+02, threshold=5.929e+02, percent-clipped=9.0 2022-12-07 14:07:36,730 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:08:03,259 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0015, 4.0167, 4.2183, 3.5322, 4.0810, 4.1115, 1.5659, 3.8092], device='cuda:0'), covar=tensor([0.0215, 0.0249, 0.0390, 0.0496, 0.0254, 0.0284, 0.3158, 0.0265], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0138, 0.0123, 0.0117, 0.0168, 0.0119, 0.0150, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:08:09,791 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8937, 0.8473, 0.7445, 0.8243, 0.8236, 0.5441, 0.9119, 0.6967], device='cuda:0'), covar=tensor([0.0634, 0.0808, 0.0343, 0.0987, 0.0738, 0.0466, 0.0584, 0.0682], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0018, 0.0019, 0.0020, 0.0026, 0.0020, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.8080e-05, 8.2820e-05, 7.3375e-05, 7.9551e-05, 8.3807e-05, 1.0024e-04, 8.6384e-05, 7.8971e-05], device='cuda:0') 2022-12-07 14:08:10,687 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:08:33,853 INFO [train.py:873] (0/4) Epoch 6, batch 1500, loss[loss=0.2137, simple_loss=0.2016, pruned_loss=0.1129, over 4975.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.1859, pruned_loss=0.07772, over 1905110.29 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 16.0 2022-12-07 14:08:46,906 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.464e+02 2.970e+02 3.813e+02 8.133e+02, threshold=5.940e+02, percent-clipped=4.0 2022-12-07 14:08:50,764 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.88 vs. limit=5.0 2022-12-07 14:09:04,269 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:10:00,569 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3339, 1.7420, 1.8104, 1.7725, 1.5606, 1.8851, 1.4429, 1.1117], device='cuda:0'), covar=tensor([0.2230, 0.1148, 0.0558, 0.0463, 0.1150, 0.0505, 0.2362, 0.3518], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0060, 0.0050, 0.0052, 0.0075, 0.0056, 0.0083, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 14:10:03,117 INFO [train.py:873] (0/4) Epoch 6, batch 1600, loss[loss=0.1662, simple_loss=0.1885, pruned_loss=0.07192, over 14284.00 frames. ], tot_loss[loss=0.171, simple_loss=0.1859, pruned_loss=0.07808, over 1896166.99 frames. ], batch size: 39, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:10:16,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.549e+01 2.367e+02 3.070e+02 3.991e+02 2.269e+03, threshold=6.141e+02, percent-clipped=9.0 2022-12-07 14:10:24,352 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7645, 0.5634, 0.6020, 0.7175, 0.7310, 0.2313, 0.5928, 0.6886], device='cuda:0'), covar=tensor([0.0226, 0.0409, 0.0133, 0.0208, 0.0134, 0.0101, 0.0442, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0018, 0.0019, 0.0020, 0.0026, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.8789e-05, 8.3125e-05, 7.4181e-05, 8.0494e-05, 8.3456e-05, 1.0148e-04, 8.7768e-05, 7.9694e-05], device='cuda:0') 2022-12-07 14:10:31,829 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9539, 1.7491, 4.3704, 4.2132, 4.1199, 4.4233, 4.0247, 4.4141], device='cuda:0'), covar=tensor([0.1155, 0.1204, 0.0062, 0.0095, 0.0121, 0.0076, 0.0101, 0.0080], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0151, 0.0104, 0.0143, 0.0118, 0.0122, 0.0094, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:10:49,218 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:03,658 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:31,578 INFO [train.py:873] (0/4) Epoch 6, batch 1700, loss[loss=0.1943, simple_loss=0.199, pruned_loss=0.09482, over 9487.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.1858, pruned_loss=0.07777, over 1934210.56 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:11:36,861 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-07 14:11:43,975 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:45,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.465e+02 3.198e+02 4.096e+02 7.326e+02, threshold=6.396e+02, percent-clipped=2.0 2022-12-07 14:11:46,572 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:11:58,800 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:12:21,441 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8525, 1.3872, 1.7934, 1.3203, 1.4618, 1.7490, 1.6291, 1.5786], device='cuda:0'), covar=tensor([0.0391, 0.0985, 0.0583, 0.0925, 0.1240, 0.0987, 0.0310, 0.1773], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0198, 0.0117, 0.0122, 0.0107, 0.0111, 0.0085, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:12:24,034 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9197, 0.8491, 0.6173, 0.9510, 1.0261, 0.3748, 1.0280, 0.8478], device='cuda:0'), covar=tensor([0.0467, 0.0525, 0.0221, 0.0488, 0.0250, 0.0369, 0.0382, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0020, 0.0018, 0.0018, 0.0020, 0.0026, 0.0020, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.9222e-05, 8.2923e-05, 7.3195e-05, 7.8624e-05, 8.2290e-05, 1.0086e-04, 8.6447e-05, 7.8950e-05], device='cuda:0') 2022-12-07 14:13:01,689 INFO [train.py:873] (0/4) Epoch 6, batch 1800, loss[loss=0.2036, simple_loss=0.2012, pruned_loss=0.103, over 6929.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.1864, pruned_loss=0.07833, over 1944981.32 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 4.0 2022-12-07 14:13:15,824 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.436e+02 3.001e+02 3.832e+02 8.283e+02, threshold=6.003e+02, percent-clipped=3.0 2022-12-07 14:13:26,807 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:13:27,561 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:13:37,876 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5303, 2.1992, 3.1743, 1.9289, 2.1192, 2.5870, 1.2746, 2.5401], device='cuda:0'), covar=tensor([0.1209, 0.1379, 0.0589, 0.2581, 0.2439, 0.1278, 0.6227, 0.1095], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0080, 0.0078, 0.0084, 0.0107, 0.0069, 0.0134, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:13:57,461 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0177, 1.4809, 3.1272, 1.4738, 3.0595, 3.1821, 2.2952, 3.3642], device='cuda:0'), covar=tensor([0.0220, 0.2527, 0.0299, 0.2044, 0.0672, 0.0326, 0.0800, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0157, 0.0136, 0.0169, 0.0155, 0.0151, 0.0123, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:13:59,691 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2022-12-07 14:14:20,068 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 14:14:29,789 INFO [train.py:873] (0/4) Epoch 6, batch 1900, loss[loss=0.1927, simple_loss=0.2121, pruned_loss=0.08661, over 13968.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.1866, pruned_loss=0.07859, over 1959950.38 frames. ], batch size: 23, lr: 1.34e-02, grad_scale: 4.0 2022-12-07 14:14:44,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.151e+02 2.287e+02 3.025e+02 3.865e+02 4.213e+03, threshold=6.050e+02, percent-clipped=8.0 2022-12-07 14:15:09,436 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 14:15:19,650 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:15:50,473 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.60 vs. limit=5.0 2022-12-07 14:15:55,860 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 14:15:58,843 INFO [train.py:873] (0/4) Epoch 6, batch 2000, loss[loss=0.2335, simple_loss=0.1998, pruned_loss=0.1337, over 1286.00 frames. ], tot_loss[loss=0.172, simple_loss=0.1868, pruned_loss=0.07866, over 1918699.27 frames. ], batch size: 100, lr: 1.34e-02, grad_scale: 8.0 2022-12-07 14:16:02,565 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:03,012 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-07 14:16:05,933 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:12,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.241e+02 2.938e+02 3.748e+02 1.021e+03, threshold=5.875e+02, percent-clipped=4.0 2022-12-07 14:16:14,115 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:25,422 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:16:33,743 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6628, 2.1807, 3.9128, 3.8786, 4.0158, 2.4850, 4.0138, 2.9699], device='cuda:0'), covar=tensor([0.0179, 0.0462, 0.0370, 0.0224, 0.0107, 0.0600, 0.0114, 0.0481], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0202, 0.0298, 0.0237, 0.0188, 0.0246, 0.0187, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 14:16:56,533 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:17:08,351 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:17:12,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 14:17:27,450 INFO [train.py:873] (0/4) Epoch 6, batch 2100, loss[loss=0.1607, simple_loss=0.183, pruned_loss=0.06923, over 14476.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.1866, pruned_loss=0.07831, over 1958301.23 frames. ], batch size: 51, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:17:42,097 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.442e+01 2.310e+02 2.903e+02 3.569e+02 8.046e+02, threshold=5.805e+02, percent-clipped=4.0 2022-12-07 14:17:53,536 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:18:35,826 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:18:41,872 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 14:18:46,872 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-40000.pt 2022-12-07 14:19:00,405 INFO [train.py:873] (0/4) Epoch 6, batch 2200, loss[loss=0.1907, simple_loss=0.1771, pruned_loss=0.1022, over 3898.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.1862, pruned_loss=0.07816, over 1951774.03 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:19:14,065 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.175e+01 2.567e+02 3.218e+02 4.341e+02 1.607e+03, threshold=6.436e+02, percent-clipped=13.0 2022-12-07 14:19:47,270 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 14:19:53,622 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7769, 1.1233, 2.0628, 1.3007, 1.9320, 1.8954, 1.5859, 1.9575], device='cuda:0'), covar=tensor([0.0371, 0.2449, 0.0366, 0.1769, 0.0462, 0.0541, 0.0988, 0.0450], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0158, 0.0139, 0.0171, 0.0156, 0.0152, 0.0127, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:20:27,391 INFO [train.py:873] (0/4) Epoch 6, batch 2300, loss[loss=0.1706, simple_loss=0.1924, pruned_loss=0.07438, over 14287.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.1857, pruned_loss=0.07726, over 1997398.64 frames. ], batch size: 44, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:20:31,756 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9292, 1.1411, 2.1306, 1.2404, 1.9790, 1.9940, 1.7342, 2.1147], device='cuda:0'), covar=tensor([0.0236, 0.1634, 0.0229, 0.1397, 0.0345, 0.0377, 0.0661, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0157, 0.0139, 0.0171, 0.0154, 0.0150, 0.0125, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:20:35,058 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:20:35,177 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3539, 2.4600, 4.2791, 4.3713, 4.4431, 2.7731, 4.4597, 3.6547], device='cuda:0'), covar=tensor([0.0119, 0.0383, 0.0381, 0.0134, 0.0099, 0.0575, 0.0078, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0201, 0.0299, 0.0237, 0.0188, 0.0249, 0.0188, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 14:20:38,457 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:20:41,820 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.659e+01 2.460e+02 3.103e+02 4.082e+02 7.462e+02, threshold=6.206e+02, percent-clipped=2.0 2022-12-07 14:20:43,651 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-07 14:20:48,951 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 14:21:18,090 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:21,624 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:42,560 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:21:57,511 INFO [train.py:873] (0/4) Epoch 6, batch 2400, loss[loss=0.1775, simple_loss=0.1637, pruned_loss=0.09566, over 2578.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.1856, pruned_loss=0.07695, over 2002273.15 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:22:03,802 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:11,490 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.289e+02 3.018e+02 4.008e+02 1.305e+03, threshold=6.036e+02, percent-clipped=3.0 2022-12-07 14:22:37,336 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:51,076 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:22:52,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4446, 1.8092, 2.3426, 2.2380, 2.4622, 2.3234, 2.2049, 2.1319], device='cuda:0'), covar=tensor([0.0286, 0.1940, 0.0409, 0.0876, 0.0300, 0.0446, 0.0391, 0.0887], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0337, 0.0376, 0.0311, 0.0357, 0.0290, 0.0336, 0.0345], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 14:22:57,523 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:11,488 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:23:25,878 INFO [train.py:873] (0/4) Epoch 6, batch 2500, loss[loss=0.1721, simple_loss=0.1566, pruned_loss=0.09386, over 2657.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.1847, pruned_loss=0.076, over 1974265.99 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:23:39,898 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.352e+02 2.913e+02 3.857e+02 7.423e+02, threshold=5.826e+02, percent-clipped=3.0 2022-12-07 14:23:43,076 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:45,731 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:23:53,912 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:24:01,610 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 14:24:14,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 14:24:36,842 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:24:54,947 INFO [train.py:873] (0/4) Epoch 6, batch 2600, loss[loss=0.1557, simple_loss=0.1816, pruned_loss=0.06492, over 14311.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.185, pruned_loss=0.07632, over 1987230.60 frames. ], batch size: 28, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:25:05,721 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:25:08,967 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.477e+02 3.292e+02 4.100e+02 6.300e+02, threshold=6.583e+02, percent-clipped=3.0 2022-12-07 14:25:20,015 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5346, 3.3115, 3.0419, 3.1184, 3.4470, 3.3799, 3.5581, 3.4322], device='cuda:0'), covar=tensor([0.0890, 0.0704, 0.2072, 0.2948, 0.0790, 0.0860, 0.0980, 0.1077], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0229, 0.0366, 0.0464, 0.0273, 0.0341, 0.0341, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:25:29,098 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 14:25:34,096 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4625, 1.7892, 2.4599, 2.4882, 2.3981, 1.8966, 2.6672, 2.0282], device='cuda:0'), covar=tensor([0.0147, 0.0329, 0.0180, 0.0134, 0.0148, 0.0412, 0.0108, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0207, 0.0304, 0.0245, 0.0191, 0.0252, 0.0193, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 14:25:42,565 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3106, 2.5909, 2.3677, 2.5301, 1.9519, 2.7027, 2.2880, 1.0154], device='cuda:0'), covar=tensor([0.3070, 0.0627, 0.1061, 0.0689, 0.1210, 0.0436, 0.1460, 0.4625], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0062, 0.0051, 0.0056, 0.0078, 0.0058, 0.0086, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:25:47,488 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:25:47,590 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:06,947 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:23,200 INFO [train.py:873] (0/4) Epoch 6, batch 2700, loss[loss=0.1526, simple_loss=0.157, pruned_loss=0.07411, over 3849.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.1851, pruned_loss=0.07679, over 1927767.46 frames. ], batch size: 100, lr: 1.33e-02, grad_scale: 8.0 2022-12-07 14:26:30,845 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:26:37,691 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.565e+02 3.016e+02 3.792e+02 1.283e+03, threshold=6.032e+02, percent-clipped=5.0 2022-12-07 14:26:55,987 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6073, 5.4010, 4.7950, 5.1884, 4.9275, 5.4838, 5.5776, 5.5200], device='cuda:0'), covar=tensor([0.0648, 0.0293, 0.1591, 0.2139, 0.0613, 0.0450, 0.0775, 0.0720], device='cuda:0'), in_proj_covar=tensor([0.0300, 0.0228, 0.0365, 0.0461, 0.0273, 0.0338, 0.0345, 0.0284], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:26:58,591 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:00,591 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:07,678 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8360, 3.5640, 3.2929, 3.4054, 3.6554, 3.6661, 3.7947, 3.7277], device='cuda:0'), covar=tensor([0.0686, 0.0599, 0.1705, 0.2605, 0.0703, 0.0721, 0.0923, 0.0840], device='cuda:0'), in_proj_covar=tensor([0.0298, 0.0227, 0.0361, 0.0456, 0.0272, 0.0337, 0.0341, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:27:19,433 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:27:52,083 INFO [train.py:873] (0/4) Epoch 6, batch 2800, loss[loss=0.1874, simple_loss=0.1738, pruned_loss=0.1004, over 2640.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.1846, pruned_loss=0.07624, over 1902468.69 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:28:05,950 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.405e+02 3.321e+02 4.166e+02 7.949e+02, threshold=6.642e+02, percent-clipped=7.0 2022-12-07 14:28:06,943 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:28:27,709 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:28:35,018 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2895, 1.6914, 2.1699, 2.1562, 2.3991, 2.1845, 2.0828, 2.0701], device='cuda:0'), covar=tensor([0.0237, 0.1152, 0.0175, 0.0489, 0.0185, 0.0331, 0.0195, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0332, 0.0378, 0.0308, 0.0351, 0.0290, 0.0341, 0.0345], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 14:28:49,646 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.8476, 5.4201, 5.1929, 5.8633, 5.4210, 5.0756, 5.8747, 5.7922], device='cuda:0'), covar=tensor([0.0492, 0.0500, 0.0430, 0.0345, 0.0466, 0.0278, 0.0422, 0.0449], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0098, 0.0114, 0.0116, 0.0116, 0.0090, 0.0130, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:28:58,690 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:29:20,279 INFO [train.py:873] (0/4) Epoch 6, batch 2900, loss[loss=0.1654, simple_loss=0.1476, pruned_loss=0.09159, over 1296.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.1849, pruned_loss=0.07704, over 1913122.23 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:29:21,383 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:29:34,595 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.325e+02 2.437e+02 2.821e+02 3.804e+02 6.190e+02, threshold=5.643e+02, percent-clipped=0.0 2022-12-07 14:29:56,133 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8635, 0.9550, 0.9324, 1.0691, 0.9248, 0.5504, 1.0029, 0.8960], device='cuda:0'), covar=tensor([0.0785, 0.0884, 0.0707, 0.0351, 0.0698, 0.0559, 0.0564, 0.0764], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0019, 0.0026, 0.0020, 0.0018], device='cuda:0'), out_proj_covar=tensor([8.1178e-05, 8.4767e-05, 7.5458e-05, 8.1362e-05, 8.3161e-05, 1.0388e-04, 8.7352e-05, 7.9981e-05], device='cuda:0') 2022-12-07 14:30:32,681 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5717, 4.4059, 4.2731, 4.6333, 4.2028, 3.8934, 4.7157, 4.4941], device='cuda:0'), covar=tensor([0.0674, 0.0505, 0.0565, 0.0581, 0.0608, 0.0541, 0.0532, 0.0702], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0098, 0.0115, 0.0117, 0.0117, 0.0090, 0.0129, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:30:48,713 INFO [train.py:873] (0/4) Epoch 6, batch 3000, loss[loss=0.1385, simple_loss=0.1522, pruned_loss=0.06235, over 4989.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.1852, pruned_loss=0.07674, over 1956192.12 frames. ], batch size: 100, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:30:48,714 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 14:31:04,487 INFO [train.py:905] (0/4) Epoch 6, validation: loss=0.1224, simple_loss=0.1659, pruned_loss=0.03945, over 857387.00 frames. 2022-12-07 14:31:04,488 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 14:31:19,219 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.206e+02 2.431e+02 3.272e+02 4.116e+02 8.676e+02, threshold=6.543e+02, percent-clipped=8.0 2022-12-07 14:31:21,557 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2022-12-07 14:31:37,483 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:31:40,106 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:31:53,090 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:00,798 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:06,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2464, 1.3359, 1.5022, 0.9258, 0.9413, 1.3101, 0.8458, 1.1286], device='cuda:0'), covar=tensor([0.1607, 0.2705, 0.0677, 0.2812, 0.2770, 0.0669, 0.2197, 0.1330], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0078, 0.0077, 0.0084, 0.0105, 0.0069, 0.0134, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:32:22,708 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:23,698 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:33,299 INFO [train.py:873] (0/4) Epoch 6, batch 3100, loss[loss=0.163, simple_loss=0.1858, pruned_loss=0.07014, over 14661.00 frames. ], tot_loss[loss=0.17, simple_loss=0.1853, pruned_loss=0.07734, over 1898959.02 frames. ], batch size: 33, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:32:36,746 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6891, 3.5816, 3.9013, 3.4097, 3.7017, 3.7035, 1.3803, 3.5284], device='cuda:0'), covar=tensor([0.0230, 0.0339, 0.0331, 0.0470, 0.0318, 0.0327, 0.3131, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0140, 0.0123, 0.0118, 0.0172, 0.0118, 0.0152, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:32:42,990 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:46,566 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:32:47,204 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.510e+02 3.195e+02 3.907e+02 1.074e+03, threshold=6.390e+02, percent-clipped=2.0 2022-12-07 14:32:48,223 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:17,855 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:33:30,768 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:40,121 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:49,422 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:33:58,627 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:01,820 INFO [train.py:873] (0/4) Epoch 6, batch 3200, loss[loss=0.1439, simple_loss=0.1671, pruned_loss=0.06041, over 13907.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.1859, pruned_loss=0.07786, over 1937133.31 frames. ], batch size: 20, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:34:16,161 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.950e+01 2.679e+02 3.334e+02 4.339e+02 1.581e+03, threshold=6.667e+02, percent-clipped=5.0 2022-12-07 14:34:22,462 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:32,072 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:35,515 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9331, 3.9958, 4.2732, 3.4112, 4.0447, 4.2829, 1.3596, 3.9208], device='cuda:0'), covar=tensor([0.0211, 0.0284, 0.0372, 0.0516, 0.0303, 0.0192, 0.3179, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0139, 0.0122, 0.0118, 0.0169, 0.0117, 0.0149, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:34:43,578 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:34:53,530 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4694, 3.8956, 3.1005, 4.7821, 4.3674, 4.3964, 3.7513, 3.3624], device='cuda:0'), covar=tensor([0.0634, 0.1174, 0.4065, 0.0345, 0.0641, 0.2123, 0.1160, 0.3524], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0307, 0.0297, 0.0195, 0.0262, 0.0263, 0.0264, 0.0288], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:35:22,837 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:35:26,750 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:35:30,959 INFO [train.py:873] (0/4) Epoch 6, batch 3300, loss[loss=0.1844, simple_loss=0.1916, pruned_loss=0.08853, over 14214.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.1853, pruned_loss=0.07697, over 1998839.08 frames. ], batch size: 94, lr: 1.32e-02, grad_scale: 8.0 2022-12-07 14:35:45,225 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.460e+02 3.115e+02 3.768e+02 7.967e+02, threshold=6.230e+02, percent-clipped=3.0 2022-12-07 14:36:04,015 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:17,012 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:42,965 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 14:36:46,077 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:36:59,351 INFO [train.py:873] (0/4) Epoch 6, batch 3400, loss[loss=0.1721, simple_loss=0.1866, pruned_loss=0.07883, over 11214.00 frames. ], tot_loss[loss=0.169, simple_loss=0.185, pruned_loss=0.07653, over 1976058.47 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:37:08,220 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:37:13,996 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.427e+02 3.173e+02 3.892e+02 5.965e+02, threshold=6.346e+02, percent-clipped=0.0 2022-12-07 14:37:39,644 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 14:38:25,193 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:38:28,733 INFO [train.py:873] (0/4) Epoch 6, batch 3500, loss[loss=0.1682, simple_loss=0.1693, pruned_loss=0.0836, over 4979.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.1848, pruned_loss=0.07626, over 1938740.91 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:38:43,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.629e+02 3.345e+02 4.291e+02 7.270e+02, threshold=6.690e+02, percent-clipped=2.0 2022-12-07 14:38:50,231 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-07 14:39:06,490 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:08,298 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:08,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4254, 2.3495, 3.1410, 2.4592, 3.3245, 3.1460, 3.1075, 2.6968], device='cuda:0'), covar=tensor([0.0424, 0.2818, 0.0677, 0.1947, 0.0669, 0.0693, 0.1170, 0.2189], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0325, 0.0369, 0.0301, 0.0349, 0.0283, 0.0341, 0.0340], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2022-12-07 14:39:34,852 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6511, 3.6124, 3.7272, 3.4006, 3.6470, 3.6356, 1.3230, 3.4053], device='cuda:0'), covar=tensor([0.0319, 0.0362, 0.0641, 0.0556, 0.0527, 0.0510, 0.3958, 0.0413], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0140, 0.0124, 0.0119, 0.0173, 0.0120, 0.0151, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:39:46,091 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7947, 1.2644, 3.6396, 1.7054, 3.6418, 3.6757, 2.8460, 4.0231], device='cuda:0'), covar=tensor([0.0185, 0.3131, 0.0340, 0.2213, 0.0415, 0.0351, 0.0561, 0.0169], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0157, 0.0141, 0.0169, 0.0157, 0.0151, 0.0123, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:39:48,949 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:39:57,987 INFO [train.py:873] (0/4) Epoch 6, batch 3600, loss[loss=0.1527, simple_loss=0.1747, pruned_loss=0.06537, over 14266.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.1844, pruned_loss=0.07517, over 2026339.94 frames. ], batch size: 25, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:40:11,736 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 2.550e+02 3.176e+02 4.214e+02 9.343e+02, threshold=6.353e+02, percent-clipped=2.0 2022-12-07 14:40:39,008 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:40:47,925 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8750, 3.6077, 3.5633, 3.8788, 3.7400, 3.5905, 3.9256, 3.3191], device='cuda:0'), covar=tensor([0.0420, 0.0814, 0.0337, 0.0436, 0.0593, 0.0799, 0.0532, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0223, 0.0152, 0.0144, 0.0154, 0.0122, 0.0227, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 14:41:26,001 INFO [train.py:873] (0/4) Epoch 6, batch 3700, loss[loss=0.1769, simple_loss=0.189, pruned_loss=0.08241, over 13539.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.1851, pruned_loss=0.0762, over 2002018.71 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 8.0 2022-12-07 14:41:34,671 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:41:39,913 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.533e+02 3.248e+02 4.162e+02 7.233e+02, threshold=6.497e+02, percent-clipped=2.0 2022-12-07 14:42:00,437 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4919, 5.2490, 5.1707, 5.5175, 5.1021, 4.6053, 5.5716, 5.4454], device='cuda:0'), covar=tensor([0.0630, 0.0527, 0.0524, 0.0554, 0.0558, 0.0435, 0.0536, 0.0559], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0104, 0.0120, 0.0121, 0.0120, 0.0095, 0.0134, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:42:06,441 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:42:18,091 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:42:40,621 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7221, 2.5433, 2.7769, 2.6336, 2.6703, 2.4543, 1.3279, 2.3645], device='cuda:0'), covar=tensor([0.0285, 0.0334, 0.0362, 0.0370, 0.0278, 0.0670, 0.2430, 0.0301], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0140, 0.0124, 0.0117, 0.0173, 0.0118, 0.0150, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:42:49,108 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:42:55,063 INFO [train.py:873] (0/4) Epoch 6, batch 3800, loss[loss=0.1563, simple_loss=0.1766, pruned_loss=0.06799, over 13544.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.1847, pruned_loss=0.07623, over 1997357.34 frames. ], batch size: 100, lr: 1.31e-02, grad_scale: 16.0 2022-12-07 14:43:00,728 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7036, 2.3972, 2.3955, 1.4920, 2.1537, 2.4683, 2.7456, 2.1871], device='cuda:0'), covar=tensor([0.0839, 0.1710, 0.1168, 0.2684, 0.1335, 0.0634, 0.0630, 0.1926], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0200, 0.0121, 0.0127, 0.0108, 0.0110, 0.0089, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:43:09,831 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.547e+02 3.241e+02 4.303e+02 1.041e+03, threshold=6.482e+02, percent-clipped=5.0 2022-12-07 14:43:33,345 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:43:34,305 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:43:39,703 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 14:44:15,992 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:44:16,094 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:44:24,885 INFO [train.py:873] (0/4) Epoch 6, batch 3900, loss[loss=0.1729, simple_loss=0.1903, pruned_loss=0.07779, over 14642.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.1846, pruned_loss=0.07529, over 2065966.86 frames. ], batch size: 33, lr: 1.31e-02, grad_scale: 16.0 2022-12-07 14:44:28,596 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 14:44:38,881 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.642e+02 3.297e+02 4.041e+02 8.886e+02, threshold=6.594e+02, percent-clipped=2.0 2022-12-07 14:44:49,805 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9115, 0.9895, 1.0948, 1.0832, 1.0472, 0.7198, 1.1392, 0.9720], device='cuda:0'), covar=tensor([0.1147, 0.0839, 0.0279, 0.1036, 0.0712, 0.0490, 0.0835, 0.0790], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0019, 0.0018, 0.0019, 0.0019, 0.0026, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.1927e-05, 8.3938e-05, 7.4536e-05, 8.2029e-05, 8.3075e-05, 1.0356e-04, 8.8705e-05, 8.0480e-05], device='cuda:0') 2022-12-07 14:44:50,493 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4780, 3.3613, 3.2200, 3.5518, 3.1497, 2.9536, 3.5711, 3.5092], device='cuda:0'), covar=tensor([0.0678, 0.0665, 0.0753, 0.0627, 0.0828, 0.0577, 0.0615, 0.0709], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0103, 0.0119, 0.0120, 0.0119, 0.0093, 0.0134, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:44:58,830 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:06,663 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:49,526 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:45:54,442 INFO [train.py:873] (0/4) Epoch 6, batch 4000, loss[loss=0.1795, simple_loss=0.1965, pruned_loss=0.08128, over 14261.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1837, pruned_loss=0.07452, over 2034871.30 frames. ], batch size: 39, lr: 1.30e-02, grad_scale: 16.0 2022-12-07 14:46:08,692 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.207e+02 2.921e+02 3.758e+02 8.130e+02, threshold=5.842e+02, percent-clipped=2.0 2022-12-07 14:47:18,138 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:47:20,225 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1729, 1.2920, 1.5118, 0.9675, 0.9579, 1.3014, 0.8351, 1.1101], device='cuda:0'), covar=tensor([0.1869, 0.1815, 0.0733, 0.2413, 0.3418, 0.0939, 0.2338, 0.1116], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0079, 0.0078, 0.0084, 0.0108, 0.0069, 0.0133, 0.0074], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:47:22,761 INFO [train.py:873] (0/4) Epoch 6, batch 4100, loss[loss=0.221, simple_loss=0.2143, pruned_loss=0.1138, over 7770.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.1842, pruned_loss=0.07463, over 2049521.65 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:47:37,463 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.509e+02 3.055e+02 4.404e+02 7.346e+02, threshold=6.111e+02, percent-clipped=4.0 2022-12-07 14:47:55,682 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:48:11,857 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:48:35,934 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0648, 2.8977, 2.7927, 3.1476, 2.6564, 2.5997, 3.0856, 3.0992], device='cuda:0'), covar=tensor([0.0809, 0.0852, 0.0927, 0.0679, 0.1191, 0.0967, 0.0834, 0.0780], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0099, 0.0115, 0.0118, 0.0116, 0.0091, 0.0131, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:48:48,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 14:48:49,663 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:48:50,455 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 14:48:51,183 INFO [train.py:873] (0/4) Epoch 6, batch 4200, loss[loss=0.153, simple_loss=0.1792, pruned_loss=0.06337, over 13836.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.1851, pruned_loss=0.07559, over 2046108.71 frames. ], batch size: 20, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:49:07,156 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.446e+02 2.869e+02 3.541e+02 9.255e+02, threshold=5.737e+02, percent-clipped=3.0 2022-12-07 14:49:19,842 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 14:50:21,091 INFO [train.py:873] (0/4) Epoch 6, batch 4300, loss[loss=0.1753, simple_loss=0.1582, pruned_loss=0.09619, over 2696.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1835, pruned_loss=0.07463, over 1935895.53 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:50:35,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.072e+01 2.450e+02 2.996e+02 3.613e+02 8.545e+02, threshold=5.992e+02, percent-clipped=0.0 2022-12-07 14:50:44,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.59 vs. limit=5.0 2022-12-07 14:50:55,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 14:51:03,207 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4129, 2.3188, 2.0167, 2.5751, 2.2843, 2.5465, 1.2460, 2.0898], device='cuda:0'), covar=tensor([0.0769, 0.0831, 0.1709, 0.0550, 0.1063, 0.0934, 0.4701, 0.0932], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0139, 0.0125, 0.0117, 0.0172, 0.0120, 0.0151, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 14:51:16,904 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9001, 1.6745, 1.9361, 1.6851, 2.1374, 1.8452, 1.7172, 1.8610], device='cuda:0'), covar=tensor([0.0362, 0.0943, 0.0146, 0.0277, 0.0195, 0.0369, 0.0171, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0326, 0.0365, 0.0307, 0.0351, 0.0288, 0.0338, 0.0335], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 14:51:18,141 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.09 vs. limit=5.0 2022-12-07 14:51:27,362 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-07 14:51:38,422 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8929, 0.9623, 0.9130, 1.0775, 0.9519, 0.6560, 1.0972, 1.0146], device='cuda:0'), covar=tensor([0.1016, 0.0714, 0.0487, 0.0742, 0.0652, 0.0503, 0.0520, 0.0576], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0019, 0.0019, 0.0026, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.3822e-05, 8.5829e-05, 7.8687e-05, 8.3153e-05, 8.4377e-05, 1.0619e-04, 8.9281e-05, 8.1060e-05], device='cuda:0') 2022-12-07 14:51:41,242 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6899, 5.5145, 5.1101, 5.8757, 5.3706, 4.8865, 5.7986, 5.7991], device='cuda:0'), covar=tensor([0.0616, 0.0455, 0.0565, 0.0336, 0.0486, 0.0376, 0.0461, 0.0416], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0098, 0.0113, 0.0115, 0.0114, 0.0090, 0.0128, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 14:51:51,140 INFO [train.py:873] (0/4) Epoch 6, batch 4400, loss[loss=0.138, simple_loss=0.1695, pruned_loss=0.05321, over 14338.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.1833, pruned_loss=0.07483, over 1905965.74 frames. ], batch size: 39, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:51:54,458 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2022-12-07 14:52:01,405 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2022-12-07 14:52:06,375 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.325e+02 2.637e+02 3.076e+02 3.933e+02 7.884e+02, threshold=6.152e+02, percent-clipped=2.0 2022-12-07 14:52:25,809 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5732, 2.6357, 2.3949, 2.7436, 2.1133, 2.9087, 2.4145, 1.0183], device='cuda:0'), covar=tensor([0.2646, 0.0707, 0.1494, 0.0539, 0.1208, 0.0386, 0.1545, 0.3880], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0063, 0.0052, 0.0055, 0.0079, 0.0058, 0.0085, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:52:36,336 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:52:46,544 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9179, 1.3415, 0.9337, 1.5828, 1.4151, 0.9404, 1.1066, 1.1130], device='cuda:0'), covar=tensor([0.1412, 0.1362, 0.1811, 0.0593, 0.1058, 0.0430, 0.0950, 0.0911], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0011, 0.0010, 0.0010, 0.0011, 0.0015, 0.0011, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.5377e-05, 6.7994e-05, 6.2081e-05, 6.1734e-05, 6.7557e-05, 9.0635e-05, 7.4554e-05, 8.6173e-05], device='cuda:0') 2022-12-07 14:52:57,963 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0199, 0.9051, 0.9383, 1.0446, 0.9874, 0.6018, 1.2478, 1.1327], device='cuda:0'), covar=tensor([0.1347, 0.0780, 0.0582, 0.0870, 0.0885, 0.0552, 0.0953, 0.1194], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0019, 0.0020, 0.0019, 0.0027, 0.0020, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.5309e-05, 8.6337e-05, 8.0591e-05, 8.5592e-05, 8.5597e-05, 1.0982e-04, 9.1879e-05, 8.2175e-05], device='cuda:0') 2022-12-07 14:53:08,825 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2721, 1.8687, 2.1352, 1.1657, 1.8588, 1.9638, 2.1456, 1.8084], device='cuda:0'), covar=tensor([0.0763, 0.1321, 0.1207, 0.2511, 0.1326, 0.0800, 0.0524, 0.2071], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0203, 0.0122, 0.0125, 0.0111, 0.0114, 0.0094, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 14:53:14,111 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 14:53:19,727 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:53:20,406 INFO [train.py:873] (0/4) Epoch 6, batch 4500, loss[loss=0.166, simple_loss=0.1904, pruned_loss=0.07082, over 14368.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.1831, pruned_loss=0.07406, over 1917583.06 frames. ], batch size: 55, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:53:35,323 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.162e+02 2.450e+02 2.856e+02 3.817e+02 6.882e+02, threshold=5.713e+02, percent-clipped=1.0 2022-12-07 14:53:40,672 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-07 14:54:02,630 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:54:22,138 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:54:33,069 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9667, 2.3640, 5.0829, 3.4876, 4.8106, 1.9755, 3.8328, 4.5735], device='cuda:0'), covar=tensor([0.0356, 0.5252, 0.0340, 0.8920, 0.0299, 0.4166, 0.1066, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0244, 0.0171, 0.0335, 0.0183, 0.0249, 0.0236, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:54:50,833 INFO [train.py:873] (0/4) Epoch 6, batch 4600, loss[loss=0.2083, simple_loss=0.1763, pruned_loss=0.1201, over 1204.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.1838, pruned_loss=0.07517, over 1935845.02 frames. ], batch size: 100, lr: 1.30e-02, grad_scale: 8.0 2022-12-07 14:55:06,686 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.342e+02 3.312e+02 4.722e+02 1.011e+03, threshold=6.623e+02, percent-clipped=9.0 2022-12-07 14:55:17,210 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:55:39,562 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5419, 1.5071, 1.2589, 1.5515, 1.1969, 1.2157, 1.5937, 1.2522], device='cuda:0'), covar=tensor([0.1040, 0.1000, 0.1256, 0.1100, 0.3686, 0.0657, 0.0421, 0.1661], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.6410e-05, 7.0283e-05, 6.4768e-05, 6.3541e-05, 6.9781e-05, 9.2033e-05, 7.7247e-05, 8.8494e-05], device='cuda:0') 2022-12-07 14:55:48,627 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2022-12-07 14:56:15,893 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:56:20,485 INFO [train.py:873] (0/4) Epoch 6, batch 4700, loss[loss=0.1844, simple_loss=0.192, pruned_loss=0.08839, over 9501.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.1845, pruned_loss=0.07555, over 1972101.95 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:56:22,499 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0214, 2.2028, 4.0326, 2.7732, 3.8609, 1.8850, 3.1246, 3.8356], device='cuda:0'), covar=tensor([0.0542, 0.4599, 0.0407, 0.9087, 0.0407, 0.4336, 0.1255, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0243, 0.0172, 0.0338, 0.0186, 0.0251, 0.0239, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:56:35,465 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.212e+02 2.979e+02 3.531e+02 6.343e+02, threshold=5.959e+02, percent-clipped=0.0 2022-12-07 14:57:06,055 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:10,585 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:15,589 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0330, 0.8720, 1.0253, 1.0362, 0.6128, 0.7106, 0.7896, 0.7106], device='cuda:0'), covar=tensor([0.0180, 0.0266, 0.0124, 0.0169, 0.0206, 0.0336, 0.0255, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.6200e-05, 6.9446e-05, 6.4064e-05, 6.2956e-05, 6.9017e-05, 9.1651e-05, 7.7183e-05, 8.7356e-05], device='cuda:0') 2022-12-07 14:57:26,095 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1958, 1.9660, 2.0962, 2.1684, 2.0961, 2.0685, 2.2433, 1.9466], device='cuda:0'), covar=tensor([0.0786, 0.1602, 0.0633, 0.0759, 0.1035, 0.0749, 0.0869, 0.0688], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0231, 0.0153, 0.0146, 0.0154, 0.0125, 0.0232, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 14:57:43,817 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 14:57:48,832 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:57:49,692 INFO [train.py:873] (0/4) Epoch 6, batch 4800, loss[loss=0.1826, simple_loss=0.1955, pruned_loss=0.08488, over 14225.00 frames. ], tot_loss[loss=0.167, simple_loss=0.1837, pruned_loss=0.07515, over 1983114.20 frames. ], batch size: 94, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:57:53,677 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6965, 1.6100, 1.3056, 1.4116, 1.5010, 1.1989, 1.1918, 0.9874], device='cuda:0'), covar=tensor([0.0457, 0.0943, 0.0889, 0.0589, 0.0656, 0.0334, 0.0333, 0.0999], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.5934e-05, 6.8835e-05, 6.3240e-05, 6.2119e-05, 6.8616e-05, 8.9013e-05, 7.6247e-05, 8.6751e-05], device='cuda:0') 2022-12-07 14:57:56,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-07 14:58:02,807 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9802, 0.8412, 0.8617, 0.8534, 1.1048, 0.5212, 1.0425, 1.0077], device='cuda:0'), covar=tensor([0.0943, 0.0662, 0.0301, 0.0499, 0.0372, 0.0464, 0.0528, 0.0410], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0019, 0.0020, 0.0020, 0.0028, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.6324e-05, 8.6389e-05, 8.1661e-05, 8.6393e-05, 8.6663e-05, 1.1225e-04, 9.2290e-05, 8.6217e-05], device='cuda:0') 2022-12-07 14:58:03,620 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0010, 0.9050, 0.9167, 1.0528, 0.8301, 0.7117, 0.8693, 0.6881], device='cuda:0'), covar=tensor([0.0234, 0.0138, 0.0164, 0.0184, 0.0392, 0.0369, 0.0384, 0.0498], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0015, 0.0012, 0.0016], device='cuda:0'), out_proj_covar=tensor([6.5826e-05, 6.8889e-05, 6.3238e-05, 6.2144e-05, 6.8600e-05, 8.8571e-05, 7.6233e-05, 8.7028e-05], device='cuda:0') 2022-12-07 14:58:05,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.352e+02 2.966e+02 3.514e+02 5.979e+02, threshold=5.933e+02, percent-clipped=1.0 2022-12-07 14:58:23,537 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:58:25,937 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 14:58:38,929 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8393, 1.2194, 3.0474, 2.8910, 3.0021, 3.0344, 2.2078, 3.0975], device='cuda:0'), covar=tensor([0.1096, 0.1354, 0.0119, 0.0249, 0.0203, 0.0112, 0.0349, 0.0127], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0158, 0.0110, 0.0152, 0.0126, 0.0127, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 14:59:00,246 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2902, 1.6440, 2.5656, 2.0081, 2.4961, 1.7091, 2.0539, 2.2427], device='cuda:0'), covar=tensor([0.1321, 0.4128, 0.0309, 0.4673, 0.0436, 0.3003, 0.1237, 0.0364], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0245, 0.0171, 0.0341, 0.0185, 0.0254, 0.0239, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 14:59:07,456 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 14:59:17,976 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 14:59:18,816 INFO [train.py:873] (0/4) Epoch 6, batch 4900, loss[loss=0.1807, simple_loss=0.1661, pruned_loss=0.09768, over 2613.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.1838, pruned_loss=0.07563, over 1886740.98 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 14:59:33,471 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.456e+02 3.213e+02 4.119e+02 7.897e+02, threshold=6.426e+02, percent-clipped=4.0 2022-12-07 14:59:40,206 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:00:38,319 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4662, 3.2497, 2.9629, 3.0645, 3.3290, 3.3001, 3.4161, 3.3700], device='cuda:0'), covar=tensor([0.0819, 0.0714, 0.2233, 0.2699, 0.0795, 0.0870, 0.1076, 0.0840], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0226, 0.0374, 0.0463, 0.0268, 0.0348, 0.0346, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:00:47,918 INFO [train.py:873] (0/4) Epoch 6, batch 5000, loss[loss=0.2259, simple_loss=0.2284, pruned_loss=0.1117, over 10342.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.1836, pruned_loss=0.07527, over 1897607.07 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:01:03,359 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.594e+01 2.343e+02 3.154e+02 4.045e+02 6.991e+02, threshold=6.308e+02, percent-clipped=1.0 2022-12-07 15:01:07,895 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4727, 1.8699, 2.3604, 2.5209, 2.3956, 1.9003, 2.5221, 2.1767], device='cuda:0'), covar=tensor([0.0141, 0.0277, 0.0175, 0.0143, 0.0141, 0.0442, 0.0103, 0.0276], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0207, 0.0313, 0.0247, 0.0198, 0.0254, 0.0199, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 15:01:30,723 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:01:33,676 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:01:48,839 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:02:16,480 INFO [train.py:873] (0/4) Epoch 6, batch 5100, loss[loss=0.1636, simple_loss=0.187, pruned_loss=0.07004, over 14278.00 frames. ], tot_loss[loss=0.169, simple_loss=0.1845, pruned_loss=0.07674, over 1872203.47 frames. ], batch size: 31, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:02:24,174 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:02:30,810 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.401e+02 2.989e+02 4.009e+02 7.834e+02, threshold=5.978e+02, percent-clipped=3.0 2022-12-07 15:02:42,274 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:03:04,929 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8784, 4.6181, 4.2514, 4.4305, 4.4390, 4.7105, 4.8392, 4.8287], device='cuda:0'), covar=tensor([0.0687, 0.0454, 0.1751, 0.2581, 0.0689, 0.0655, 0.0746, 0.0728], device='cuda:0'), in_proj_covar=tensor([0.0305, 0.0229, 0.0372, 0.0467, 0.0270, 0.0346, 0.0347, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:03:40,326 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:03:42,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3296, 2.0720, 5.1315, 4.6773, 4.5462, 5.1998, 4.9204, 5.2585], device='cuda:0'), covar=tensor([0.1034, 0.1056, 0.0055, 0.0089, 0.0114, 0.0075, 0.0061, 0.0076], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0156, 0.0109, 0.0149, 0.0124, 0.0125, 0.0098, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:03:45,518 INFO [train.py:873] (0/4) Epoch 6, batch 5200, loss[loss=0.1759, simple_loss=0.1853, pruned_loss=0.08324, over 9470.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.1841, pruned_loss=0.07557, over 1955499.82 frames. ], batch size: 100, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:04:01,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.408e+02 3.132e+02 3.708e+02 6.690e+02, threshold=6.264e+02, percent-clipped=2.0 2022-12-07 15:04:07,529 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:04:32,842 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0362, 2.5291, 2.8017, 2.8799, 2.8842, 3.1058, 3.0637, 2.4141], device='cuda:0'), covar=tensor([0.0871, 0.2315, 0.1259, 0.1139, 0.1313, 0.0516, 0.1096, 0.1254], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0230, 0.0152, 0.0145, 0.0151, 0.0121, 0.0226, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:04:50,244 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:05:10,342 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1571, 3.1727, 3.4341, 3.2040, 3.2193, 2.9853, 1.3381, 3.0388], device='cuda:0'), covar=tensor([0.0305, 0.0385, 0.0373, 0.0339, 0.0358, 0.0667, 0.3168, 0.0312], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0147, 0.0126, 0.0119, 0.0175, 0.0119, 0.0153, 0.0163], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:05:15,648 INFO [train.py:873] (0/4) Epoch 6, batch 5300, loss[loss=0.1428, simple_loss=0.1665, pruned_loss=0.05956, over 13605.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1833, pruned_loss=0.0748, over 1939489.05 frames. ], batch size: 17, lr: 1.29e-02, grad_scale: 8.0 2022-12-07 15:05:29,869 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.504e+02 3.046e+02 3.756e+02 7.408e+02, threshold=6.093e+02, percent-clipped=3.0 2022-12-07 15:06:00,715 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:43,045 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5277, 2.3220, 2.3317, 1.4572, 2.2310, 2.4219, 2.5924, 2.0867], device='cuda:0'), covar=tensor([0.0755, 0.1429, 0.1078, 0.2482, 0.1093, 0.0641, 0.0534, 0.1913], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0196, 0.0119, 0.0126, 0.0110, 0.0110, 0.0090, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:06:43,856 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:44,599 INFO [train.py:873] (0/4) Epoch 6, batch 5400, loss[loss=0.1653, simple_loss=0.1545, pruned_loss=0.08808, over 2666.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.1829, pruned_loss=0.07432, over 2008707.97 frames. ], batch size: 100, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:06:47,973 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:06:59,606 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.808e+01 2.467e+02 2.944e+02 3.777e+02 9.316e+02, threshold=5.888e+02, percent-clipped=3.0 2022-12-07 15:07:06,171 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:08:01,478 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0387, 0.8698, 0.8279, 0.9248, 0.9875, 0.5660, 0.9472, 0.8747], device='cuda:0'), covar=tensor([0.0533, 0.0515, 0.0304, 0.0464, 0.0477, 0.0544, 0.0735, 0.0452], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0019, 0.0020, 0.0027, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.5288e-05, 8.7357e-05, 8.2650e-05, 8.5288e-05, 8.7903e-05, 1.1089e-04, 9.1510e-05, 8.4659e-05], device='cuda:0') 2022-12-07 15:08:07,158 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3744, 2.1161, 2.6502, 1.6775, 1.8059, 2.3287, 1.2351, 2.3368], device='cuda:0'), covar=tensor([0.1053, 0.1523, 0.0757, 0.2328, 0.2797, 0.0951, 0.4960, 0.0906], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0081, 0.0078, 0.0086, 0.0112, 0.0070, 0.0135, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:08:08,989 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:08:14,525 INFO [train.py:873] (0/4) Epoch 6, batch 5500, loss[loss=0.1634, simple_loss=0.1781, pruned_loss=0.07433, over 14512.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.1827, pruned_loss=0.07381, over 2030490.58 frames. ], batch size: 51, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:08:28,980 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.280e+02 2.901e+02 3.837e+02 7.523e+02, threshold=5.802e+02, percent-clipped=2.0 2022-12-07 15:08:52,281 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:09:02,912 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7532, 1.8050, 1.8835, 1.4408, 1.5896, 1.1597, 1.2686, 1.0623], device='cuda:0'), covar=tensor([0.0364, 0.0921, 0.0530, 0.0655, 0.0570, 0.0497, 0.0344, 0.0813], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0016, 0.0012, 0.0017], device='cuda:0'), out_proj_covar=tensor([6.8484e-05, 7.2406e-05, 6.4631e-05, 6.6033e-05, 7.1686e-05, 9.4617e-05, 7.8630e-05, 9.2179e-05], device='cuda:0') 2022-12-07 15:09:43,893 INFO [train.py:873] (0/4) Epoch 6, batch 5600, loss[loss=0.139, simple_loss=0.1702, pruned_loss=0.05393, over 14307.00 frames. ], tot_loss[loss=0.165, simple_loss=0.1828, pruned_loss=0.07363, over 2009998.47 frames. ], batch size: 39, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:09:59,513 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.262e+02 2.794e+02 3.488e+02 5.213e+02, threshold=5.588e+02, percent-clipped=0.0 2022-12-07 15:10:02,820 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2022-12-07 15:10:04,395 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9493, 0.8744, 0.8658, 0.9343, 1.0430, 0.6356, 1.0461, 0.9935], device='cuda:0'), covar=tensor([0.0926, 0.0777, 0.0398, 0.1070, 0.0634, 0.0548, 0.0739, 0.0803], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0019, 0.0019, 0.0019, 0.0020, 0.0027, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.5385e-05, 8.5657e-05, 8.1868e-05, 8.4359e-05, 8.7370e-05, 1.1024e-04, 9.1255e-05, 8.5513e-05], device='cuda:0') 2022-12-07 15:10:32,680 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6389, 1.3281, 3.6780, 1.5227, 3.5689, 3.6832, 2.5517, 3.8698], device='cuda:0'), covar=tensor([0.0233, 0.3030, 0.0311, 0.2360, 0.0435, 0.0329, 0.0725, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0160, 0.0142, 0.0171, 0.0157, 0.0157, 0.0125, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:11:13,946 INFO [train.py:873] (0/4) Epoch 6, batch 5700, loss[loss=0.1817, simple_loss=0.1748, pruned_loss=0.09432, over 2665.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.1833, pruned_loss=0.07471, over 1937736.77 frames. ], batch size: 100, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:11:16,543 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:17,446 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:28,309 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.452e+02 3.294e+02 4.312e+02 1.114e+03, threshold=6.587e+02, percent-clipped=10.0 2022-12-07 15:11:34,595 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:11:42,629 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7911, 2.0980, 2.6348, 2.3076, 2.6573, 2.5373, 2.4777, 2.2953], device='cuda:0'), covar=tensor([0.0323, 0.1977, 0.0689, 0.1346, 0.0413, 0.0740, 0.0755, 0.1582], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0333, 0.0378, 0.0313, 0.0358, 0.0304, 0.0358, 0.0344], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:11:44,602 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2161, 4.7614, 4.7318, 5.1953, 4.8175, 4.3526, 5.2332, 5.0292], device='cuda:0'), covar=tensor([0.0582, 0.0518, 0.0584, 0.0543, 0.0570, 0.0442, 0.0531, 0.0717], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0103, 0.0115, 0.0120, 0.0120, 0.0093, 0.0132, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 15:11:52,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2022-12-07 15:11:59,846 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:01,748 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9296, 1.6243, 1.8777, 1.7311, 2.0629, 1.8055, 1.6548, 1.7714], device='cuda:0'), covar=tensor([0.0273, 0.0968, 0.0134, 0.0292, 0.0147, 0.0432, 0.0127, 0.0238], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0330, 0.0376, 0.0311, 0.0356, 0.0304, 0.0355, 0.0344], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:12:10,431 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:17,476 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:12:33,696 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1901, 4.6877, 4.6454, 5.2109, 4.8345, 4.2762, 5.1283, 4.3847], device='cuda:0'), covar=tensor([0.0270, 0.0954, 0.0324, 0.0392, 0.0650, 0.0368, 0.0506, 0.0436], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0230, 0.0152, 0.0145, 0.0154, 0.0122, 0.0229, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:12:42,159 INFO [train.py:873] (0/4) Epoch 6, batch 5800, loss[loss=0.184, simple_loss=0.1883, pruned_loss=0.08982, over 11962.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.1828, pruned_loss=0.07448, over 1937071.13 frames. ], batch size: 100, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:12:44,927 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0303, 2.0663, 1.9771, 2.1796, 1.7081, 1.9293, 2.1054, 2.1115], device='cuda:0'), covar=tensor([0.1044, 0.1030, 0.1083, 0.0865, 0.1507, 0.0889, 0.1120, 0.0997], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0101, 0.0114, 0.0118, 0.0119, 0.0092, 0.0130, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 15:12:57,492 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.484e+02 3.216e+02 3.939e+02 1.250e+03, threshold=6.432e+02, percent-clipped=4.0 2022-12-07 15:14:12,023 INFO [train.py:873] (0/4) Epoch 6, batch 5900, loss[loss=0.1709, simple_loss=0.1921, pruned_loss=0.07486, over 14265.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.1829, pruned_loss=0.07438, over 1991754.78 frames. ], batch size: 76, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:14:15,408 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.18 vs. limit=2.0 2022-12-07 15:14:27,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.477e+02 2.996e+02 3.769e+02 7.369e+02, threshold=5.992e+02, percent-clipped=1.0 2022-12-07 15:14:29,752 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6182, 1.2243, 0.9316, 1.3928, 1.2063, 0.9320, 1.5513, 1.1718], device='cuda:0'), covar=tensor([0.0880, 0.1737, 0.1900, 0.1418, 0.1062, 0.0705, 0.0589, 0.1492], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0010, 0.0011, 0.0012, 0.0016, 0.0013, 0.0017], device='cuda:0'), out_proj_covar=tensor([7.1277e-05, 7.6709e-05, 6.8234e-05, 6.9712e-05, 7.4707e-05, 1.0133e-04, 8.2387e-05, 9.6781e-05], device='cuda:0') 2022-12-07 15:14:47,608 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2022-12-07 15:15:20,315 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:15:42,277 INFO [train.py:873] (0/4) Epoch 6, batch 6000, loss[loss=0.1639, simple_loss=0.1909, pruned_loss=0.06846, over 14296.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.1828, pruned_loss=0.07468, over 1956010.98 frames. ], batch size: 25, lr: 1.28e-02, grad_scale: 8.0 2022-12-07 15:15:42,278 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 15:15:51,412 INFO [train.py:905] (0/4) Epoch 6, validation: loss=0.1218, simple_loss=0.1652, pruned_loss=0.03921, over 857387.00 frames. 2022-12-07 15:15:51,412 INFO [train.py:906] (0/4) Maximum memory allocated so far is 17804MB 2022-12-07 15:16:06,715 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.267e+02 3.074e+02 3.851e+02 8.509e+02, threshold=6.148e+02, percent-clipped=4.0 2022-12-07 15:16:24,593 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:16:31,613 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9123, 2.2003, 3.8323, 4.1285, 4.0238, 2.4389, 4.0247, 3.1596], device='cuda:0'), covar=tensor([0.0160, 0.0460, 0.0398, 0.0170, 0.0138, 0.0688, 0.0127, 0.0438], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0209, 0.0313, 0.0250, 0.0199, 0.0252, 0.0206, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 15:16:44,605 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:17:09,914 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5049, 2.1780, 4.3837, 4.6834, 4.7827, 2.7025, 4.4947, 3.8176], device='cuda:0'), covar=tensor([0.0107, 0.0455, 0.0360, 0.0108, 0.0082, 0.0605, 0.0103, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0211, 0.0314, 0.0249, 0.0200, 0.0253, 0.0206, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 15:17:10,142 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.88 vs. limit=5.0 2022-12-07 15:17:19,325 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:17:21,000 INFO [train.py:873] (0/4) Epoch 6, batch 6100, loss[loss=0.1523, simple_loss=0.146, pruned_loss=0.07936, over 2618.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.1837, pruned_loss=0.07553, over 1965074.22 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 16.0 2022-12-07 15:17:35,710 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.736e+02 3.463e+02 4.201e+02 1.121e+03, threshold=6.926e+02, percent-clipped=3.0 2022-12-07 15:18:03,846 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:18:14,102 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:18:15,669 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2874, 1.4664, 4.0568, 1.9900, 4.0991, 4.2849, 3.6677, 4.6214], device='cuda:0'), covar=tensor([0.0213, 0.2935, 0.0312, 0.2002, 0.0307, 0.0261, 0.0386, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0159, 0.0142, 0.0170, 0.0157, 0.0155, 0.0125, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:18:50,028 INFO [train.py:873] (0/4) Epoch 6, batch 6200, loss[loss=0.1815, simple_loss=0.1964, pruned_loss=0.08329, over 14268.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.1828, pruned_loss=0.07463, over 1945989.45 frames. ], batch size: 37, lr: 1.27e-02, grad_scale: 16.0 2022-12-07 15:18:58,964 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:19:06,617 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.517e+02 3.185e+02 3.672e+02 8.437e+02, threshold=6.370e+02, percent-clipped=2.0 2022-12-07 15:20:19,412 INFO [train.py:873] (0/4) Epoch 6, batch 6300, loss[loss=0.1528, simple_loss=0.1788, pruned_loss=0.06337, over 14453.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.1817, pruned_loss=0.07287, over 1966363.50 frames. ], batch size: 51, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:20:35,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 2.319e+02 2.963e+02 3.987e+02 7.339e+02, threshold=5.926e+02, percent-clipped=1.0 2022-12-07 15:20:35,353 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8661, 0.8653, 0.9670, 1.0728, 1.0353, 0.5393, 0.9769, 1.0518], device='cuda:0'), covar=tensor([0.1102, 0.1012, 0.0451, 0.0581, 0.0526, 0.0664, 0.0750, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0020, 0.0019, 0.0020, 0.0029, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.8526e-05, 9.1176e-05, 8.5994e-05, 8.5747e-05, 9.0801e-05, 1.1599e-04, 9.3394e-05, 8.7909e-05], device='cuda:0') 2022-12-07 15:20:48,187 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:21:11,807 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:21:48,488 INFO [train.py:873] (0/4) Epoch 6, batch 6400, loss[loss=0.253, simple_loss=0.211, pruned_loss=0.1475, over 1311.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.1819, pruned_loss=0.07343, over 1961254.45 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:21:55,377 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:22:04,827 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.640e+02 3.353e+02 4.203e+02 1.187e+03, threshold=6.705e+02, percent-clipped=10.0 2022-12-07 15:22:15,971 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5953, 1.2178, 2.0670, 1.9303, 1.9675, 2.0989, 1.4971, 2.0713], device='cuda:0'), covar=tensor([0.0430, 0.0778, 0.0115, 0.0220, 0.0206, 0.0099, 0.0283, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0158, 0.0108, 0.0151, 0.0124, 0.0125, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:22:37,058 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:23:17,613 INFO [train.py:873] (0/4) Epoch 6, batch 6500, loss[loss=0.161, simple_loss=0.1832, pruned_loss=0.06942, over 14190.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.1829, pruned_loss=0.07396, over 2029718.96 frames. ], batch size: 37, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:23:21,308 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:23:33,244 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.290e+02 3.052e+02 3.998e+02 7.186e+02, threshold=6.105e+02, percent-clipped=1.0 2022-12-07 15:23:53,872 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2022-12-07 15:24:46,954 INFO [train.py:873] (0/4) Epoch 6, batch 6600, loss[loss=0.1415, simple_loss=0.1659, pruned_loss=0.05852, over 13946.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.182, pruned_loss=0.07361, over 1984730.62 frames. ], batch size: 22, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:24:48,057 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3582, 3.1234, 3.8245, 2.7806, 2.3537, 3.0783, 1.6398, 3.1075], device='cuda:0'), covar=tensor([0.1099, 0.1169, 0.0849, 0.1850, 0.2579, 0.1035, 0.5176, 0.1943], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0080, 0.0078, 0.0082, 0.0108, 0.0068, 0.0128, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:25:03,289 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.314e+02 2.475e+02 3.026e+02 3.737e+02 6.916e+02, threshold=6.052e+02, percent-clipped=2.0 2022-12-07 15:25:15,252 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:25:36,938 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3017, 1.1722, 4.1106, 1.6139, 4.0424, 4.2983, 3.6689, 4.6305], device='cuda:0'), covar=tensor([0.0229, 0.3384, 0.0450, 0.2682, 0.0349, 0.0429, 0.0483, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0160, 0.0144, 0.0171, 0.0157, 0.0156, 0.0130, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:25:58,402 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:25:59,764 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2022-12-07 15:26:16,138 INFO [train.py:873] (0/4) Epoch 6, batch 6700, loss[loss=0.1919, simple_loss=0.1874, pruned_loss=0.09819, over 6050.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.1819, pruned_loss=0.07381, over 1964008.45 frames. ], batch size: 100, lr: 1.27e-02, grad_scale: 8.0 2022-12-07 15:26:30,060 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:26:31,879 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.579e+02 3.217e+02 4.461e+02 7.506e+02, threshold=6.434e+02, percent-clipped=5.0 2022-12-07 15:26:51,109 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:04,052 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:05,270 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2022-12-07 15:27:13,271 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4685, 2.1721, 4.9944, 4.5930, 4.6016, 5.0968, 4.9354, 5.1883], device='cuda:0'), covar=tensor([0.1052, 0.1120, 0.0057, 0.0109, 0.0101, 0.0063, 0.0049, 0.0072], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0109, 0.0152, 0.0125, 0.0125, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:27:16,951 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8168, 1.2031, 2.0726, 1.2389, 1.9974, 2.0676, 1.7290, 2.0298], device='cuda:0'), covar=tensor([0.0265, 0.1612, 0.0284, 0.1606, 0.0378, 0.0329, 0.0709, 0.0365], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0158, 0.0144, 0.0169, 0.0156, 0.0155, 0.0130, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:27:24,034 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1055, 1.8659, 2.0056, 2.0604, 1.9727, 1.8891, 2.1224, 1.6736], device='cuda:0'), covar=tensor([0.0795, 0.1591, 0.0699, 0.0864, 0.1145, 0.0763, 0.0934, 0.0848], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0233, 0.0152, 0.0146, 0.0156, 0.0121, 0.0229, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:27:25,055 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:45,455 INFO [train.py:873] (0/4) Epoch 6, batch 6800, loss[loss=0.1279, simple_loss=0.1624, pruned_loss=0.04667, over 14285.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.1813, pruned_loss=0.07305, over 1961847.06 frames. ], batch size: 63, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:27:45,628 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:45,647 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8753, 2.7250, 1.9851, 2.9212, 2.6415, 2.7949, 2.4231, 2.1125], device='cuda:0'), covar=tensor([0.0585, 0.1086, 0.3273, 0.0326, 0.0751, 0.0731, 0.1296, 0.3052], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0297, 0.0285, 0.0202, 0.0261, 0.0266, 0.0250, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:27:47,290 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:27:49,129 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:28:01,814 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.228e+02 2.960e+02 3.530e+02 6.520e+02, threshold=5.920e+02, percent-clipped=1.0 2022-12-07 15:28:32,301 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:28:44,663 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7632, 1.2993, 2.7325, 2.5985, 2.7402, 2.7280, 2.1138, 2.8194], device='cuda:0'), covar=tensor([0.0753, 0.0991, 0.0097, 0.0219, 0.0195, 0.0088, 0.0258, 0.0110], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0156, 0.0109, 0.0152, 0.0124, 0.0125, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:29:15,080 INFO [train.py:873] (0/4) Epoch 6, batch 6900, loss[loss=0.1726, simple_loss=0.1992, pruned_loss=0.07303, over 14466.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.1816, pruned_loss=0.07338, over 1927335.71 frames. ], batch size: 51, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:29:23,120 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5972, 3.2750, 4.3109, 3.2916, 4.3698, 4.3232, 4.2119, 3.9660], device='cuda:0'), covar=tensor([0.0447, 0.2359, 0.0805, 0.1739, 0.0623, 0.0554, 0.1410, 0.1594], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0331, 0.0374, 0.0310, 0.0356, 0.0304, 0.0346, 0.0343], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:29:31,171 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.434e+02 2.991e+02 3.865e+02 6.867e+02, threshold=5.982e+02, percent-clipped=5.0 2022-12-07 15:29:46,265 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-07 15:29:46,548 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9977, 1.8449, 1.9992, 2.0609, 1.9617, 1.6567, 1.3280, 1.8017], device='cuda:0'), covar=tensor([0.0397, 0.0433, 0.0459, 0.0332, 0.0341, 0.1047, 0.1829, 0.0403], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0142, 0.0125, 0.0118, 0.0171, 0.0117, 0.0150, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:29:53,264 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2022-12-07 15:30:08,675 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1657, 1.3497, 1.4887, 1.0100, 0.9477, 1.3064, 0.9187, 1.3300], device='cuda:0'), covar=tensor([0.1568, 0.1812, 0.0562, 0.1823, 0.2213, 0.0814, 0.1794, 0.1020], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0082, 0.0079, 0.0085, 0.0107, 0.0069, 0.0132, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:30:43,883 INFO [train.py:873] (0/4) Epoch 6, batch 7000, loss[loss=0.1678, simple_loss=0.1869, pruned_loss=0.07428, over 14355.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.1826, pruned_loss=0.07451, over 1939350.71 frames. ], batch size: 73, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:30:56,411 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 15:31:00,498 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.539e+02 3.130e+02 3.727e+02 8.173e+02, threshold=6.260e+02, percent-clipped=3.0 2022-12-07 15:31:10,025 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2014, 2.4442, 2.1941, 2.4987, 1.7615, 2.6051, 2.3888, 0.9891], device='cuda:0'), covar=tensor([0.2002, 0.0729, 0.1038, 0.0650, 0.1327, 0.0479, 0.1268, 0.3798], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0062, 0.0050, 0.0054, 0.0080, 0.0057, 0.0083, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:31:36,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2022-12-07 15:31:47,387 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:32:08,018 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:32:12,182 INFO [train.py:873] (0/4) Epoch 6, batch 7100, loss[loss=0.1905, simple_loss=0.1882, pruned_loss=0.09643, over 6000.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1833, pruned_loss=0.07472, over 1950375.36 frames. ], batch size: 100, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:32:20,787 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6004, 3.2182, 2.8575, 2.0054, 3.0520, 3.4089, 3.7528, 2.6122], device='cuda:0'), covar=tensor([0.0974, 0.3331, 0.1892, 0.3587, 0.1253, 0.0725, 0.0942, 0.2535], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0198, 0.0126, 0.0128, 0.0114, 0.0112, 0.0095, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:32:27,515 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.599e+02 3.099e+02 3.894e+02 6.870e+02, threshold=6.198e+02, percent-clipped=1.0 2022-12-07 15:32:57,922 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3342, 2.0960, 4.4433, 3.0133, 4.3103, 1.9721, 3.2003, 4.0927], device='cuda:0'), covar=tensor([0.0463, 0.5201, 0.0262, 0.8470, 0.0322, 0.3886, 0.1354, 0.0355], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0242, 0.0170, 0.0332, 0.0185, 0.0248, 0.0236, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:33:14,114 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9158, 1.5650, 1.9032, 1.7609, 2.0416, 1.8003, 1.6857, 1.8560], device='cuda:0'), covar=tensor([0.0245, 0.1132, 0.0123, 0.0250, 0.0169, 0.0517, 0.0121, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0298, 0.0329, 0.0374, 0.0310, 0.0356, 0.0304, 0.0346, 0.0338], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:33:31,399 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-45000.pt 2022-12-07 15:33:44,151 INFO [train.py:873] (0/4) Epoch 6, batch 7200, loss[loss=0.1698, simple_loss=0.1927, pruned_loss=0.07341, over 14596.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.1833, pruned_loss=0.07474, over 1951789.06 frames. ], batch size: 43, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:34:00,689 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.403e+02 2.886e+02 3.979e+02 7.470e+02, threshold=5.772e+02, percent-clipped=2.0 2022-12-07 15:34:06,013 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9937, 2.0196, 1.9129, 2.1145, 1.7375, 1.9572, 2.0036, 2.0649], device='cuda:0'), covar=tensor([0.0798, 0.0822, 0.1031, 0.0637, 0.1392, 0.0902, 0.0981, 0.0756], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0104, 0.0120, 0.0123, 0.0124, 0.0096, 0.0131, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 15:34:24,857 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:34:25,828 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6227, 2.4792, 4.6790, 2.8858, 4.4822, 2.0786, 3.4583, 4.3751], device='cuda:0'), covar=tensor([0.0397, 0.4483, 0.0214, 0.9582, 0.0338, 0.3840, 0.1284, 0.0279], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0247, 0.0173, 0.0336, 0.0188, 0.0250, 0.0239, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:34:46,863 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:35:13,051 INFO [train.py:873] (0/4) Epoch 6, batch 7300, loss[loss=0.1577, simple_loss=0.1747, pruned_loss=0.07037, over 14281.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.1827, pruned_loss=0.07377, over 2039730.02 frames. ], batch size: 44, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:35:16,724 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2526, 3.2869, 3.4854, 3.1067, 3.3561, 3.0307, 1.3387, 3.1427], device='cuda:0'), covar=tensor([0.0270, 0.0308, 0.0345, 0.0446, 0.0291, 0.0500, 0.2995, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0143, 0.0123, 0.0117, 0.0170, 0.0114, 0.0149, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:35:18,617 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 15:35:29,218 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.544e+02 3.092e+02 3.955e+02 6.821e+02, threshold=6.183e+02, percent-clipped=3.0 2022-12-07 15:35:41,005 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:07,525 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5078, 2.4156, 1.8426, 2.5098, 2.3372, 2.4182, 2.1920, 2.0504], device='cuda:0'), covar=tensor([0.0450, 0.0684, 0.2214, 0.0324, 0.0977, 0.0468, 0.1049, 0.1309], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0294, 0.0285, 0.0202, 0.0265, 0.0269, 0.0251, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:36:17,661 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:37,771 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:36:41,982 INFO [train.py:873] (0/4) Epoch 6, batch 7400, loss[loss=0.183, simple_loss=0.1771, pruned_loss=0.09447, over 3830.00 frames. ], tot_loss[loss=0.165, simple_loss=0.1824, pruned_loss=0.07379, over 1990521.66 frames. ], batch size: 100, lr: 1.26e-02, grad_scale: 8.0 2022-12-07 15:36:58,367 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.594e+02 3.335e+02 4.109e+02 6.658e+02, threshold=6.669e+02, percent-clipped=6.0 2022-12-07 15:37:00,154 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:37:08,078 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 15:37:08,464 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0104, 3.0143, 2.3572, 2.4421, 2.9055, 2.9627, 3.1038, 2.9739], device='cuda:0'), covar=tensor([0.1470, 0.0875, 0.3151, 0.4363, 0.1353, 0.1360, 0.1590, 0.1498], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0234, 0.0385, 0.0488, 0.0284, 0.0362, 0.0359, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:37:20,785 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:37:42,985 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 15:37:52,224 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0923, 2.9792, 2.8723, 3.2046, 2.7075, 2.6360, 3.1108, 3.0840], device='cuda:0'), covar=tensor([0.0691, 0.0660, 0.0862, 0.0580, 0.0990, 0.0896, 0.0738, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0102, 0.0117, 0.0120, 0.0120, 0.0094, 0.0128, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 15:38:12,294 INFO [train.py:873] (0/4) Epoch 6, batch 7500, loss[loss=0.2192, simple_loss=0.2123, pruned_loss=0.113, over 10383.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.1822, pruned_loss=0.07395, over 1945606.58 frames. ], batch size: 100, lr: 1.25e-02, grad_scale: 8.0 2022-12-07 15:38:27,609 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.117e+02 2.437e+02 3.096e+02 3.781e+02 7.631e+02, threshold=6.193e+02, percent-clipped=3.0 2022-12-07 15:38:44,183 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:38:49,349 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0381, 1.9179, 1.9926, 2.0915, 1.9728, 1.7168, 1.3004, 1.7360], device='cuda:0'), covar=tensor([0.0452, 0.0415, 0.0543, 0.0276, 0.0355, 0.0933, 0.1896, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0149, 0.0128, 0.0120, 0.0175, 0.0119, 0.0154, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:38:59,487 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-6.pt 2022-12-07 15:39:51,269 INFO [train.py:873] (0/4) Epoch 7, batch 0, loss[loss=0.1665, simple_loss=0.1897, pruned_loss=0.07169, over 14209.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.1897, pruned_loss=0.07169, over 14209.00 frames. ], batch size: 35, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:39:51,269 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 15:39:59,859 INFO [train.py:905] (0/4) Epoch 7, validation: loss=0.1305, simple_loss=0.175, pruned_loss=0.04304, over 857387.00 frames. 2022-12-07 15:39:59,860 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 15:40:33,654 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 15:40:36,177 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 15:40:51,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.719e+01 2.285e+02 3.118e+02 4.102e+02 1.207e+03, threshold=6.237e+02, percent-clipped=4.0 2022-12-07 15:40:59,116 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:41:31,135 INFO [train.py:873] (0/4) Epoch 7, batch 100, loss[loss=0.1685, simple_loss=0.189, pruned_loss=0.07401, over 14314.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.1838, pruned_loss=0.07337, over 914913.74 frames. ], batch size: 28, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:42:21,346 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.542e+02 3.020e+02 3.979e+02 7.369e+02, threshold=6.041e+02, percent-clipped=5.0 2022-12-07 15:43:00,539 INFO [train.py:873] (0/4) Epoch 7, batch 200, loss[loss=0.1627, simple_loss=0.1896, pruned_loss=0.06785, over 14520.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.1824, pruned_loss=0.07288, over 1348912.09 frames. ], batch size: 34, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:43:43,893 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2122, 3.3296, 3.0949, 3.3514, 2.4387, 3.4255, 3.0260, 1.4142], device='cuda:0'), covar=tensor([0.2901, 0.1289, 0.1811, 0.1226, 0.1325, 0.0525, 0.1446, 0.3590], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0061, 0.0051, 0.0054, 0.0079, 0.0057, 0.0083, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:43:50,944 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.408e+02 2.969e+02 3.747e+02 6.560e+02, threshold=5.938e+02, percent-clipped=2.0 2022-12-07 15:44:25,035 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.73 vs. limit=5.0 2022-12-07 15:44:30,681 INFO [train.py:873] (0/4) Epoch 7, batch 300, loss[loss=0.1783, simple_loss=0.1983, pruned_loss=0.07912, over 13522.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.181, pruned_loss=0.07136, over 1602817.38 frames. ], batch size: 100, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:44:49,424 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4826, 1.2223, 1.2614, 1.4996, 1.3552, 0.5481, 1.5826, 1.5432], device='cuda:0'), covar=tensor([0.2043, 0.0704, 0.1501, 0.1912, 0.2761, 0.0703, 0.0972, 0.1063], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0019, 0.0020, 0.0020, 0.0028, 0.0020, 0.0021], device='cuda:0'), out_proj_covar=tensor([8.9382e-05, 8.8975e-05, 8.6395e-05, 8.9976e-05, 9.0203e-05, 1.1590e-04, 9.4890e-05, 9.2127e-05], device='cuda:0') 2022-12-07 15:44:56,182 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:44:58,115 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 15:45:05,221 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 15:45:20,483 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.117e+02 2.543e+02 3.312e+02 3.999e+02 6.909e+02, threshold=6.625e+02, percent-clipped=2.0 2022-12-07 15:45:26,822 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2618, 2.1638, 2.2305, 1.5157, 2.0002, 2.3031, 2.3080, 2.0431], device='cuda:0'), covar=tensor([0.0589, 0.0916, 0.0806, 0.1797, 0.1001, 0.0567, 0.0395, 0.1597], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0193, 0.0124, 0.0127, 0.0111, 0.0113, 0.0095, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:45:27,779 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:48,349 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:50,950 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:45:59,487 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 15:45:59,857 INFO [train.py:873] (0/4) Epoch 7, batch 400, loss[loss=0.1803, simple_loss=0.1989, pruned_loss=0.08091, over 14203.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.1805, pruned_loss=0.07197, over 1755436.53 frames. ], batch size: 89, lr: 1.17e-02, grad_scale: 8.0 2022-12-07 15:46:01,044 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:46:10,776 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:46:46,790 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2022-12-07 15:46:50,565 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.431e+02 3.067e+02 3.597e+02 6.934e+02, threshold=6.134e+02, percent-clipped=1.0 2022-12-07 15:46:53,632 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8966, 0.7295, 0.6255, 0.8654, 0.7020, 0.4905, 0.5633, 0.6410], device='cuda:0'), covar=tensor([0.0191, 0.0156, 0.0178, 0.0148, 0.0353, 0.0499, 0.0265, 0.0533], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0011, 0.0011, 0.0016, 0.0013, 0.0017], device='cuda:0'), out_proj_covar=tensor([7.1144e-05, 7.7754e-05, 7.0002e-05, 7.1078e-05, 7.1490e-05, 1.0454e-04, 8.6631e-05, 9.9212e-05], device='cuda:0') 2022-12-07 15:46:56,410 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:47:17,537 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:47:30,184 INFO [train.py:873] (0/4) Epoch 7, batch 500, loss[loss=0.1603, simple_loss=0.1898, pruned_loss=0.06542, over 13810.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.1804, pruned_loss=0.07221, over 1844187.02 frames. ], batch size: 23, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:47:32,698 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 15:47:36,878 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0526, 1.5491, 4.4657, 4.2142, 4.2366, 4.6331, 4.2029, 4.6388], device='cuda:0'), covar=tensor([0.1195, 0.1393, 0.0111, 0.0148, 0.0117, 0.0079, 0.0134, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0157, 0.0111, 0.0153, 0.0125, 0.0127, 0.0101, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:48:12,889 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:48:21,753 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.455e+02 3.322e+02 4.008e+02 8.135e+02, threshold=6.645e+02, percent-clipped=7.0 2022-12-07 15:48:59,906 INFO [train.py:873] (0/4) Epoch 7, batch 600, loss[loss=0.1296, simple_loss=0.1565, pruned_loss=0.0513, over 13964.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.1804, pruned_loss=0.072, over 1890146.65 frames. ], batch size: 20, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:49:21,765 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8072, 1.8640, 4.4826, 2.0767, 4.2287, 4.4901, 4.3000, 5.2489], device='cuda:0'), covar=tensor([0.0174, 0.2656, 0.0407, 0.2122, 0.0280, 0.0471, 0.0253, 0.0108], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0159, 0.0148, 0.0169, 0.0158, 0.0160, 0.0131, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:49:27,341 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 15:49:30,612 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 15:49:32,406 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 15:49:49,711 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.803e+01 2.347e+02 3.029e+02 4.189e+02 7.825e+02, threshold=6.059e+02, percent-clipped=1.0 2022-12-07 15:50:09,871 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 15:50:15,098 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:50:28,496 INFO [train.py:873] (0/4) Epoch 7, batch 700, loss[loss=0.1677, simple_loss=0.1887, pruned_loss=0.07336, over 14420.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.1804, pruned_loss=0.07157, over 1929469.37 frames. ], batch size: 53, lr: 1.17e-02, grad_scale: 4.0 2022-12-07 15:50:31,236 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1231, 2.4254, 2.1667, 2.3274, 1.9134, 2.4523, 2.1653, 1.0369], device='cuda:0'), covar=tensor([0.2183, 0.0598, 0.1231, 0.0750, 0.1099, 0.0534, 0.1280, 0.3595], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0062, 0.0050, 0.0054, 0.0078, 0.0057, 0.0083, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 15:50:33,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-07 15:51:18,729 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.605e+01 2.609e+02 3.339e+02 4.151e+02 6.051e+02, threshold=6.678e+02, percent-clipped=0.0 2022-12-07 15:51:18,879 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:51:56,447 INFO [train.py:873] (0/4) Epoch 7, batch 800, loss[loss=0.1438, simple_loss=0.1793, pruned_loss=0.05412, over 14098.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.1799, pruned_loss=0.07157, over 1955453.10 frames. ], batch size: 29, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:52:25,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2022-12-07 15:52:34,722 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:52:47,697 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.967e+01 2.522e+02 3.122e+02 4.007e+02 6.928e+02, threshold=6.244e+02, percent-clipped=1.0 2022-12-07 15:53:26,504 INFO [train.py:873] (0/4) Epoch 7, batch 900, loss[loss=0.2065, simple_loss=0.2, pruned_loss=0.1064, over 7794.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.1805, pruned_loss=0.07227, over 1943057.70 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:53:29,267 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:54:06,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2294, 2.1234, 2.4646, 1.5197, 1.8116, 2.2347, 1.1876, 2.1802], device='cuda:0'), covar=tensor([0.0881, 0.1553, 0.0616, 0.2561, 0.2872, 0.0782, 0.4864, 0.0838], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0083, 0.0078, 0.0086, 0.0110, 0.0069, 0.0131, 0.0077], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 15:54:16,878 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.538e+02 3.268e+02 4.038e+02 9.582e+02, threshold=6.536e+02, percent-clipped=3.0 2022-12-07 15:54:23,256 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:54:42,199 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:54:55,207 INFO [train.py:873] (0/4) Epoch 7, batch 1000, loss[loss=0.1611, simple_loss=0.1696, pruned_loss=0.0763, over 5958.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.1799, pruned_loss=0.07153, over 1970106.01 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:54:57,628 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 15:55:04,054 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0434, 1.9434, 1.6700, 1.9777, 1.9269, 2.0875, 1.8450, 1.7618], device='cuda:0'), covar=tensor([0.0428, 0.0702, 0.1734, 0.0337, 0.0518, 0.0264, 0.1022, 0.0622], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0291, 0.0276, 0.0199, 0.0260, 0.0260, 0.0248, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:55:24,801 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:55:25,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2022-12-07 15:55:46,198 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.372e+02 2.921e+02 3.865e+02 1.419e+03, threshold=5.842e+02, percent-clipped=9.0 2022-12-07 15:55:46,367 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:56:15,754 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2022-12-07 15:56:24,646 INFO [train.py:873] (0/4) Epoch 7, batch 1100, loss[loss=0.1561, simple_loss=0.1803, pruned_loss=0.06594, over 14245.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.1797, pruned_loss=0.07173, over 1936695.52 frames. ], batch size: 63, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:56:29,025 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:56:48,165 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.29 vs. limit=2.0 2022-12-07 15:57:01,949 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1164, 1.5262, 3.9890, 1.6771, 3.8786, 4.0201, 3.1635, 4.4354], device='cuda:0'), covar=tensor([0.0153, 0.2657, 0.0371, 0.2179, 0.0350, 0.0318, 0.0551, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0158, 0.0147, 0.0168, 0.0159, 0.0158, 0.0128, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:57:01,970 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:57:13,605 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5602, 1.2022, 2.0606, 1.8918, 1.9882, 2.0324, 1.5583, 2.0763], device='cuda:0'), covar=tensor([0.0415, 0.0814, 0.0119, 0.0273, 0.0244, 0.0120, 0.0310, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0156, 0.0110, 0.0153, 0.0126, 0.0127, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 15:57:15,198 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.418e+02 3.079e+02 3.635e+02 7.997e+02, threshold=6.158e+02, percent-clipped=1.0 2022-12-07 15:57:28,787 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8655, 3.6429, 3.3888, 3.5151, 3.7872, 3.7986, 3.8596, 3.8185], device='cuda:0'), covar=tensor([0.0731, 0.0642, 0.2125, 0.2302, 0.0743, 0.0660, 0.0887, 0.0877], device='cuda:0'), in_proj_covar=tensor([0.0318, 0.0233, 0.0391, 0.0487, 0.0283, 0.0360, 0.0343, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 15:57:45,029 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:57:53,827 INFO [train.py:873] (0/4) Epoch 7, batch 1200, loss[loss=0.1486, simple_loss=0.1729, pruned_loss=0.0622, over 14288.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.1804, pruned_loss=0.07254, over 1953482.79 frames. ], batch size: 60, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:58:13,376 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:58:20,083 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:58:44,672 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.081e+01 2.469e+02 2.984e+02 3.753e+02 8.629e+02, threshold=5.968e+02, percent-clipped=3.0 2022-12-07 15:58:46,480 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:59:08,440 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 15:59:11,734 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3592, 3.6289, 2.7875, 4.6719, 4.2216, 4.2787, 3.4464, 3.0265], device='cuda:0'), covar=tensor([0.0532, 0.1414, 0.4537, 0.0485, 0.0641, 0.1664, 0.1396, 0.3798], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0295, 0.0279, 0.0203, 0.0259, 0.0261, 0.0248, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:59:14,296 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 15:59:18,699 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4202, 2.3593, 2.4697, 2.4932, 2.4222, 2.1611, 1.3955, 2.1666], device='cuda:0'), covar=tensor([0.0398, 0.0401, 0.0513, 0.0320, 0.0321, 0.0941, 0.2500, 0.0374], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0147, 0.0128, 0.0119, 0.0174, 0.0121, 0.0150, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 15:59:23,322 INFO [train.py:873] (0/4) Epoch 7, batch 1300, loss[loss=0.1543, simple_loss=0.1832, pruned_loss=0.06275, over 14543.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.1804, pruned_loss=0.07185, over 1945461.04 frames. ], batch size: 43, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 15:59:39,809 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 16:00:12,958 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3945, 2.2900, 1.8364, 2.4449, 2.2435, 2.3069, 2.1738, 1.9796], device='cuda:0'), covar=tensor([0.0428, 0.0871, 0.2200, 0.0412, 0.0924, 0.0482, 0.1081, 0.1425], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0300, 0.0285, 0.0208, 0.0263, 0.0265, 0.0252, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:00:14,512 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.325e+02 3.068e+02 3.769e+02 6.043e+02, threshold=6.136e+02, percent-clipped=2.0 2022-12-07 16:00:53,005 INFO [train.py:873] (0/4) Epoch 7, batch 1400, loss[loss=0.1577, simple_loss=0.1822, pruned_loss=0.06666, over 14270.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.1804, pruned_loss=0.07252, over 1869168.52 frames. ], batch size: 76, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 16:01:04,973 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2177, 1.5744, 2.3529, 1.9637, 2.1700, 1.7051, 1.9219, 2.1903], device='cuda:0'), covar=tensor([0.1154, 0.2811, 0.0201, 0.2056, 0.0366, 0.1833, 0.0885, 0.0381], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0237, 0.0166, 0.0325, 0.0189, 0.0243, 0.0231, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:01:27,153 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:01:45,246 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.517e+02 3.054e+02 3.788e+02 7.041e+02, threshold=6.108e+02, percent-clipped=3.0 2022-12-07 16:02:19,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 16:02:21,991 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:02:23,385 INFO [train.py:873] (0/4) Epoch 7, batch 1500, loss[loss=0.1757, simple_loss=0.1627, pruned_loss=0.09437, over 2650.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.1793, pruned_loss=0.07111, over 1925451.40 frames. ], batch size: 100, lr: 1.16e-02, grad_scale: 8.0 2022-12-07 16:03:07,545 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9014, 2.2666, 3.2747, 2.3001, 2.1348, 2.6871, 1.2500, 2.6785], device='cuda:0'), covar=tensor([0.1086, 0.1771, 0.0700, 0.2446, 0.3583, 0.1195, 0.6454, 0.1315], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0082, 0.0079, 0.0087, 0.0110, 0.0070, 0.0132, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:03:13,535 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.064e+01 2.462e+02 3.053e+02 4.274e+02 1.237e+03, threshold=6.105e+02, percent-clipped=6.0 2022-12-07 16:03:15,338 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:03:20,790 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3885, 1.4931, 1.2697, 1.1190, 1.5959, 1.0747, 1.2764, 0.8227], device='cuda:0'), covar=tensor([0.0467, 0.1017, 0.0862, 0.1668, 0.0768, 0.0406, 0.0543, 0.1300], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0011, 0.0017, 0.0013, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.3630e-05, 7.9960e-05, 7.2000e-05, 7.2399e-05, 7.3677e-05, 1.0747e-04, 9.1366e-05, 1.0311e-04], device='cuda:0') 2022-12-07 16:03:32,138 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:03:38,431 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:03:51,574 INFO [train.py:873] (0/4) Epoch 7, batch 1600, loss[loss=0.1839, simple_loss=0.1952, pruned_loss=0.08631, over 13925.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.1803, pruned_loss=0.07148, over 2004230.60 frames. ], batch size: 23, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:03:57,839 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:04:14,424 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2022-12-07 16:04:42,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 2.534e+02 3.085e+02 3.879e+02 6.356e+02, threshold=6.171e+02, percent-clipped=1.0 2022-12-07 16:04:43,177 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8618, 1.5520, 3.6359, 3.4069, 3.4876, 3.5838, 3.0857, 3.6777], device='cuda:0'), covar=tensor([0.1186, 0.1241, 0.0085, 0.0161, 0.0160, 0.0098, 0.0165, 0.0101], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0155, 0.0111, 0.0153, 0.0126, 0.0127, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:05:21,094 INFO [train.py:873] (0/4) Epoch 7, batch 1700, loss[loss=0.1657, simple_loss=0.1604, pruned_loss=0.08545, over 2621.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.1797, pruned_loss=0.07071, over 2004619.22 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:06:11,978 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.426e+02 3.035e+02 3.683e+02 7.018e+02, threshold=6.070e+02, percent-clipped=2.0 2022-12-07 16:06:24,325 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8667, 4.4331, 4.3581, 4.7678, 4.4567, 4.1923, 4.7837, 3.9995], device='cuda:0'), covar=tensor([0.0261, 0.0813, 0.0329, 0.0376, 0.0708, 0.0503, 0.0443, 0.0491], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0234, 0.0152, 0.0147, 0.0155, 0.0124, 0.0232, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:06:44,204 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:06:50,173 INFO [train.py:873] (0/4) Epoch 7, batch 1800, loss[loss=0.1778, simple_loss=0.19, pruned_loss=0.08279, over 14221.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.1803, pruned_loss=0.07106, over 2040297.74 frames. ], batch size: 35, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:07:40,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.187e+02 2.439e+02 3.276e+02 4.235e+02 9.028e+02, threshold=6.551e+02, percent-clipped=6.0 2022-12-07 16:07:50,947 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:07:59,647 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:04,892 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1916, 4.1988, 4.5010, 3.7665, 4.1745, 4.5191, 1.5559, 4.0215], device='cuda:0'), covar=tensor([0.0190, 0.0287, 0.0297, 0.0437, 0.0285, 0.0148, 0.3077, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0147, 0.0126, 0.0120, 0.0173, 0.0121, 0.0151, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:08:05,791 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:19,481 INFO [train.py:873] (0/4) Epoch 7, batch 1900, loss[loss=0.1839, simple_loss=0.2037, pruned_loss=0.08206, over 13968.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.1804, pruned_loss=0.07122, over 2028338.73 frames. ], batch size: 23, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:08:23,881 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0369, 1.7609, 2.1103, 1.5000, 1.7773, 1.9814, 2.0713, 1.7869], device='cuda:0'), covar=tensor([0.0643, 0.0899, 0.0808, 0.1663, 0.0962, 0.0686, 0.0528, 0.1422], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0193, 0.0124, 0.0126, 0.0111, 0.0116, 0.0096, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:08:42,242 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:44,903 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:08:48,286 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:09:09,836 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-07 16:09:10,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.352e+02 2.909e+02 3.754e+02 8.188e+02, threshold=5.818e+02, percent-clipped=1.0 2022-12-07 16:09:13,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 16:09:39,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2083, 1.3526, 2.5628, 1.4371, 2.5035, 2.4110, 1.9044, 2.5062], device='cuda:0'), covar=tensor([0.0296, 0.2292, 0.0331, 0.1740, 0.0391, 0.0525, 0.1000, 0.0338], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0161, 0.0147, 0.0168, 0.0162, 0.0159, 0.0131, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:09:47,745 INFO [train.py:873] (0/4) Epoch 7, batch 2000, loss[loss=0.1531, simple_loss=0.1727, pruned_loss=0.06676, over 12760.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.1799, pruned_loss=0.07144, over 2006997.65 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:09:53,026 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:10:10,368 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3895, 3.6687, 2.9298, 4.5562, 4.2506, 4.2951, 3.7777, 3.1856], device='cuda:0'), covar=tensor([0.0613, 0.1416, 0.4682, 0.0264, 0.0662, 0.1097, 0.1091, 0.3857], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0303, 0.0286, 0.0204, 0.0270, 0.0273, 0.0253, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:10:33,661 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1644, 4.2013, 4.4627, 3.7421, 4.2828, 4.5525, 1.3873, 4.0817], device='cuda:0'), covar=tensor([0.0208, 0.0241, 0.0384, 0.0412, 0.0271, 0.0248, 0.3290, 0.0241], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0147, 0.0127, 0.0120, 0.0172, 0.0121, 0.0151, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:10:40,010 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.312e+02 2.798e+02 3.545e+02 9.094e+02, threshold=5.596e+02, percent-clipped=2.0 2022-12-07 16:10:48,723 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:11:11,282 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:11:13,820 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-07 16:11:18,001 INFO [train.py:873] (0/4) Epoch 7, batch 2100, loss[loss=0.1552, simple_loss=0.1747, pruned_loss=0.06785, over 10349.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.1791, pruned_loss=0.07124, over 1912474.27 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:11:33,195 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4050, 3.9921, 4.0138, 4.4827, 3.9673, 3.5343, 4.4178, 4.3494], device='cuda:0'), covar=tensor([0.0557, 0.0682, 0.0650, 0.0428, 0.0689, 0.0653, 0.0586, 0.0607], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0103, 0.0118, 0.0121, 0.0122, 0.0094, 0.0133, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 16:11:43,150 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.66 vs. limit=5.0 2022-12-07 16:11:53,884 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:11:53,993 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0490, 1.8946, 4.9278, 4.4349, 4.3142, 4.9350, 4.7283, 4.9282], device='cuda:0'), covar=tensor([0.1178, 0.1160, 0.0054, 0.0126, 0.0134, 0.0071, 0.0053, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0154, 0.0110, 0.0153, 0.0126, 0.0127, 0.0102, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:12:09,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.308e+02 2.873e+02 3.601e+02 7.245e+02, threshold=5.745e+02, percent-clipped=7.0 2022-12-07 16:12:23,919 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2520, 1.9026, 2.2707, 2.4489, 2.1255, 1.8235, 2.4197, 2.1379], device='cuda:0'), covar=tensor([0.0116, 0.0287, 0.0147, 0.0098, 0.0159, 0.0374, 0.0113, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0216, 0.0327, 0.0261, 0.0208, 0.0261, 0.0220, 0.0251], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 16:12:46,253 INFO [train.py:873] (0/4) Epoch 7, batch 2200, loss[loss=0.1828, simple_loss=0.1732, pruned_loss=0.09616, over 3898.00 frames. ], tot_loss[loss=0.162, simple_loss=0.1799, pruned_loss=0.07203, over 1945867.73 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:12:55,062 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9896, 1.5743, 1.9724, 1.4061, 1.5884, 1.9608, 1.9248, 1.6621], device='cuda:0'), covar=tensor([0.0696, 0.1014, 0.0836, 0.1455, 0.1178, 0.0734, 0.0473, 0.1691], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0197, 0.0125, 0.0129, 0.0113, 0.0118, 0.0098, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0007, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:13:08,060 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:13:37,950 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.347e+02 2.892e+02 3.477e+02 7.491e+02, threshold=5.783e+02, percent-clipped=4.0 2022-12-07 16:13:55,096 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9966, 1.8823, 1.5346, 2.0649, 1.7722, 2.0206, 1.8464, 1.8253], device='cuda:0'), covar=tensor([0.0643, 0.0886, 0.2066, 0.0380, 0.0745, 0.0317, 0.1120, 0.0529], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0308, 0.0286, 0.0207, 0.0272, 0.0272, 0.0252, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:14:15,730 INFO [train.py:873] (0/4) Epoch 7, batch 2300, loss[loss=0.209, simple_loss=0.1833, pruned_loss=0.1173, over 1247.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.1799, pruned_loss=0.07178, over 1923855.18 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:14:28,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-07 16:14:39,684 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5406, 1.4735, 2.8058, 1.5418, 2.8057, 2.6957, 1.9090, 2.8276], device='cuda:0'), covar=tensor([0.0216, 0.2073, 0.0252, 0.1496, 0.0274, 0.0372, 0.0865, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0159, 0.0147, 0.0164, 0.0158, 0.0157, 0.0129, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:15:02,608 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1801, 1.0834, 1.0049, 1.1422, 1.1681, 0.7324, 1.0513, 1.3248], device='cuda:0'), covar=tensor([0.0541, 0.0911, 0.0635, 0.0777, 0.1462, 0.0706, 0.0812, 0.0404], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0019, 0.0019, 0.0019, 0.0028, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.7522e-05, 8.9653e-05, 8.7971e-05, 8.9492e-05, 8.9341e-05, 1.1635e-04, 9.4188e-05, 9.0525e-05], device='cuda:0') 2022-12-07 16:15:07,554 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.381e+02 2.948e+02 3.680e+02 8.732e+02, threshold=5.896e+02, percent-clipped=3.0 2022-12-07 16:15:10,957 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:15:13,652 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:15:17,146 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7128, 2.8547, 2.6658, 2.8705, 2.3035, 3.0965, 2.6358, 1.3814], device='cuda:0'), covar=tensor([0.3041, 0.0954, 0.1420, 0.0876, 0.1190, 0.0787, 0.1428, 0.3394], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0063, 0.0051, 0.0055, 0.0079, 0.0060, 0.0085, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:15:45,313 INFO [train.py:873] (0/4) Epoch 7, batch 2400, loss[loss=0.2004, simple_loss=0.1697, pruned_loss=0.1156, over 1291.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.1794, pruned_loss=0.0707, over 1958078.58 frames. ], batch size: 100, lr: 1.15e-02, grad_scale: 8.0 2022-12-07 16:15:58,190 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:16:08,557 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:16:17,328 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:16:22,633 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4680, 2.4086, 2.4915, 2.4580, 2.4046, 2.0559, 1.3565, 2.1938], device='cuda:0'), covar=tensor([0.0288, 0.0308, 0.0384, 0.0255, 0.0285, 0.0957, 0.2086, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0147, 0.0126, 0.0120, 0.0174, 0.0122, 0.0151, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:16:36,797 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.462e+02 3.088e+02 3.941e+02 7.678e+02, threshold=6.176e+02, percent-clipped=7.0 2022-12-07 16:16:48,482 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.44 vs. limit=5.0 2022-12-07 16:16:53,191 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:17:12,050 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:17:14,362 INFO [train.py:873] (0/4) Epoch 7, batch 2500, loss[loss=0.1569, simple_loss=0.1739, pruned_loss=0.06991, over 14256.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.1794, pruned_loss=0.07083, over 1972982.23 frames. ], batch size: 80, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:17:35,610 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:17:55,277 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2022-12-07 16:18:06,392 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.559e+02 3.205e+02 4.003e+02 7.455e+02, threshold=6.409e+02, percent-clipped=4.0 2022-12-07 16:18:18,471 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:18:43,889 INFO [train.py:873] (0/4) Epoch 7, batch 2600, loss[loss=0.1302, simple_loss=0.1651, pruned_loss=0.04767, over 13959.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.1786, pruned_loss=0.07034, over 1917430.50 frames. ], batch size: 26, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:19:36,189 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.385e+02 3.008e+02 3.884e+02 8.359e+02, threshold=6.016e+02, percent-clipped=2.0 2022-12-07 16:19:40,098 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:10,306 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5227, 2.1781, 2.3209, 1.4437, 2.0904, 2.3112, 2.6099, 2.1233], device='cuda:0'), covar=tensor([0.0672, 0.1549, 0.1246, 0.2799, 0.1209, 0.0729, 0.0594, 0.1576], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0196, 0.0126, 0.0133, 0.0114, 0.0117, 0.0098, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:20:13,185 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:13,909 INFO [train.py:873] (0/4) Epoch 7, batch 2700, loss[loss=0.1839, simple_loss=0.1882, pruned_loss=0.08979, over 7765.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.1789, pruned_loss=0.07038, over 1976316.21 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:20:23,342 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:33,200 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:20:33,715 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-07 16:21:01,797 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2071, 2.6993, 3.9694, 4.1662, 4.3543, 2.6131, 4.2424, 3.3829], device='cuda:0'), covar=tensor([0.0165, 0.0446, 0.0509, 0.0225, 0.0119, 0.0739, 0.0163, 0.0436], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0217, 0.0329, 0.0266, 0.0212, 0.0267, 0.0221, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-07 16:21:06,887 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.439e+01 2.366e+02 3.097e+02 3.751e+02 6.035e+02, threshold=6.194e+02, percent-clipped=1.0 2022-12-07 16:21:08,877 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:21:17,952 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:21:24,406 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6345, 3.3851, 3.0305, 2.2476, 2.8735, 3.3673, 3.5727, 2.8075], device='cuda:0'), covar=tensor([0.0572, 0.1963, 0.1250, 0.2160, 0.1280, 0.0525, 0.0973, 0.1534], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0196, 0.0127, 0.0131, 0.0114, 0.0117, 0.0097, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:21:37,460 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 16:21:40,900 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9375, 2.1649, 2.2404, 2.1278, 1.7955, 2.3292, 1.9509, 1.0587], device='cuda:0'), covar=tensor([0.2417, 0.1157, 0.0646, 0.0494, 0.1132, 0.0524, 0.1427, 0.3068], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0063, 0.0052, 0.0055, 0.0080, 0.0060, 0.0087, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0005, 0.0005], device='cuda:0') 2022-12-07 16:21:44,250 INFO [train.py:873] (0/4) Epoch 7, batch 2800, loss[loss=0.1766, simple_loss=0.1932, pruned_loss=0.07995, over 14241.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1789, pruned_loss=0.07009, over 2029970.12 frames. ], batch size: 69, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:21:56,837 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.58 vs. limit=5.0 2022-12-07 16:22:19,967 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:22:34,041 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1999, 1.9995, 2.1074, 2.1674, 2.0966, 2.1166, 2.2657, 1.8910], device='cuda:0'), covar=tensor([0.0525, 0.1108, 0.0558, 0.0643, 0.0774, 0.0504, 0.0631, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0237, 0.0155, 0.0148, 0.0159, 0.0124, 0.0233, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:22:36,452 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.244e+02 3.117e+02 3.930e+02 6.165e+02, threshold=6.234e+02, percent-clipped=0.0 2022-12-07 16:23:14,251 INFO [train.py:873] (0/4) Epoch 7, batch 2900, loss[loss=0.1528, simple_loss=0.1585, pruned_loss=0.07359, over 3829.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.1788, pruned_loss=0.07038, over 1971143.83 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:23:14,690 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:23:27,138 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 16:24:01,585 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:24:07,736 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.377e+02 3.206e+02 3.983e+02 6.235e+02, threshold=6.412e+02, percent-clipped=1.0 2022-12-07 16:24:46,197 INFO [train.py:873] (0/4) Epoch 7, batch 3000, loss[loss=0.1411, simple_loss=0.1693, pruned_loss=0.05641, over 13957.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.1789, pruned_loss=0.07008, over 1992487.57 frames. ], batch size: 26, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:24:46,197 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 16:25:07,553 INFO [train.py:905] (0/4) Epoch 7, validation: loss=0.1228, simple_loss=0.1657, pruned_loss=0.03995, over 857387.00 frames. 2022-12-07 16:25:07,554 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 16:25:07,690 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2063, 5.0351, 4.5037, 4.7779, 4.7121, 5.0703, 5.2333, 5.2413], device='cuda:0'), covar=tensor([0.0698, 0.0443, 0.2165, 0.2469, 0.0725, 0.0608, 0.0853, 0.0712], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0230, 0.0391, 0.0486, 0.0282, 0.0353, 0.0348, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:25:17,408 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1182, 1.0070, 1.0109, 1.2230, 1.2588, 0.6516, 1.2145, 1.3315], device='cuda:0'), covar=tensor([0.1060, 0.0799, 0.0449, 0.0626, 0.0843, 0.0643, 0.0653, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0021, 0.0021, 0.0020, 0.0028, 0.0020, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.3446e-05, 9.1584e-05, 9.2939e-05, 9.4806e-05, 9.2343e-05, 1.2083e-04, 9.8631e-05, 9.3810e-05], device='cuda:0') 2022-12-07 16:25:19,094 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:23,204 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3464, 2.9158, 2.7650, 1.8130, 2.7245, 3.0132, 3.3143, 2.6374], device='cuda:0'), covar=tensor([0.0653, 0.2098, 0.1583, 0.2791, 0.1015, 0.0738, 0.0978, 0.1770], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0195, 0.0125, 0.0128, 0.0113, 0.0118, 0.0096, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:25:27,609 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:30,245 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:54,426 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:25:58,983 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:01,391 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.229e+02 2.374e+02 2.963e+02 3.800e+02 1.337e+03, threshold=5.925e+02, percent-clipped=7.0 2022-12-07 16:26:10,818 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:12,865 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:21,507 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8694, 4.5126, 4.4790, 4.8346, 4.3825, 3.9090, 4.8675, 4.7441], device='cuda:0'), covar=tensor([0.0560, 0.0506, 0.0617, 0.0633, 0.0754, 0.0616, 0.0604, 0.0691], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0102, 0.0116, 0.0121, 0.0120, 0.0095, 0.0131, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 16:26:25,984 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:32,552 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:26:39,375 INFO [train.py:873] (0/4) Epoch 7, batch 3100, loss[loss=0.1668, simple_loss=0.1894, pruned_loss=0.07207, over 14591.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.1782, pruned_loss=0.06962, over 1955485.66 frames. ], batch size: 24, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:26:49,180 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:26:56,197 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:27:15,595 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:27:24,381 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2864, 1.0455, 0.9922, 1.2476, 0.9268, 0.6846, 1.2228, 1.3587], device='cuda:0'), covar=tensor([0.1757, 0.1223, 0.1743, 0.1007, 0.1544, 0.0782, 0.1355, 0.0873], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0021, 0.0021, 0.0020, 0.0029, 0.0021, 0.0022], device='cuda:0'), out_proj_covar=tensor([9.4520e-05, 9.3591e-05, 9.4359e-05, 9.5465e-05, 9.2880e-05, 1.2282e-04, 1.0132e-04, 9.8339e-05], device='cuda:0') 2022-12-07 16:27:31,119 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.509e+02 3.187e+02 4.348e+02 7.597e+02, threshold=6.375e+02, percent-clipped=5.0 2022-12-07 16:27:35,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2022-12-07 16:27:54,147 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6848, 5.0371, 5.0998, 5.6081, 5.2360, 4.5628, 5.5614, 4.6121], device='cuda:0'), covar=tensor([0.0259, 0.0863, 0.0297, 0.0400, 0.0713, 0.0357, 0.0427, 0.0502], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0237, 0.0157, 0.0150, 0.0159, 0.0124, 0.0236, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:28:04,843 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 16:28:09,138 INFO [train.py:873] (0/4) Epoch 7, batch 3200, loss[loss=0.1556, simple_loss=0.1748, pruned_loss=0.06821, over 10307.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.1793, pruned_loss=0.07073, over 1956851.15 frames. ], batch size: 100, lr: 1.14e-02, grad_scale: 8.0 2022-12-07 16:28:14,304 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1019, 3.8378, 3.7248, 4.1366, 3.9022, 3.6267, 4.2378, 3.4539], device='cuda:0'), covar=tensor([0.0554, 0.1050, 0.0426, 0.0484, 0.0833, 0.1272, 0.0525, 0.0676], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0239, 0.0158, 0.0151, 0.0159, 0.0126, 0.0236, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:28:19,947 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1698, 0.9731, 0.9164, 1.1193, 1.0886, 0.6963, 1.2109, 1.1825], device='cuda:0'), covar=tensor([0.0819, 0.1205, 0.0972, 0.0581, 0.0704, 0.0737, 0.1096, 0.0553], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0021, 0.0021, 0.0020, 0.0029, 0.0022, 0.0022], device='cuda:0'), out_proj_covar=tensor([9.6064e-05, 9.6566e-05, 9.6160e-05, 9.7890e-05, 9.3714e-05, 1.2456e-04, 1.0359e-04, 9.9793e-05], device='cuda:0') 2022-12-07 16:28:40,735 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9707, 3.4014, 2.6358, 4.0673, 3.8579, 3.8916, 3.2372, 2.6911], device='cuda:0'), covar=tensor([0.0579, 0.1400, 0.4058, 0.0387, 0.0791, 0.1060, 0.1230, 0.4071], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0306, 0.0284, 0.0208, 0.0274, 0.0275, 0.0253, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:29:02,495 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.245e+02 2.769e+02 3.901e+02 6.467e+02, threshold=5.537e+02, percent-clipped=1.0 2022-12-07 16:29:16,772 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1562, 3.0203, 2.6598, 2.8213, 3.0982, 3.0685, 3.1585, 3.1821], device='cuda:0'), covar=tensor([0.0855, 0.0677, 0.2174, 0.2439, 0.0714, 0.0730, 0.1001, 0.0718], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0227, 0.0382, 0.0481, 0.0278, 0.0349, 0.0341, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:29:40,601 INFO [train.py:873] (0/4) Epoch 7, batch 3300, loss[loss=0.2401, simple_loss=0.1962, pruned_loss=0.142, over 1197.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.1784, pruned_loss=0.07022, over 1972096.61 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:29:42,620 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4165, 2.2175, 3.4184, 3.5266, 3.4694, 2.2244, 3.3201, 2.5644], device='cuda:0'), covar=tensor([0.0201, 0.0454, 0.0347, 0.0238, 0.0176, 0.0734, 0.0206, 0.0560], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0219, 0.0331, 0.0269, 0.0215, 0.0267, 0.0224, 0.0255], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:29:46,873 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:30:30,577 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:30:32,953 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.201e+02 2.834e+02 3.394e+02 6.106e+02, threshold=5.668e+02, percent-clipped=1.0 2022-12-07 16:30:38,896 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8764, 1.7293, 4.5273, 1.7815, 4.1831, 4.6389, 4.5505, 5.2110], device='cuda:0'), covar=tensor([0.0168, 0.2764, 0.0344, 0.2290, 0.0374, 0.0341, 0.0232, 0.0127], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0160, 0.0148, 0.0167, 0.0159, 0.0160, 0.0131, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:30:53,186 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:09,739 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1160, 2.1156, 2.4387, 1.6565, 1.7810, 2.3389, 1.1901, 2.2135], device='cuda:0'), covar=tensor([0.1082, 0.1586, 0.0688, 0.2770, 0.2966, 0.0676, 0.4381, 0.0803], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0088, 0.0080, 0.0090, 0.0114, 0.0072, 0.0131, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:31:10,410 INFO [train.py:873] (0/4) Epoch 7, batch 3400, loss[loss=0.1668, simple_loss=0.1773, pruned_loss=0.07814, over 5994.00 frames. ], tot_loss[loss=0.1586, simple_loss=0.1775, pruned_loss=0.06988, over 1912224.38 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:31:13,833 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:16,508 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:31:26,906 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7302, 3.4672, 3.5157, 3.7879, 3.3855, 2.9919, 3.8005, 3.7061], device='cuda:0'), covar=tensor([0.0714, 0.0838, 0.0666, 0.0737, 0.0907, 0.0863, 0.0691, 0.0765], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0107, 0.0122, 0.0127, 0.0124, 0.0098, 0.0137, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 16:31:54,698 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:32:04,512 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.534e+01 2.454e+02 2.984e+02 3.725e+02 7.522e+02, threshold=5.968e+02, percent-clipped=2.0 2022-12-07 16:32:09,976 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:32:13,952 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2022-12-07 16:32:36,372 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:32:40,437 INFO [train.py:873] (0/4) Epoch 7, batch 3500, loss[loss=0.153, simple_loss=0.1785, pruned_loss=0.0638, over 14221.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.1776, pruned_loss=0.06999, over 1903422.02 frames. ], batch size: 60, lr: 1.13e-02, grad_scale: 4.0 2022-12-07 16:32:48,516 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:33:03,227 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:33:18,184 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:33:31,827 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.610e+02 3.212e+02 4.326e+02 7.674e+02, threshold=6.425e+02, percent-clipped=4.0 2022-12-07 16:34:08,387 INFO [train.py:873] (0/4) Epoch 7, batch 3600, loss[loss=0.128, simple_loss=0.1597, pruned_loss=0.04817, over 14262.00 frames. ], tot_loss[loss=0.1574, simple_loss=0.1766, pruned_loss=0.06907, over 1895416.31 frames. ], batch size: 63, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:34:14,446 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:34:44,396 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9632, 4.6569, 4.4947, 5.1112, 4.6235, 4.0800, 5.1140, 5.0099], device='cuda:0'), covar=tensor([0.0680, 0.0575, 0.0674, 0.0459, 0.0758, 0.0526, 0.0551, 0.0531], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0105, 0.0121, 0.0125, 0.0122, 0.0096, 0.0135, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 16:34:58,398 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:35:01,629 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.848e+01 2.321e+02 2.995e+02 3.518e+02 8.417e+02, threshold=5.990e+02, percent-clipped=3.0 2022-12-07 16:35:20,071 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:35:35,299 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 16:35:38,277 INFO [train.py:873] (0/4) Epoch 7, batch 3700, loss[loss=0.2074, simple_loss=0.2016, pruned_loss=0.1066, over 7758.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.177, pruned_loss=0.06898, over 1927585.63 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:35:43,634 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:02,766 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:26,218 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:36:30,493 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.323e+02 3.026e+02 3.847e+02 8.325e+02, threshold=6.052e+02, percent-clipped=4.0 2022-12-07 16:36:54,358 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-07 16:37:07,567 INFO [train.py:873] (0/4) Epoch 7, batch 3800, loss[loss=0.1524, simple_loss=0.1729, pruned_loss=0.06594, over 13544.00 frames. ], tot_loss[loss=0.158, simple_loss=0.1776, pruned_loss=0.06918, over 1958116.29 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:37:11,422 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:37:26,394 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:37:54,380 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9738, 2.2172, 2.7948, 2.4935, 2.7847, 2.7699, 2.8185, 2.4775], device='cuda:0'), covar=tensor([0.0563, 0.2873, 0.0819, 0.1698, 0.0498, 0.0783, 0.0917, 0.1896], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0330, 0.0386, 0.0312, 0.0366, 0.0304, 0.0362, 0.0337], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:38:01,122 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.147e+01 2.278e+02 2.831e+02 3.417e+02 7.148e+02, threshold=5.663e+02, percent-clipped=1.0 2022-12-07 16:38:04,719 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:38:37,955 INFO [train.py:873] (0/4) Epoch 7, batch 3900, loss[loss=0.1845, simple_loss=0.1929, pruned_loss=0.08806, over 10334.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.1772, pruned_loss=0.0691, over 1973471.55 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:38:40,193 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-07 16:38:59,128 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:39:03,183 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-07 16:39:30,001 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.072e+02 2.772e+02 3.353e+02 6.720e+02, threshold=5.544e+02, percent-clipped=1.0 2022-12-07 16:39:32,475 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-07 16:39:35,974 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0799, 4.6335, 4.5550, 5.0620, 4.7547, 4.4313, 5.0734, 4.2535], device='cuda:0'), covar=tensor([0.0346, 0.1062, 0.0328, 0.0400, 0.0703, 0.0423, 0.0465, 0.0509], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0238, 0.0159, 0.0150, 0.0156, 0.0125, 0.0237, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:40:05,914 INFO [train.py:873] (0/4) Epoch 7, batch 4000, loss[loss=0.219, simple_loss=0.178, pruned_loss=0.1299, over 1237.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.1769, pruned_loss=0.06924, over 1910597.66 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:40:49,575 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:40:58,226 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1589, 2.6970, 3.8599, 3.0560, 3.8530, 3.8361, 3.7184, 3.2963], device='cuda:0'), covar=tensor([0.0543, 0.3128, 0.0925, 0.2074, 0.0823, 0.0787, 0.1558, 0.2486], device='cuda:0'), in_proj_covar=tensor([0.0307, 0.0320, 0.0372, 0.0304, 0.0356, 0.0298, 0.0351, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:40:58,793 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.814e+01 2.339e+02 2.890e+02 3.562e+02 6.238e+02, threshold=5.779e+02, percent-clipped=1.0 2022-12-07 16:40:58,962 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:41:34,939 INFO [train.py:873] (0/4) Epoch 7, batch 4100, loss[loss=0.1616, simple_loss=0.1779, pruned_loss=0.07264, over 13536.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.1772, pruned_loss=0.06891, over 1975948.47 frames. ], batch size: 100, lr: 1.13e-02, grad_scale: 8.0 2022-12-07 16:41:38,600 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:41:43,915 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:41:52,405 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:41:53,234 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:42:00,549 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7318, 2.6145, 3.5002, 2.2932, 2.4054, 2.6321, 1.5732, 3.0225], device='cuda:0'), covar=tensor([0.1268, 0.1985, 0.0588, 0.2327, 0.2242, 0.1176, 0.4666, 0.0904], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0087, 0.0076, 0.0087, 0.0110, 0.0070, 0.0128, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:42:16,489 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0794, 0.8999, 1.1917, 1.2209, 1.3107, 0.6202, 1.2054, 1.1408], device='cuda:0'), covar=tensor([0.0695, 0.0969, 0.0551, 0.0699, 0.0853, 0.0612, 0.0813, 0.0797], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0020, 0.0020, 0.0028, 0.0021, 0.0021], device='cuda:0'), out_proj_covar=tensor([9.3050e-05, 9.4381e-05, 9.3506e-05, 9.4757e-05, 9.4560e-05, 1.2038e-04, 1.0139e-04, 9.7538e-05], device='cuda:0') 2022-12-07 16:42:20,799 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:42:26,663 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.128e+02 2.350e+02 2.777e+02 3.332e+02 5.900e+02, threshold=5.554e+02, percent-clipped=2.0 2022-12-07 16:42:35,642 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:43:03,202 INFO [train.py:873] (0/4) Epoch 7, batch 4200, loss[loss=0.133, simple_loss=0.1616, pruned_loss=0.05223, over 14019.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.177, pruned_loss=0.0684, over 2032276.60 frames. ], batch size: 19, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:43:19,650 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6764, 0.6790, 0.6955, 0.7050, 0.8162, 0.0958, 0.6559, 0.6957], device='cuda:0'), covar=tensor([0.0263, 0.0237, 0.0248, 0.0235, 0.0156, 0.0191, 0.0553, 0.0225], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0021, 0.0021, 0.0021, 0.0029, 0.0021, 0.0021], device='cuda:0'), out_proj_covar=tensor([9.4045e-05, 9.5833e-05, 9.4754e-05, 9.6965e-05, 9.6993e-05, 1.2280e-04, 1.0254e-04, 9.8832e-05], device='cuda:0') 2022-12-07 16:43:20,487 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:43:28,065 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2022-12-07 16:43:36,474 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-07 16:43:57,174 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.513e+02 3.123e+02 4.078e+02 1.360e+03, threshold=6.245e+02, percent-clipped=14.0 2022-12-07 16:44:32,551 INFO [train.py:873] (0/4) Epoch 7, batch 4300, loss[loss=0.1522, simple_loss=0.1824, pruned_loss=0.06106, over 14270.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.1776, pruned_loss=0.06908, over 2039922.33 frames. ], batch size: 57, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:45:22,446 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2406, 1.3133, 1.4436, 0.9604, 0.8649, 1.3290, 0.7863, 1.2053], device='cuda:0'), covar=tensor([0.1247, 0.1743, 0.0689, 0.2129, 0.2858, 0.0657, 0.2173, 0.0986], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0086, 0.0076, 0.0086, 0.0111, 0.0069, 0.0128, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:45:25,736 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.563e+02 3.565e+02 4.633e+02 7.506e+02, threshold=7.130e+02, percent-clipped=5.0 2022-12-07 16:45:56,355 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:02,011 INFO [train.py:873] (0/4) Epoch 7, batch 4400, loss[loss=0.1484, simple_loss=0.1794, pruned_loss=0.05869, over 14268.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.1771, pruned_loss=0.06835, over 2054101.59 frames. ], batch size: 57, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:46:06,640 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:14,501 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5795, 4.4285, 3.8978, 4.1172, 4.2070, 4.3868, 4.5383, 4.5539], device='cuda:0'), covar=tensor([0.0698, 0.0325, 0.1946, 0.2414, 0.0769, 0.0684, 0.0728, 0.0650], device='cuda:0'), in_proj_covar=tensor([0.0320, 0.0229, 0.0392, 0.0487, 0.0291, 0.0359, 0.0344, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:46:15,377 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:46:50,914 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:46:55,944 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.417e+01 2.268e+02 2.875e+02 3.664e+02 5.837e+02, threshold=5.750e+02, percent-clipped=0.0 2022-12-07 16:47:03,411 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2313, 1.0884, 1.4779, 1.3181, 1.5254, 0.6498, 1.4003, 1.6611], device='cuda:0'), covar=tensor([0.1566, 0.2349, 0.1351, 0.3989, 0.3150, 0.0660, 0.1140, 0.0795], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0020, 0.0020, 0.0028, 0.0021, 0.0021], device='cuda:0'), out_proj_covar=tensor([9.2933e-05, 9.4255e-05, 9.3964e-05, 9.4686e-05, 9.4444e-05, 1.2121e-04, 1.0030e-04, 9.7069e-05], device='cuda:0') 2022-12-07 16:47:31,485 INFO [train.py:873] (0/4) Epoch 7, batch 4500, loss[loss=0.1551, simple_loss=0.146, pruned_loss=0.08208, over 1243.00 frames. ], tot_loss[loss=0.1552, simple_loss=0.1758, pruned_loss=0.06732, over 2020012.42 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:47:48,469 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:48:11,251 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 16:48:24,474 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6314, 1.2211, 3.6524, 1.3637, 3.5352, 3.7391, 2.7232, 3.8641], device='cuda:0'), covar=tensor([0.0260, 0.3464, 0.0405, 0.2770, 0.0612, 0.0383, 0.0762, 0.0241], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0158, 0.0147, 0.0169, 0.0159, 0.0160, 0.0129, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 16:48:25,119 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.380e+02 2.362e+02 2.957e+02 3.544e+02 6.398e+02, threshold=5.915e+02, percent-clipped=1.0 2022-12-07 16:48:31,718 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:49:00,839 INFO [train.py:873] (0/4) Epoch 7, batch 4600, loss[loss=0.1606, simple_loss=0.1843, pruned_loss=0.06839, over 14028.00 frames. ], tot_loss[loss=0.1562, simple_loss=0.1766, pruned_loss=0.06793, over 1963087.37 frames. ], batch size: 26, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:49:25,398 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-50000.pt 2022-12-07 16:49:58,510 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.255e+02 2.301e+02 2.897e+02 3.880e+02 6.571e+02, threshold=5.793e+02, percent-clipped=2.0 2022-12-07 16:50:01,702 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-07 16:50:27,410 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2295, 1.3915, 1.3507, 1.1763, 1.2393, 1.0362, 1.0235, 0.7706], device='cuda:0'), covar=tensor([0.0703, 0.0554, 0.1056, 0.0775, 0.0721, 0.0409, 0.0370, 0.1367], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0010, 0.0010, 0.0012, 0.0017, 0.0013, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.4374e-05, 8.0605e-05, 7.3306e-05, 7.4282e-05, 7.8186e-05, 1.1196e-04, 9.3505e-05, 1.0731e-04], device='cuda:0') 2022-12-07 16:50:32,304 INFO [train.py:873] (0/4) Epoch 7, batch 4700, loss[loss=0.1884, simple_loss=0.1917, pruned_loss=0.09252, over 7690.00 frames. ], tot_loss[loss=0.158, simple_loss=0.1777, pruned_loss=0.06912, over 1982422.49 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 4.0 2022-12-07 16:50:37,341 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:50:45,855 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 16:51:15,735 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:19,180 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:20,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:51:25,948 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.013e+02 2.271e+02 2.971e+02 3.641e+02 8.155e+02, threshold=5.943e+02, percent-clipped=3.0 2022-12-07 16:51:27,666 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:51:54,625 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8772, 1.4517, 1.8026, 1.3122, 1.4687, 1.8249, 1.7037, 1.5185], device='cuda:0'), covar=tensor([0.0445, 0.0541, 0.0488, 0.0742, 0.1336, 0.0361, 0.0321, 0.1408], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0198, 0.0127, 0.0130, 0.0118, 0.0120, 0.0099, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 16:52:00,690 INFO [train.py:873] (0/4) Epoch 7, batch 4800, loss[loss=0.119, simple_loss=0.1448, pruned_loss=0.04654, over 14299.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.1771, pruned_loss=0.06892, over 1989092.29 frames. ], batch size: 18, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:52:14,816 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:52:21,946 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:52:54,463 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.403e+02 3.099e+02 4.047e+02 7.554e+02, threshold=6.197e+02, percent-clipped=4.0 2022-12-07 16:52:55,787 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2678, 3.1112, 2.8371, 2.9311, 3.2154, 3.1674, 3.3037, 3.2367], device='cuda:0'), covar=tensor([0.1077, 0.0818, 0.2636, 0.3521, 0.0820, 0.1010, 0.1224, 0.1098], device='cuda:0'), in_proj_covar=tensor([0.0326, 0.0233, 0.0388, 0.0488, 0.0286, 0.0359, 0.0353, 0.0309], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:53:15,863 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 16:53:29,364 INFO [train.py:873] (0/4) Epoch 7, batch 4900, loss[loss=0.1256, simple_loss=0.1541, pruned_loss=0.04855, over 14057.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.1782, pruned_loss=0.07003, over 1977692.14 frames. ], batch size: 22, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:53:56,957 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2022-12-07 16:54:23,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.354e+02 3.057e+02 3.894e+02 8.460e+02, threshold=6.115e+02, percent-clipped=5.0 2022-12-07 16:54:49,283 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1127, 1.1782, 1.1592, 1.1340, 1.3208, 0.6301, 1.1485, 1.4449], device='cuda:0'), covar=tensor([0.1002, 0.1043, 0.0801, 0.2504, 0.1325, 0.0730, 0.3506, 0.0676], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0020, 0.0020, 0.0020, 0.0028, 0.0021, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.1914e-05, 9.3816e-05, 9.2309e-05, 9.2703e-05, 9.3581e-05, 1.2064e-04, 1.0011e-04, 9.5787e-05], device='cuda:0') 2022-12-07 16:54:58,365 INFO [train.py:873] (0/4) Epoch 7, batch 5000, loss[loss=0.1572, simple_loss=0.1546, pruned_loss=0.07994, over 2674.00 frames. ], tot_loss[loss=0.16, simple_loss=0.1785, pruned_loss=0.07073, over 1943456.83 frames. ], batch size: 100, lr: 1.12e-02, grad_scale: 8.0 2022-12-07 16:55:42,266 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:55:52,925 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.513e+02 3.076e+02 4.134e+02 7.861e+02, threshold=6.153e+02, percent-clipped=5.0 2022-12-07 16:56:01,998 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-07 16:56:05,261 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 16:56:24,876 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:56:27,487 INFO [train.py:873] (0/4) Epoch 7, batch 5100, loss[loss=0.1708, simple_loss=0.1801, pruned_loss=0.08075, over 11200.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.1774, pruned_loss=0.0696, over 1921277.06 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 16:56:28,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7530, 3.5551, 3.2513, 3.3905, 3.6278, 3.6554, 3.7626, 3.7347], device='cuda:0'), covar=tensor([0.0918, 0.0577, 0.2047, 0.2587, 0.0775, 0.0789, 0.0950, 0.0851], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0238, 0.0393, 0.0498, 0.0292, 0.0365, 0.0361, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 16:56:37,624 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:57:10,444 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0010, 3.6898, 3.6674, 3.9969, 3.8083, 3.5176, 4.0263, 3.4298], device='cuda:0'), covar=tensor([0.0429, 0.0891, 0.0360, 0.0424, 0.0655, 0.1192, 0.0490, 0.0478], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0238, 0.0163, 0.0157, 0.0158, 0.0128, 0.0237, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:57:18,094 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 16:57:20,713 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.452e+02 3.012e+02 3.839e+02 6.174e+02, threshold=6.024e+02, percent-clipped=1.0 2022-12-07 16:57:23,545 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2031, 1.9675, 2.0984, 2.1871, 2.1277, 2.0711, 2.2474, 1.8150], device='cuda:0'), covar=tensor([0.0645, 0.1232, 0.0639, 0.0828, 0.0868, 0.0641, 0.0810, 0.0774], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0240, 0.0164, 0.0158, 0.0160, 0.0129, 0.0239, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 16:57:32,036 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:57:38,058 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 16:57:55,458 INFO [train.py:873] (0/4) Epoch 7, batch 5200, loss[loss=0.138, simple_loss=0.1656, pruned_loss=0.05524, over 14027.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.1777, pruned_loss=0.06905, over 2044722.00 frames. ], batch size: 22, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 16:58:17,146 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5512, 3.1714, 2.3571, 3.6242, 3.4697, 3.4155, 3.0345, 2.3970], device='cuda:0'), covar=tensor([0.0630, 0.1475, 0.4048, 0.0404, 0.0658, 0.1420, 0.1288, 0.4183], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0304, 0.0283, 0.0209, 0.0274, 0.0273, 0.0254, 0.0269], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 16:58:26,292 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 16:58:49,801 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.234e+02 2.881e+02 3.626e+02 9.172e+02, threshold=5.763e+02, percent-clipped=1.0 2022-12-07 16:59:09,981 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8551, 4.5608, 4.4798, 4.8954, 4.3326, 4.0253, 4.9388, 4.7891], device='cuda:0'), covar=tensor([0.0772, 0.0555, 0.0678, 0.0657, 0.0834, 0.0559, 0.0558, 0.0775], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0108, 0.0123, 0.0125, 0.0127, 0.0096, 0.0136, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 16:59:24,246 INFO [train.py:873] (0/4) Epoch 7, batch 5300, loss[loss=0.1584, simple_loss=0.184, pruned_loss=0.06635, over 14087.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.1785, pruned_loss=0.06983, over 1993689.27 frames. ], batch size: 29, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 16:59:49,246 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4597, 1.5792, 4.2359, 1.9472, 4.1266, 4.3980, 3.6256, 4.7484], device='cuda:0'), covar=tensor([0.0160, 0.2861, 0.0301, 0.2064, 0.0268, 0.0240, 0.0437, 0.0121], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0159, 0.0149, 0.0168, 0.0160, 0.0161, 0.0131, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:00:04,530 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 17:00:16,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 2.441e+02 3.172e+02 3.955e+02 6.877e+02, threshold=6.344e+02, percent-clipped=8.0 2022-12-07 17:00:49,950 INFO [train.py:873] (0/4) Epoch 7, batch 5400, loss[loss=0.1543, simple_loss=0.1466, pruned_loss=0.08101, over 2666.00 frames. ], tot_loss[loss=0.1578, simple_loss=0.1778, pruned_loss=0.06891, over 1981703.47 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:00:58,073 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:00:59,861 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:01:20,863 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:01:22,684 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3459, 1.2757, 1.3432, 1.1934, 1.0282, 0.8469, 0.7291, 0.7657], device='cuda:0'), covar=tensor([0.0244, 0.0529, 0.0239, 0.0308, 0.0353, 0.0310, 0.0244, 0.0756], device='cuda:0'), in_proj_covar=tensor([0.0011, 0.0012, 0.0011, 0.0011, 0.0012, 0.0018, 0.0014, 0.0018], device='cuda:0'), out_proj_covar=tensor([7.6287e-05, 8.3168e-05, 7.5630e-05, 7.7563e-05, 7.9355e-05, 1.1582e-04, 9.5871e-05, 1.1080e-04], device='cuda:0') 2022-12-07 17:01:30,760 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-07 17:01:41,533 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:01:43,629 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.469e+02 3.143e+02 3.850e+02 7.689e+02, threshold=6.287e+02, percent-clipped=2.0 2022-12-07 17:01:51,284 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:02:00,051 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:02:10,865 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0517, 1.2341, 1.7619, 0.9714, 1.3424, 0.9476, 1.1140, 0.9277], device='cuda:0'), covar=tensor([0.0725, 0.1305, 0.0791, 0.1336, 0.1180, 0.0487, 0.0762, 0.1279], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0014, 0.0019], device='cuda:0'), out_proj_covar=tensor([7.8205e-05, 8.4983e-05, 7.7942e-05, 7.9471e-05, 8.1353e-05, 1.1879e-04, 9.8982e-05, 1.1371e-04], device='cuda:0') 2022-12-07 17:02:14,147 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:02:17,732 INFO [train.py:873] (0/4) Epoch 7, batch 5500, loss[loss=0.167, simple_loss=0.1839, pruned_loss=0.07503, over 14249.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.1779, pruned_loss=0.06976, over 1971981.04 frames. ], batch size: 69, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:02:37,535 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2022-12-07 17:02:42,265 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:02:43,109 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:02:45,933 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6360, 2.6190, 1.8924, 2.7316, 2.4568, 2.5915, 2.3447, 2.1608], device='cuda:0'), covar=tensor([0.0694, 0.1141, 0.3166, 0.0404, 0.1117, 0.0662, 0.1563, 0.2644], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0302, 0.0284, 0.0207, 0.0277, 0.0272, 0.0252, 0.0269], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:03:10,733 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.288e+02 2.905e+02 3.483e+02 6.290e+02, threshold=5.809e+02, percent-clipped=1.0 2022-12-07 17:03:21,370 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:03:44,815 INFO [train.py:873] (0/4) Epoch 7, batch 5600, loss[loss=0.1399, simple_loss=0.1747, pruned_loss=0.05254, over 14348.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.1779, pruned_loss=0.06988, over 1952983.97 frames. ], batch size: 31, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:04:08,813 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-07 17:04:15,897 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:04:38,633 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.390e+02 3.229e+02 4.054e+02 7.897e+02, threshold=6.457e+02, percent-clipped=5.0 2022-12-07 17:04:47,584 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1047, 3.8713, 3.5905, 3.7455, 3.9490, 4.0313, 4.1348, 4.0862], device='cuda:0'), covar=tensor([0.0882, 0.0611, 0.2259, 0.2784, 0.0748, 0.0681, 0.0948, 0.0836], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0235, 0.0386, 0.0486, 0.0288, 0.0357, 0.0349, 0.0307], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:05:03,864 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:05:13,365 INFO [train.py:873] (0/4) Epoch 7, batch 5700, loss[loss=0.1857, simple_loss=0.1682, pruned_loss=0.1016, over 1244.00 frames. ], tot_loss[loss=0.1571, simple_loss=0.177, pruned_loss=0.06863, over 1976172.14 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:05:27,100 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3939, 1.3459, 3.4008, 1.5171, 3.3024, 3.5574, 2.4790, 3.7487], device='cuda:0'), covar=tensor([0.0225, 0.3035, 0.0375, 0.2216, 0.0699, 0.0312, 0.0758, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0159, 0.0148, 0.0169, 0.0161, 0.0160, 0.0128, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:05:55,389 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-07 17:05:56,810 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:06:05,990 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.301e+02 2.940e+02 3.681e+02 9.913e+02, threshold=5.881e+02, percent-clipped=3.0 2022-12-07 17:06:09,495 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:06:31,989 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:06:39,082 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:06:39,723 INFO [train.py:873] (0/4) Epoch 7, batch 5800, loss[loss=0.1956, simple_loss=0.1921, pruned_loss=0.09956, over 11962.00 frames. ], tot_loss[loss=0.1584, simple_loss=0.1777, pruned_loss=0.06952, over 1970816.40 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:06:51,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8076, 1.3771, 3.5829, 3.3396, 3.4831, 3.5951, 2.8237, 3.6244], device='cuda:0'), covar=tensor([0.1288, 0.1522, 0.0102, 0.0194, 0.0165, 0.0110, 0.0250, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0114, 0.0156, 0.0130, 0.0128, 0.0106, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:07:05,751 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:07:12,850 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9523, 1.3775, 3.8361, 1.5515, 3.8354, 3.8904, 2.9469, 4.2800], device='cuda:0'), covar=tensor([0.0203, 0.3299, 0.0356, 0.2503, 0.0423, 0.0399, 0.0552, 0.0133], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0161, 0.0148, 0.0171, 0.0162, 0.0163, 0.0129, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:07:23,312 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1832, 4.9569, 4.5240, 4.7892, 4.7781, 5.0525, 5.1936, 5.1502], device='cuda:0'), covar=tensor([0.0766, 0.0327, 0.1991, 0.2246, 0.0730, 0.0579, 0.0704, 0.0783], device='cuda:0'), in_proj_covar=tensor([0.0328, 0.0235, 0.0391, 0.0489, 0.0289, 0.0356, 0.0350, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:07:33,316 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:07:33,900 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.265e+02 2.888e+02 3.967e+02 8.048e+02, threshold=5.777e+02, percent-clipped=2.0 2022-12-07 17:07:34,120 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0219, 3.4528, 4.2064, 3.0069, 2.5527, 3.5416, 2.0207, 3.5280], device='cuda:0'), covar=tensor([0.1630, 0.1019, 0.0512, 0.2461, 0.2516, 0.0884, 0.4372, 0.0778], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0084, 0.0078, 0.0087, 0.0107, 0.0070, 0.0127, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:07:48,087 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:08:08,702 INFO [train.py:873] (0/4) Epoch 7, batch 5900, loss[loss=0.1471, simple_loss=0.165, pruned_loss=0.06455, over 5979.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.1768, pruned_loss=0.06852, over 2023242.33 frames. ], batch size: 100, lr: 1.11e-02, grad_scale: 8.0 2022-12-07 17:08:34,078 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:08:56,743 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6197, 3.8730, 4.4320, 4.7020, 4.3473, 4.0477, 4.5717, 3.9046], device='cuda:0'), covar=tensor([0.0764, 0.1909, 0.0719, 0.0774, 0.1096, 0.0794, 0.0910, 0.0946], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0232, 0.0161, 0.0152, 0.0157, 0.0125, 0.0233, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:09:01,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.784e+01 2.421e+02 2.907e+02 3.542e+02 5.776e+02, threshold=5.815e+02, percent-clipped=0.0 2022-12-07 17:09:26,091 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7174, 2.7288, 2.5680, 2.8835, 2.4266, 2.5413, 2.8459, 2.8130], device='cuda:0'), covar=tensor([0.0827, 0.0840, 0.0865, 0.0732, 0.0939, 0.0730, 0.0788, 0.0886], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0111, 0.0124, 0.0129, 0.0127, 0.0098, 0.0140, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:09:35,645 INFO [train.py:873] (0/4) Epoch 7, batch 6000, loss[loss=0.1667, simple_loss=0.1676, pruned_loss=0.08294, over 4991.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.1765, pruned_loss=0.06812, over 2017381.13 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:09:35,645 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 17:09:52,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2198, 2.2135, 4.6087, 4.2883, 4.2623, 4.6989, 4.4510, 4.7937], device='cuda:0'), covar=tensor([0.1359, 0.1237, 0.0079, 0.0153, 0.0154, 0.0086, 0.0071, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0158, 0.0115, 0.0157, 0.0130, 0.0129, 0.0105, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:09:54,555 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4995, 2.2576, 2.2089, 1.6911, 1.8847, 2.3796, 2.2358, 1.8574], device='cuda:0'), covar=tensor([0.0277, 0.0373, 0.0301, 0.0549, 0.0745, 0.0261, 0.0243, 0.0987], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0194, 0.0123, 0.0126, 0.0117, 0.0116, 0.0096, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 17:09:56,740 INFO [train.py:905] (0/4) Epoch 7, validation: loss=0.1227, simple_loss=0.1653, pruned_loss=0.04007, over 857387.00 frames. 2022-12-07 17:09:56,740 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 17:10:04,650 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:13,333 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:36,276 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:49,689 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 2.553e+02 3.284e+02 4.114e+02 1.252e+03, threshold=6.568e+02, percent-clipped=8.0 2022-12-07 17:10:53,286 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:10:57,480 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:03,118 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-07 17:11:06,347 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:15,868 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:11:23,460 INFO [train.py:873] (0/4) Epoch 7, batch 6100, loss[loss=0.1602, simple_loss=0.1828, pruned_loss=0.06885, over 14509.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.1776, pruned_loss=0.06907, over 1985118.00 frames. ], batch size: 34, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:11:34,636 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:38,101 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3769, 1.2178, 3.5285, 1.4504, 3.3755, 3.4698, 2.3799, 3.6985], device='cuda:0'), covar=tensor([0.0227, 0.3145, 0.0286, 0.2319, 0.0619, 0.0333, 0.0786, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0157, 0.0145, 0.0167, 0.0160, 0.0159, 0.0128, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:11:42,576 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:11:57,330 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:12:10,909 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 17:12:16,320 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.483e+02 3.170e+02 3.695e+02 8.405e+02, threshold=6.340e+02, percent-clipped=2.0 2022-12-07 17:12:35,844 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:12:41,494 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7298, 3.4893, 3.2441, 3.4068, 3.6885, 3.6698, 3.7551, 3.6862], device='cuda:0'), covar=tensor([0.0992, 0.0882, 0.2795, 0.3208, 0.0848, 0.0885, 0.1161, 0.1133], device='cuda:0'), in_proj_covar=tensor([0.0337, 0.0237, 0.0403, 0.0499, 0.0290, 0.0370, 0.0363, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:12:50,517 INFO [train.py:873] (0/4) Epoch 7, batch 6200, loss[loss=0.1638, simple_loss=0.1913, pruned_loss=0.06813, over 14023.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.177, pruned_loss=0.06835, over 2028204.81 frames. ], batch size: 22, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:13:12,855 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3199, 3.1194, 3.0707, 3.4669, 3.0303, 2.7965, 3.3880, 3.4214], device='cuda:0'), covar=tensor([0.0709, 0.1026, 0.0880, 0.0644, 0.0953, 0.0891, 0.0748, 0.0715], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0109, 0.0122, 0.0126, 0.0124, 0.0096, 0.0136, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:13:12,871 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4799, 3.1726, 3.1842, 3.4485, 3.2749, 3.4768, 3.5437, 2.9031], device='cuda:0'), covar=tensor([0.0442, 0.1216, 0.0498, 0.0581, 0.0829, 0.0368, 0.0614, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0235, 0.0163, 0.0154, 0.0159, 0.0127, 0.0237, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:13:16,641 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:13:39,969 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 17:13:44,670 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.329e+02 2.979e+02 3.715e+02 8.886e+02, threshold=5.958e+02, percent-clipped=2.0 2022-12-07 17:13:45,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7991, 2.4538, 3.4898, 2.7183, 3.6411, 3.4662, 3.3465, 3.0887], device='cuda:0'), covar=tensor([0.0588, 0.3540, 0.1183, 0.2213, 0.1044, 0.0877, 0.1879, 0.2218], device='cuda:0'), in_proj_covar=tensor([0.0304, 0.0326, 0.0379, 0.0301, 0.0365, 0.0296, 0.0351, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:13:58,640 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:14:04,984 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1509, 1.9383, 2.0312, 2.1525, 2.0649, 2.0464, 2.2014, 1.8503], device='cuda:0'), covar=tensor([0.0698, 0.1377, 0.0856, 0.0841, 0.1013, 0.0750, 0.0920, 0.0693], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0236, 0.0163, 0.0154, 0.0159, 0.0128, 0.0238, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:14:19,069 INFO [train.py:873] (0/4) Epoch 7, batch 6300, loss[loss=0.1799, simple_loss=0.1719, pruned_loss=0.09392, over 3884.00 frames. ], tot_loss[loss=0.1559, simple_loss=0.1763, pruned_loss=0.06778, over 1995521.16 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:14:52,282 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0086, 2.0643, 1.8431, 2.1522, 1.6666, 1.9234, 2.0521, 2.1125], device='cuda:0'), covar=tensor([0.0984, 0.1059, 0.1154, 0.0968, 0.1463, 0.0945, 0.1180, 0.1077], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0108, 0.0121, 0.0127, 0.0124, 0.0096, 0.0136, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:14:58,591 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:12,417 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.206e+02 2.744e+02 3.241e+02 7.232e+02, threshold=5.488e+02, percent-clipped=1.0 2022-12-07 17:15:16,218 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:17,023 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0771, 3.0182, 2.8295, 3.2115, 2.7039, 2.6749, 3.1397, 3.1675], device='cuda:0'), covar=tensor([0.0776, 0.0767, 0.0799, 0.0643, 0.1046, 0.0859, 0.0784, 0.0681], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0108, 0.0121, 0.0127, 0.0124, 0.0096, 0.0137, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:15:23,974 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:24,630 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:40,162 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:15:46,560 INFO [train.py:873] (0/4) Epoch 7, batch 6400, loss[loss=0.1101, simple_loss=0.1402, pruned_loss=0.03995, over 11352.00 frames. ], tot_loss[loss=0.1549, simple_loss=0.1753, pruned_loss=0.06719, over 1933026.96 frames. ], batch size: 14, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:16:17,552 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:16:31,223 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.18 vs. limit=5.0 2022-12-07 17:16:35,053 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:16:39,967 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.371e+02 3.096e+02 4.074e+02 1.030e+03, threshold=6.192e+02, percent-clipped=3.0 2022-12-07 17:16:55,656 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:17:03,265 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 17:17:12,025 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7843, 0.8344, 0.8047, 0.8448, 1.1119, 0.5259, 0.9626, 0.7973], device='cuda:0'), covar=tensor([0.0593, 0.0601, 0.0578, 0.0663, 0.0341, 0.0589, 0.0466, 0.0551], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0021, 0.0021, 0.0021, 0.0030, 0.0021, 0.0021], device='cuda:0'), out_proj_covar=tensor([9.7302e-05, 1.0040e-04, 9.9805e-05, 9.8785e-05, 9.9175e-05, 1.2916e-04, 1.0247e-04, 1.0015e-04], device='cuda:0') 2022-12-07 17:17:14,510 INFO [train.py:873] (0/4) Epoch 7, batch 6500, loss[loss=0.1534, simple_loss=0.1773, pruned_loss=0.06472, over 14323.00 frames. ], tot_loss[loss=0.156, simple_loss=0.1763, pruned_loss=0.06783, over 1979696.31 frames. ], batch size: 60, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:17:17,360 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:18:07,971 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.306e+02 2.865e+02 3.658e+02 7.175e+02, threshold=5.730e+02, percent-clipped=1.0 2022-12-07 17:18:28,931 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2022-12-07 17:18:41,743 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9355, 4.2817, 3.2052, 5.0705, 4.4987, 4.9173, 4.2095, 3.5366], device='cuda:0'), covar=tensor([0.0539, 0.1217, 0.4691, 0.1050, 0.1258, 0.1042, 0.1238, 0.3886], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0297, 0.0281, 0.0210, 0.0270, 0.0269, 0.0252, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:18:42,428 INFO [train.py:873] (0/4) Epoch 7, batch 6600, loss[loss=0.1274, simple_loss=0.1645, pruned_loss=0.04509, over 14009.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.1763, pruned_loss=0.06827, over 1974827.06 frames. ], batch size: 19, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:19:12,864 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2785, 3.9743, 3.8910, 4.2969, 4.0791, 3.7892, 4.3209, 3.6520], device='cuda:0'), covar=tensor([0.0415, 0.0850, 0.0360, 0.0404, 0.0687, 0.0845, 0.0508, 0.0481], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0233, 0.0163, 0.0154, 0.0158, 0.0129, 0.0241, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:19:35,920 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.181e+02 2.663e+02 3.409e+02 8.463e+02, threshold=5.326e+02, percent-clipped=3.0 2022-12-07 17:19:39,823 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:19:41,586 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:19:48,552 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:10,104 INFO [train.py:873] (0/4) Epoch 7, batch 6700, loss[loss=0.1315, simple_loss=0.1348, pruned_loss=0.06411, over 2617.00 frames. ], tot_loss[loss=0.157, simple_loss=0.1768, pruned_loss=0.06855, over 1974785.69 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 16.0 2022-12-07 17:20:21,591 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:30,300 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:34,983 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:20:36,519 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:21:03,676 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 2.435e+02 3.169e+02 4.506e+02 1.155e+03, threshold=6.339e+02, percent-clipped=9.0 2022-12-07 17:21:18,726 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:21:38,016 INFO [train.py:873] (0/4) Epoch 7, batch 6800, loss[loss=0.1598, simple_loss=0.1649, pruned_loss=0.07734, over 3927.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.1761, pruned_loss=0.0684, over 1924731.21 frames. ], batch size: 100, lr: 1.10e-02, grad_scale: 16.0 2022-12-07 17:21:55,470 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:21:57,975 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2747, 1.3093, 1.4617, 0.9796, 0.9733, 1.2981, 0.7958, 1.2087], device='cuda:0'), covar=tensor([0.1511, 0.2124, 0.0534, 0.1977, 0.2721, 0.0896, 0.1957, 0.1101], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0085, 0.0079, 0.0087, 0.0108, 0.0072, 0.0128, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:22:00,774 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:22:05,035 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 17:22:22,248 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.35 vs. limit=2.0 2022-12-07 17:22:32,532 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.590e+02 3.090e+02 3.890e+02 7.235e+02, threshold=6.180e+02, percent-clipped=2.0 2022-12-07 17:22:49,421 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:23:05,569 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2022-12-07 17:23:05,760 INFO [train.py:873] (0/4) Epoch 7, batch 6900, loss[loss=0.168, simple_loss=0.1817, pruned_loss=0.07711, over 14281.00 frames. ], tot_loss[loss=0.1571, simple_loss=0.1763, pruned_loss=0.06892, over 1932835.16 frames. ], batch size: 60, lr: 1.10e-02, grad_scale: 8.0 2022-12-07 17:23:25,311 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0765, 3.8198, 3.5942, 3.7215, 3.9440, 3.9543, 4.1334, 4.0765], device='cuda:0'), covar=tensor([0.0832, 0.0590, 0.1683, 0.2765, 0.0655, 0.0776, 0.0810, 0.0777], device='cuda:0'), in_proj_covar=tensor([0.0325, 0.0232, 0.0388, 0.0486, 0.0283, 0.0357, 0.0346, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:23:40,130 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:24:00,341 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.451e+02 3.077e+02 3.642e+02 5.911e+02, threshold=6.154e+02, percent-clipped=0.0 2022-12-07 17:24:02,263 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7928, 3.1866, 4.4235, 3.2960, 4.5187, 4.2253, 4.1054, 3.8118], device='cuda:0'), covar=tensor([0.0418, 0.2745, 0.0843, 0.1893, 0.0791, 0.0842, 0.1611, 0.1962], device='cuda:0'), in_proj_covar=tensor([0.0311, 0.0331, 0.0382, 0.0306, 0.0371, 0.0301, 0.0352, 0.0333], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:24:20,405 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8435, 2.0701, 2.6159, 2.2538, 2.6193, 2.5659, 2.4109, 2.2396], device='cuda:0'), covar=tensor([0.0639, 0.2976, 0.0888, 0.2077, 0.0685, 0.0889, 0.1005, 0.1719], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0332, 0.0385, 0.0310, 0.0375, 0.0302, 0.0355, 0.0334], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:24:33,250 INFO [train.py:873] (0/4) Epoch 7, batch 7000, loss[loss=0.2271, simple_loss=0.1903, pruned_loss=0.1319, over 1193.00 frames. ], tot_loss[loss=0.1563, simple_loss=0.1759, pruned_loss=0.06834, over 1932626.52 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:24:33,387 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:24:36,430 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2022-12-07 17:24:53,410 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:25:00,095 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:25:26,019 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 17:25:27,918 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.412e+02 3.030e+02 3.769e+02 8.602e+02, threshold=6.059e+02, percent-clipped=2.0 2022-12-07 17:25:41,994 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:26:01,212 INFO [train.py:873] (0/4) Epoch 7, batch 7100, loss[loss=0.1401, simple_loss=0.1678, pruned_loss=0.05621, over 14239.00 frames. ], tot_loss[loss=0.1546, simple_loss=0.1754, pruned_loss=0.06694, over 1948368.81 frames. ], batch size: 69, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:26:04,672 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7971, 1.2311, 1.7466, 1.1501, 1.3623, 1.8589, 1.5504, 1.4847], device='cuda:0'), covar=tensor([0.0531, 0.0923, 0.0585, 0.1020, 0.1091, 0.0763, 0.0500, 0.1481], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0192, 0.0124, 0.0127, 0.0116, 0.0118, 0.0099, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 17:26:53,042 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8242, 2.5859, 2.6893, 2.8416, 2.7764, 2.7816, 2.9139, 2.4259], device='cuda:0'), covar=tensor([0.0599, 0.1238, 0.0550, 0.0603, 0.0740, 0.0496, 0.0716, 0.0656], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0242, 0.0166, 0.0158, 0.0161, 0.0132, 0.0244, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:26:56,577 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.463e+02 3.010e+02 3.918e+02 9.338e+02, threshold=6.019e+02, percent-clipped=5.0 2022-12-07 17:26:58,014 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.87 vs. limit=5.0 2022-12-07 17:27:08,055 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:27:13,420 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:27:23,145 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.32 vs. limit=2.0 2022-12-07 17:27:29,108 INFO [train.py:873] (0/4) Epoch 7, batch 7200, loss[loss=0.1891, simple_loss=0.166, pruned_loss=0.1061, over 2576.00 frames. ], tot_loss[loss=0.155, simple_loss=0.1752, pruned_loss=0.06742, over 1907620.04 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:28:07,075 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:28:13,748 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 17:28:23,146 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2022-12-07 17:28:24,357 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.010e+02 2.308e+02 2.919e+02 3.605e+02 5.197e+02, threshold=5.838e+02, percent-clipped=0.0 2022-12-07 17:28:41,632 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-07 17:28:52,746 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:28:56,838 INFO [train.py:873] (0/4) Epoch 7, batch 7300, loss[loss=0.1469, simple_loss=0.1718, pruned_loss=0.06106, over 14292.00 frames. ], tot_loss[loss=0.1547, simple_loss=0.1756, pruned_loss=0.06692, over 2000666.12 frames. ], batch size: 76, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:29:16,728 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:29:47,018 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8714, 1.4357, 1.8602, 2.0633, 1.4118, 1.7238, 1.7739, 1.9490], device='cuda:0'), covar=tensor([0.0055, 0.0111, 0.0055, 0.0032, 0.0106, 0.0121, 0.0070, 0.0046], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0221, 0.0327, 0.0270, 0.0218, 0.0264, 0.0233, 0.0255], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:29:48,598 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.77 vs. limit=2.0 2022-12-07 17:29:51,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.341e+02 3.110e+02 4.071e+02 1.233e+03, threshold=6.219e+02, percent-clipped=5.0 2022-12-07 17:29:58,700 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:30:05,580 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:30:24,391 INFO [train.py:873] (0/4) Epoch 7, batch 7400, loss[loss=0.1995, simple_loss=0.1978, pruned_loss=0.1006, over 9485.00 frames. ], tot_loss[loss=0.1536, simple_loss=0.175, pruned_loss=0.06605, over 1994354.47 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:31:00,177 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:31:01,085 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5785, 2.1675, 3.4829, 3.6444, 3.5977, 2.1868, 3.5331, 2.6519], device='cuda:0'), covar=tensor([0.0191, 0.0497, 0.0450, 0.0255, 0.0172, 0.0776, 0.0194, 0.0556], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0222, 0.0329, 0.0272, 0.0220, 0.0266, 0.0237, 0.0256], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:31:04,045 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 17:31:19,607 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.081e+01 2.208e+02 2.890e+02 3.954e+02 8.452e+02, threshold=5.779e+02, percent-clipped=3.0 2022-12-07 17:31:31,432 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:31:52,179 INFO [train.py:873] (0/4) Epoch 7, batch 7500, loss[loss=0.1707, simple_loss=0.1808, pruned_loss=0.08035, over 6000.00 frames. ], tot_loss[loss=0.1536, simple_loss=0.1748, pruned_loss=0.06615, over 1950933.28 frames. ], batch size: 100, lr: 1.09e-02, grad_scale: 8.0 2022-12-07 17:31:59,266 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6959, 2.9977, 4.2042, 3.2946, 4.3141, 4.3178, 4.1736, 3.8728], device='cuda:0'), covar=tensor([0.0437, 0.2866, 0.1290, 0.1903, 0.0868, 0.0703, 0.1789, 0.1897], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0322, 0.0378, 0.0304, 0.0365, 0.0300, 0.0346, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:32:13,091 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:32:24,629 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:32:29,806 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8893, 1.5356, 1.8606, 2.0520, 1.4184, 1.7257, 1.7001, 1.8811], device='cuda:0'), covar=tensor([0.0053, 0.0128, 0.0055, 0.0038, 0.0108, 0.0131, 0.0098, 0.0061], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0221, 0.0332, 0.0274, 0.0221, 0.0267, 0.0237, 0.0255], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:32:36,263 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0779, 1.8884, 2.0433, 2.0813, 2.0229, 2.0017, 2.1530, 1.8293], device='cuda:0'), covar=tensor([0.0842, 0.1301, 0.0614, 0.0754, 0.0917, 0.0697, 0.0917, 0.0733], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0240, 0.0164, 0.0155, 0.0158, 0.0129, 0.0245, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:32:36,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.73 vs. limit=5.0 2022-12-07 17:32:39,008 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-7.pt 2022-12-07 17:33:18,930 INFO [train.py:873] (0/4) Epoch 8, batch 0, loss[loss=0.1929, simple_loss=0.2053, pruned_loss=0.09026, over 14364.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2053, pruned_loss=0.09026, over 14364.00 frames. ], batch size: 73, lr: 1.03e-02, grad_scale: 8.0 2022-12-07 17:33:18,931 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 17:33:26,264 INFO [train.py:905] (0/4) Epoch 8, validation: loss=0.1282, simple_loss=0.1716, pruned_loss=0.04242, over 857387.00 frames. 2022-12-07 17:33:26,265 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 17:33:27,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.973e+01 1.673e+02 2.697e+02 3.774e+02 7.847e+02, threshold=5.395e+02, percent-clipped=3.0 2022-12-07 17:33:38,430 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9214, 3.8024, 3.3360, 2.5762, 3.2498, 3.6353, 4.2030, 3.0805], device='cuda:0'), covar=tensor([0.0674, 0.1808, 0.1131, 0.2204, 0.0962, 0.0526, 0.0738, 0.1546], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0192, 0.0123, 0.0126, 0.0117, 0.0120, 0.0100, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 17:33:56,360 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:34:38,729 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:34:55,991 INFO [train.py:873] (0/4) Epoch 8, batch 100, loss[loss=0.1579, simple_loss=0.1471, pruned_loss=0.08436, over 2625.00 frames. ], tot_loss[loss=0.1522, simple_loss=0.175, pruned_loss=0.06464, over 886054.27 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:34:56,725 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.496e+02 3.096e+02 4.127e+02 8.461e+02, threshold=6.192e+02, percent-clipped=10.0 2022-12-07 17:35:13,520 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:35:26,068 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2678, 2.9676, 3.8502, 2.4774, 2.3835, 3.2496, 1.5250, 2.9347], device='cuda:0'), covar=tensor([0.1463, 0.1940, 0.0908, 0.3184, 0.3362, 0.1652, 0.7022, 0.1572], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0090, 0.0081, 0.0091, 0.0112, 0.0074, 0.0133, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:35:59,476 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:36:00,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.47 vs. limit=5.0 2022-12-07 17:36:07,748 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:36:23,089 INFO [train.py:873] (0/4) Epoch 8, batch 200, loss[loss=0.1994, simple_loss=0.2019, pruned_loss=0.09847, over 8626.00 frames. ], tot_loss[loss=0.1534, simple_loss=0.1745, pruned_loss=0.06613, over 1264850.53 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:36:23,899 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.183e+02 2.918e+02 3.622e+02 1.019e+03, threshold=5.836e+02, percent-clipped=2.0 2022-12-07 17:36:47,891 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2905, 1.2953, 1.4386, 1.1264, 1.2905, 0.9001, 0.9960, 0.6827], device='cuda:0'), covar=tensor([0.0400, 0.0737, 0.0420, 0.0330, 0.0449, 0.0426, 0.0382, 0.0794], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0014, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.1522e-05, 8.7044e-05, 7.8391e-05, 8.0861e-05, 8.3586e-05, 1.2069e-04, 1.0116e-04, 1.1568e-04], device='cuda:0') 2022-12-07 17:37:30,576 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 17:37:52,051 INFO [train.py:873] (0/4) Epoch 8, batch 300, loss[loss=0.1471, simple_loss=0.1743, pruned_loss=0.05993, over 13930.00 frames. ], tot_loss[loss=0.1542, simple_loss=0.1748, pruned_loss=0.06682, over 1530682.21 frames. ], batch size: 23, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:37:52,977 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.915e+01 2.253e+02 2.827e+02 3.791e+02 8.401e+02, threshold=5.654e+02, percent-clipped=6.0 2022-12-07 17:37:57,605 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:38:03,645 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5203, 2.2300, 3.1418, 1.9687, 2.0395, 2.6601, 1.2591, 2.7201], device='cuda:0'), covar=tensor([0.1024, 0.1950, 0.0576, 0.2618, 0.2820, 0.1007, 0.5769, 0.1054], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0089, 0.0080, 0.0089, 0.0112, 0.0073, 0.0131, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:38:12,809 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:38:46,613 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9686, 1.9945, 1.7244, 2.1017, 1.8842, 2.0049, 1.8845, 1.7500], device='cuda:0'), covar=tensor([0.0753, 0.0581, 0.1749, 0.0343, 0.0718, 0.0360, 0.1059, 0.0684], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0302, 0.0287, 0.0214, 0.0279, 0.0274, 0.0256, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 17:38:50,109 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:38:51,849 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:02,156 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:07,180 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8838, 2.6557, 2.7633, 2.8794, 2.8294, 2.7768, 2.9624, 2.4983], device='cuda:0'), covar=tensor([0.0521, 0.1022, 0.0438, 0.0553, 0.0658, 0.0470, 0.0627, 0.0619], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0241, 0.0165, 0.0156, 0.0161, 0.0128, 0.0244, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:39:18,813 INFO [train.py:873] (0/4) Epoch 8, batch 400, loss[loss=0.1973, simple_loss=0.2015, pruned_loss=0.09658, over 7792.00 frames. ], tot_loss[loss=0.1537, simple_loss=0.1745, pruned_loss=0.06644, over 1739843.66 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:39:19,679 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.324e+02 2.854e+02 3.563e+02 7.977e+02, threshold=5.709e+02, percent-clipped=8.0 2022-12-07 17:39:37,765 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0472, 1.9982, 2.0404, 2.0931, 1.9901, 1.6996, 1.3474, 1.8095], device='cuda:0'), covar=tensor([0.0359, 0.0332, 0.0464, 0.0266, 0.0333, 0.0988, 0.1718, 0.0359], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0150, 0.0129, 0.0125, 0.0181, 0.0123, 0.0152, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:39:41,670 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-07 17:39:43,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6258, 2.7478, 2.6769, 2.8553, 2.0191, 2.9590, 2.7778, 1.3130], device='cuda:0'), covar=tensor([0.2737, 0.0833, 0.1398, 0.0830, 0.1327, 0.0685, 0.1227, 0.3312], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0064, 0.0053, 0.0054, 0.0081, 0.0061, 0.0085, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 17:39:45,353 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:46,662 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-07 17:39:47,065 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:39:48,196 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2022-12-07 17:39:54,877 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:13,403 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7830, 2.6182, 2.6171, 2.8994, 2.4443, 2.4407, 2.7908, 2.8095], device='cuda:0'), covar=tensor([0.0773, 0.0900, 0.0923, 0.0595, 0.1157, 0.0835, 0.0759, 0.0732], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0110, 0.0121, 0.0125, 0.0125, 0.0096, 0.0136, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:40:23,159 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:24,146 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8373, 1.7837, 3.0525, 2.2472, 2.9583, 1.8285, 2.3766, 2.7482], device='cuda:0'), covar=tensor([0.0829, 0.4509, 0.0359, 0.5032, 0.0662, 0.3498, 0.1305, 0.0638], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0236, 0.0179, 0.0317, 0.0199, 0.0243, 0.0230, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:40:26,487 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:40:40,761 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:40:47,040 INFO [train.py:873] (0/4) Epoch 8, batch 500, loss[loss=0.1569, simple_loss=0.1573, pruned_loss=0.07825, over 2548.00 frames. ], tot_loss[loss=0.1547, simple_loss=0.1753, pruned_loss=0.06708, over 1792902.26 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:40:48,831 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.493e+02 3.208e+02 4.073e+02 9.438e+02, threshold=6.416e+02, percent-clipped=6.0 2022-12-07 17:41:05,151 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:41:38,998 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5393, 1.7223, 2.7822, 2.0652, 2.6502, 1.7068, 2.1031, 2.5371], device='cuda:0'), covar=tensor([0.1175, 0.4029, 0.0333, 0.4885, 0.0686, 0.3184, 0.1417, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0237, 0.0181, 0.0320, 0.0201, 0.0243, 0.0233, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:41:45,195 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8708, 1.6549, 1.9670, 1.7747, 2.0876, 1.7778, 1.6673, 1.8736], device='cuda:0'), covar=tensor([0.0325, 0.0940, 0.0149, 0.0262, 0.0195, 0.0426, 0.0150, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0315, 0.0324, 0.0380, 0.0307, 0.0366, 0.0299, 0.0349, 0.0323], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:42:13,824 INFO [train.py:873] (0/4) Epoch 8, batch 600, loss[loss=0.1701, simple_loss=0.1571, pruned_loss=0.09153, over 2653.00 frames. ], tot_loss[loss=0.1536, simple_loss=0.1751, pruned_loss=0.06605, over 1897394.43 frames. ], batch size: 100, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:42:15,675 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.499e+01 2.137e+02 2.538e+02 3.360e+02 7.193e+02, threshold=5.075e+02, percent-clipped=1.0 2022-12-07 17:43:08,699 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:43:09,224 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 17:43:42,308 INFO [train.py:873] (0/4) Epoch 8, batch 700, loss[loss=0.1662, simple_loss=0.1878, pruned_loss=0.07226, over 14299.00 frames. ], tot_loss[loss=0.1533, simple_loss=0.1745, pruned_loss=0.06608, over 1942266.01 frames. ], batch size: 37, lr: 1.02e-02, grad_scale: 4.0 2022-12-07 17:43:44,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.400e+02 3.018e+02 4.076e+02 1.039e+03, threshold=6.035e+02, percent-clipped=12.0 2022-12-07 17:43:57,213 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 17:44:04,150 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:10,932 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.16 vs. limit=2.0 2022-12-07 17:44:13,765 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:29,670 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8996, 3.6078, 3.4282, 3.4666, 3.7349, 3.7212, 3.8477, 3.8939], device='cuda:0'), covar=tensor([0.0974, 0.0685, 0.2178, 0.3230, 0.0876, 0.0955, 0.1136, 0.0796], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0238, 0.0404, 0.0503, 0.0297, 0.0368, 0.0364, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:44:49,975 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:44:59,965 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:45:03,460 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4535, 2.8802, 2.6731, 1.8475, 2.7379, 3.1322, 3.4679, 2.4331], device='cuda:0'), covar=tensor([0.0735, 0.2606, 0.1464, 0.2830, 0.1271, 0.1012, 0.1140, 0.2250], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0191, 0.0125, 0.0126, 0.0119, 0.0123, 0.0102, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 17:45:05,979 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5349, 4.9407, 4.9084, 5.4413, 5.1519, 4.5369, 5.3994, 4.3527], device='cuda:0'), covar=tensor([0.0298, 0.1040, 0.0280, 0.0383, 0.0604, 0.0398, 0.0468, 0.0545], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0239, 0.0166, 0.0155, 0.0161, 0.0129, 0.0244, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 17:45:10,197 INFO [train.py:873] (0/4) Epoch 8, batch 800, loss[loss=0.1726, simple_loss=0.1942, pruned_loss=0.07548, over 13988.00 frames. ], tot_loss[loss=0.1526, simple_loss=0.1742, pruned_loss=0.06555, over 2031980.91 frames. ], batch size: 22, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:45:11,875 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.469e+01 2.335e+02 2.764e+02 3.237e+02 6.593e+02, threshold=5.529e+02, percent-clipped=1.0 2022-12-07 17:45:21,762 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3334, 1.8068, 1.7799, 1.9122, 1.7580, 1.9903, 1.5410, 1.1275], device='cuda:0'), covar=tensor([0.1684, 0.0996, 0.0555, 0.0338, 0.1147, 0.0508, 0.1916, 0.2615], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0065, 0.0054, 0.0055, 0.0081, 0.0061, 0.0086, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 17:45:32,793 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:45:53,527 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0097, 2.0179, 2.0614, 2.0840, 2.0099, 1.7045, 1.2829, 1.8439], device='cuda:0'), covar=tensor([0.0399, 0.0340, 0.0384, 0.0312, 0.0333, 0.0951, 0.1830, 0.0372], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0149, 0.0129, 0.0125, 0.0180, 0.0121, 0.0151, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:45:58,431 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2022-12-07 17:46:28,195 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2841, 1.4304, 1.3418, 1.1921, 1.0205, 0.7321, 0.5996, 0.8309], device='cuda:0'), covar=tensor([0.0185, 0.0167, 0.0224, 0.0267, 0.0278, 0.0329, 0.0256, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0015, 0.0018], device='cuda:0'), out_proj_covar=tensor([8.3125e-05, 8.7753e-05, 7.8211e-05, 8.3278e-05, 8.2289e-05, 1.2000e-04, 1.0279e-04, 1.1381e-04], device='cuda:0') 2022-12-07 17:46:38,789 INFO [train.py:873] (0/4) Epoch 8, batch 900, loss[loss=0.161, simple_loss=0.1802, pruned_loss=0.07089, over 14182.00 frames. ], tot_loss[loss=0.1523, simple_loss=0.174, pruned_loss=0.06534, over 1982070.41 frames. ], batch size: 89, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:46:40,851 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.164e+02 2.734e+02 3.492e+02 7.190e+02, threshold=5.467e+02, percent-clipped=1.0 2022-12-07 17:46:46,820 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6247, 3.3190, 3.2407, 3.5655, 3.4047, 3.5568, 3.6723, 2.9835], device='cuda:0'), covar=tensor([0.0461, 0.1153, 0.0462, 0.0504, 0.0886, 0.0404, 0.0578, 0.0655], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0243, 0.0167, 0.0157, 0.0163, 0.0133, 0.0246, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 17:46:47,619 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6650, 4.3626, 4.0251, 4.2411, 4.4028, 4.4653, 4.6516, 4.6163], device='cuda:0'), covar=tensor([0.0712, 0.0539, 0.2273, 0.2706, 0.0765, 0.0787, 0.0912, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0329, 0.0236, 0.0406, 0.0504, 0.0293, 0.0367, 0.0367, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:46:48,647 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3599, 3.8324, 2.8856, 4.5253, 4.1362, 4.3391, 3.4765, 3.1671], device='cuda:0'), covar=tensor([0.0662, 0.1248, 0.4249, 0.0763, 0.0907, 0.1250, 0.1383, 0.3793], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0296, 0.0283, 0.0215, 0.0279, 0.0272, 0.0252, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:47:32,838 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:05,623 INFO [train.py:873] (0/4) Epoch 8, batch 1000, loss[loss=0.1275, simple_loss=0.159, pruned_loss=0.04797, over 14539.00 frames. ], tot_loss[loss=0.1528, simple_loss=0.1745, pruned_loss=0.06558, over 1986782.08 frames. ], batch size: 24, lr: 1.02e-02, grad_scale: 8.0 2022-12-07 17:48:07,331 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.511e+01 2.177e+02 2.724e+02 3.766e+02 8.147e+02, threshold=5.449e+02, percent-clipped=5.0 2022-12-07 17:48:14,321 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:27,852 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:48:33,142 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2563, 2.8893, 2.7427, 2.1004, 2.7993, 2.9487, 3.1433, 2.4605], device='cuda:0'), covar=tensor([0.0787, 0.1993, 0.1266, 0.2231, 0.1011, 0.0817, 0.0976, 0.1807], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0190, 0.0124, 0.0126, 0.0116, 0.0121, 0.0100, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 17:48:37,336 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:49:09,285 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:49:18,691 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:49:20,499 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8392, 1.5643, 2.0533, 1.6458, 1.9190, 1.4756, 1.6748, 1.7161], device='cuda:0'), covar=tensor([0.1405, 0.1859, 0.0222, 0.1431, 0.0688, 0.0925, 0.0894, 0.0370], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0232, 0.0180, 0.0319, 0.0202, 0.0241, 0.0228, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:49:22,047 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 17:49:22,935 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1111, 2.2547, 2.2647, 2.4464, 2.0063, 2.6125, 2.1036, 1.1938], device='cuda:0'), covar=tensor([0.2293, 0.1065, 0.1363, 0.0781, 0.1337, 0.0501, 0.1720, 0.3509], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0066, 0.0055, 0.0056, 0.0082, 0.0062, 0.0088, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 17:49:25,659 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5861, 1.4647, 2.8479, 1.5184, 2.8030, 2.7499, 2.1076, 2.8637], device='cuda:0'), covar=tensor([0.0257, 0.2210, 0.0255, 0.1744, 0.0278, 0.0386, 0.0791, 0.0185], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0160, 0.0150, 0.0167, 0.0163, 0.0163, 0.0131, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 17:49:32,902 INFO [train.py:873] (0/4) Epoch 8, batch 1100, loss[loss=0.151, simple_loss=0.1738, pruned_loss=0.06405, over 14171.00 frames. ], tot_loss[loss=0.1529, simple_loss=0.1741, pruned_loss=0.0659, over 1952612.66 frames. ], batch size: 99, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:49:34,931 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.125e+02 2.772e+02 3.608e+02 1.248e+03, threshold=5.543e+02, percent-clipped=5.0 2022-12-07 17:49:36,312 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.08 vs. limit=2.0 2022-12-07 17:50:04,032 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 17:50:21,940 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 17:50:59,597 INFO [train.py:873] (0/4) Epoch 8, batch 1200, loss[loss=0.1515, simple_loss=0.1589, pruned_loss=0.07201, over 5939.00 frames. ], tot_loss[loss=0.153, simple_loss=0.1743, pruned_loss=0.06587, over 1923380.84 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:51:00,665 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3401, 1.1771, 1.1451, 0.8729, 1.2211, 0.7195, 1.2556, 1.3038], device='cuda:0'), covar=tensor([0.0825, 0.1215, 0.0991, 0.2437, 0.1090, 0.0841, 0.0638, 0.0730], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0023, 0.0021, 0.0023, 0.0033, 0.0023, 0.0022], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 17:51:02,201 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.508e+02 3.088e+02 3.577e+02 7.086e+02, threshold=6.177e+02, percent-clipped=4.0 2022-12-07 17:52:27,388 INFO [train.py:873] (0/4) Epoch 8, batch 1300, loss[loss=0.142, simple_loss=0.1715, pruned_loss=0.05622, over 14336.00 frames. ], tot_loss[loss=0.1532, simple_loss=0.1744, pruned_loss=0.066, over 1932472.04 frames. ], batch size: 55, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:52:30,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.196e+02 2.858e+02 3.785e+02 8.723e+02, threshold=5.716e+02, percent-clipped=2.0 2022-12-07 17:53:45,500 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2189, 3.8620, 3.9336, 4.3030, 3.8167, 3.4893, 4.3118, 4.1617], device='cuda:0'), covar=tensor([0.0687, 0.0764, 0.0706, 0.0521, 0.0760, 0.0729, 0.0604, 0.0667], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0112, 0.0126, 0.0130, 0.0129, 0.0100, 0.0142, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 17:53:54,866 INFO [train.py:873] (0/4) Epoch 8, batch 1400, loss[loss=0.1207, simple_loss=0.1492, pruned_loss=0.04609, over 14469.00 frames. ], tot_loss[loss=0.1525, simple_loss=0.1737, pruned_loss=0.06563, over 1913199.25 frames. ], batch size: 18, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:53:57,393 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.292e+02 2.799e+02 3.377e+02 6.597e+02, threshold=5.599e+02, percent-clipped=2.0 2022-12-07 17:54:01,936 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6092, 2.0125, 3.6404, 2.5674, 3.5780, 1.8352, 2.8148, 3.5386], device='cuda:0'), covar=tensor([0.0692, 0.4709, 0.0374, 0.7204, 0.0579, 0.4141, 0.1425, 0.0376], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0235, 0.0179, 0.0313, 0.0202, 0.0239, 0.0227, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 17:54:20,064 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.36 vs. limit=5.0 2022-12-07 17:54:47,122 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8083, 1.7048, 1.9081, 2.0717, 1.5422, 1.7205, 1.9867, 1.8611], device='cuda:0'), covar=tensor([0.0071, 0.0099, 0.0052, 0.0039, 0.0109, 0.0175, 0.0068, 0.0059], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0226, 0.0342, 0.0279, 0.0223, 0.0273, 0.0244, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 17:55:22,065 INFO [train.py:873] (0/4) Epoch 8, batch 1500, loss[loss=0.1954, simple_loss=0.1708, pruned_loss=0.11, over 1208.00 frames. ], tot_loss[loss=0.151, simple_loss=0.1729, pruned_loss=0.06458, over 1947484.79 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:55:25,014 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.978e+01 2.195e+02 2.660e+02 3.596e+02 7.442e+02, threshold=5.321e+02, percent-clipped=3.0 2022-12-07 17:56:49,480 INFO [train.py:873] (0/4) Epoch 8, batch 1600, loss[loss=0.1932, simple_loss=0.1884, pruned_loss=0.09905, over 7804.00 frames. ], tot_loss[loss=0.1517, simple_loss=0.173, pruned_loss=0.06516, over 1930284.07 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:56:51,887 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.201e+02 2.581e+02 3.079e+02 3.798e+02 6.375e+02, threshold=6.158e+02, percent-clipped=8.0 2022-12-07 17:58:17,001 INFO [train.py:873] (0/4) Epoch 8, batch 1700, loss[loss=0.1317, simple_loss=0.1544, pruned_loss=0.05448, over 13664.00 frames. ], tot_loss[loss=0.1526, simple_loss=0.1738, pruned_loss=0.06574, over 1942403.33 frames. ], batch size: 17, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:58:19,792 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.375e+02 2.929e+02 3.612e+02 8.453e+02, threshold=5.858e+02, percent-clipped=6.0 2022-12-07 17:58:28,730 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2022-12-07 17:58:31,864 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9807, 0.9608, 0.9049, 0.9397, 1.1791, 0.5634, 0.9443, 1.0963], device='cuda:0'), covar=tensor([0.0594, 0.1056, 0.0630, 0.0573, 0.0651, 0.0930, 0.0650, 0.0467], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0022, 0.0020, 0.0021, 0.0031, 0.0021, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.0264e-04, 1.0278e-04, 1.0370e-04, 9.9657e-05, 1.0354e-04, 1.3549e-04, 1.0585e-04, 1.0649e-04], device='cuda:0') 2022-12-07 17:58:42,694 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:59:36,602 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 17:59:44,435 INFO [train.py:873] (0/4) Epoch 8, batch 1800, loss[loss=0.1475, simple_loss=0.1653, pruned_loss=0.06483, over 12784.00 frames. ], tot_loss[loss=0.1528, simple_loss=0.1742, pruned_loss=0.06571, over 1986870.62 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 17:59:46,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.304e+02 2.792e+02 3.452e+02 5.919e+02, threshold=5.584e+02, percent-clipped=1.0 2022-12-07 18:00:50,615 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 18:01:11,086 INFO [train.py:873] (0/4) Epoch 8, batch 1900, loss[loss=0.1413, simple_loss=0.1685, pruned_loss=0.05705, over 14259.00 frames. ], tot_loss[loss=0.152, simple_loss=0.1733, pruned_loss=0.06534, over 1950425.34 frames. ], batch size: 69, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:01:13,579 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.876e+01 2.431e+02 3.063e+02 3.945e+02 8.424e+02, threshold=6.126e+02, percent-clipped=6.0 2022-12-07 18:01:28,906 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9095, 5.3544, 5.3547, 5.8539, 5.4020, 4.8024, 5.8277, 4.7112], device='cuda:0'), covar=tensor([0.0276, 0.0769, 0.0201, 0.0339, 0.0715, 0.0299, 0.0427, 0.0498], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0243, 0.0167, 0.0159, 0.0167, 0.0133, 0.0249, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:01:33,225 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0251, 1.9771, 1.7050, 1.6529, 1.9980, 1.9723, 1.9777, 1.9879], device='cuda:0'), covar=tensor([0.1237, 0.1272, 0.3137, 0.4130, 0.1313, 0.1350, 0.2134, 0.1415], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0243, 0.0403, 0.0510, 0.0298, 0.0375, 0.0365, 0.0315], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:01:35,132 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:01:35,915 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5490, 1.5968, 4.3077, 1.9915, 4.1533, 4.4786, 4.0911, 4.9144], device='cuda:0'), covar=tensor([0.0189, 0.3110, 0.0395, 0.2417, 0.0339, 0.0353, 0.0303, 0.0152], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0157, 0.0149, 0.0167, 0.0161, 0.0163, 0.0131, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:02:29,215 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 18:02:39,003 INFO [train.py:873] (0/4) Epoch 8, batch 2000, loss[loss=0.1558, simple_loss=0.1759, pruned_loss=0.06787, over 10318.00 frames. ], tot_loss[loss=0.1519, simple_loss=0.1735, pruned_loss=0.06513, over 1978530.63 frames. ], batch size: 100, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:02:39,122 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0709, 3.7799, 3.5167, 3.6876, 3.8841, 3.9355, 4.0748, 4.0416], device='cuda:0'), covar=tensor([0.0765, 0.0667, 0.1896, 0.2791, 0.0663, 0.0742, 0.0856, 0.0815], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0243, 0.0401, 0.0510, 0.0296, 0.0374, 0.0363, 0.0315], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:02:41,554 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.941e+01 2.269e+02 2.935e+02 3.791e+02 7.498e+02, threshold=5.870e+02, percent-clipped=3.0 2022-12-07 18:02:44,232 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5341, 5.0910, 5.0197, 5.4525, 5.1525, 4.5974, 5.5239, 4.4184], device='cuda:0'), covar=tensor([0.0325, 0.0806, 0.0307, 0.0429, 0.0658, 0.0390, 0.0398, 0.0484], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0242, 0.0166, 0.0158, 0.0164, 0.0132, 0.0247, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:03:18,501 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8868, 4.4499, 4.4345, 4.7659, 4.5898, 4.2399, 4.8314, 4.0796], device='cuda:0'), covar=tensor([0.0323, 0.1055, 0.0319, 0.0530, 0.0712, 0.0606, 0.0512, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0244, 0.0167, 0.0159, 0.0166, 0.0133, 0.0249, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:03:36,508 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-55000.pt 2022-12-07 18:03:57,253 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 18:03:58,405 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-07 18:04:02,928 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2759, 2.9935, 2.7652, 2.9341, 3.2073, 3.1561, 3.2377, 3.2183], device='cuda:0'), covar=tensor([0.0924, 0.0871, 0.2558, 0.3163, 0.0984, 0.0927, 0.1282, 0.1029], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0243, 0.0401, 0.0507, 0.0294, 0.0373, 0.0363, 0.0313], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:04:09,836 INFO [train.py:873] (0/4) Epoch 8, batch 2100, loss[loss=0.1759, simple_loss=0.1968, pruned_loss=0.07744, over 14265.00 frames. ], tot_loss[loss=0.1517, simple_loss=0.1735, pruned_loss=0.06494, over 1993737.08 frames. ], batch size: 69, lr: 1.01e-02, grad_scale: 8.0 2022-12-07 18:04:12,640 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.381e+02 2.883e+02 3.565e+02 6.243e+02, threshold=5.765e+02, percent-clipped=1.0 2022-12-07 18:05:01,746 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8479, 4.5111, 4.2832, 4.3894, 4.5778, 4.7074, 4.8829, 4.8518], device='cuda:0'), covar=tensor([0.0727, 0.0536, 0.1951, 0.2906, 0.0617, 0.0683, 0.0762, 0.0689], device='cuda:0'), in_proj_covar=tensor([0.0328, 0.0239, 0.0395, 0.0502, 0.0290, 0.0366, 0.0355, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:05:37,625 INFO [train.py:873] (0/4) Epoch 8, batch 2200, loss[loss=0.1471, simple_loss=0.1724, pruned_loss=0.06085, over 14265.00 frames. ], tot_loss[loss=0.1518, simple_loss=0.1732, pruned_loss=0.06518, over 1996447.27 frames. ], batch size: 76, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:05:39,870 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.363e+02 3.011e+02 4.055e+02 7.336e+02, threshold=6.021e+02, percent-clipped=10.0 2022-12-07 18:05:48,530 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:06:40,787 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9870, 2.0609, 1.8611, 2.1264, 1.6719, 1.9434, 2.0624, 2.0796], device='cuda:0'), covar=tensor([0.1001, 0.0934, 0.1091, 0.0784, 0.1217, 0.0835, 0.1017, 0.0880], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0112, 0.0125, 0.0127, 0.0129, 0.0100, 0.0142, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 18:06:43,415 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:06:44,915 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:06:51,027 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 18:07:05,469 INFO [train.py:873] (0/4) Epoch 8, batch 2300, loss[loss=0.1951, simple_loss=0.1816, pruned_loss=0.1042, over 1313.00 frames. ], tot_loss[loss=0.1513, simple_loss=0.1729, pruned_loss=0.06484, over 2047866.73 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:07:08,045 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.336e+02 2.949e+02 3.747e+02 7.724e+02, threshold=5.898e+02, percent-clipped=2.0 2022-12-07 18:07:31,119 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2856, 1.8171, 2.2364, 1.9406, 2.3719, 2.0724, 1.9920, 2.0864], device='cuda:0'), covar=tensor([0.0348, 0.1673, 0.0297, 0.0652, 0.0278, 0.0566, 0.0303, 0.0536], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0325, 0.0385, 0.0305, 0.0371, 0.0308, 0.0360, 0.0330], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:07:39,106 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:08:21,831 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:08:34,795 INFO [train.py:873] (0/4) Epoch 8, batch 2400, loss[loss=0.1416, simple_loss=0.1575, pruned_loss=0.06287, over 5960.00 frames. ], tot_loss[loss=0.152, simple_loss=0.1733, pruned_loss=0.06536, over 2025443.30 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:08:37,246 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.285e+02 2.947e+02 4.114e+02 7.749e+02, threshold=5.894e+02, percent-clipped=7.0 2022-12-07 18:09:04,224 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:09:30,457 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4629, 4.4165, 4.7126, 4.1839, 4.4706, 4.8467, 1.7486, 4.1911], device='cuda:0'), covar=tensor([0.0203, 0.0237, 0.0409, 0.0348, 0.0353, 0.0121, 0.3158, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0150, 0.0131, 0.0125, 0.0181, 0.0122, 0.0152, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:10:03,414 INFO [train.py:873] (0/4) Epoch 8, batch 2500, loss[loss=0.1271, simple_loss=0.147, pruned_loss=0.05358, over 6937.00 frames. ], tot_loss[loss=0.1508, simple_loss=0.1726, pruned_loss=0.06448, over 1938578.78 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:10:05,936 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 2.135e+02 2.789e+02 3.558e+02 8.555e+02, threshold=5.578e+02, percent-clipped=3.0 2022-12-07 18:10:53,008 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 18:11:03,333 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:11:06,239 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-07 18:11:15,513 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:11:29,790 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0478, 2.1248, 3.1281, 3.1434, 3.1174, 2.1230, 3.1309, 2.3664], device='cuda:0'), covar=tensor([0.0239, 0.0514, 0.0369, 0.0222, 0.0222, 0.0781, 0.0211, 0.0576], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0227, 0.0345, 0.0284, 0.0228, 0.0277, 0.0246, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:11:30,375 INFO [train.py:873] (0/4) Epoch 8, batch 2600, loss[loss=0.1581, simple_loss=0.1817, pruned_loss=0.06725, over 13549.00 frames. ], tot_loss[loss=0.1516, simple_loss=0.1731, pruned_loss=0.06506, over 1896096.45 frames. ], batch size: 100, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:11:32,918 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.392e+02 2.780e+02 3.615e+02 5.645e+02, threshold=5.559e+02, percent-clipped=1.0 2022-12-07 18:11:57,670 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:11:58,688 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:12:57,537 INFO [train.py:873] (0/4) Epoch 8, batch 2700, loss[loss=0.1127, simple_loss=0.1548, pruned_loss=0.0353, over 14561.00 frames. ], tot_loss[loss=0.1511, simple_loss=0.1728, pruned_loss=0.06466, over 1932334.86 frames. ], batch size: 43, lr: 1.00e-02, grad_scale: 8.0 2022-12-07 18:13:00,127 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.995e+01 2.381e+02 3.010e+02 3.860e+02 1.123e+03, threshold=6.020e+02, percent-clipped=6.0 2022-12-07 18:13:21,705 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 18:14:27,408 INFO [train.py:873] (0/4) Epoch 8, batch 2800, loss[loss=0.1809, simple_loss=0.1694, pruned_loss=0.09617, over 1168.00 frames. ], tot_loss[loss=0.1516, simple_loss=0.1732, pruned_loss=0.06502, over 1933664.18 frames. ], batch size: 100, lr: 9.99e-03, grad_scale: 8.0 2022-12-07 18:14:30,868 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.464e+02 2.891e+02 3.921e+02 7.148e+02, threshold=5.781e+02, percent-clipped=1.0 2022-12-07 18:14:38,321 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:28,254 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:29,038 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9223, 4.6226, 4.2507, 4.4382, 4.5189, 4.7193, 4.8495, 4.8281], device='cuda:0'), covar=tensor([0.0587, 0.0405, 0.1904, 0.2602, 0.0619, 0.0621, 0.0810, 0.0724], device='cuda:0'), in_proj_covar=tensor([0.0334, 0.0242, 0.0395, 0.0503, 0.0289, 0.0369, 0.0364, 0.0320], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:15:32,974 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:15:46,181 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7874, 1.3308, 3.5820, 1.6778, 3.7389, 3.8898, 2.7152, 4.1169], device='cuda:0'), covar=tensor([0.0219, 0.3134, 0.0455, 0.2229, 0.0397, 0.0329, 0.0815, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0159, 0.0152, 0.0166, 0.0165, 0.0164, 0.0131, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:15:54,463 INFO [train.py:873] (0/4) Epoch 8, batch 2900, loss[loss=0.1553, simple_loss=0.1673, pruned_loss=0.07169, over 4960.00 frames. ], tot_loss[loss=0.1524, simple_loss=0.1735, pruned_loss=0.06563, over 1969494.14 frames. ], batch size: 100, lr: 9.99e-03, grad_scale: 8.0 2022-12-07 18:15:57,949 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.733e+01 2.419e+02 2.995e+02 3.748e+02 5.974e+02, threshold=5.989e+02, percent-clipped=2.0 2022-12-07 18:15:59,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7325, 0.7682, 0.6712, 0.6760, 0.6429, 0.4127, 0.6480, 0.5579], device='cuda:0'), covar=tensor([0.0156, 0.0126, 0.0106, 0.0109, 0.0169, 0.0356, 0.0204, 0.0382], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0011, 0.0012, 0.0018, 0.0015, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.3144e-05, 9.1035e-05, 8.1118e-05, 8.4368e-05, 8.4648e-05, 1.2238e-04, 1.0309e-04, 1.1847e-04], device='cuda:0') 2022-12-07 18:16:06,772 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:09,825 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:15,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-07 18:16:22,889 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:16:55,047 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1743, 4.2591, 4.7047, 3.8530, 4.3977, 4.6445, 1.7557, 4.1850], device='cuda:0'), covar=tensor([0.0255, 0.0330, 0.0302, 0.0487, 0.0352, 0.0187, 0.3136, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0151, 0.0133, 0.0127, 0.0183, 0.0122, 0.0152, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:16:59,477 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:17:04,318 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:17:22,325 INFO [train.py:873] (0/4) Epoch 8, batch 3000, loss[loss=0.1439, simple_loss=0.1557, pruned_loss=0.06607, over 3869.00 frames. ], tot_loss[loss=0.1513, simple_loss=0.1727, pruned_loss=0.06496, over 1967689.02 frames. ], batch size: 100, lr: 9.98e-03, grad_scale: 8.0 2022-12-07 18:17:22,326 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 18:17:42,269 INFO [train.py:905] (0/4) Epoch 8, validation: loss=0.1226, simple_loss=0.1659, pruned_loss=0.03968, over 857387.00 frames. 2022-12-07 18:17:42,270 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 18:17:45,688 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 2.436e+02 3.032e+02 3.912e+02 1.128e+03, threshold=6.064e+02, percent-clipped=7.0 2022-12-07 18:18:10,528 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-07 18:18:15,649 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-07 18:18:51,546 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 18:18:55,473 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7411, 3.3302, 2.5831, 3.8725, 3.7232, 3.6687, 3.1479, 2.6001], device='cuda:0'), covar=tensor([0.0652, 0.1475, 0.3663, 0.0394, 0.0784, 0.1182, 0.1284, 0.3947], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0294, 0.0274, 0.0217, 0.0278, 0.0273, 0.0255, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:19:11,702 INFO [train.py:873] (0/4) Epoch 8, batch 3100, loss[loss=0.1357, simple_loss=0.1569, pruned_loss=0.05726, over 6896.00 frames. ], tot_loss[loss=0.1493, simple_loss=0.1716, pruned_loss=0.06355, over 1926577.23 frames. ], batch size: 100, lr: 9.97e-03, grad_scale: 8.0 2022-12-07 18:19:15,078 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.402e+01 2.241e+02 2.784e+02 3.482e+02 7.776e+02, threshold=5.568e+02, percent-clipped=3.0 2022-12-07 18:20:11,548 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:20:33,703 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4474, 1.6812, 1.3442, 1.0462, 1.7929, 0.6100, 1.4933, 1.6595], device='cuda:0'), covar=tensor([0.1587, 0.0985, 0.0945, 0.2159, 0.1214, 0.0908, 0.1607, 0.0785], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0022, 0.0020, 0.0022, 0.0031, 0.0022, 0.0022], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:20:38,655 INFO [train.py:873] (0/4) Epoch 8, batch 3200, loss[loss=0.1575, simple_loss=0.171, pruned_loss=0.07201, over 6969.00 frames. ], tot_loss[loss=0.1506, simple_loss=0.1724, pruned_loss=0.06439, over 1957883.72 frames. ], batch size: 100, lr: 9.96e-03, grad_scale: 8.0 2022-12-07 18:20:42,716 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.292e+02 3.048e+02 3.641e+02 9.234e+02, threshold=6.095e+02, percent-clipped=4.0 2022-12-07 18:20:47,467 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:20:50,278 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2022-12-07 18:21:15,994 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2445, 2.8866, 2.1825, 3.2867, 2.9852, 3.2048, 2.7600, 2.2209], device='cuda:0'), covar=tensor([0.0712, 0.1506, 0.3648, 0.0422, 0.1023, 0.0810, 0.1443, 0.4009], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0298, 0.0276, 0.0219, 0.0280, 0.0275, 0.0257, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:21:22,780 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9251, 1.0473, 0.8645, 0.9313, 1.0225, 0.4597, 0.7904, 0.9594], device='cuda:0'), covar=tensor([0.0510, 0.0354, 0.0428, 0.0279, 0.0200, 0.0422, 0.0842, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0021, 0.0022, 0.0021, 0.0022, 0.0032, 0.0022, 0.0022], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:21:40,224 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:21:42,071 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:22:05,431 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0407, 1.6128, 4.2599, 3.9449, 4.0486, 4.4280, 3.6942, 4.3991], device='cuda:0'), covar=tensor([0.1260, 0.1444, 0.0081, 0.0165, 0.0152, 0.0081, 0.0218, 0.0091], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0152, 0.0114, 0.0158, 0.0133, 0.0128, 0.0109, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:22:08,270 INFO [train.py:873] (0/4) Epoch 8, batch 3300, loss[loss=0.146, simple_loss=0.1724, pruned_loss=0.05982, over 14173.00 frames. ], tot_loss[loss=0.1508, simple_loss=0.1727, pruned_loss=0.06448, over 1967419.80 frames. ], batch size: 89, lr: 9.95e-03, grad_scale: 8.0 2022-12-07 18:22:12,320 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.398e+02 3.040e+02 3.643e+02 7.318e+02, threshold=6.080e+02, percent-clipped=1.0 2022-12-07 18:22:55,045 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1154, 2.9897, 2.9057, 2.8811, 2.5281, 3.3386, 2.7280, 1.4811], device='cuda:0'), covar=tensor([0.2188, 0.1031, 0.1270, 0.1988, 0.0983, 0.0418, 0.1486, 0.3212], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0067, 0.0055, 0.0058, 0.0084, 0.0064, 0.0089, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 18:23:12,326 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1892, 3.0041, 2.6463, 2.8452, 3.0609, 3.0905, 3.1545, 3.1270], device='cuda:0'), covar=tensor([0.0863, 0.0910, 0.2952, 0.3067, 0.0949, 0.1054, 0.1286, 0.1052], device='cuda:0'), in_proj_covar=tensor([0.0334, 0.0243, 0.0405, 0.0510, 0.0289, 0.0372, 0.0369, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:23:23,549 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5817, 2.9551, 4.2598, 3.3652, 4.3241, 4.1311, 4.0272, 3.7097], device='cuda:0'), covar=tensor([0.0602, 0.3257, 0.0818, 0.1831, 0.0842, 0.0817, 0.2167, 0.2025], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0322, 0.0387, 0.0300, 0.0367, 0.0302, 0.0358, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:23:35,253 INFO [train.py:873] (0/4) Epoch 8, batch 3400, loss[loss=0.1703, simple_loss=0.1787, pruned_loss=0.08099, over 5994.00 frames. ], tot_loss[loss=0.1492, simple_loss=0.1715, pruned_loss=0.06344, over 1958482.70 frames. ], batch size: 100, lr: 9.94e-03, grad_scale: 8.0 2022-12-07 18:23:35,481 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6625, 1.8034, 1.7526, 1.1078, 1.8119, 0.7540, 1.7030, 1.7199], device='cuda:0'), covar=tensor([0.0910, 0.1502, 0.1209, 0.3096, 0.1231, 0.0800, 0.1655, 0.1074], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0023, 0.0021, 0.0022, 0.0031, 0.0023, 0.0022], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:23:39,443 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.408e+02 2.894e+02 3.622e+02 6.016e+02, threshold=5.789e+02, percent-clipped=0.0 2022-12-07 18:23:50,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9436, 3.7299, 3.6616, 3.9939, 3.6181, 3.2804, 4.0063, 3.8907], device='cuda:0'), covar=tensor([0.0720, 0.0703, 0.0719, 0.0655, 0.0817, 0.0746, 0.0607, 0.0812], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0113, 0.0124, 0.0130, 0.0128, 0.0101, 0.0142, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 18:24:20,011 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 18:24:36,645 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:25:01,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5262, 1.5424, 2.7723, 1.5683, 2.8167, 2.7749, 2.2198, 2.8965], device='cuda:0'), covar=tensor([0.0257, 0.1950, 0.0251, 0.1530, 0.0257, 0.0372, 0.0767, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0159, 0.0150, 0.0168, 0.0162, 0.0165, 0.0130, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:25:03,323 INFO [train.py:873] (0/4) Epoch 8, batch 3500, loss[loss=0.1357, simple_loss=0.1658, pruned_loss=0.05277, over 14221.00 frames. ], tot_loss[loss=0.1499, simple_loss=0.1722, pruned_loss=0.06377, over 2016203.36 frames. ], batch size: 60, lr: 9.93e-03, grad_scale: 8.0 2022-12-07 18:25:06,214 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3833, 1.6428, 2.5501, 1.9003, 2.4513, 1.6030, 2.0771, 2.3453], device='cuda:0'), covar=tensor([0.1504, 0.4747, 0.0521, 0.4930, 0.0961, 0.3656, 0.1418, 0.0706], device='cuda:0'), in_proj_covar=tensor([0.0233, 0.0235, 0.0186, 0.0316, 0.0206, 0.0239, 0.0222, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:25:07,749 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.219e+02 2.782e+02 3.549e+02 6.206e+02, threshold=5.564e+02, percent-clipped=1.0 2022-12-07 18:25:18,106 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:01,185 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:03,943 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:20,865 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 18:26:30,711 INFO [train.py:873] (0/4) Epoch 8, batch 3600, loss[loss=0.1585, simple_loss=0.1698, pruned_loss=0.07357, over 5988.00 frames. ], tot_loss[loss=0.1504, simple_loss=0.173, pruned_loss=0.06392, over 2091917.94 frames. ], batch size: 100, lr: 9.92e-03, grad_scale: 8.0 2022-12-07 18:26:35,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.906e+01 2.280e+02 2.872e+02 3.796e+02 9.491e+02, threshold=5.743e+02, percent-clipped=4.0 2022-12-07 18:26:36,114 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8479, 1.4924, 1.8820, 2.0939, 1.4475, 1.7320, 1.8241, 1.9071], device='cuda:0'), covar=tensor([0.0066, 0.0097, 0.0067, 0.0041, 0.0127, 0.0163, 0.0065, 0.0057], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0226, 0.0340, 0.0282, 0.0227, 0.0274, 0.0250, 0.0261], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:26:45,862 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-07 18:26:46,143 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:26:51,448 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 18:27:14,253 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:27:50,290 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2022-12-07 18:27:59,551 INFO [train.py:873] (0/4) Epoch 8, batch 3700, loss[loss=0.1467, simple_loss=0.1404, pruned_loss=0.07647, over 2614.00 frames. ], tot_loss[loss=0.1513, simple_loss=0.1733, pruned_loss=0.06467, over 2030876.41 frames. ], batch size: 100, lr: 9.92e-03, grad_scale: 8.0 2022-12-07 18:28:03,912 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.170e+02 2.783e+02 3.492e+02 8.002e+02, threshold=5.566e+02, percent-clipped=4.0 2022-12-07 18:28:13,404 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5665, 1.4412, 1.5161, 1.3228, 1.7480, 0.8255, 1.4436, 1.7162], device='cuda:0'), covar=tensor([0.2118, 0.1199, 0.1039, 0.3613, 0.2158, 0.0772, 0.1570, 0.1071], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0023, 0.0021, 0.0022, 0.0031, 0.0023, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:29:00,366 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-07 18:29:00,374 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.75 vs. limit=5.0 2022-12-07 18:29:25,509 INFO [train.py:873] (0/4) Epoch 8, batch 3800, loss[loss=0.1543, simple_loss=0.1687, pruned_loss=0.06998, over 5990.00 frames. ], tot_loss[loss=0.1511, simple_loss=0.1729, pruned_loss=0.06464, over 1981540.13 frames. ], batch size: 100, lr: 9.91e-03, grad_scale: 8.0 2022-12-07 18:29:29,583 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.607e+02 3.160e+02 3.908e+02 8.359e+02, threshold=6.320e+02, percent-clipped=5.0 2022-12-07 18:30:24,867 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:30:26,512 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:30:36,824 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9279, 1.8119, 3.1903, 2.2422, 2.9430, 1.8399, 2.4465, 2.9069], device='cuda:0'), covar=tensor([0.0876, 0.4328, 0.0416, 0.5016, 0.0683, 0.3417, 0.1284, 0.0695], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0230, 0.0184, 0.0312, 0.0203, 0.0233, 0.0223, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:30:54,315 INFO [train.py:873] (0/4) Epoch 8, batch 3900, loss[loss=0.2079, simple_loss=0.1754, pruned_loss=0.1202, over 1205.00 frames. ], tot_loss[loss=0.1493, simple_loss=0.1715, pruned_loss=0.06354, over 2007495.69 frames. ], batch size: 100, lr: 9.90e-03, grad_scale: 8.0 2022-12-07 18:30:58,940 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.187e+02 2.677e+02 3.401e+02 6.261e+02, threshold=5.355e+02, percent-clipped=0.0 2022-12-07 18:31:02,297 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0273, 1.0746, 0.8266, 0.9737, 1.2371, 0.6804, 1.0108, 0.8892], device='cuda:0'), covar=tensor([0.0683, 0.0966, 0.0781, 0.0594, 0.0640, 0.0715, 0.0732, 0.0782], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0023, 0.0021, 0.0022, 0.0031, 0.0023, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:31:06,459 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:31:12,402 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0891, 2.0168, 1.7619, 1.8216, 2.0314, 2.0448, 2.0687, 2.0277], device='cuda:0'), covar=tensor([0.1252, 0.1088, 0.2828, 0.3409, 0.1110, 0.1115, 0.1746, 0.1347], device='cuda:0'), in_proj_covar=tensor([0.0339, 0.0245, 0.0406, 0.0514, 0.0294, 0.0376, 0.0370, 0.0326], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:31:19,806 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:31:33,512 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:31:58,958 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:32:22,732 INFO [train.py:873] (0/4) Epoch 8, batch 4000, loss[loss=0.156, simple_loss=0.143, pruned_loss=0.08446, over 1193.00 frames. ], tot_loss[loss=0.1495, simple_loss=0.1715, pruned_loss=0.06371, over 1984642.36 frames. ], batch size: 100, lr: 9.89e-03, grad_scale: 8.0 2022-12-07 18:32:23,756 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9017, 5.6452, 5.3802, 6.0572, 5.5205, 5.1894, 5.9922, 5.8311], device='cuda:0'), covar=tensor([0.0573, 0.0403, 0.0524, 0.0283, 0.0492, 0.0349, 0.0421, 0.0446], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0109, 0.0119, 0.0124, 0.0122, 0.0096, 0.0136, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 18:32:27,090 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 2.242e+02 2.985e+02 3.676e+02 7.027e+02, threshold=5.970e+02, percent-clipped=5.0 2022-12-07 18:32:44,626 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2022-12-07 18:32:53,001 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:14,706 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:16,336 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:33:19,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5076, 4.2375, 3.9468, 4.1125, 4.2450, 4.3548, 4.4574, 4.4199], device='cuda:0'), covar=tensor([0.0674, 0.0535, 0.2009, 0.2791, 0.0670, 0.0668, 0.0895, 0.0839], device='cuda:0'), in_proj_covar=tensor([0.0335, 0.0242, 0.0400, 0.0511, 0.0291, 0.0370, 0.0368, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:33:38,963 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7517, 0.9339, 0.7578, 0.7600, 0.6872, 0.5552, 0.7197, 0.7503], device='cuda:0'), covar=tensor([0.0333, 0.0391, 0.0342, 0.0310, 0.0443, 0.0690, 0.0440, 0.0738], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0012, 0.0012, 0.0019, 0.0015, 0.0020], device='cuda:0'), out_proj_covar=tensor([8.8927e-05, 9.6540e-05, 8.5066e-05, 9.0780e-05, 8.7031e-05, 1.3127e-04, 1.1003e-04, 1.2535e-04], device='cuda:0') 2022-12-07 18:33:51,681 INFO [train.py:873] (0/4) Epoch 8, batch 4100, loss[loss=0.1412, simple_loss=0.1681, pruned_loss=0.05708, over 14313.00 frames. ], tot_loss[loss=0.1502, simple_loss=0.1719, pruned_loss=0.06424, over 1969907.37 frames. ], batch size: 44, lr: 9.88e-03, grad_scale: 8.0 2022-12-07 18:33:56,137 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 2.377e+02 3.103e+02 3.855e+02 1.239e+03, threshold=6.205e+02, percent-clipped=7.0 2022-12-07 18:34:08,895 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:34:10,602 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:35:04,459 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0859, 1.7924, 4.5510, 4.2178, 4.1574, 4.6801, 4.1237, 4.6416], device='cuda:0'), covar=tensor([0.1286, 0.1391, 0.0082, 0.0151, 0.0161, 0.0089, 0.0180, 0.0096], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0157, 0.0117, 0.0161, 0.0137, 0.0130, 0.0110, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:35:20,884 INFO [train.py:873] (0/4) Epoch 8, batch 4200, loss[loss=0.1476, simple_loss=0.1603, pruned_loss=0.06741, over 3891.00 frames. ], tot_loss[loss=0.1501, simple_loss=0.1728, pruned_loss=0.06374, over 2060085.56 frames. ], batch size: 100, lr: 9.87e-03, grad_scale: 8.0 2022-12-07 18:35:21,854 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9451, 4.1038, 4.3229, 3.7199, 4.0900, 4.3695, 1.5956, 3.8426], device='cuda:0'), covar=tensor([0.0238, 0.0269, 0.0308, 0.0466, 0.0296, 0.0197, 0.3095, 0.0260], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0150, 0.0130, 0.0127, 0.0180, 0.0122, 0.0150, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:35:25,040 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.260e+02 2.916e+02 3.546e+02 6.128e+02, threshold=5.833e+02, percent-clipped=0.0 2022-12-07 18:35:42,116 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:35:59,644 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:36:06,025 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:15,622 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:31,658 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8929, 3.1096, 2.8853, 2.9903, 2.4808, 3.0985, 2.8623, 1.3481], device='cuda:0'), covar=tensor([0.2921, 0.0691, 0.1072, 0.0783, 0.1026, 0.0621, 0.1474, 0.3555], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0066, 0.0055, 0.0057, 0.0084, 0.0063, 0.0086, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 18:36:41,988 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 18:36:48,820 INFO [train.py:873] (0/4) Epoch 8, batch 4300, loss[loss=0.1292, simple_loss=0.163, pruned_loss=0.04772, over 14081.00 frames. ], tot_loss[loss=0.15, simple_loss=0.1727, pruned_loss=0.06368, over 2037379.51 frames. ], batch size: 29, lr: 9.86e-03, grad_scale: 8.0 2022-12-07 18:36:53,098 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 2.274e+02 2.646e+02 3.498e+02 8.636e+02, threshold=5.293e+02, percent-clipped=3.0 2022-12-07 18:36:57,018 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:36:59,616 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:37:08,883 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:37:14,087 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:37:47,785 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9970, 2.1812, 2.9571, 3.0813, 2.9683, 2.1518, 2.9909, 2.3920], device='cuda:0'), covar=tensor([0.0207, 0.0492, 0.0359, 0.0229, 0.0204, 0.0719, 0.0180, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0231, 0.0347, 0.0290, 0.0231, 0.0277, 0.0252, 0.0266], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:37:49,486 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:10,617 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6120, 3.4388, 3.2074, 3.2869, 3.5633, 3.5181, 3.6140, 3.5828], device='cuda:0'), covar=tensor([0.1065, 0.0657, 0.1911, 0.2868, 0.0695, 0.0847, 0.1165, 0.0904], device='cuda:0'), in_proj_covar=tensor([0.0344, 0.0244, 0.0401, 0.0515, 0.0294, 0.0376, 0.0373, 0.0323], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:38:15,734 INFO [train.py:873] (0/4) Epoch 8, batch 4400, loss[loss=0.1473, simple_loss=0.1792, pruned_loss=0.05764, over 14281.00 frames. ], tot_loss[loss=0.1505, simple_loss=0.1729, pruned_loss=0.06404, over 2035174.93 frames. ], batch size: 44, lr: 9.86e-03, grad_scale: 8.0 2022-12-07 18:38:19,656 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.383e+01 2.226e+02 2.824e+02 3.594e+02 6.936e+02, threshold=5.649e+02, percent-clipped=3.0 2022-12-07 18:38:27,823 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:29,520 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:38:51,584 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2022-12-07 18:39:10,898 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:39:36,433 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4700, 3.2007, 3.2196, 3.4467, 3.3287, 3.4315, 3.5247, 2.9536], device='cuda:0'), covar=tensor([0.0395, 0.1006, 0.0434, 0.0525, 0.0656, 0.0327, 0.0577, 0.0570], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0241, 0.0166, 0.0159, 0.0160, 0.0132, 0.0249, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:39:43,841 INFO [train.py:873] (0/4) Epoch 8, batch 4500, loss[loss=0.1532, simple_loss=0.1762, pruned_loss=0.0651, over 14269.00 frames. ], tot_loss[loss=0.1485, simple_loss=0.1714, pruned_loss=0.06279, over 1983029.90 frames. ], batch size: 46, lr: 9.85e-03, grad_scale: 8.0 2022-12-07 18:39:47,867 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.294e+02 2.953e+02 3.643e+02 6.271e+02, threshold=5.907e+02, percent-clipped=3.0 2022-12-07 18:39:50,714 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5519, 3.4107, 3.0857, 2.2111, 3.0391, 3.2676, 3.5934, 2.8477], device='cuda:0'), covar=tensor([0.0605, 0.1588, 0.1061, 0.2190, 0.0735, 0.0700, 0.0750, 0.1430], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0189, 0.0128, 0.0127, 0.0121, 0.0127, 0.0104, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005], device='cuda:0') 2022-12-07 18:39:57,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-07 18:39:59,198 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1159, 2.0051, 1.7297, 1.7743, 2.0150, 2.0240, 2.0558, 1.9978], device='cuda:0'), covar=tensor([0.1282, 0.0969, 0.2897, 0.3575, 0.1359, 0.1108, 0.1744, 0.1399], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0247, 0.0408, 0.0518, 0.0299, 0.0382, 0.0377, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:40:04,309 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:04,355 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:07,052 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:17,609 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8939, 1.6396, 3.6087, 3.4641, 3.5448, 3.6814, 3.1473, 3.6984], device='cuda:0'), covar=tensor([0.1233, 0.1277, 0.0109, 0.0182, 0.0166, 0.0121, 0.0187, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0154, 0.0116, 0.0159, 0.0135, 0.0131, 0.0108, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:40:35,502 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:40:45,757 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:00,129 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:10,290 INFO [train.py:873] (0/4) Epoch 8, batch 4600, loss[loss=0.1248, simple_loss=0.1635, pruned_loss=0.043, over 14459.00 frames. ], tot_loss[loss=0.1476, simple_loss=0.1709, pruned_loss=0.06212, over 1935361.76 frames. ], batch size: 24, lr: 9.84e-03, grad_scale: 8.0 2022-12-07 18:41:14,686 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.523e+02 3.027e+02 3.692e+02 7.181e+02, threshold=6.055e+02, percent-clipped=3.0 2022-12-07 18:41:16,375 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:25,833 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 18:41:29,286 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:41:35,979 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:07,618 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:18,291 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:38,955 INFO [train.py:873] (0/4) Epoch 8, batch 4700, loss[loss=0.1634, simple_loss=0.188, pruned_loss=0.06943, over 14154.00 frames. ], tot_loss[loss=0.1475, simple_loss=0.1709, pruned_loss=0.062, over 1942104.52 frames. ], batch size: 84, lr: 9.83e-03, grad_scale: 8.0 2022-12-07 18:42:43,279 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.451e+02 2.895e+02 3.498e+02 7.656e+02, threshold=5.791e+02, percent-clipped=2.0 2022-12-07 18:42:51,279 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:42:53,027 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:33,472 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:35,271 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:43:37,129 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1475, 3.9366, 3.5525, 3.7618, 3.9595, 4.0453, 4.1073, 4.1297], device='cuda:0'), covar=tensor([0.0803, 0.0484, 0.2078, 0.2576, 0.0840, 0.0713, 0.1108, 0.0734], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0245, 0.0409, 0.0520, 0.0299, 0.0382, 0.0370, 0.0326], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:44:05,135 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:44:06,679 INFO [train.py:873] (0/4) Epoch 8, batch 4800, loss[loss=0.1687, simple_loss=0.1544, pruned_loss=0.09144, over 2622.00 frames. ], tot_loss[loss=0.1482, simple_loss=0.1711, pruned_loss=0.06258, over 1948112.37 frames. ], batch size: 100, lr: 9.82e-03, grad_scale: 8.0 2022-12-07 18:44:10,753 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.410e+02 2.943e+02 4.035e+02 9.695e+02, threshold=5.886e+02, percent-clipped=7.0 2022-12-07 18:44:22,372 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2848, 2.2965, 2.8909, 1.8533, 1.8792, 2.4717, 1.2846, 2.5059], device='cuda:0'), covar=tensor([0.1682, 0.1450, 0.0718, 0.2458, 0.2860, 0.1072, 0.5440, 0.1038], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0089, 0.0083, 0.0090, 0.0113, 0.0076, 0.0130, 0.0080], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:44:23,481 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:44:31,496 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5670, 1.9533, 2.6835, 2.6843, 2.5024, 1.9708, 2.7321, 2.1845], device='cuda:0'), covar=tensor([0.0210, 0.0509, 0.0266, 0.0198, 0.0260, 0.0724, 0.0173, 0.0471], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0231, 0.0346, 0.0290, 0.0230, 0.0276, 0.0252, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:44:52,441 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1855, 3.6507, 3.9773, 4.1612, 4.0082, 3.8279, 4.1155, 3.4659], device='cuda:0'), covar=tensor([0.0913, 0.2012, 0.0764, 0.0908, 0.1249, 0.1104, 0.1371, 0.1043], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0246, 0.0168, 0.0160, 0.0164, 0.0133, 0.0253, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:44:58,922 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:20,528 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:25,656 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 18:45:36,247 INFO [train.py:873] (0/4) Epoch 8, batch 4900, loss[loss=0.1462, simple_loss=0.1759, pruned_loss=0.05824, over 14292.00 frames. ], tot_loss[loss=0.1489, simple_loss=0.1717, pruned_loss=0.06303, over 1967641.69 frames. ], batch size: 35, lr: 9.81e-03, grad_scale: 8.0 2022-12-07 18:45:40,367 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.301e+02 2.954e+02 3.459e+02 6.054e+02, threshold=5.907e+02, percent-clipped=1.0 2022-12-07 18:45:42,322 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:50,054 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:45:51,623 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 18:45:58,757 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9377, 2.2870, 3.3237, 2.2892, 2.0147, 2.7949, 1.6292, 2.7132], device='cuda:0'), covar=tensor([0.0786, 0.1406, 0.0540, 0.1608, 0.2612, 0.0845, 0.4190, 0.1217], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0089, 0.0083, 0.0089, 0.0113, 0.0075, 0.0129, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:46:06,813 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6791, 1.3001, 1.6280, 1.2433, 1.6050, 0.7928, 1.5927, 1.6590], device='cuda:0'), covar=tensor([0.1682, 0.0876, 0.0893, 0.1589, 0.1833, 0.0590, 0.1305, 0.0709], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0022, 0.0021, 0.0023, 0.0032, 0.0023, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:46:11,650 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6227, 1.1548, 2.0227, 1.9050, 1.9646, 2.0615, 1.5425, 2.0543], device='cuda:0'), covar=tensor([0.0555, 0.0835, 0.0127, 0.0290, 0.0321, 0.0153, 0.0366, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0156, 0.0117, 0.0161, 0.0137, 0.0131, 0.0109, 0.0115], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:46:24,220 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:33,373 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:34,124 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:46:37,502 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6894, 3.7332, 3.9259, 3.5109, 3.8083, 3.8853, 1.5221, 3.5711], device='cuda:0'), covar=tensor([0.0253, 0.0278, 0.0358, 0.0462, 0.0297, 0.0283, 0.3072, 0.0264], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0153, 0.0132, 0.0128, 0.0183, 0.0125, 0.0150, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:46:55,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2022-12-07 18:47:03,931 INFO [train.py:873] (0/4) Epoch 8, batch 5000, loss[loss=0.1585, simple_loss=0.1755, pruned_loss=0.07076, over 9502.00 frames. ], tot_loss[loss=0.1486, simple_loss=0.1717, pruned_loss=0.06275, over 1948079.29 frames. ], batch size: 100, lr: 9.80e-03, grad_scale: 8.0 2022-12-07 18:47:08,079 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.289e+02 3.008e+02 3.693e+02 7.100e+02, threshold=6.016e+02, percent-clipped=4.0 2022-12-07 18:47:14,181 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5830, 2.4806, 2.1710, 2.2584, 2.5078, 2.5147, 2.5492, 2.5003], device='cuda:0'), covar=tensor([0.0890, 0.0750, 0.2235, 0.2679, 0.1043, 0.1131, 0.1292, 0.1011], device='cuda:0'), in_proj_covar=tensor([0.0335, 0.0240, 0.0401, 0.0506, 0.0294, 0.0377, 0.0363, 0.0323], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:47:15,004 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:47:23,392 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-07 18:47:40,002 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.57 vs. limit=5.0 2022-12-07 18:47:43,277 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 18:48:15,960 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8519, 2.7534, 2.0061, 2.8943, 2.6659, 2.7910, 2.3810, 2.2839], device='cuda:0'), covar=tensor([0.0645, 0.1217, 0.3047, 0.0557, 0.0856, 0.0896, 0.1489, 0.2759], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0292, 0.0276, 0.0220, 0.0280, 0.0274, 0.0252, 0.0263], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:48:31,876 INFO [train.py:873] (0/4) Epoch 8, batch 5100, loss[loss=0.1533, simple_loss=0.1746, pruned_loss=0.06604, over 14380.00 frames. ], tot_loss[loss=0.1489, simple_loss=0.1715, pruned_loss=0.06312, over 1931669.11 frames. ], batch size: 53, lr: 9.80e-03, grad_scale: 8.0 2022-12-07 18:48:36,398 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.346e+02 3.025e+02 3.888e+02 7.693e+02, threshold=6.049e+02, percent-clipped=2.0 2022-12-07 18:48:48,752 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:04,030 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1417, 3.3124, 2.6814, 4.2892, 4.0476, 4.1218, 3.3790, 2.7464], device='cuda:0'), covar=tensor([0.0721, 0.1677, 0.5077, 0.0521, 0.0935, 0.1684, 0.1662, 0.4609], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0293, 0.0277, 0.0220, 0.0280, 0.0273, 0.0252, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:49:15,929 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2022-12-07 18:49:19,601 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:19,705 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2367, 1.0664, 1.0390, 1.0579, 1.1836, 0.5833, 1.1180, 1.1341], device='cuda:0'), covar=tensor([0.0808, 0.0722, 0.0842, 0.0882, 0.0849, 0.0652, 0.0862, 0.0884], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0023, 0.0022, 0.0023, 0.0033, 0.0023, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:49:30,686 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:45,335 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:49:59,577 INFO [train.py:873] (0/4) Epoch 8, batch 5200, loss[loss=0.1724, simple_loss=0.161, pruned_loss=0.09185, over 2614.00 frames. ], tot_loss[loss=0.1478, simple_loss=0.1709, pruned_loss=0.06237, over 1938929.26 frames. ], batch size: 100, lr: 9.79e-03, grad_scale: 16.0 2022-12-07 18:50:04,075 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 2.199e+02 2.811e+02 3.642e+02 7.366e+02, threshold=5.622e+02, percent-clipped=3.0 2022-12-07 18:50:05,146 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:07,814 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:08,007 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2022-12-07 18:50:13,800 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:25,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7354, 1.4367, 2.9052, 1.4587, 3.0047, 2.9048, 2.0349, 3.0806], device='cuda:0'), covar=tensor([0.0239, 0.2481, 0.0297, 0.1931, 0.0272, 0.0338, 0.0876, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0156, 0.0150, 0.0167, 0.0163, 0.0164, 0.0131, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:50:26,681 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:56,058 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:50:59,062 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:51:01,540 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:51:27,322 INFO [train.py:873] (0/4) Epoch 8, batch 5300, loss[loss=0.1623, simple_loss=0.1537, pruned_loss=0.08541, over 2579.00 frames. ], tot_loss[loss=0.149, simple_loss=0.1714, pruned_loss=0.06331, over 1898236.63 frames. ], batch size: 100, lr: 9.78e-03, grad_scale: 8.0 2022-12-07 18:51:32,495 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.430e+02 3.290e+02 3.970e+02 9.558e+02, threshold=6.580e+02, percent-clipped=6.0 2022-12-07 18:51:35,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6997, 5.3646, 5.0689, 5.7722, 5.2969, 5.0053, 5.7622, 5.5883], device='cuda:0'), covar=tensor([0.0483, 0.0599, 0.0621, 0.0430, 0.0492, 0.0311, 0.0485, 0.0630], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0118, 0.0128, 0.0134, 0.0130, 0.0103, 0.0146, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 18:52:55,482 INFO [train.py:873] (0/4) Epoch 8, batch 5400, loss[loss=0.1219, simple_loss=0.1517, pruned_loss=0.04612, over 13961.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.1717, pruned_loss=0.0626, over 2019705.72 frames. ], batch size: 19, lr: 9.77e-03, grad_scale: 8.0 2022-12-07 18:53:00,690 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.324e+02 2.944e+02 3.627e+02 6.890e+02, threshold=5.887e+02, percent-clipped=1.0 2022-12-07 18:53:43,213 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:54:15,553 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=14.26 vs. limit=5.0 2022-12-07 18:54:16,326 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7488, 2.1632, 3.7616, 3.8288, 3.7829, 2.3531, 3.7781, 2.8348], device='cuda:0'), covar=tensor([0.0269, 0.0676, 0.0565, 0.0307, 0.0216, 0.0948, 0.0251, 0.0684], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0231, 0.0348, 0.0292, 0.0232, 0.0278, 0.0252, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:54:23,919 INFO [train.py:873] (0/4) Epoch 8, batch 5500, loss[loss=0.1642, simple_loss=0.1881, pruned_loss=0.07016, over 14262.00 frames. ], tot_loss[loss=0.1469, simple_loss=0.1706, pruned_loss=0.06156, over 1981842.46 frames. ], batch size: 39, lr: 9.76e-03, grad_scale: 8.0 2022-12-07 18:54:26,070 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:54:29,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.638e+01 2.350e+02 2.867e+02 3.886e+02 1.534e+03, threshold=5.734e+02, percent-clipped=10.0 2022-12-07 18:54:34,967 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6764, 0.5922, 0.6082, 0.6799, 0.6601, 0.3708, 0.5179, 0.4619], device='cuda:0'), covar=tensor([0.0169, 0.0168, 0.0105, 0.0111, 0.0207, 0.0361, 0.0198, 0.0356], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0011, 0.0018, 0.0015, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.4388e-05, 9.0539e-05, 8.1237e-05, 8.5876e-05, 8.3655e-05, 1.2418e-04, 1.0446e-04, 1.1905e-04], device='cuda:0') 2022-12-07 18:54:57,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3286, 4.2781, 4.6036, 3.9326, 4.3384, 4.5847, 1.6358, 4.1444], device='cuda:0'), covar=tensor([0.0220, 0.0302, 0.0316, 0.0462, 0.0335, 0.0195, 0.3177, 0.0269], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0154, 0.0131, 0.0129, 0.0185, 0.0128, 0.0153, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:55:18,776 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:55:21,725 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:55:36,825 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5137, 3.2323, 3.1683, 3.4621, 3.3404, 3.4239, 3.5061, 2.9089], device='cuda:0'), covar=tensor([0.0393, 0.1034, 0.0431, 0.0486, 0.0731, 0.0387, 0.0642, 0.0603], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0246, 0.0169, 0.0159, 0.0165, 0.0134, 0.0251, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:55:52,341 INFO [train.py:873] (0/4) Epoch 8, batch 5600, loss[loss=0.1531, simple_loss=0.1624, pruned_loss=0.07188, over 3871.00 frames. ], tot_loss[loss=0.1479, simple_loss=0.171, pruned_loss=0.06238, over 1977467.71 frames. ], batch size: 100, lr: 9.75e-03, grad_scale: 8.0 2022-12-07 18:55:57,836 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.292e+02 2.814e+02 3.387e+02 5.498e+02, threshold=5.627e+02, percent-clipped=0.0 2022-12-07 18:56:06,725 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8369, 0.7406, 0.6351, 0.7382, 0.7793, 0.2511, 0.7300, 0.7455], device='cuda:0'), covar=tensor([0.0235, 0.0433, 0.0309, 0.0199, 0.0181, 0.0185, 0.0614, 0.0276], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0022, 0.0022, 0.0023, 0.0032, 0.0023, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 18:56:20,517 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4105, 2.3331, 2.5085, 2.5120, 2.4231, 2.1583, 1.3978, 2.2119], device='cuda:0'), covar=tensor([0.0394, 0.0495, 0.0452, 0.0294, 0.0363, 0.1097, 0.2216, 0.0374], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0131, 0.0130, 0.0188, 0.0129, 0.0156, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:56:52,143 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.76 vs. limit=5.0 2022-12-07 18:57:21,321 INFO [train.py:873] (0/4) Epoch 8, batch 5700, loss[loss=0.154, simple_loss=0.1832, pruned_loss=0.06243, over 14506.00 frames. ], tot_loss[loss=0.1475, simple_loss=0.1705, pruned_loss=0.06221, over 2027281.35 frames. ], batch size: 49, lr: 9.75e-03, grad_scale: 8.0 2022-12-07 18:57:21,391 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2478, 2.9697, 3.0557, 3.1970, 3.1428, 3.1783, 3.3146, 2.7171], device='cuda:0'), covar=tensor([0.0490, 0.1083, 0.0431, 0.0568, 0.0787, 0.0472, 0.0617, 0.0619], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0248, 0.0170, 0.0161, 0.0166, 0.0135, 0.0254, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 18:57:26,806 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.316e+02 2.828e+02 3.502e+02 7.696e+02, threshold=5.656e+02, percent-clipped=3.0 2022-12-07 18:57:53,818 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8323, 1.6763, 2.9594, 2.0279, 2.8861, 1.6623, 2.1139, 2.6817], device='cuda:0'), covar=tensor([0.1213, 0.4906, 0.0704, 0.6453, 0.0865, 0.4367, 0.1639, 0.0823], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0237, 0.0187, 0.0321, 0.0208, 0.0240, 0.0229, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 18:58:02,119 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:58:29,421 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2587, 1.3637, 1.3436, 1.1321, 1.1989, 0.8374, 1.1155, 0.8502], device='cuda:0'), covar=tensor([0.0242, 0.0372, 0.0346, 0.0337, 0.0349, 0.0396, 0.0296, 0.0583], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.7002e-05, 9.3783e-05, 8.2948e-05, 8.8550e-05, 8.6849e-05, 1.2949e-04, 1.0857e-04, 1.2291e-04], device='cuda:0') 2022-12-07 18:58:49,961 INFO [train.py:873] (0/4) Epoch 8, batch 5800, loss[loss=0.1365, simple_loss=0.1688, pruned_loss=0.05211, over 14363.00 frames. ], tot_loss[loss=0.1477, simple_loss=0.1704, pruned_loss=0.06255, over 1969977.01 frames. ], batch size: 41, lr: 9.74e-03, grad_scale: 8.0 2022-12-07 18:58:51,060 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3348, 1.9277, 2.2472, 2.3454, 2.1810, 1.8720, 2.3421, 2.1769], device='cuda:0'), covar=tensor([0.0199, 0.0384, 0.0195, 0.0145, 0.0205, 0.0453, 0.0170, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0231, 0.0348, 0.0291, 0.0231, 0.0277, 0.0251, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 18:58:55,173 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.215e+02 2.766e+02 3.426e+02 6.991e+02, threshold=5.532e+02, percent-clipped=1.0 2022-12-07 18:58:56,302 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:01,620 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2235, 2.8645, 3.6780, 2.5105, 2.0811, 3.0808, 1.5339, 3.3951], device='cuda:0'), covar=tensor([0.0652, 0.1143, 0.0526, 0.1631, 0.2438, 0.0794, 0.4298, 0.0594], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0089, 0.0083, 0.0089, 0.0111, 0.0076, 0.0127, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 18:59:07,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.02 vs. limit=2.0 2022-12-07 18:59:25,769 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:35,894 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4624, 4.2267, 4.1422, 4.5085, 4.0851, 3.8482, 4.4957, 4.4116], device='cuda:0'), covar=tensor([0.0606, 0.0687, 0.0740, 0.0543, 0.0769, 0.0571, 0.0600, 0.0634], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0119, 0.0128, 0.0136, 0.0131, 0.0103, 0.0146, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 18:59:46,034 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 18:59:48,892 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:19,697 INFO [train.py:873] (0/4) Epoch 8, batch 5900, loss[loss=0.1629, simple_loss=0.1449, pruned_loss=0.09041, over 1372.00 frames. ], tot_loss[loss=0.1464, simple_loss=0.1693, pruned_loss=0.06175, over 1918896.93 frames. ], batch size: 100, lr: 9.73e-03, grad_scale: 8.0 2022-12-07 19:00:20,794 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:25,227 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.290e+02 2.920e+02 3.537e+02 8.146e+02, threshold=5.840e+02, percent-clipped=6.0 2022-12-07 19:00:29,004 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:31,563 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:00:37,383 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2022-12-07 19:00:43,540 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5182, 3.3393, 3.0196, 2.0070, 3.0554, 3.3213, 3.6143, 2.7623], device='cuda:0'), covar=tensor([0.0663, 0.1538, 0.1097, 0.2090, 0.0837, 0.0544, 0.0887, 0.1417], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0186, 0.0129, 0.0125, 0.0122, 0.0128, 0.0104, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:01:09,529 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.75 vs. limit=5.0 2022-12-07 19:01:22,581 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:01:47,423 INFO [train.py:873] (0/4) Epoch 8, batch 6000, loss[loss=0.1532, simple_loss=0.1821, pruned_loss=0.06214, over 14296.00 frames. ], tot_loss[loss=0.1471, simple_loss=0.1702, pruned_loss=0.06203, over 1995207.09 frames. ], batch size: 25, lr: 9.72e-03, grad_scale: 8.0 2022-12-07 19:01:47,423 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 19:02:10,414 INFO [train.py:905] (0/4) Epoch 8, validation: loss=0.1228, simple_loss=0.1649, pruned_loss=0.04039, over 857387.00 frames. 2022-12-07 19:02:10,414 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 19:02:16,060 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.300e+02 3.027e+02 3.751e+02 1.020e+03, threshold=6.055e+02, percent-clipped=3.0 2022-12-07 19:02:22,942 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=13.89 vs. limit=5.0 2022-12-07 19:02:26,518 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.41 vs. limit=5.0 2022-12-07 19:02:39,024 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:03:38,336 INFO [train.py:873] (0/4) Epoch 8, batch 6100, loss[loss=0.1777, simple_loss=0.1847, pruned_loss=0.08536, over 7803.00 frames. ], tot_loss[loss=0.1481, simple_loss=0.1712, pruned_loss=0.06253, over 2050031.64 frames. ], batch size: 100, lr: 9.71e-03, grad_scale: 8.0 2022-12-07 19:03:40,136 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:03:43,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.326e+02 2.830e+02 3.436e+02 6.113e+02, threshold=5.661e+02, percent-clipped=1.0 2022-12-07 19:03:49,652 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8524, 3.4262, 4.1504, 2.8570, 2.4813, 3.2637, 1.6176, 3.0814], device='cuda:0'), covar=tensor([0.1451, 0.1145, 0.0644, 0.1989, 0.2627, 0.2090, 0.4842, 0.2350], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0090, 0.0082, 0.0089, 0.0111, 0.0076, 0.0128, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:04:03,199 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2403, 1.2933, 1.4697, 0.8799, 0.8716, 1.4055, 0.7857, 1.1366], device='cuda:0'), covar=tensor([0.1036, 0.2062, 0.0547, 0.2353, 0.2859, 0.0575, 0.2063, 0.0926], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0090, 0.0082, 0.0089, 0.0111, 0.0077, 0.0128, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:05:02,938 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:05:06,212 INFO [train.py:873] (0/4) Epoch 8, batch 6200, loss[loss=0.1336, simple_loss=0.1611, pruned_loss=0.05307, over 14248.00 frames. ], tot_loss[loss=0.1482, simple_loss=0.1712, pruned_loss=0.06256, over 2025836.91 frames. ], batch size: 57, lr: 9.71e-03, grad_scale: 8.0 2022-12-07 19:05:11,776 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.400e+01 2.402e+02 3.111e+02 3.796e+02 6.098e+02, threshold=6.221e+02, percent-clipped=5.0 2022-12-07 19:05:26,012 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2924, 3.3199, 3.3528, 3.1327, 3.2732, 3.1522, 1.2379, 3.1494], device='cuda:0'), covar=tensor([0.0420, 0.0509, 0.0662, 0.0643, 0.0644, 0.0747, 0.4070, 0.0452], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0156, 0.0133, 0.0130, 0.0188, 0.0129, 0.0154, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:05:32,165 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:05:51,846 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-07 19:06:26,869 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:06:35,505 INFO [train.py:873] (0/4) Epoch 8, batch 6300, loss[loss=0.1965, simple_loss=0.198, pruned_loss=0.09745, over 8638.00 frames. ], tot_loss[loss=0.1484, simple_loss=0.1708, pruned_loss=0.06301, over 1924261.19 frames. ], batch size: 100, lr: 9.70e-03, grad_scale: 8.0 2022-12-07 19:06:38,445 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-07 19:06:40,625 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.206e+02 2.756e+02 3.412e+02 6.293e+02, threshold=5.512e+02, percent-clipped=2.0 2022-12-07 19:07:00,045 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:07:15,060 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0743, 1.2094, 1.1269, 0.9762, 1.2417, 0.5524, 1.1118, 1.0642], device='cuda:0'), covar=tensor([0.0876, 0.0658, 0.0661, 0.0625, 0.0758, 0.0768, 0.0733, 0.0629], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0022, 0.0022, 0.0021, 0.0023, 0.0031, 0.0022, 0.0022], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 19:07:27,664 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 19:08:02,870 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 19:08:04,135 INFO [train.py:873] (0/4) Epoch 8, batch 6400, loss[loss=0.1252, simple_loss=0.1568, pruned_loss=0.04683, over 13883.00 frames. ], tot_loss[loss=0.1481, simple_loss=0.1706, pruned_loss=0.0628, over 1906589.16 frames. ], batch size: 23, lr: 9.69e-03, grad_scale: 8.0 2022-12-07 19:08:06,064 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:08:09,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.267e+01 2.304e+02 2.776e+02 3.758e+02 5.907e+02, threshold=5.553e+02, percent-clipped=4.0 2022-12-07 19:08:24,589 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 19:08:48,455 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:09:08,911 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.01 vs. limit=5.0 2022-12-07 19:09:10,822 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2022-12-07 19:09:23,854 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4043, 1.4744, 4.2497, 1.8241, 4.0182, 4.2728, 3.9502, 4.7644], device='cuda:0'), covar=tensor([0.0218, 0.3253, 0.0360, 0.2587, 0.0440, 0.0493, 0.0389, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0158, 0.0151, 0.0170, 0.0164, 0.0166, 0.0131, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:09:29,197 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:09:32,600 INFO [train.py:873] (0/4) Epoch 8, batch 6500, loss[loss=0.1319, simple_loss=0.1471, pruned_loss=0.05835, over 6955.00 frames. ], tot_loss[loss=0.1495, simple_loss=0.1719, pruned_loss=0.06352, over 1944445.54 frames. ], batch size: 100, lr: 9.68e-03, grad_scale: 8.0 2022-12-07 19:09:37,915 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.343e+02 2.966e+02 3.672e+02 1.051e+03, threshold=5.932e+02, percent-clipped=2.0 2022-12-07 19:09:57,745 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.67 vs. limit=5.0 2022-12-07 19:10:00,829 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2022-12-07 19:10:10,824 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:31,102 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:46,630 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:10:59,606 INFO [train.py:873] (0/4) Epoch 8, batch 6600, loss[loss=0.1973, simple_loss=0.1743, pruned_loss=0.1101, over 1224.00 frames. ], tot_loss[loss=0.1477, simple_loss=0.1709, pruned_loss=0.06228, over 1964487.64 frames. ], batch size: 100, lr: 9.67e-03, grad_scale: 8.0 2022-12-07 19:11:04,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.283e+02 2.818e+02 3.412e+02 6.206e+02, threshold=5.635e+02, percent-clipped=2.0 2022-12-07 19:11:23,660 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:11:24,598 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:11:47,677 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6353, 4.2479, 4.2180, 4.6321, 4.3423, 4.1010, 4.6617, 3.9182], device='cuda:0'), covar=tensor([0.0319, 0.0929, 0.0362, 0.0382, 0.0743, 0.0597, 0.0523, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0239, 0.0164, 0.0157, 0.0161, 0.0126, 0.0245, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 19:12:05,780 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:12:27,867 INFO [train.py:873] (0/4) Epoch 8, batch 6700, loss[loss=0.1274, simple_loss=0.1556, pruned_loss=0.04956, over 14586.00 frames. ], tot_loss[loss=0.1483, simple_loss=0.1712, pruned_loss=0.06268, over 1981223.00 frames. ], batch size: 34, lr: 9.66e-03, grad_scale: 8.0 2022-12-07 19:12:29,397 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-07 19:12:32,694 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.361e+02 2.930e+02 3.597e+02 6.848e+02, threshold=5.861e+02, percent-clipped=3.0 2022-12-07 19:13:16,978 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 19:13:55,715 INFO [train.py:873] (0/4) Epoch 8, batch 6800, loss[loss=0.1759, simple_loss=0.1899, pruned_loss=0.08097, over 14197.00 frames. ], tot_loss[loss=0.1468, simple_loss=0.17, pruned_loss=0.06181, over 1975821.19 frames. ], batch size: 69, lr: 9.66e-03, grad_scale: 8.0 2022-12-07 19:14:01,555 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.278e+02 2.761e+02 3.768e+02 7.478e+02, threshold=5.523e+02, percent-clipped=3.0 2022-12-07 19:14:46,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8200, 2.3748, 3.5397, 2.6583, 3.6863, 3.4411, 3.3427, 2.8546], device='cuda:0'), covar=tensor([0.0713, 0.3164, 0.1040, 0.2226, 0.0852, 0.0887, 0.1519, 0.2306], device='cuda:0'), in_proj_covar=tensor([0.0327, 0.0323, 0.0397, 0.0307, 0.0377, 0.0309, 0.0360, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 19:14:56,166 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4097, 4.1621, 3.9291, 3.9941, 4.2375, 4.3156, 4.4181, 4.4068], device='cuda:0'), covar=tensor([0.0862, 0.0627, 0.2099, 0.2936, 0.0679, 0.0726, 0.1098, 0.0770], device='cuda:0'), in_proj_covar=tensor([0.0339, 0.0241, 0.0405, 0.0517, 0.0291, 0.0382, 0.0369, 0.0328], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 19:15:12,023 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:25,705 INFO [train.py:873] (0/4) Epoch 8, batch 6900, loss[loss=0.1739, simple_loss=0.1832, pruned_loss=0.08227, over 9493.00 frames. ], tot_loss[loss=0.1475, simple_loss=0.1705, pruned_loss=0.06225, over 1985159.94 frames. ], batch size: 100, lr: 9.65e-03, grad_scale: 8.0 2022-12-07 19:15:30,915 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.523e+02 3.039e+02 3.738e+02 1.408e+03, threshold=6.079e+02, percent-clipped=6.0 2022-12-07 19:15:31,924 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4268, 4.5211, 4.8719, 4.1996, 4.6893, 4.9299, 1.6733, 4.2159], device='cuda:0'), covar=tensor([0.0218, 0.0275, 0.0382, 0.0422, 0.0261, 0.0159, 0.3246, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0134, 0.0130, 0.0186, 0.0129, 0.0155, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:15:36,958 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:45,649 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:15:54,680 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:13,248 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:31,391 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:16:53,932 INFO [train.py:873] (0/4) Epoch 8, batch 7000, loss[loss=0.1658, simple_loss=0.1535, pruned_loss=0.08905, over 1188.00 frames. ], tot_loss[loss=0.1467, simple_loss=0.17, pruned_loss=0.06176, over 1974022.65 frames. ], batch size: 100, lr: 9.64e-03, grad_scale: 8.0 2022-12-07 19:16:59,264 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.338e+02 3.066e+02 3.722e+02 8.266e+02, threshold=6.131e+02, percent-clipped=2.0 2022-12-07 19:17:08,028 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:17:18,296 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9176, 2.9689, 3.1206, 2.9329, 3.0057, 2.9297, 1.4224, 2.7218], device='cuda:0'), covar=tensor([0.0331, 0.0360, 0.0449, 0.0445, 0.0353, 0.0597, 0.2876, 0.0391], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0134, 0.0131, 0.0188, 0.0127, 0.0155, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:17:41,985 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5124, 1.8082, 1.8660, 1.9600, 1.7748, 2.0164, 1.5894, 1.2542], device='cuda:0'), covar=tensor([0.1219, 0.0795, 0.0727, 0.0485, 0.0960, 0.0472, 0.1764, 0.2327], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0067, 0.0056, 0.0057, 0.0085, 0.0065, 0.0089, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:17:51,565 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9185, 1.8848, 1.5729, 2.0555, 1.8587, 1.8940, 1.8128, 1.7483], device='cuda:0'), covar=tensor([0.0580, 0.0687, 0.1781, 0.0419, 0.0696, 0.0439, 0.0991, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0296, 0.0273, 0.0226, 0.0292, 0.0279, 0.0257, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 19:17:52,392 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-60000.pt 2022-12-07 19:18:27,445 INFO [train.py:873] (0/4) Epoch 8, batch 7100, loss[loss=0.1815, simple_loss=0.1856, pruned_loss=0.08871, over 7774.00 frames. ], tot_loss[loss=0.1459, simple_loss=0.1695, pruned_loss=0.06118, over 1996412.89 frames. ], batch size: 100, lr: 9.63e-03, grad_scale: 8.0 2022-12-07 19:18:28,999 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-07 19:18:31,866 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:18:32,560 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.178e+02 2.684e+02 3.271e+02 7.336e+02, threshold=5.369e+02, percent-clipped=1.0 2022-12-07 19:18:42,324 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:18:54,028 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:05,177 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:24,535 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 19:19:35,585 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 19:19:46,370 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:53,836 INFO [train.py:873] (0/4) Epoch 8, batch 7200, loss[loss=0.1515, simple_loss=0.1743, pruned_loss=0.06432, over 13871.00 frames. ], tot_loss[loss=0.1468, simple_loss=0.17, pruned_loss=0.06185, over 1988765.97 frames. ], batch size: 23, lr: 9.63e-03, grad_scale: 8.0 2022-12-07 19:19:54,431 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2022-12-07 19:19:57,705 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:19:59,258 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.579e+01 2.475e+02 2.936e+02 3.540e+02 1.198e+03, threshold=5.873e+02, percent-clipped=8.0 2022-12-07 19:20:07,285 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 19:20:14,375 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:49,269 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:54,600 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:20:56,408 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:21:22,474 INFO [train.py:873] (0/4) Epoch 8, batch 7300, loss[loss=0.1059, simple_loss=0.1425, pruned_loss=0.03468, over 13889.00 frames. ], tot_loss[loss=0.145, simple_loss=0.1685, pruned_loss=0.06074, over 1979409.06 frames. ], batch size: 20, lr: 9.62e-03, grad_scale: 16.0 2022-12-07 19:21:28,209 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.538e+01 2.297e+02 3.128e+02 4.022e+02 9.031e+02, threshold=6.256e+02, percent-clipped=2.0 2022-12-07 19:21:30,656 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:21:43,244 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:22:49,539 INFO [train.py:873] (0/4) Epoch 8, batch 7400, loss[loss=0.1474, simple_loss=0.1413, pruned_loss=0.07679, over 2639.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1692, pruned_loss=0.06162, over 1916581.50 frames. ], batch size: 100, lr: 9.61e-03, grad_scale: 8.0 2022-12-07 19:22:56,039 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.246e+02 2.782e+02 3.435e+02 7.808e+02, threshold=5.565e+02, percent-clipped=3.0 2022-12-07 19:23:21,197 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2022-12-07 19:23:43,972 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 19:23:53,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0582, 4.1017, 4.3729, 3.7689, 4.1234, 4.3620, 1.4677, 3.8791], device='cuda:0'), covar=tensor([0.0205, 0.0340, 0.0331, 0.0500, 0.0307, 0.0204, 0.3334, 0.0265], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0160, 0.0137, 0.0133, 0.0190, 0.0131, 0.0158, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:23:53,483 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0540, 1.3770, 3.8928, 1.7163, 3.9405, 4.0044, 3.0206, 4.4233], device='cuda:0'), covar=tensor([0.0211, 0.3205, 0.0381, 0.2330, 0.0383, 0.0363, 0.0652, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0158, 0.0154, 0.0168, 0.0166, 0.0167, 0.0133, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:23:54,314 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 19:24:06,057 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:24:17,783 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:24:18,542 INFO [train.py:873] (0/4) Epoch 8, batch 7500, loss[loss=0.1473, simple_loss=0.1744, pruned_loss=0.06011, over 14265.00 frames. ], tot_loss[loss=0.1472, simple_loss=0.1697, pruned_loss=0.06231, over 1844938.29 frames. ], batch size: 80, lr: 9.60e-03, grad_scale: 8.0 2022-12-07 19:24:21,182 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9381, 2.9641, 2.8442, 2.9550, 2.9039, 2.8150, 1.2047, 2.6305], device='cuda:0'), covar=tensor([0.0432, 0.0545, 0.0747, 0.0557, 0.0604, 0.0950, 0.3735, 0.0560], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0136, 0.0131, 0.0189, 0.0129, 0.0156, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:24:24,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.510e+02 2.850e+02 3.552e+02 6.747e+02, threshold=5.701e+02, percent-clipped=5.0 2022-12-07 19:24:38,125 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4569, 2.2089, 2.4292, 1.5884, 2.1062, 2.3876, 2.5218, 2.0962], device='cuda:0'), covar=tensor([0.0820, 0.1305, 0.1056, 0.2071, 0.1183, 0.0737, 0.0767, 0.1759], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0186, 0.0128, 0.0123, 0.0123, 0.0127, 0.0104, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:25:05,342 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-8.pt 2022-12-07 19:25:46,334 INFO [train.py:873] (0/4) Epoch 9, batch 0, loss[loss=0.1707, simple_loss=0.186, pruned_loss=0.07771, over 14161.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.186, pruned_loss=0.07771, over 14161.00 frames. ], batch size: 99, lr: 9.08e-03, grad_scale: 8.0 2022-12-07 19:25:46,334 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 19:25:53,547 INFO [train.py:905] (0/4) Epoch 9, validation: loss=0.1275, simple_loss=0.1706, pruned_loss=0.04216, over 857387.00 frames. 2022-12-07 19:25:53,548 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 19:26:00,141 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:33,659 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.503e+01 2.217e+02 2.837e+02 3.988e+02 6.329e+02, threshold=5.675e+02, percent-clipped=3.0 2022-12-07 19:26:36,462 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:43,321 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:45,465 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:26:50,488 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-07 19:27:19,861 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:27:23,501 INFO [train.py:873] (0/4) Epoch 9, batch 100, loss[loss=0.1595, simple_loss=0.1786, pruned_loss=0.07018, over 13528.00 frames. ], tot_loss[loss=0.1464, simple_loss=0.1709, pruned_loss=0.06096, over 881840.08 frames. ], batch size: 100, lr: 9.08e-03, grad_scale: 8.0 2022-12-07 19:27:30,703 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9663, 5.4347, 5.4706, 5.9069, 5.5154, 5.0053, 5.8489, 4.7436], device='cuda:0'), covar=tensor([0.0237, 0.0707, 0.0236, 0.0333, 0.0672, 0.0285, 0.0422, 0.0491], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0242, 0.0164, 0.0160, 0.0162, 0.0132, 0.0248, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 19:28:03,331 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.443e+02 2.952e+02 3.884e+02 8.156e+02, threshold=5.905e+02, percent-clipped=9.0 2022-12-07 19:28:12,644 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 19:28:13,919 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7072, 4.3399, 4.2876, 4.7332, 4.3560, 3.8378, 4.7515, 4.5961], device='cuda:0'), covar=tensor([0.0609, 0.0783, 0.0708, 0.0474, 0.0641, 0.0591, 0.0507, 0.0656], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0117, 0.0127, 0.0132, 0.0130, 0.0104, 0.0144, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 19:28:35,755 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6163, 2.5280, 2.7461, 2.7514, 2.6677, 2.3899, 1.4025, 2.4165], device='cuda:0'), covar=tensor([0.0413, 0.0468, 0.0479, 0.0401, 0.0371, 0.1075, 0.2683, 0.0377], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0135, 0.0132, 0.0188, 0.0131, 0.0158, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 19:28:46,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 19:28:51,077 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:28:51,890 INFO [train.py:873] (0/4) Epoch 9, batch 200, loss[loss=0.1366, simple_loss=0.1592, pruned_loss=0.05702, over 13957.00 frames. ], tot_loss[loss=0.147, simple_loss=0.1703, pruned_loss=0.0618, over 1347604.18 frames. ], batch size: 19, lr: 9.07e-03, grad_scale: 8.0 2022-12-07 19:29:01,480 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:10,966 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:13,791 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:24,484 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-07 19:29:24,921 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:32,172 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.211e+01 2.496e+02 3.042e+02 4.274e+02 1.188e+03, threshold=6.084e+02, percent-clipped=4.0 2022-12-07 19:29:33,102 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:44,434 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:29:55,091 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6581, 3.4621, 3.4159, 3.7491, 3.3565, 3.0556, 3.7082, 3.6427], device='cuda:0'), covar=tensor([0.0700, 0.0882, 0.0797, 0.0586, 0.0896, 0.0659, 0.0714, 0.0733], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0117, 0.0126, 0.0132, 0.0129, 0.0104, 0.0143, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-07 19:29:55,861 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:04,721 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:07,093 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:30:20,886 INFO [train.py:873] (0/4) Epoch 9, batch 300, loss[loss=0.139, simple_loss=0.1678, pruned_loss=0.05513, over 14227.00 frames. ], tot_loss[loss=0.1481, simple_loss=0.1708, pruned_loss=0.06265, over 1641338.19 frames. ], batch size: 60, lr: 9.06e-03, grad_scale: 4.0 2022-12-07 19:30:49,717 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4870, 1.6660, 2.6750, 2.1483, 2.5353, 1.6320, 2.1062, 2.3868], device='cuda:0'), covar=tensor([0.1113, 0.4040, 0.0312, 0.4284, 0.0684, 0.2943, 0.1118, 0.0603], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0229, 0.0184, 0.0311, 0.0209, 0.0232, 0.0224, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 19:30:57,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2022-12-07 19:31:01,444 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.299e+02 2.848e+02 3.507e+02 1.292e+03, threshold=5.696e+02, percent-clipped=2.0 2022-12-07 19:31:11,125 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:31:38,599 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.17 vs. limit=5.0 2022-12-07 19:31:49,307 INFO [train.py:873] (0/4) Epoch 9, batch 400, loss[loss=0.1454, simple_loss=0.175, pruned_loss=0.05787, over 14206.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1695, pruned_loss=0.0614, over 1763956.57 frames. ], batch size: 60, lr: 9.06e-03, grad_scale: 8.0 2022-12-07 19:31:53,808 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:32:07,440 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8393, 0.9549, 0.7855, 0.8137, 0.8909, 0.4025, 0.7407, 0.8331], device='cuda:0'), covar=tensor([0.0451, 0.0615, 0.0570, 0.0455, 0.0602, 0.0586, 0.0854, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0024, 0.0021, 0.0023, 0.0033, 0.0023, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 19:32:10,056 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5259, 1.4271, 3.8028, 1.5679, 3.5730, 3.8168, 3.0235, 3.9114], device='cuda:0'), covar=tensor([0.0386, 0.4068, 0.0458, 0.2880, 0.0658, 0.0400, 0.0631, 0.0351], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0156, 0.0151, 0.0165, 0.0166, 0.0162, 0.0132, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:32:15,296 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5022, 4.1793, 4.0229, 4.5061, 4.2115, 3.9445, 4.5182, 3.8218], device='cuda:0'), covar=tensor([0.0393, 0.0958, 0.0352, 0.0438, 0.0729, 0.0815, 0.0585, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0245, 0.0167, 0.0163, 0.0163, 0.0134, 0.0250, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 19:32:29,748 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.639e+01 2.151e+02 2.975e+02 3.759e+02 9.050e+02, threshold=5.950e+02, percent-clipped=6.0 2022-12-07 19:33:01,374 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1579, 1.6555, 1.4505, 1.5353, 1.3105, 1.4293, 1.4516, 1.3458], device='cuda:0'), covar=tensor([0.1070, 0.0971, 0.0906, 0.0494, 0.0562, 0.0562, 0.0483, 0.0730], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0012, 0.0012, 0.0019, 0.0015, 0.0019], device='cuda:0'), out_proj_covar=tensor([8.9671e-05, 9.5886e-05, 8.4972e-05, 9.0581e-05, 8.6427e-05, 1.3197e-04, 1.1040e-04, 1.2410e-04], device='cuda:0') 2022-12-07 19:33:18,101 INFO [train.py:873] (0/4) Epoch 9, batch 500, loss[loss=0.1875, simple_loss=0.168, pruned_loss=0.1036, over 1273.00 frames. ], tot_loss[loss=0.1466, simple_loss=0.1697, pruned_loss=0.06175, over 1809670.37 frames. ], batch size: 100, lr: 9.05e-03, grad_scale: 8.0 2022-12-07 19:33:59,480 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 2.291e+02 2.820e+02 3.548e+02 5.650e+02, threshold=5.639e+02, percent-clipped=0.0 2022-12-07 19:34:04,742 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1865, 3.5273, 4.2947, 3.0414, 2.6742, 3.4815, 2.0891, 3.8017], device='cuda:0'), covar=tensor([0.0909, 0.0986, 0.0450, 0.2546, 0.2170, 0.1309, 0.4001, 0.0622], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0089, 0.0082, 0.0089, 0.0109, 0.0076, 0.0127, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:34:14,667 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:34:25,962 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:34:30,207 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 19:34:45,880 INFO [train.py:873] (0/4) Epoch 9, batch 600, loss[loss=0.1144, simple_loss=0.1492, pruned_loss=0.03983, over 13964.00 frames. ], tot_loss[loss=0.1475, simple_loss=0.1703, pruned_loss=0.06233, over 1853331.64 frames. ], batch size: 20, lr: 9.04e-03, grad_scale: 8.0 2022-12-07 19:35:08,496 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:35:25,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.178e+01 2.284e+02 2.726e+02 3.406e+02 8.336e+02, threshold=5.452e+02, percent-clipped=5.0 2022-12-07 19:35:57,824 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 19:36:13,113 INFO [train.py:873] (0/4) Epoch 9, batch 700, loss[loss=0.1775, simple_loss=0.1714, pruned_loss=0.09177, over 3871.00 frames. ], tot_loss[loss=0.1466, simple_loss=0.1698, pruned_loss=0.06167, over 1918190.13 frames. ], batch size: 100, lr: 9.03e-03, grad_scale: 8.0 2022-12-07 19:36:44,332 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3395, 1.3062, 3.3465, 1.3922, 3.2819, 3.4426, 2.2994, 3.6768], device='cuda:0'), covar=tensor([0.0236, 0.3038, 0.0401, 0.2320, 0.0691, 0.0372, 0.0876, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0158, 0.0155, 0.0169, 0.0168, 0.0167, 0.0134, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:36:54,420 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.220e+02 2.923e+02 3.750e+02 6.888e+02, threshold=5.845e+02, percent-clipped=6.0 2022-12-07 19:37:17,891 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 19:37:41,703 INFO [train.py:873] (0/4) Epoch 9, batch 800, loss[loss=0.1302, simple_loss=0.1655, pruned_loss=0.04741, over 13918.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1694, pruned_loss=0.0615, over 1926841.97 frames. ], batch size: 26, lr: 9.03e-03, grad_scale: 8.0 2022-12-07 19:38:22,020 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 2.254e+02 2.833e+02 3.522e+02 6.123e+02, threshold=5.665e+02, percent-clipped=1.0 2022-12-07 19:38:49,295 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:39:09,293 INFO [train.py:873] (0/4) Epoch 9, batch 900, loss[loss=0.1641, simple_loss=0.1444, pruned_loss=0.09194, over 1249.00 frames. ], tot_loss[loss=0.1462, simple_loss=0.1697, pruned_loss=0.06133, over 1931020.82 frames. ], batch size: 100, lr: 9.02e-03, grad_scale: 8.0 2022-12-07 19:39:21,480 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3656, 4.8095, 4.9143, 5.3494, 5.0066, 4.5139, 5.2873, 4.4463], device='cuda:0'), covar=tensor([0.0257, 0.0937, 0.0256, 0.0386, 0.0599, 0.0374, 0.0473, 0.0421], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0246, 0.0169, 0.0165, 0.0165, 0.0133, 0.0251, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 19:39:27,300 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:39:31,525 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:39:42,858 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4136, 2.1822, 3.4029, 3.4637, 3.3936, 2.3665, 3.3266, 2.7114], device='cuda:0'), covar=tensor([0.0232, 0.0539, 0.0518, 0.0269, 0.0218, 0.0787, 0.0239, 0.0548], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0232, 0.0348, 0.0294, 0.0236, 0.0281, 0.0258, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:39:50,649 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.290e+02 3.289e+02 4.035e+02 9.123e+02, threshold=6.578e+02, percent-clipped=5.0 2022-12-07 19:40:02,621 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 19:40:37,281 INFO [train.py:873] (0/4) Epoch 9, batch 1000, loss[loss=0.1452, simple_loss=0.1626, pruned_loss=0.0639, over 4955.00 frames. ], tot_loss[loss=0.1465, simple_loss=0.1697, pruned_loss=0.06168, over 1937225.59 frames. ], batch size: 100, lr: 9.01e-03, grad_scale: 8.0 2022-12-07 19:41:18,075 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.101e+02 2.828e+02 3.770e+02 7.982e+02, threshold=5.656e+02, percent-clipped=1.0 2022-12-07 19:41:27,155 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8968, 1.7776, 2.0250, 1.7759, 2.0898, 1.8330, 1.6289, 1.8617], device='cuda:0'), covar=tensor([0.0446, 0.0982, 0.0233, 0.0372, 0.0219, 0.0554, 0.0203, 0.0333], device='cuda:0'), in_proj_covar=tensor([0.0328, 0.0320, 0.0394, 0.0307, 0.0373, 0.0309, 0.0361, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 19:42:05,669 INFO [train.py:873] (0/4) Epoch 9, batch 1100, loss[loss=0.1347, simple_loss=0.1635, pruned_loss=0.05302, over 14207.00 frames. ], tot_loss[loss=0.1458, simple_loss=0.1692, pruned_loss=0.06118, over 1935478.95 frames. ], batch size: 25, lr: 9.00e-03, grad_scale: 8.0 2022-12-07 19:42:13,902 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.15 vs. limit=2.0 2022-12-07 19:42:46,207 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.543e+02 3.100e+02 3.894e+02 7.246e+02, threshold=6.200e+02, percent-clipped=7.0 2022-12-07 19:42:49,131 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 19:43:32,678 INFO [train.py:873] (0/4) Epoch 9, batch 1200, loss[loss=0.1252, simple_loss=0.1627, pruned_loss=0.04382, over 14473.00 frames. ], tot_loss[loss=0.1452, simple_loss=0.1692, pruned_loss=0.06065, over 1990200.03 frames. ], batch size: 24, lr: 9.00e-03, grad_scale: 8.0 2022-12-07 19:43:49,885 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:03,914 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:13,227 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.189e+02 2.580e+02 3.174e+02 5.497e+02, threshold=5.160e+02, percent-clipped=0.0 2022-12-07 19:44:13,469 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:32,972 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:56,469 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:44:59,741 INFO [train.py:873] (0/4) Epoch 9, batch 1300, loss[loss=0.1375, simple_loss=0.1706, pruned_loss=0.05217, over 14284.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1688, pruned_loss=0.0604, over 2006483.07 frames. ], batch size: 80, lr: 8.99e-03, grad_scale: 8.0 2022-12-07 19:45:07,016 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:45:41,029 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.302e+02 2.881e+02 3.480e+02 7.451e+02, threshold=5.762e+02, percent-clipped=5.0 2022-12-07 19:45:43,021 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2022-12-07 19:46:19,491 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.81 vs. limit=5.0 2022-12-07 19:46:28,405 INFO [train.py:873] (0/4) Epoch 9, batch 1400, loss[loss=0.1954, simple_loss=0.1974, pruned_loss=0.09677, over 7746.00 frames. ], tot_loss[loss=0.1445, simple_loss=0.1688, pruned_loss=0.06012, over 2008914.60 frames. ], batch size: 100, lr: 8.98e-03, grad_scale: 8.0 2022-12-07 19:46:42,964 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:47:08,293 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.184e+02 2.770e+02 3.491e+02 7.127e+02, threshold=5.539e+02, percent-clipped=2.0 2022-12-07 19:47:13,852 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6192, 3.9430, 3.3390, 4.9158, 4.2632, 4.7216, 3.9631, 3.5007], device='cuda:0'), covar=tensor([0.0577, 0.1281, 0.3880, 0.0538, 0.1502, 0.1658, 0.1205, 0.3580], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0295, 0.0277, 0.0232, 0.0295, 0.0277, 0.0255, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 19:47:37,255 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:47:52,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-07 19:47:56,375 INFO [train.py:873] (0/4) Epoch 9, batch 1500, loss[loss=0.1341, simple_loss=0.1678, pruned_loss=0.05023, over 14531.00 frames. ], tot_loss[loss=0.1428, simple_loss=0.1674, pruned_loss=0.05915, over 1964435.34 frames. ], batch size: 43, lr: 8.98e-03, grad_scale: 8.0 2022-12-07 19:48:11,232 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2022-12-07 19:48:19,403 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2022-12-07 19:48:37,572 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.345e+02 2.854e+02 3.466e+02 6.884e+02, threshold=5.708e+02, percent-clipped=1.0 2022-12-07 19:48:52,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2794, 1.5817, 3.8749, 1.7425, 4.0337, 4.0826, 3.4665, 4.6401], device='cuda:0'), covar=tensor([0.0202, 0.3032, 0.0440, 0.2367, 0.0397, 0.0419, 0.0572, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0156, 0.0153, 0.0170, 0.0167, 0.0167, 0.0134, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:48:56,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2580, 3.0588, 5.4019, 3.6522, 4.8604, 2.5597, 4.1621, 4.8063], device='cuda:0'), covar=tensor([0.0285, 0.3280, 0.0134, 0.6678, 0.0402, 0.3045, 0.0776, 0.0264], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0231, 0.0188, 0.0310, 0.0213, 0.0230, 0.0222, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 19:49:18,252 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:49:20,980 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 19:49:25,699 INFO [train.py:873] (0/4) Epoch 9, batch 1600, loss[loss=0.1461, simple_loss=0.1679, pruned_loss=0.06212, over 14244.00 frames. ], tot_loss[loss=0.1433, simple_loss=0.1674, pruned_loss=0.05956, over 1930108.54 frames. ], batch size: 25, lr: 8.97e-03, grad_scale: 8.0 2022-12-07 19:49:27,553 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:49:44,204 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 19:49:48,636 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:50:07,093 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.173e+02 2.069e+02 2.487e+02 3.323e+02 7.707e+02, threshold=4.973e+02, percent-clipped=4.0 2022-12-07 19:50:29,891 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.33 vs. limit=2.0 2022-12-07 19:50:42,628 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:50:51,482 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2022-12-07 19:50:54,732 INFO [train.py:873] (0/4) Epoch 9, batch 1700, loss[loss=0.1337, simple_loss=0.1426, pruned_loss=0.06238, over 3842.00 frames. ], tot_loss[loss=0.1431, simple_loss=0.1675, pruned_loss=0.05934, over 1986192.12 frames. ], batch size: 100, lr: 8.96e-03, grad_scale: 8.0 2022-12-07 19:51:10,790 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-07 19:51:35,609 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.316e+02 3.145e+02 3.790e+02 7.968e+02, threshold=6.289e+02, percent-clipped=7.0 2022-12-07 19:51:55,848 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7550, 3.8437, 4.0720, 3.6118, 3.9272, 3.9337, 1.5939, 3.7193], device='cuda:0'), covar=tensor([0.0291, 0.0330, 0.0387, 0.0448, 0.0318, 0.0368, 0.3105, 0.0287], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0154, 0.0130, 0.0127, 0.0185, 0.0126, 0.0151, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:51:59,355 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:52:01,978 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:52:23,526 INFO [train.py:873] (0/4) Epoch 9, batch 1800, loss[loss=0.1047, simple_loss=0.1315, pruned_loss=0.03901, over 10776.00 frames. ], tot_loss[loss=0.1426, simple_loss=0.1673, pruned_loss=0.05895, over 1953846.51 frames. ], batch size: 13, lr: 8.95e-03, grad_scale: 8.0 2022-12-07 19:52:56,826 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:52:57,604 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2605, 1.3194, 1.4549, 0.8937, 0.9253, 1.2448, 0.8632, 1.1564], device='cuda:0'), covar=tensor([0.1019, 0.2491, 0.0656, 0.2592, 0.3168, 0.0782, 0.2484, 0.1231], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0088, 0.0081, 0.0088, 0.0106, 0.0075, 0.0125, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 19:53:04,487 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.235e+02 2.778e+02 3.404e+02 6.725e+02, threshold=5.556e+02, percent-clipped=1.0 2022-12-07 19:53:30,797 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7996, 2.7807, 2.9590, 2.9177, 2.8555, 2.6148, 1.4525, 2.6446], device='cuda:0'), covar=tensor([0.0343, 0.0392, 0.0349, 0.0305, 0.0349, 0.0911, 0.2549, 0.0299], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0154, 0.0129, 0.0125, 0.0184, 0.0124, 0.0152, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:53:32,472 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7054, 2.0556, 2.0030, 2.1111, 1.8467, 2.1402, 1.7609, 1.2481], device='cuda:0'), covar=tensor([0.1280, 0.0934, 0.0824, 0.0646, 0.1109, 0.0600, 0.1533, 0.2872], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0067, 0.0056, 0.0057, 0.0087, 0.0066, 0.0090, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:53:44,450 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:47,062 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:52,234 INFO [train.py:873] (0/4) Epoch 9, batch 1900, loss[loss=0.1642, simple_loss=0.1825, pruned_loss=0.07301, over 9465.00 frames. ], tot_loss[loss=0.1443, simple_loss=0.168, pruned_loss=0.06031, over 1950725.13 frames. ], batch size: 100, lr: 8.95e-03, grad_scale: 8.0 2022-12-07 19:53:54,385 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:53:58,381 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9750, 0.9560, 0.9661, 0.7728, 0.8145, 0.5089, 0.7001, 0.7454], device='cuda:0'), covar=tensor([0.0163, 0.0186, 0.0120, 0.0194, 0.0171, 0.0434, 0.0251, 0.0424], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0013, 0.0011, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.0694e-05, 9.8538e-05, 8.7196e-05, 9.5785e-05, 8.9928e-05, 1.3620e-04, 1.1565e-04, 1.2960e-04], device='cuda:0') 2022-12-07 19:53:59,531 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 19:54:26,815 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:54:33,571 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.190e+02 2.770e+02 3.608e+02 5.661e+02, threshold=5.540e+02, percent-clipped=1.0 2022-12-07 19:54:37,191 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:54:41,703 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:55:04,161 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:55:08,002 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 19:55:12,169 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4202, 2.3806, 2.4884, 2.5190, 2.4266, 2.0315, 1.3925, 2.2090], device='cuda:0'), covar=tensor([0.0436, 0.0513, 0.0483, 0.0343, 0.0477, 0.1463, 0.2476, 0.0403], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0132, 0.0128, 0.0188, 0.0128, 0.0154, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 19:55:18,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2022-12-07 19:55:20,941 INFO [train.py:873] (0/4) Epoch 9, batch 2000, loss[loss=0.1596, simple_loss=0.1499, pruned_loss=0.08464, over 2569.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1689, pruned_loss=0.06034, over 1994305.90 frames. ], batch size: 100, lr: 8.94e-03, grad_scale: 8.0 2022-12-07 19:56:01,037 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 2.315e+02 2.841e+02 3.698e+02 9.209e+02, threshold=5.682e+02, percent-clipped=7.0 2022-12-07 19:56:02,994 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:11,092 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.28 vs. limit=5.0 2022-12-07 19:56:24,862 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:37,491 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:47,847 INFO [train.py:873] (0/4) Epoch 9, batch 2100, loss[loss=0.1167, simple_loss=0.1533, pruned_loss=0.04006, over 14245.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1687, pruned_loss=0.06045, over 1964066.80 frames. ], batch size: 46, lr: 8.93e-03, grad_scale: 8.0 2022-12-07 19:56:56,301 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:56:58,987 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4766, 1.1230, 1.1172, 0.9186, 0.8536, 1.2304, 1.3736, 1.2298], device='cuda:0'), covar=tensor([0.0880, 0.0816, 0.1210, 0.1058, 0.1364, 0.0682, 0.0707, 0.1061], device='cuda:0'), in_proj_covar=tensor([0.0012, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.3457e-05, 1.0000e-04, 8.8823e-05, 9.6975e-05, 9.1744e-05, 1.3842e-04, 1.1726e-04, 1.3246e-04], device='cuda:0') 2022-12-07 19:57:07,124 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:10,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7910, 2.3369, 2.6411, 1.5795, 2.3965, 2.4037, 2.8807, 2.3103], device='cuda:0'), covar=tensor([0.0744, 0.1466, 0.0945, 0.2176, 0.1264, 0.0950, 0.0603, 0.1479], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0187, 0.0127, 0.0123, 0.0124, 0.0130, 0.0106, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:57:12,379 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8202, 3.5993, 3.3060, 2.4146, 3.3183, 3.4265, 3.9968, 3.0339], device='cuda:0'), covar=tensor([0.0508, 0.1817, 0.1023, 0.1976, 0.0884, 0.0680, 0.0797, 0.1523], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0187, 0.0127, 0.0123, 0.0124, 0.0130, 0.0106, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 19:57:16,356 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:24,293 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1355, 3.5253, 2.8406, 4.2158, 4.0459, 4.1243, 3.3956, 2.9936], device='cuda:0'), covar=tensor([0.0653, 0.1287, 0.3582, 0.0418, 0.0789, 0.0885, 0.1283, 0.3264], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0295, 0.0276, 0.0232, 0.0293, 0.0278, 0.0254, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 19:57:28,900 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.143e+02 2.774e+02 3.519e+02 5.598e+02, threshold=5.548e+02, percent-clipped=0.0 2022-12-07 19:57:31,622 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:57:48,815 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:58:15,319 INFO [train.py:873] (0/4) Epoch 9, batch 2200, loss[loss=0.2086, simple_loss=0.207, pruned_loss=0.1051, over 8594.00 frames. ], tot_loss[loss=0.1456, simple_loss=0.1691, pruned_loss=0.06105, over 1954631.62 frames. ], batch size: 100, lr: 8.93e-03, grad_scale: 8.0 2022-12-07 19:58:42,068 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:58:56,803 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.557e+02 3.266e+02 4.324e+02 1.294e+03, threshold=6.533e+02, percent-clipped=13.0 2022-12-07 19:58:59,466 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:21,898 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:26,020 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 19:59:41,694 INFO [train.py:873] (0/4) Epoch 9, batch 2300, loss[loss=0.1673, simple_loss=0.1747, pruned_loss=0.07993, over 6908.00 frames. ], tot_loss[loss=0.1446, simple_loss=0.1684, pruned_loss=0.06038, over 1975469.15 frames. ], batch size: 100, lr: 8.92e-03, grad_scale: 8.0 2022-12-07 20:00:07,219 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:00:14,364 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:00:23,176 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.029e+02 2.536e+02 3.250e+02 8.518e+02, threshold=5.072e+02, percent-clipped=1.0 2022-12-07 20:00:56,166 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-07 20:01:09,227 INFO [train.py:873] (0/4) Epoch 9, batch 2400, loss[loss=0.1542, simple_loss=0.1581, pruned_loss=0.0751, over 4939.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1688, pruned_loss=0.06036, over 2003913.37 frames. ], batch size: 100, lr: 8.91e-03, grad_scale: 8.0 2022-12-07 20:01:10,170 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1736, 3.9186, 3.6365, 3.8161, 4.0119, 4.0553, 4.1898, 4.1065], device='cuda:0'), covar=tensor([0.0833, 0.0617, 0.1952, 0.2623, 0.0666, 0.0797, 0.0962, 0.0960], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0244, 0.0411, 0.0521, 0.0301, 0.0386, 0.0378, 0.0340], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:01:12,642 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:36,865 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:41,416 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7459, 1.3360, 3.6044, 1.5698, 3.6050, 3.8440, 2.7728, 4.0419], device='cuda:0'), covar=tensor([0.0207, 0.2958, 0.0408, 0.2275, 0.0591, 0.0343, 0.0801, 0.0159], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0156, 0.0155, 0.0170, 0.0166, 0.0165, 0.0134, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:01:47,424 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:01:49,791 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.251e+02 2.910e+02 3.795e+02 1.080e+03, threshold=5.819e+02, percent-clipped=6.0 2022-12-07 20:01:52,795 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2022-12-07 20:02:18,739 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:02:26,501 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2022-12-07 20:02:35,717 INFO [train.py:873] (0/4) Epoch 9, batch 2500, loss[loss=0.1343, simple_loss=0.1716, pruned_loss=0.04856, over 14448.00 frames. ], tot_loss[loss=0.1448, simple_loss=0.1691, pruned_loss=0.06025, over 1999869.11 frames. ], batch size: 24, lr: 8.90e-03, grad_scale: 8.0 2022-12-07 20:02:58,145 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:03:17,451 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.289e+02 2.848e+02 3.593e+02 5.714e+02, threshold=5.696e+02, percent-clipped=0.0 2022-12-07 20:03:20,531 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:03:23,838 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2213, 1.3441, 1.3707, 1.1724, 1.2342, 0.6554, 0.9711, 1.2275], device='cuda:0'), covar=tensor([0.1324, 0.0846, 0.0953, 0.1291, 0.1218, 0.1103, 0.2271, 0.1086], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0022, 0.0021, 0.0023, 0.0033, 0.0023, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2022-12-07 20:04:02,991 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:04:04,719 INFO [train.py:873] (0/4) Epoch 9, batch 2600, loss[loss=0.2064, simple_loss=0.1759, pruned_loss=0.1184, over 1219.00 frames. ], tot_loss[loss=0.144, simple_loss=0.1685, pruned_loss=0.0597, over 2018210.76 frames. ], batch size: 100, lr: 8.90e-03, grad_scale: 8.0 2022-12-07 20:04:25,157 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3533, 1.9720, 3.5355, 2.4309, 3.3434, 1.8416, 2.5248, 3.2822], device='cuda:0'), covar=tensor([0.0663, 0.4651, 0.0373, 0.6762, 0.0591, 0.4032, 0.1495, 0.0562], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0228, 0.0190, 0.0308, 0.0211, 0.0230, 0.0219, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:04:32,672 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:04:45,881 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.604e+02 3.223e+02 3.963e+02 1.117e+03, threshold=6.445e+02, percent-clipped=7.0 2022-12-07 20:05:14,926 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3466, 3.6783, 3.0263, 4.4818, 4.1240, 4.1750, 3.6872, 3.0732], device='cuda:0'), covar=tensor([0.0643, 0.1388, 0.3878, 0.0493, 0.0899, 0.1864, 0.1243, 0.3656], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0294, 0.0271, 0.0228, 0.0286, 0.0278, 0.0250, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:05:32,671 INFO [train.py:873] (0/4) Epoch 9, batch 2700, loss[loss=0.1384, simple_loss=0.1684, pruned_loss=0.05423, over 14295.00 frames. ], tot_loss[loss=0.143, simple_loss=0.1682, pruned_loss=0.05886, over 2054367.27 frames. ], batch size: 25, lr: 8.89e-03, grad_scale: 8.0 2022-12-07 20:05:36,730 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:05:54,614 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:11,781 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:14,513 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.383e+02 3.002e+02 3.565e+02 6.845e+02, threshold=6.004e+02, percent-clipped=1.0 2022-12-07 20:06:19,150 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:41,502 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5323, 1.8879, 1.9427, 1.9463, 1.7846, 2.0113, 1.6514, 1.2711], device='cuda:0'), covar=tensor([0.1487, 0.1158, 0.0670, 0.0557, 0.1144, 0.0616, 0.2153, 0.2712], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0069, 0.0056, 0.0058, 0.0087, 0.0066, 0.0091, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:06:47,901 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:06:54,053 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:07:00,707 INFO [train.py:873] (0/4) Epoch 9, batch 2800, loss[loss=0.168, simple_loss=0.1843, pruned_loss=0.07586, over 14241.00 frames. ], tot_loss[loss=0.1439, simple_loss=0.1685, pruned_loss=0.05965, over 1996082.90 frames. ], batch size: 69, lr: 8.88e-03, grad_scale: 8.0 2022-12-07 20:07:15,715 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3217, 2.2724, 2.6871, 1.7745, 1.7624, 2.4223, 1.3456, 2.4316], device='cuda:0'), covar=tensor([0.1110, 0.1719, 0.0809, 0.2138, 0.2837, 0.1140, 0.4561, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0090, 0.0084, 0.0089, 0.0108, 0.0078, 0.0126, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:07:22,806 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:07:42,345 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 2.099e+02 2.541e+02 3.230e+02 5.695e+02, threshold=5.082e+02, percent-clipped=0.0 2022-12-07 20:08:04,952 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:25,422 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9848, 1.8029, 3.1991, 2.2777, 3.0241, 1.8441, 2.5244, 2.9519], device='cuda:0'), covar=tensor([0.0942, 0.4429, 0.0498, 0.6403, 0.0718, 0.3547, 0.1174, 0.0753], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0227, 0.0191, 0.0309, 0.0212, 0.0231, 0.0218, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:08:27,862 INFO [train.py:873] (0/4) Epoch 9, batch 2900, loss[loss=0.1289, simple_loss=0.1645, pruned_loss=0.04665, over 14575.00 frames. ], tot_loss[loss=0.1432, simple_loss=0.1677, pruned_loss=0.05936, over 1951001.90 frames. ], batch size: 22, lr: 8.88e-03, grad_scale: 8.0 2022-12-07 20:08:48,396 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:50,056 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:08:55,956 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:09,074 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.350e+02 2.806e+02 3.593e+02 6.969e+02, threshold=5.612e+02, percent-clipped=3.0 2022-12-07 20:09:37,898 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:41,624 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:43,302 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:09:54,006 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8646, 2.6125, 4.9333, 3.0479, 4.5692, 2.1619, 3.4633, 4.5158], device='cuda:0'), covar=tensor([0.0520, 0.4784, 0.0385, 1.0041, 0.0547, 0.4657, 0.1460, 0.0360], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0228, 0.0190, 0.0308, 0.0210, 0.0232, 0.0218, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:09:55,508 INFO [train.py:873] (0/4) Epoch 9, batch 3000, loss[loss=0.1216, simple_loss=0.1502, pruned_loss=0.04647, over 14128.00 frames. ], tot_loss[loss=0.1434, simple_loss=0.168, pruned_loss=0.05938, over 1980573.71 frames. ], batch size: 19, lr: 8.87e-03, grad_scale: 8.0 2022-12-07 20:09:55,508 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 20:10:05,839 INFO [train.py:905] (0/4) Epoch 9, validation: loss=0.124, simple_loss=0.1667, pruned_loss=0.04063, over 857387.00 frames. 2022-12-07 20:10:05,840 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 20:10:21,179 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7089, 2.7033, 4.5447, 4.5782, 4.5034, 2.7746, 4.5381, 3.7773], device='cuda:0'), covar=tensor([0.0161, 0.0570, 0.0418, 0.0238, 0.0195, 0.0917, 0.0213, 0.0469], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0237, 0.0348, 0.0298, 0.0240, 0.0283, 0.0265, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:10:46,173 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.483e+02 2.932e+02 3.564e+02 6.010e+02, threshold=5.864e+02, percent-clipped=1.0 2022-12-07 20:11:08,759 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:15,606 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:17,920 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9871, 3.7086, 3.4747, 3.6492, 3.8879, 3.9232, 4.0208, 3.9983], device='cuda:0'), covar=tensor([0.0909, 0.0628, 0.1972, 0.2462, 0.0755, 0.0791, 0.0952, 0.0807], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0239, 0.0402, 0.0515, 0.0297, 0.0387, 0.0378, 0.0336], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:11:23,833 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:11:33,373 INFO [train.py:873] (0/4) Epoch 9, batch 3100, loss[loss=0.1568, simple_loss=0.1807, pruned_loss=0.06651, over 8650.00 frames. ], tot_loss[loss=0.1437, simple_loss=0.1682, pruned_loss=0.05962, over 1972483.92 frames. ], batch size: 100, lr: 8.86e-03, grad_scale: 8.0 2022-12-07 20:11:48,410 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-07 20:12:03,466 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:12:15,373 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.304e+02 2.853e+02 3.296e+02 8.516e+02, threshold=5.706e+02, percent-clipped=4.0 2022-12-07 20:12:18,633 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:12:31,303 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9131, 4.5556, 4.2280, 4.3960, 4.5268, 4.8001, 4.8356, 4.8786], device='cuda:0'), covar=tensor([0.0791, 0.0503, 0.2129, 0.2822, 0.0713, 0.0693, 0.0911, 0.0783], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0243, 0.0408, 0.0528, 0.0303, 0.0392, 0.0384, 0.0341], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:12:35,654 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 20:13:02,478 INFO [train.py:873] (0/4) Epoch 9, batch 3200, loss[loss=0.1471, simple_loss=0.1745, pruned_loss=0.05988, over 14405.00 frames. ], tot_loss[loss=0.1456, simple_loss=0.1692, pruned_loss=0.06098, over 1937841.69 frames. ], batch size: 53, lr: 8.86e-03, grad_scale: 8.0 2022-12-07 20:13:43,328 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 2.396e+02 2.898e+02 3.837e+02 7.541e+02, threshold=5.796e+02, percent-clipped=6.0 2022-12-07 20:13:47,651 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2539, 2.5182, 4.2889, 4.2996, 4.2514, 2.4989, 4.2449, 3.2613], device='cuda:0'), covar=tensor([0.0231, 0.0638, 0.0451, 0.0271, 0.0216, 0.1000, 0.0235, 0.0594], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0236, 0.0348, 0.0299, 0.0240, 0.0284, 0.0265, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:14:11,313 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:14:13,013 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:14:18,686 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8381, 2.5038, 3.6369, 2.7586, 3.7120, 3.5582, 3.3742, 3.0272], device='cuda:0'), covar=tensor([0.0715, 0.2724, 0.0969, 0.1884, 0.0687, 0.0815, 0.1867, 0.1832], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0318, 0.0391, 0.0304, 0.0377, 0.0309, 0.0364, 0.0325], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:14:29,970 INFO [train.py:873] (0/4) Epoch 9, batch 3300, loss[loss=0.1127, simple_loss=0.1262, pruned_loss=0.0496, over 2578.00 frames. ], tot_loss[loss=0.1444, simple_loss=0.1688, pruned_loss=0.06001, over 1994539.73 frames. ], batch size: 100, lr: 8.85e-03, grad_scale: 8.0 2022-12-07 20:14:32,926 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2202, 1.6080, 1.7976, 1.7221, 1.6434, 1.8154, 1.4037, 1.2284], device='cuda:0'), covar=tensor([0.1588, 0.1184, 0.0550, 0.0530, 0.1169, 0.0533, 0.1880, 0.2487], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0070, 0.0055, 0.0058, 0.0087, 0.0067, 0.0092, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:14:34,632 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1805, 1.6045, 1.7465, 1.7010, 1.5942, 1.7828, 1.3595, 1.2930], device='cuda:0'), covar=tensor([0.1400, 0.0773, 0.0293, 0.0405, 0.0984, 0.0560, 0.1615, 0.1651], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0069, 0.0055, 0.0058, 0.0087, 0.0067, 0.0092, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:14:41,646 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:10,959 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.144e+02 2.539e+02 3.073e+02 7.001e+02, threshold=5.078e+02, percent-clipped=2.0 2022-12-07 20:15:11,965 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:14,353 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2022-12-07 20:15:35,035 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:40,211 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:15:46,076 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3791, 1.0300, 1.3363, 0.8915, 1.0757, 1.4547, 1.0787, 1.0979], device='cuda:0'), covar=tensor([0.0491, 0.1003, 0.0784, 0.0581, 0.1207, 0.0785, 0.0481, 0.1380], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0185, 0.0131, 0.0124, 0.0124, 0.0131, 0.0107, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:15:49,731 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4351, 2.3749, 2.5185, 2.4584, 2.4428, 2.0316, 1.3688, 2.2425], device='cuda:0'), covar=tensor([0.0454, 0.0509, 0.0502, 0.0427, 0.0449, 0.1539, 0.2563, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0161, 0.0133, 0.0130, 0.0189, 0.0130, 0.0157, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:15:56,935 INFO [train.py:873] (0/4) Epoch 9, batch 3400, loss[loss=0.1574, simple_loss=0.1718, pruned_loss=0.07145, over 10332.00 frames. ], tot_loss[loss=0.1435, simple_loss=0.1681, pruned_loss=0.0595, over 1934674.21 frames. ], batch size: 100, lr: 8.84e-03, grad_scale: 8.0 2022-12-07 20:16:05,380 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:14,178 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:16,111 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-07 20:16:20,872 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8923, 1.3729, 3.0622, 1.4507, 3.1634, 2.9994, 2.1553, 3.1694], device='cuda:0'), covar=tensor([0.0253, 0.2580, 0.0316, 0.2072, 0.0291, 0.0426, 0.1075, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0155, 0.0156, 0.0171, 0.0168, 0.0165, 0.0134, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:16:21,656 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:21,704 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:36,672 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:16:38,551 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.128e+02 2.612e+02 3.647e+02 5.829e+02, threshold=5.224e+02, percent-clipped=2.0 2022-12-07 20:17:06,697 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8286, 1.3881, 3.3596, 3.0818, 3.2535, 3.3556, 2.6728, 3.3576], device='cuda:0'), covar=tensor([0.1323, 0.1541, 0.0119, 0.0275, 0.0212, 0.0128, 0.0315, 0.0145], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0156, 0.0118, 0.0160, 0.0136, 0.0132, 0.0113, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:17:07,606 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:17:25,060 INFO [train.py:873] (0/4) Epoch 9, batch 3500, loss[loss=0.1441, simple_loss=0.1707, pruned_loss=0.05873, over 14441.00 frames. ], tot_loss[loss=0.1424, simple_loss=0.1672, pruned_loss=0.05881, over 1940278.33 frames. ], batch size: 51, lr: 8.83e-03, grad_scale: 8.0 2022-12-07 20:17:48,235 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:06,279 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.173e+02 2.574e+02 3.096e+02 7.573e+02, threshold=5.148e+02, percent-clipped=3.0 2022-12-07 20:18:11,881 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2841, 1.2638, 3.3830, 1.4379, 3.1316, 3.3543, 2.1586, 3.5373], device='cuda:0'), covar=tensor([0.0299, 0.3381, 0.0399, 0.2559, 0.0790, 0.0465, 0.1134, 0.0280], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0155, 0.0154, 0.0169, 0.0166, 0.0164, 0.0134, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:18:32,008 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4566, 1.6093, 2.8275, 2.0804, 2.5529, 1.7246, 2.1024, 2.3342], device='cuda:0'), covar=tensor([0.1455, 0.4890, 0.0451, 0.5171, 0.0955, 0.3966, 0.1651, 0.0838], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0231, 0.0192, 0.0309, 0.0213, 0.0231, 0.0220, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:18:33,705 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:35,455 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:35,838 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 20:18:40,761 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:18:51,457 INFO [train.py:873] (0/4) Epoch 9, batch 3600, loss[loss=0.1232, simple_loss=0.152, pruned_loss=0.04725, over 14170.00 frames. ], tot_loss[loss=0.1425, simple_loss=0.1667, pruned_loss=0.05916, over 1951892.29 frames. ], batch size: 37, lr: 8.83e-03, grad_scale: 8.0 2022-12-07 20:18:59,845 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3558, 2.1032, 3.3261, 3.4862, 3.4080, 2.1920, 3.3048, 2.6346], device='cuda:0'), covar=tensor([0.0267, 0.0630, 0.0505, 0.0275, 0.0243, 0.0936, 0.0305, 0.0619], device='cuda:0'), in_proj_covar=tensor([0.0257, 0.0234, 0.0346, 0.0295, 0.0238, 0.0282, 0.0264, 0.0263], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:19:04,813 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3068, 4.3935, 4.6469, 4.0192, 4.3735, 4.6522, 1.6981, 4.1941], device='cuda:0'), covar=tensor([0.0249, 0.0264, 0.0329, 0.0501, 0.0263, 0.0169, 0.3253, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0158, 0.0131, 0.0127, 0.0186, 0.0127, 0.0154, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:19:15,029 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:19:16,705 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:19:20,158 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 20:19:32,556 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.135e+01 2.573e+02 3.171e+02 4.040e+02 8.310e+02, threshold=6.341e+02, percent-clipped=10.0 2022-12-07 20:19:38,183 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2767, 1.4921, 2.6370, 1.9082, 2.4241, 1.5656, 2.1092, 2.2669], device='cuda:0'), covar=tensor([0.1091, 0.4957, 0.0545, 0.4902, 0.0946, 0.4129, 0.1332, 0.0727], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0229, 0.0192, 0.0308, 0.0213, 0.0231, 0.0219, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:19:51,901 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:19:56,486 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1134, 3.5634, 2.8159, 4.2811, 3.9862, 4.0337, 3.3122, 2.7935], device='cuda:0'), covar=tensor([0.0862, 0.1419, 0.4158, 0.0502, 0.1071, 0.2334, 0.1430, 0.3832], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0293, 0.0273, 0.0231, 0.0293, 0.0286, 0.0257, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 20:20:18,991 INFO [train.py:873] (0/4) Epoch 9, batch 3700, loss[loss=0.1536, simple_loss=0.1446, pruned_loss=0.08127, over 1227.00 frames. ], tot_loss[loss=0.1433, simple_loss=0.1674, pruned_loss=0.0596, over 1960454.49 frames. ], batch size: 100, lr: 8.82e-03, grad_scale: 8.0 2022-12-07 20:20:22,563 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:44,492 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:50,612 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:20:54,373 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-07 20:20:59,017 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:00,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.090e+02 2.323e+02 2.858e+02 3.638e+02 5.583e+02, threshold=5.716e+02, percent-clipped=0.0 2022-12-07 20:21:14,397 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3775, 1.4853, 1.5071, 1.6011, 1.4576, 1.0898, 1.1365, 0.9298], device='cuda:0'), covar=tensor([0.0631, 0.0499, 0.0443, 0.0416, 0.0571, 0.0298, 0.0328, 0.0711], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.4664e-05, 1.0160e-04, 8.9716e-05, 9.7764e-05, 9.3491e-05, 1.4169e-04, 1.1752e-04, 1.3392e-04], device='cuda:0') 2022-12-07 20:21:25,187 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:25,988 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:34,187 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4314, 1.4979, 1.3717, 1.5149, 1.2258, 1.2241, 1.1499, 1.0177], device='cuda:0'), covar=tensor([0.0524, 0.0546, 0.0761, 0.0475, 0.0580, 0.0398, 0.0277, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0014, 0.0012, 0.0013, 0.0012, 0.0020, 0.0016, 0.0020], device='cuda:0'), out_proj_covar=tensor([9.4901e-05, 1.0181e-04, 9.0122e-05, 9.8019e-05, 9.3524e-05, 1.4221e-04, 1.1761e-04, 1.3443e-04], device='cuda:0') 2022-12-07 20:21:41,382 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:43,985 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:21:47,277 INFO [train.py:873] (0/4) Epoch 9, batch 3800, loss[loss=0.1342, simple_loss=0.1607, pruned_loss=0.05389, over 13544.00 frames. ], tot_loss[loss=0.1431, simple_loss=0.1675, pruned_loss=0.05936, over 1999358.01 frames. ], batch size: 100, lr: 8.81e-03, grad_scale: 8.0 2022-12-07 20:22:28,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.269e+02 2.644e+02 3.472e+02 7.922e+02, threshold=5.288e+02, percent-clipped=2.0 2022-12-07 20:22:29,677 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:22:59,870 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:23:00,760 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8367, 1.2649, 2.0737, 1.1784, 1.9915, 2.0337, 1.6736, 2.1174], device='cuda:0'), covar=tensor([0.0308, 0.1848, 0.0291, 0.1716, 0.0405, 0.0477, 0.0923, 0.0303], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0156, 0.0156, 0.0168, 0.0167, 0.0166, 0.0134, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:23:01,745 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8185, 2.6809, 1.9373, 2.8393, 2.5907, 2.6997, 2.4101, 2.1626], device='cuda:0'), covar=tensor([0.1256, 0.1495, 0.3709, 0.0743, 0.1354, 0.1083, 0.1730, 0.3392], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0297, 0.0277, 0.0231, 0.0294, 0.0288, 0.0258, 0.0263], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 20:23:04,137 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2022-12-07 20:23:11,173 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4709, 1.9409, 2.4328, 2.5311, 2.3663, 1.9961, 2.5961, 2.1036], device='cuda:0'), covar=tensor([0.0294, 0.0570, 0.0347, 0.0288, 0.0319, 0.0769, 0.0261, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0240, 0.0355, 0.0301, 0.0241, 0.0288, 0.0271, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:23:16,036 INFO [train.py:873] (0/4) Epoch 9, batch 3900, loss[loss=0.1449, simple_loss=0.1713, pruned_loss=0.05925, over 14642.00 frames. ], tot_loss[loss=0.1432, simple_loss=0.1675, pruned_loss=0.05942, over 1939356.47 frames. ], batch size: 33, lr: 8.81e-03, grad_scale: 8.0 2022-12-07 20:23:23,816 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 20:23:24,367 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:23:57,837 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.154e+02 2.432e+02 3.241e+02 6.644e+02, threshold=4.864e+02, percent-clipped=3.0 2022-12-07 20:24:17,333 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:24:18,094 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:24:45,108 INFO [train.py:873] (0/4) Epoch 9, batch 4000, loss[loss=0.1362, simple_loss=0.1684, pruned_loss=0.05199, over 14151.00 frames. ], tot_loss[loss=0.143, simple_loss=0.1675, pruned_loss=0.05924, over 1945799.53 frames. ], batch size: 84, lr: 8.80e-03, grad_scale: 8.0 2022-12-07 20:24:45,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2092, 2.1556, 5.0251, 4.5118, 4.4318, 5.1691, 4.9159, 5.1694], device='cuda:0'), covar=tensor([0.1231, 0.1217, 0.0068, 0.0134, 0.0154, 0.0078, 0.0082, 0.0100], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0119, 0.0162, 0.0139, 0.0134, 0.0114, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:24:48,764 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:24:58,383 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-07 20:25:00,317 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:11,186 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:21,022 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3322, 2.9964, 2.9018, 1.9318, 2.7862, 3.0608, 3.3364, 2.4470], device='cuda:0'), covar=tensor([0.0643, 0.1214, 0.1092, 0.1810, 0.0926, 0.0443, 0.0701, 0.1603], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0186, 0.0132, 0.0125, 0.0125, 0.0134, 0.0108, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:25:26,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.384e+02 2.847e+02 3.821e+02 7.495e+02, threshold=5.695e+02, percent-clipped=8.0 2022-12-07 20:25:31,523 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:25:51,461 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:06,368 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:07,575 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:13,039 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 20:26:14,196 INFO [train.py:873] (0/4) Epoch 9, batch 4100, loss[loss=0.173, simple_loss=0.19, pruned_loss=0.07802, over 14160.00 frames. ], tot_loss[loss=0.1429, simple_loss=0.1675, pruned_loss=0.05914, over 1976375.48 frames. ], batch size: 99, lr: 8.79e-03, grad_scale: 8.0 2022-12-07 20:26:34,414 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:26:55,696 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.345e+01 2.214e+02 2.881e+02 3.617e+02 6.891e+02, threshold=5.761e+02, percent-clipped=1.0 2022-12-07 20:26:59,532 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:01,300 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:12,852 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2022-12-07 20:27:24,518 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 20:27:26,384 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:40,699 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 20:27:41,954 INFO [train.py:873] (0/4) Epoch 9, batch 4200, loss[loss=0.1221, simple_loss=0.1545, pruned_loss=0.04485, over 14019.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1674, pruned_loss=0.05865, over 2005138.27 frames. ], batch size: 19, lr: 8.79e-03, grad_scale: 8.0 2022-12-07 20:27:45,406 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:47,023 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:27:52,555 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:04,691 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-07 20:28:07,767 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:23,777 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.280e+02 2.745e+02 3.613e+02 6.444e+02, threshold=5.491e+02, percent-clipped=3.0 2022-12-07 20:28:23,986 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:28,269 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2392, 1.8933, 2.3035, 1.4329, 1.9967, 2.2214, 2.3201, 1.9825], device='cuda:0'), covar=tensor([0.0683, 0.0817, 0.0848, 0.1736, 0.1072, 0.0646, 0.0520, 0.1504], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0185, 0.0131, 0.0124, 0.0124, 0.0132, 0.0107, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:28:30,329 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:32,018 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9327, 1.5567, 1.9292, 1.2906, 1.6431, 1.9684, 1.8636, 1.6176], device='cuda:0'), covar=tensor([0.0768, 0.1132, 0.0847, 0.1344, 0.1197, 0.0636, 0.0562, 0.1603], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0185, 0.0131, 0.0125, 0.0124, 0.0132, 0.0107, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:28:40,213 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:28:50,300 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 20:29:01,583 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7103, 2.6146, 4.7589, 3.1000, 4.3723, 2.2837, 3.6994, 4.4406], device='cuda:0'), covar=tensor([0.0418, 0.3749, 0.0331, 0.7046, 0.0448, 0.3438, 0.1046, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0227, 0.0194, 0.0308, 0.0215, 0.0233, 0.0223, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:29:09,365 INFO [train.py:873] (0/4) Epoch 9, batch 4300, loss[loss=0.1495, simple_loss=0.1727, pruned_loss=0.06311, over 10376.00 frames. ], tot_loss[loss=0.1436, simple_loss=0.1679, pruned_loss=0.05958, over 1996737.48 frames. ], batch size: 100, lr: 8.78e-03, grad_scale: 8.0 2022-12-07 20:29:17,569 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:23,529 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:30,631 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:29:51,886 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.326e+02 2.910e+02 3.508e+02 7.763e+02, threshold=5.820e+02, percent-clipped=3.0 2022-12-07 20:30:09,860 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1897, 2.1394, 4.2064, 2.8137, 3.9400, 1.9364, 3.1040, 4.0081], device='cuda:0'), covar=tensor([0.0537, 0.4200, 0.0377, 0.6882, 0.0574, 0.3704, 0.1258, 0.0353], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0229, 0.0194, 0.0308, 0.0215, 0.0233, 0.0222, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:30:23,327 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 20:30:29,595 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:30:37,025 INFO [train.py:873] (0/4) Epoch 9, batch 4400, loss[loss=0.1478, simple_loss=0.1641, pruned_loss=0.06578, over 6006.00 frames. ], tot_loss[loss=0.1419, simple_loss=0.167, pruned_loss=0.05846, over 2025176.73 frames. ], batch size: 100, lr: 8.77e-03, grad_scale: 8.0 2022-12-07 20:31:10,758 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:31:15,371 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5788, 3.6467, 3.8369, 3.5000, 3.7100, 3.7353, 1.4494, 3.4891], device='cuda:0'), covar=tensor([0.0288, 0.0358, 0.0420, 0.0489, 0.0333, 0.0412, 0.3340, 0.0295], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0158, 0.0134, 0.0127, 0.0189, 0.0128, 0.0154, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:31:15,475 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7380, 1.9517, 3.7841, 2.6303, 3.6537, 1.9290, 2.8600, 3.6003], device='cuda:0'), covar=tensor([0.0713, 0.4875, 0.0572, 0.6955, 0.0556, 0.3825, 0.1500, 0.0439], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0228, 0.0193, 0.0305, 0.0215, 0.0231, 0.0219, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:31:18,743 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.162e+02 2.178e+02 2.781e+02 3.445e+02 8.866e+02, threshold=5.563e+02, percent-clipped=5.0 2022-12-07 20:31:18,861 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:31:29,631 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.58 vs. limit=2.0 2022-12-07 20:31:45,934 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:04,726 INFO [train.py:873] (0/4) Epoch 9, batch 4500, loss[loss=0.1353, simple_loss=0.1639, pruned_loss=0.05336, over 14500.00 frames. ], tot_loss[loss=0.1403, simple_loss=0.1657, pruned_loss=0.05742, over 1973255.51 frames. ], batch size: 51, lr: 8.77e-03, grad_scale: 8.0 2022-12-07 20:32:07,629 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-65000.pt 2022-12-07 20:32:11,683 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:14,130 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:43,527 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:32:44,217 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5031, 3.2143, 3.2168, 3.4246, 3.3072, 3.4391, 3.5003, 2.9158], device='cuda:0'), covar=tensor([0.0472, 0.1010, 0.0447, 0.0558, 0.0802, 0.0376, 0.0606, 0.0655], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0253, 0.0175, 0.0170, 0.0170, 0.0140, 0.0260, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 20:32:45,191 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2198, 3.7326, 2.9885, 4.4430, 4.1082, 4.1433, 3.7182, 3.0709], device='cuda:0'), covar=tensor([0.0949, 0.1469, 0.4212, 0.0592, 0.0922, 0.1991, 0.1217, 0.3871], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0299, 0.0277, 0.0231, 0.0295, 0.0289, 0.0257, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 20:32:50,125 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.276e+02 2.761e+02 3.494e+02 6.954e+02, threshold=5.522e+02, percent-clipped=1.0 2022-12-07 20:32:53,590 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:32:58,614 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:02,169 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:35,181 INFO [train.py:873] (0/4) Epoch 9, batch 4600, loss[loss=0.1631, simple_loss=0.1758, pruned_loss=0.0752, over 7803.00 frames. ], tot_loss[loss=0.1409, simple_loss=0.1662, pruned_loss=0.05779, over 1985274.57 frames. ], batch size: 100, lr: 8.76e-03, grad_scale: 8.0 2022-12-07 20:33:39,018 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:45,152 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:52,363 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:33:56,562 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:34:17,223 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.207e+01 2.339e+02 3.104e+02 3.874e+02 1.034e+03, threshold=6.208e+02, percent-clipped=6.0 2022-12-07 20:34:38,794 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:34:50,348 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0059, 1.0629, 0.9069, 0.9824, 1.1322, 0.3862, 0.8682, 1.0464], device='cuda:0'), covar=tensor([0.0397, 0.0481, 0.0274, 0.0361, 0.0233, 0.0274, 0.0893, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0024, 0.0022, 0.0024, 0.0034, 0.0023, 0.0025], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 20:35:03,850 INFO [train.py:873] (0/4) Epoch 9, batch 4700, loss[loss=0.1782, simple_loss=0.1704, pruned_loss=0.09295, over 2611.00 frames. ], tot_loss[loss=0.1424, simple_loss=0.167, pruned_loss=0.05889, over 1920184.38 frames. ], batch size: 100, lr: 8.75e-03, grad_scale: 8.0 2022-12-07 20:35:31,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2022-12-07 20:35:36,793 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8855, 1.5414, 3.2712, 3.0039, 3.1606, 3.3172, 2.5643, 3.2928], device='cuda:0'), covar=tensor([0.1187, 0.1307, 0.0112, 0.0226, 0.0219, 0.0120, 0.0297, 0.0131], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0156, 0.0120, 0.0164, 0.0140, 0.0135, 0.0114, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:35:46,990 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.284e+02 2.894e+02 3.637e+02 5.922e+02, threshold=5.788e+02, percent-clipped=0.0 2022-12-07 20:35:47,131 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:21,034 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1589, 1.0205, 1.0572, 1.0443, 1.1708, 0.6081, 0.9694, 1.0490], device='cuda:0'), covar=tensor([0.0528, 0.0819, 0.0876, 0.0524, 0.0610, 0.0728, 0.0881, 0.0722], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0024, 0.0022, 0.0024, 0.0034, 0.0023, 0.0025], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 20:36:21,896 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:28,419 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:31,781 INFO [train.py:873] (0/4) Epoch 9, batch 4800, loss[loss=0.1981, simple_loss=0.1653, pruned_loss=0.1155, over 1322.00 frames. ], tot_loss[loss=0.1421, simple_loss=0.1664, pruned_loss=0.05889, over 1923507.91 frames. ], batch size: 100, lr: 8.75e-03, grad_scale: 8.0 2022-12-07 20:36:38,336 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:36:58,484 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7602, 3.7548, 3.6185, 3.4926, 2.6761, 3.6804, 3.7172, 1.8475], device='cuda:0'), covar=tensor([0.2203, 0.0944, 0.1767, 0.1191, 0.1248, 0.0418, 0.1206, 0.2980], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0070, 0.0055, 0.0059, 0.0088, 0.0067, 0.0092, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 20:37:02,587 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:37:03,528 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5343, 3.1535, 2.4755, 3.6684, 3.4773, 3.4313, 3.0586, 2.4553], device='cuda:0'), covar=tensor([0.1236, 0.1835, 0.4704, 0.0664, 0.0978, 0.1934, 0.1603, 0.5045], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0292, 0.0273, 0.0228, 0.0291, 0.0285, 0.0253, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 20:37:05,127 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8087, 2.1765, 2.8751, 2.9632, 2.7814, 2.1522, 2.9099, 2.4500], device='cuda:0'), covar=tensor([0.0215, 0.0490, 0.0322, 0.0231, 0.0241, 0.0668, 0.0206, 0.0448], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0238, 0.0349, 0.0302, 0.0241, 0.0284, 0.0269, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:37:13,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.162e+02 2.794e+02 3.355e+02 5.841e+02, threshold=5.587e+02, percent-clipped=1.0 2022-12-07 20:37:14,947 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:20,542 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:26,003 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:37:58,499 INFO [train.py:873] (0/4) Epoch 9, batch 4900, loss[loss=0.1356, simple_loss=0.1709, pruned_loss=0.05017, over 14289.00 frames. ], tot_loss[loss=0.1423, simple_loss=0.1672, pruned_loss=0.05875, over 1967723.67 frames. ], batch size: 25, lr: 8.74e-03, grad_scale: 8.0 2022-12-07 20:38:02,295 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:07,251 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:08,185 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:10,657 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:40,336 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.955e+01 2.399e+02 2.803e+02 3.609e+02 7.825e+02, threshold=5.606e+02, percent-clipped=2.0 2022-12-07 20:38:43,762 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:49,492 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:38:59,913 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-07 20:39:25,267 INFO [train.py:873] (0/4) Epoch 9, batch 5000, loss[loss=0.1356, simple_loss=0.163, pruned_loss=0.05413, over 14240.00 frames. ], tot_loss[loss=0.1419, simple_loss=0.167, pruned_loss=0.05835, over 1964315.57 frames. ], batch size: 76, lr: 8.73e-03, grad_scale: 8.0 2022-12-07 20:39:56,532 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:40:07,679 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.681e+01 2.367e+02 2.989e+02 3.809e+02 1.039e+03, threshold=5.978e+02, percent-clipped=5.0 2022-12-07 20:40:40,459 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4347, 3.4989, 3.6450, 3.2745, 3.5228, 3.3355, 1.4720, 3.3176], device='cuda:0'), covar=tensor([0.0289, 0.0311, 0.0355, 0.0459, 0.0302, 0.0528, 0.3020, 0.0264], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0156, 0.0132, 0.0127, 0.0188, 0.0128, 0.0153, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:40:50,524 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:40:52,233 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2301, 1.7497, 3.4493, 2.3692, 3.2546, 1.7892, 2.5747, 3.1439], device='cuda:0'), covar=tensor([0.0863, 0.4633, 0.0496, 0.6814, 0.0717, 0.3826, 0.1462, 0.0787], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0224, 0.0194, 0.0308, 0.0216, 0.0230, 0.0220, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:40:53,898 INFO [train.py:873] (0/4) Epoch 9, batch 5100, loss[loss=0.1689, simple_loss=0.155, pruned_loss=0.0914, over 2609.00 frames. ], tot_loss[loss=0.1409, simple_loss=0.1665, pruned_loss=0.05771, over 1992892.65 frames. ], batch size: 100, lr: 8.73e-03, grad_scale: 8.0 2022-12-07 20:41:25,181 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:41:32,988 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:41:36,452 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.306e+02 2.958e+02 3.852e+02 6.809e+02, threshold=5.917e+02, percent-clipped=2.0 2022-12-07 20:42:07,903 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:42:22,489 INFO [train.py:873] (0/4) Epoch 9, batch 5200, loss[loss=0.139, simple_loss=0.1712, pruned_loss=0.05344, over 14523.00 frames. ], tot_loss[loss=0.1424, simple_loss=0.1677, pruned_loss=0.05861, over 2016610.37 frames. ], batch size: 34, lr: 8.72e-03, grad_scale: 8.0 2022-12-07 20:42:26,176 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8622, 1.3242, 2.0546, 1.2820, 2.0430, 2.0678, 1.6245, 2.1321], device='cuda:0'), covar=tensor([0.0289, 0.1682, 0.0345, 0.1545, 0.0418, 0.0553, 0.1089, 0.0315], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0156, 0.0155, 0.0169, 0.0166, 0.0168, 0.0134, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:42:35,325 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:42:50,635 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-07 20:43:02,987 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.08 vs. limit=2.0 2022-12-07 20:43:04,810 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.172e+02 3.153e+02 3.803e+02 8.139e+02, threshold=6.306e+02, percent-clipped=3.0 2022-12-07 20:43:17,697 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:43:50,698 INFO [train.py:873] (0/4) Epoch 9, batch 5300, loss[loss=0.1303, simple_loss=0.1638, pruned_loss=0.04842, over 14250.00 frames. ], tot_loss[loss=0.1428, simple_loss=0.1678, pruned_loss=0.05894, over 1987622.24 frames. ], batch size: 39, lr: 8.71e-03, grad_scale: 8.0 2022-12-07 20:43:55,955 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4419, 2.4290, 1.9697, 2.5007, 2.2700, 2.3604, 2.1931, 2.0786], device='cuda:0'), covar=tensor([0.0547, 0.0626, 0.1498, 0.0413, 0.0609, 0.0485, 0.1050, 0.1364], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0294, 0.0274, 0.0228, 0.0289, 0.0284, 0.0256, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 20:44:26,688 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:44:32,252 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.588e+01 2.128e+02 2.955e+02 3.598e+02 1.062e+03, threshold=5.909e+02, percent-clipped=4.0 2022-12-07 20:44:56,689 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-07 20:45:09,776 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:45:17,383 INFO [train.py:873] (0/4) Epoch 9, batch 5400, loss[loss=0.1346, simple_loss=0.166, pruned_loss=0.05159, over 14228.00 frames. ], tot_loss[loss=0.1426, simple_loss=0.1675, pruned_loss=0.05883, over 1943138.19 frames. ], batch size: 80, lr: 8.71e-03, grad_scale: 8.0 2022-12-07 20:45:19,399 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:45:39,214 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0015, 1.9402, 1.9049, 2.0475, 1.9035, 1.7409, 1.3089, 1.6967], device='cuda:0'), covar=tensor([0.0594, 0.0650, 0.0735, 0.0396, 0.0629, 0.1433, 0.2678, 0.0578], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0156, 0.0131, 0.0127, 0.0188, 0.0129, 0.0153, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:45:39,298 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8271, 1.6960, 1.9207, 1.7275, 2.0908, 1.8050, 1.7188, 1.7887], device='cuda:0'), covar=tensor([0.0627, 0.1196, 0.0221, 0.0415, 0.0246, 0.0668, 0.0224, 0.0275], device='cuda:0'), in_proj_covar=tensor([0.0337, 0.0317, 0.0398, 0.0308, 0.0379, 0.0314, 0.0361, 0.0321], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 20:45:56,470 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:46:00,157 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.009e+02 2.028e+02 2.560e+02 3.349e+02 6.726e+02, threshold=5.119e+02, percent-clipped=1.0 2022-12-07 20:46:38,491 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:46:45,301 INFO [train.py:873] (0/4) Epoch 9, batch 5500, loss[loss=0.1765, simple_loss=0.1548, pruned_loss=0.09908, over 1242.00 frames. ], tot_loss[loss=0.1414, simple_loss=0.1669, pruned_loss=0.05795, over 1992541.11 frames. ], batch size: 100, lr: 8.70e-03, grad_scale: 8.0 2022-12-07 20:47:14,135 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-07 20:47:14,337 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6624, 4.5402, 4.3711, 4.7359, 4.4048, 4.1344, 4.7789, 4.6076], device='cuda:0'), covar=tensor([0.0671, 0.0583, 0.0781, 0.0543, 0.0589, 0.0547, 0.0596, 0.0647], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0119, 0.0131, 0.0137, 0.0132, 0.0107, 0.0151, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 20:47:20,797 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 20:47:27,195 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.224e+02 2.797e+02 3.647e+02 1.046e+03, threshold=5.594e+02, percent-clipped=7.0 2022-12-07 20:48:02,427 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.49 vs. limit=2.0 2022-12-07 20:48:11,087 INFO [train.py:873] (0/4) Epoch 9, batch 5600, loss[loss=0.1403, simple_loss=0.165, pruned_loss=0.05777, over 14164.00 frames. ], tot_loss[loss=0.1447, simple_loss=0.1686, pruned_loss=0.06039, over 1907181.17 frames. ], batch size: 99, lr: 8.69e-03, grad_scale: 8.0 2022-12-07 20:48:52,439 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:48:53,078 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.262e+02 2.866e+02 3.556e+02 6.245e+02, threshold=5.731e+02, percent-clipped=4.0 2022-12-07 20:49:07,493 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:49:17,888 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6039, 3.7053, 3.8485, 3.4099, 3.7370, 3.6551, 1.4936, 3.5356], device='cuda:0'), covar=tensor([0.0293, 0.0291, 0.0388, 0.0483, 0.0323, 0.0423, 0.3259, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0156, 0.0130, 0.0127, 0.0186, 0.0128, 0.0153, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:49:32,026 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:49:37,654 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:49:40,171 INFO [train.py:873] (0/4) Epoch 9, batch 5700, loss[loss=0.121, simple_loss=0.1531, pruned_loss=0.04445, over 14212.00 frames. ], tot_loss[loss=0.1434, simple_loss=0.168, pruned_loss=0.05944, over 1938331.05 frames. ], batch size: 89, lr: 8.69e-03, grad_scale: 8.0 2022-12-07 20:49:47,572 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:50:01,181 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:50:07,697 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4301, 4.4668, 4.8370, 3.9750, 4.6054, 4.9450, 1.7023, 4.3213], device='cuda:0'), covar=tensor([0.0265, 0.0295, 0.0323, 0.0385, 0.0312, 0.0139, 0.3242, 0.0287], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0155, 0.0129, 0.0127, 0.0185, 0.0127, 0.0152, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:50:15,192 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:50:23,973 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.292e+02 2.851e+02 3.352e+02 5.991e+02, threshold=5.702e+02, percent-clipped=1.0 2022-12-07 20:50:34,213 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:51:08,643 INFO [train.py:873] (0/4) Epoch 9, batch 5800, loss[loss=0.1852, simple_loss=0.194, pruned_loss=0.08823, over 9503.00 frames. ], tot_loss[loss=0.1412, simple_loss=0.167, pruned_loss=0.0577, over 1939001.27 frames. ], batch size: 100, lr: 8.68e-03, grad_scale: 4.0 2022-12-07 20:51:28,592 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:51:51,563 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.104e+02 2.481e+02 2.990e+02 4.125e+02 8.241e+02, threshold=5.981e+02, percent-clipped=12.0 2022-12-07 20:52:16,161 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4321, 2.3824, 2.5106, 2.5336, 2.4881, 2.1659, 1.4618, 2.2088], device='cuda:0'), covar=tensor([0.0389, 0.0403, 0.0454, 0.0337, 0.0359, 0.1023, 0.2154, 0.0351], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0156, 0.0129, 0.0128, 0.0186, 0.0128, 0.0153, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:52:36,916 INFO [train.py:873] (0/4) Epoch 9, batch 5900, loss[loss=0.1564, simple_loss=0.142, pruned_loss=0.08536, over 1372.00 frames. ], tot_loss[loss=0.1407, simple_loss=0.1661, pruned_loss=0.05763, over 1912484.40 frames. ], batch size: 100, lr: 8.67e-03, grad_scale: 4.0 2022-12-07 20:53:19,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.452e+02 3.007e+02 3.915e+02 6.171e+02, threshold=6.013e+02, percent-clipped=2.0 2022-12-07 20:54:02,030 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:54:04,616 INFO [train.py:873] (0/4) Epoch 9, batch 6000, loss[loss=0.1454, simple_loss=0.156, pruned_loss=0.0674, over 4991.00 frames. ], tot_loss[loss=0.1398, simple_loss=0.1658, pruned_loss=0.05689, over 1987505.27 frames. ], batch size: 100, lr: 8.67e-03, grad_scale: 8.0 2022-12-07 20:54:04,617 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 20:54:16,409 INFO [train.py:905] (0/4) Epoch 9, validation: loss=0.1244, simple_loss=0.166, pruned_loss=0.04137, over 857387.00 frames. 2022-12-07 20:54:16,409 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 20:54:19,127 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:54:32,905 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:54:37,339 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:54:56,083 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 20:54:59,453 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.842e+01 2.449e+02 2.904e+02 3.531e+02 7.481e+02, threshold=5.809e+02, percent-clipped=4.0 2022-12-07 20:55:30,031 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9527, 1.3829, 3.8534, 1.6323, 3.8823, 4.0548, 3.0453, 4.3290], device='cuda:0'), covar=tensor([0.0207, 0.3092, 0.0412, 0.2259, 0.0360, 0.0318, 0.0627, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0159, 0.0158, 0.0170, 0.0169, 0.0171, 0.0137, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 20:55:31,863 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:55:38,137 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0652, 4.8496, 4.6713, 5.0715, 4.5782, 4.4554, 5.1252, 4.9784], device='cuda:0'), covar=tensor([0.0563, 0.0565, 0.0748, 0.0569, 0.0705, 0.0485, 0.0572, 0.0621], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0120, 0.0132, 0.0139, 0.0132, 0.0108, 0.0152, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 20:55:45,079 INFO [train.py:873] (0/4) Epoch 9, batch 6100, loss[loss=0.1346, simple_loss=0.1342, pruned_loss=0.06752, over 2618.00 frames. ], tot_loss[loss=0.1394, simple_loss=0.1651, pruned_loss=0.05682, over 1957042.29 frames. ], batch size: 100, lr: 8.66e-03, grad_scale: 8.0 2022-12-07 20:56:00,272 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:56:07,178 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9472, 3.6618, 3.6512, 3.9413, 3.7391, 3.4725, 4.0237, 3.3232], device='cuda:0'), covar=tensor([0.0515, 0.0918, 0.0415, 0.0507, 0.0820, 0.1325, 0.0577, 0.0674], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0253, 0.0177, 0.0173, 0.0172, 0.0138, 0.0259, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 20:56:28,746 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.301e+02 2.963e+02 3.616e+02 7.389e+02, threshold=5.926e+02, percent-clipped=2.0 2022-12-07 20:57:13,995 INFO [train.py:873] (0/4) Epoch 9, batch 6200, loss[loss=0.1388, simple_loss=0.1691, pruned_loss=0.05428, over 14413.00 frames. ], tot_loss[loss=0.1389, simple_loss=0.1647, pruned_loss=0.05655, over 1920193.16 frames. ], batch size: 41, lr: 8.66e-03, grad_scale: 8.0 2022-12-07 20:57:44,148 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:57:57,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.480e+01 2.278e+02 2.813e+02 3.570e+02 7.547e+02, threshold=5.626e+02, percent-clipped=4.0 2022-12-07 20:58:34,797 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2022-12-07 20:58:37,965 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:58:42,095 INFO [train.py:873] (0/4) Epoch 9, batch 6300, loss[loss=0.1375, simple_loss=0.1696, pruned_loss=0.05272, over 14071.00 frames. ], tot_loss[loss=0.139, simple_loss=0.1649, pruned_loss=0.05651, over 1968255.18 frames. ], batch size: 29, lr: 8.65e-03, grad_scale: 8.0 2022-12-07 20:58:45,138 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 20:58:47,103 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2022-12-07 20:58:59,248 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:59:24,046 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7358, 1.5985, 1.8379, 2.0826, 1.4112, 1.6240, 1.8340, 1.9759], device='cuda:0'), covar=tensor([0.0105, 0.0146, 0.0093, 0.0072, 0.0174, 0.0209, 0.0116, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0239, 0.0355, 0.0300, 0.0241, 0.0289, 0.0270, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 20:59:25,549 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.076e+02 2.178e+02 2.874e+02 3.614e+02 8.081e+02, threshold=5.749e+02, percent-clipped=6.0 2022-12-07 20:59:27,374 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 20:59:42,253 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 20:59:53,387 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:00:10,824 INFO [train.py:873] (0/4) Epoch 9, batch 6400, loss[loss=0.1469, simple_loss=0.1717, pruned_loss=0.06107, over 13530.00 frames. ], tot_loss[loss=0.14, simple_loss=0.1656, pruned_loss=0.05719, over 1937289.80 frames. ], batch size: 100, lr: 8.64e-03, grad_scale: 8.0 2022-12-07 21:00:26,482 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:00:54,936 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 2.352e+02 2.835e+02 3.387e+02 8.466e+02, threshold=5.670e+02, percent-clipped=3.0 2022-12-07 21:01:04,705 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:01:09,015 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:01:40,465 INFO [train.py:873] (0/4) Epoch 9, batch 6500, loss[loss=0.1387, simple_loss=0.1661, pruned_loss=0.05562, over 14576.00 frames. ], tot_loss[loss=0.1399, simple_loss=0.1654, pruned_loss=0.05718, over 1940805.85 frames. ], batch size: 49, lr: 8.64e-03, grad_scale: 8.0 2022-12-07 21:01:59,331 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:02:24,260 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.174e+02 2.311e+02 2.795e+02 3.453e+02 6.517e+02, threshold=5.590e+02, percent-clipped=2.0 2022-12-07 21:02:40,715 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-07 21:02:53,475 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6401, 2.1126, 3.5387, 3.7473, 3.6991, 2.3771, 3.5741, 2.7644], device='cuda:0'), covar=tensor([0.0309, 0.0741, 0.0689, 0.0371, 0.0257, 0.0986, 0.0318, 0.0685], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0239, 0.0358, 0.0300, 0.0242, 0.0291, 0.0268, 0.0269], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:02:57,871 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:03:01,184 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:03:10,249 INFO [train.py:873] (0/4) Epoch 9, batch 6600, loss[loss=0.1236, simple_loss=0.1578, pruned_loss=0.04472, over 14350.00 frames. ], tot_loss[loss=0.1411, simple_loss=0.1656, pruned_loss=0.05825, over 1877166.62 frames. ], batch size: 31, lr: 8.63e-03, grad_scale: 8.0 2022-12-07 21:03:16,148 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-07 21:03:53,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8183, 3.5978, 3.3122, 3.4452, 3.7132, 3.7075, 3.8041, 3.7728], device='cuda:0'), covar=tensor([0.0923, 0.0624, 0.2351, 0.2743, 0.0766, 0.0890, 0.1193, 0.0938], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0247, 0.0416, 0.0534, 0.0311, 0.0400, 0.0389, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:03:53,401 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:03:54,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.185e+02 2.883e+02 3.643e+02 7.006e+02, threshold=5.766e+02, percent-clipped=2.0 2022-12-07 21:04:20,080 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-07 21:04:20,674 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3323, 4.0605, 3.7724, 3.9399, 4.0630, 4.1947, 4.3002, 4.2744], device='cuda:0'), covar=tensor([0.0731, 0.0540, 0.2066, 0.2586, 0.0758, 0.0792, 0.0962, 0.0794], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0247, 0.0413, 0.0531, 0.0309, 0.0399, 0.0386, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:04:22,386 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:04:37,373 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0040, 3.5205, 2.7245, 4.1645, 4.0026, 4.0656, 3.4043, 2.8864], device='cuda:0'), covar=tensor([0.1106, 0.1605, 0.4433, 0.0604, 0.0922, 0.1252, 0.1463, 0.3960], device='cuda:0'), in_proj_covar=tensor([0.0264, 0.0304, 0.0281, 0.0237, 0.0305, 0.0294, 0.0258, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 21:04:40,785 INFO [train.py:873] (0/4) Epoch 9, batch 6700, loss[loss=0.1162, simple_loss=0.1556, pruned_loss=0.03841, over 14166.00 frames. ], tot_loss[loss=0.1405, simple_loss=0.1659, pruned_loss=0.05748, over 1913196.40 frames. ], batch size: 35, lr: 8.62e-03, grad_scale: 8.0 2022-12-07 21:05:02,730 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0536, 1.7173, 2.0956, 1.3902, 1.7738, 2.0457, 1.9964, 1.7827], device='cuda:0'), covar=tensor([0.0768, 0.0947, 0.0818, 0.1492, 0.1262, 0.0768, 0.0528, 0.1601], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0182, 0.0131, 0.0124, 0.0125, 0.0135, 0.0111, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0005], device='cuda:0') 2022-12-07 21:05:06,515 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:05:07,513 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8975, 1.2028, 2.0709, 1.2901, 2.0305, 2.0598, 1.7132, 2.1265], device='cuda:0'), covar=tensor([0.0315, 0.1719, 0.0338, 0.1345, 0.0401, 0.0445, 0.0868, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0160, 0.0160, 0.0173, 0.0172, 0.0171, 0.0137, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 21:05:19,053 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9268, 2.0220, 4.0121, 2.7795, 3.8085, 1.8464, 2.9493, 3.7544], device='cuda:0'), covar=tensor([0.0621, 0.4432, 0.0659, 0.5930, 0.0697, 0.4000, 0.1358, 0.0664], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0224, 0.0192, 0.0303, 0.0212, 0.0230, 0.0220, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:05:24,548 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.308e+02 2.851e+02 3.420e+02 6.457e+02, threshold=5.703e+02, percent-clipped=2.0 2022-12-07 21:05:28,520 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-07 21:06:10,531 INFO [train.py:873] (0/4) Epoch 9, batch 6800, loss[loss=0.1635, simple_loss=0.1796, pruned_loss=0.07366, over 11983.00 frames. ], tot_loss[loss=0.1416, simple_loss=0.1667, pruned_loss=0.05825, over 1936790.21 frames. ], batch size: 100, lr: 8.62e-03, grad_scale: 8.0 2022-12-07 21:06:25,244 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:06:54,534 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.355e+02 2.871e+02 3.825e+02 6.310e+02, threshold=5.742e+02, percent-clipped=3.0 2022-12-07 21:07:31,787 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:07:40,406 INFO [train.py:873] (0/4) Epoch 9, batch 6900, loss[loss=0.1528, simple_loss=0.1627, pruned_loss=0.07147, over 7768.00 frames. ], tot_loss[loss=0.1408, simple_loss=0.1657, pruned_loss=0.05799, over 1917563.95 frames. ], batch size: 100, lr: 8.61e-03, grad_scale: 8.0 2022-12-07 21:07:41,954 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-07 21:08:00,062 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7061, 0.7801, 0.5565, 0.7517, 0.8991, 0.1782, 0.6485, 0.7198], device='cuda:0'), covar=tensor([0.0174, 0.0347, 0.0329, 0.0245, 0.0194, 0.0170, 0.0540, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0025, 0.0022, 0.0024, 0.0034, 0.0024, 0.0025], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 21:08:06,279 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6724, 3.3696, 3.3257, 3.7028, 3.4003, 3.5739, 3.7007, 3.0717], device='cuda:0'), covar=tensor([0.0396, 0.1153, 0.0441, 0.0464, 0.0897, 0.0396, 0.0686, 0.0648], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0255, 0.0178, 0.0173, 0.0170, 0.0140, 0.0261, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 21:08:14,506 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:08:17,945 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:08:24,082 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.180e+02 2.447e+02 3.065e+02 3.875e+02 6.583e+02, threshold=6.130e+02, percent-clipped=2.0 2022-12-07 21:08:34,989 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0347, 1.6378, 4.1387, 4.0322, 3.9224, 4.3369, 3.7887, 4.2074], device='cuda:0'), covar=tensor([0.1498, 0.1663, 0.0175, 0.0226, 0.0242, 0.0151, 0.0201, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0122, 0.0165, 0.0141, 0.0135, 0.0114, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:09:09,666 INFO [train.py:873] (0/4) Epoch 9, batch 7000, loss[loss=0.1307, simple_loss=0.1591, pruned_loss=0.05115, over 14256.00 frames. ], tot_loss[loss=0.1418, simple_loss=0.1661, pruned_loss=0.05872, over 1856706.79 frames. ], batch size: 60, lr: 8.60e-03, grad_scale: 8.0 2022-12-07 21:09:24,682 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:09:53,897 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.336e+02 2.835e+02 3.380e+02 6.445e+02, threshold=5.670e+02, percent-clipped=1.0 2022-12-07 21:09:55,840 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9763, 1.9655, 4.5382, 2.2703, 4.3459, 4.7327, 4.4167, 5.3693], device='cuda:0'), covar=tensor([0.0171, 0.2742, 0.0345, 0.1997, 0.0290, 0.0380, 0.0289, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0157, 0.0157, 0.0168, 0.0170, 0.0168, 0.0136, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 21:10:19,400 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:10:40,050 INFO [train.py:873] (0/4) Epoch 9, batch 7100, loss[loss=0.1531, simple_loss=0.1483, pruned_loss=0.07893, over 2671.00 frames. ], tot_loss[loss=0.1412, simple_loss=0.1657, pruned_loss=0.05833, over 1852646.92 frames. ], batch size: 100, lr: 8.60e-03, grad_scale: 8.0 2022-12-07 21:10:55,409 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:11:18,378 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.25 vs. limit=5.0 2022-12-07 21:11:27,148 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.435e+02 3.463e+02 4.295e+02 1.529e+03, threshold=6.926e+02, percent-clipped=8.0 2022-12-07 21:11:42,158 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:00,604 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:11,910 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:16,513 INFO [train.py:873] (0/4) Epoch 9, batch 7200, loss[loss=0.1284, simple_loss=0.1576, pruned_loss=0.0496, over 14634.00 frames. ], tot_loss[loss=0.1417, simple_loss=0.166, pruned_loss=0.05868, over 1889892.37 frames. ], batch size: 22, lr: 8.59e-03, grad_scale: 8.0 2022-12-07 21:12:18,622 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7787, 0.7265, 0.6568, 0.7861, 0.6526, 0.4690, 0.4348, 0.5542], device='cuda:0'), covar=tensor([0.0098, 0.0124, 0.0119, 0.0120, 0.0175, 0.0294, 0.0214, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:0'), out_proj_covar=tensor([9.9603e-05, 1.1008e-04, 9.4371e-05, 1.0299e-04, 1.0138e-04, 1.5070e-04, 1.2727e-04, 1.4569e-04], device='cuda:0') 2022-12-07 21:12:21,848 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3798, 1.3937, 1.3880, 1.1580, 1.1721, 0.8874, 0.7756, 0.8467], device='cuda:0'), covar=tensor([0.0153, 0.0318, 0.0132, 0.0175, 0.0224, 0.0294, 0.0215, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:0'), out_proj_covar=tensor([9.9683e-05, 1.1014e-04, 9.4444e-05, 1.0308e-04, 1.0142e-04, 1.5076e-04, 1.2736e-04, 1.4579e-04], device='cuda:0') 2022-12-07 21:12:57,417 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:12:59,656 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:13:04,123 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.252e+02 2.773e+02 3.514e+02 7.822e+02, threshold=5.546e+02, percent-clipped=1.0 2022-12-07 21:13:11,065 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:13:14,926 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.53 vs. limit=2.0 2022-12-07 21:13:43,166 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:13:52,642 INFO [train.py:873] (0/4) Epoch 9, batch 7300, loss[loss=0.1391, simple_loss=0.1654, pruned_loss=0.05638, over 14400.00 frames. ], tot_loss[loss=0.1412, simple_loss=0.1656, pruned_loss=0.05842, over 1929393.36 frames. ], batch size: 41, lr: 8.58e-03, grad_scale: 8.0 2022-12-07 21:14:09,938 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.19 vs. limit=5.0 2022-12-07 21:14:39,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.094e+02 2.117e+02 2.726e+02 3.356e+02 6.858e+02, threshold=5.453e+02, percent-clipped=1.0 2022-12-07 21:15:01,812 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:15:17,084 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0476, 1.0917, 1.0240, 0.8998, 0.8331, 0.6104, 0.6982, 0.6901], device='cuda:0'), covar=tensor([0.0116, 0.0122, 0.0121, 0.0116, 0.0173, 0.0323, 0.0229, 0.0365], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.0046e-04, 1.0996e-04, 9.5558e-05, 1.0268e-04, 1.0047e-04, 1.5083e-04, 1.2694e-04, 1.4584e-04], device='cuda:0') 2022-12-07 21:15:28,019 INFO [train.py:873] (0/4) Epoch 9, batch 7400, loss[loss=0.1401, simple_loss=0.1375, pruned_loss=0.07138, over 2639.00 frames. ], tot_loss[loss=0.1408, simple_loss=0.1656, pruned_loss=0.05801, over 2013620.03 frames. ], batch size: 100, lr: 8.58e-03, grad_scale: 8.0 2022-12-07 21:15:40,857 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7766, 1.5753, 2.0667, 1.6614, 1.9510, 1.4811, 1.6743, 1.8864], device='cuda:0'), covar=tensor([0.2209, 0.2711, 0.0350, 0.1675, 0.0898, 0.1421, 0.1133, 0.0670], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0221, 0.0191, 0.0303, 0.0212, 0.0228, 0.0218, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:16:15,149 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.396e+02 2.952e+02 3.564e+02 5.471e+02, threshold=5.904e+02, percent-clipped=1.0 2022-12-07 21:16:47,299 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:02,497 INFO [train.py:873] (0/4) Epoch 9, batch 7500, loss[loss=0.1307, simple_loss=0.1586, pruned_loss=0.05145, over 14154.00 frames. ], tot_loss[loss=0.1413, simple_loss=0.1661, pruned_loss=0.05828, over 1975795.40 frames. ], batch size: 35, lr: 8.57e-03, grad_scale: 8.0 2022-12-07 21:17:38,173 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:42,609 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:17:44,527 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.588e+01 2.266e+02 2.777e+02 3.907e+02 7.467e+02, threshold=5.554e+02, percent-clipped=5.0 2022-12-07 21:17:45,640 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:17:52,896 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-9.pt 2022-12-07 21:18:40,276 INFO [train.py:873] (0/4) Epoch 10, batch 0, loss[loss=0.1811, simple_loss=0.2009, pruned_loss=0.08067, over 14485.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2009, pruned_loss=0.08067, over 14485.00 frames. ], batch size: 49, lr: 8.15e-03, grad_scale: 8.0 2022-12-07 21:18:40,278 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 21:18:48,290 INFO [train.py:905] (0/4) Epoch 10, validation: loss=0.1297, simple_loss=0.1728, pruned_loss=0.04327, over 857387.00 frames. 2022-12-07 21:18:48,290 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 21:19:16,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8913, 3.6038, 3.6719, 3.9547, 3.5460, 3.1491, 3.9357, 3.8583], device='cuda:0'), covar=tensor([0.0675, 0.0796, 0.0748, 0.0634, 0.0884, 0.0733, 0.0625, 0.0721], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0118, 0.0127, 0.0134, 0.0131, 0.0106, 0.0149, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 21:19:16,367 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:19:39,109 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0600, 2.8260, 2.5925, 2.7554, 2.9578, 2.9594, 3.0102, 2.9951], device='cuda:0'), covar=tensor([0.1070, 0.1074, 0.2626, 0.3222, 0.0982, 0.1039, 0.1512, 0.1013], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0254, 0.0417, 0.0531, 0.0313, 0.0400, 0.0390, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:19:42,345 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9656, 3.7449, 3.7060, 4.0959, 3.5836, 3.3128, 4.0430, 3.9665], device='cuda:0'), covar=tensor([0.0674, 0.0871, 0.0730, 0.0486, 0.0829, 0.0686, 0.0634, 0.0670], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0119, 0.0127, 0.0135, 0.0131, 0.0106, 0.0149, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 21:19:46,918 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4948, 1.8223, 1.6379, 1.5181, 1.5359, 0.9611, 0.6814, 0.9169], device='cuda:0'), covar=tensor([0.0267, 0.0509, 0.0298, 0.0643, 0.0347, 0.0283, 0.0293, 0.0584], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:0'), out_proj_covar=tensor([9.9646e-05, 1.0990e-04, 9.6060e-05, 1.0201e-04, 1.0100e-04, 1.4967e-04, 1.2693e-04, 1.4472e-04], device='cuda:0') 2022-12-07 21:20:06,702 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7988, 3.0633, 4.4964, 3.3567, 4.4014, 4.2481, 4.2522, 3.8531], device='cuda:0'), covar=tensor([0.0510, 0.3059, 0.1077, 0.1873, 0.0975, 0.0823, 0.1673, 0.1822], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0317, 0.0397, 0.0299, 0.0370, 0.0311, 0.0361, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:20:12,822 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.622e+01 2.492e+02 3.120e+02 3.862e+02 1.100e+03, threshold=6.240e+02, percent-clipped=9.0 2022-12-07 21:20:13,988 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:20:24,791 INFO [train.py:873] (0/4) Epoch 10, batch 100, loss[loss=0.1214, simple_loss=0.1382, pruned_loss=0.05232, over 3874.00 frames. ], tot_loss[loss=0.1416, simple_loss=0.1665, pruned_loss=0.05836, over 869028.87 frames. ], batch size: 100, lr: 8.14e-03, grad_scale: 4.0 2022-12-07 21:20:34,368 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:21:19,969 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:21:48,935 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.529e+02 3.051e+02 3.928e+02 8.118e+02, threshold=6.102e+02, percent-clipped=5.0 2022-12-07 21:21:59,799 INFO [train.py:873] (0/4) Epoch 10, batch 200, loss[loss=0.1142, simple_loss=0.1455, pruned_loss=0.04141, over 14184.00 frames. ], tot_loss[loss=0.1395, simple_loss=0.1652, pruned_loss=0.0569, over 1292333.87 frames. ], batch size: 37, lr: 8.14e-03, grad_scale: 4.0 2022-12-07 21:22:19,083 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1841, 3.4367, 3.2421, 3.3805, 2.5182, 3.4857, 3.1755, 1.6186], device='cuda:0'), covar=tensor([0.2676, 0.0930, 0.1297, 0.1056, 0.1088, 0.0471, 0.1198, 0.2841], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0071, 0.0057, 0.0060, 0.0087, 0.0068, 0.0093, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 21:22:39,711 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5884, 2.0250, 2.5525, 2.6970, 2.5266, 2.0152, 2.6338, 2.2011], device='cuda:0'), covar=tensor([0.0304, 0.0634, 0.0422, 0.0289, 0.0403, 0.0867, 0.0286, 0.0545], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0240, 0.0359, 0.0298, 0.0243, 0.0292, 0.0273, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:22:48,995 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7706, 1.1384, 1.2964, 1.2544, 1.1325, 1.3598, 1.1238, 0.9043], device='cuda:0'), covar=tensor([0.2220, 0.0888, 0.0445, 0.0282, 0.0998, 0.0535, 0.1645, 0.1200], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0072, 0.0058, 0.0060, 0.0087, 0.0068, 0.0093, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:0') 2022-12-07 21:23:09,109 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8568, 2.6634, 2.7000, 2.8508, 2.7725, 2.7901, 2.9529, 2.4531], device='cuda:0'), covar=tensor([0.0831, 0.1117, 0.0598, 0.0656, 0.0991, 0.0587, 0.0682, 0.0665], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0249, 0.0174, 0.0169, 0.0168, 0.0135, 0.0254, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 21:23:12,796 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:13,597 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:23,110 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 2.072e+02 2.682e+02 3.537e+02 7.107e+02, threshold=5.365e+02, percent-clipped=2.0 2022-12-07 21:23:24,244 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:23:33,060 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6335, 3.2309, 2.5665, 3.7705, 3.6539, 3.5944, 3.2608, 2.5719], device='cuda:0'), covar=tensor([0.0855, 0.1559, 0.3888, 0.0563, 0.0748, 0.1279, 0.1139, 0.4077], device='cuda:0'), in_proj_covar=tensor([0.0260, 0.0298, 0.0274, 0.0235, 0.0296, 0.0288, 0.0258, 0.0263], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 21:23:34,540 INFO [train.py:873] (0/4) Epoch 10, batch 300, loss[loss=0.1473, simple_loss=0.138, pruned_loss=0.07834, over 2598.00 frames. ], tot_loss[loss=0.1401, simple_loss=0.1651, pruned_loss=0.05754, over 1506505.64 frames. ], batch size: 100, lr: 8.13e-03, grad_scale: 4.0 2022-12-07 21:23:57,993 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:10,431 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:13,323 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:24:55,068 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1434, 1.9994, 1.7842, 1.8575, 2.0855, 2.0522, 2.1112, 2.0886], device='cuda:0'), covar=tensor([0.1204, 0.0920, 0.2516, 0.3067, 0.0984, 0.1037, 0.1419, 0.1092], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0248, 0.0416, 0.0528, 0.0308, 0.0398, 0.0386, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:24:56,049 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:24:59,359 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.338e+02 2.858e+02 3.586e+02 7.542e+02, threshold=5.716e+02, percent-clipped=9.0 2022-12-07 21:25:10,938 INFO [train.py:873] (0/4) Epoch 10, batch 400, loss[loss=0.1343, simple_loss=0.1649, pruned_loss=0.05185, over 6939.00 frames. ], tot_loss[loss=0.1397, simple_loss=0.1652, pruned_loss=0.05713, over 1684262.90 frames. ], batch size: 100, lr: 8.12e-03, grad_scale: 8.0 2022-12-07 21:25:12,076 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:25:20,279 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2059, 2.7827, 4.0013, 2.8153, 4.0288, 3.9447, 3.8239, 3.2015], device='cuda:0'), covar=tensor([0.0654, 0.3301, 0.1062, 0.2187, 0.0847, 0.0819, 0.1749, 0.2997], device='cuda:0'), in_proj_covar=tensor([0.0337, 0.0320, 0.0402, 0.0305, 0.0373, 0.0315, 0.0364, 0.0320], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:26:12,282 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4450, 2.0682, 3.6323, 2.5655, 3.5022, 1.9755, 2.7480, 3.3766], device='cuda:0'), covar=tensor([0.0694, 0.4281, 0.0456, 0.6300, 0.0595, 0.3647, 0.1326, 0.0519], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0225, 0.0194, 0.0302, 0.0216, 0.0226, 0.0219, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:26:35,458 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.435e+02 2.815e+02 3.563e+02 7.495e+02, threshold=5.630e+02, percent-clipped=4.0 2022-12-07 21:26:47,612 INFO [train.py:873] (0/4) Epoch 10, batch 500, loss[loss=0.1495, simple_loss=0.1525, pruned_loss=0.07321, over 3895.00 frames. ], tot_loss[loss=0.1399, simple_loss=0.1653, pruned_loss=0.05724, over 1795848.74 frames. ], batch size: 100, lr: 8.12e-03, grad_scale: 8.0 2022-12-07 21:27:05,886 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:27:33,178 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1147, 2.0134, 1.7503, 1.7848, 2.0678, 2.0176, 2.1044, 2.0436], device='cuda:0'), covar=tensor([0.1357, 0.0900, 0.2847, 0.3338, 0.1092, 0.1174, 0.1443, 0.1279], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0250, 0.0419, 0.0532, 0.0312, 0.0402, 0.0390, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:28:01,939 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:03,532 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:10,691 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.029e+02 2.141e+02 2.717e+02 3.371e+02 6.429e+02, threshold=5.434e+02, percent-clipped=1.0 2022-12-07 21:28:18,680 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7544, 2.4032, 2.5777, 1.7592, 2.3375, 2.6432, 2.8465, 2.2758], device='cuda:0'), covar=tensor([0.0699, 0.1078, 0.1090, 0.1863, 0.0924, 0.0559, 0.0466, 0.1598], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0188, 0.0135, 0.0128, 0.0128, 0.0138, 0.0114, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-07 21:28:20,760 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 21:28:21,922 INFO [train.py:873] (0/4) Epoch 10, batch 600, loss[loss=0.1783, simple_loss=0.1968, pruned_loss=0.07991, over 14301.00 frames. ], tot_loss[loss=0.1393, simple_loss=0.1652, pruned_loss=0.05667, over 1858194.18 frames. ], batch size: 35, lr: 8.11e-03, grad_scale: 8.0 2022-12-07 21:28:46,815 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:28:46,951 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:29:16,343 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1585, 1.8952, 3.2693, 2.3132, 3.2274, 1.8535, 2.5079, 3.1446], device='cuda:0'), covar=tensor([0.0754, 0.4589, 0.0642, 0.6100, 0.0704, 0.3963, 0.1543, 0.0544], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0225, 0.0196, 0.0304, 0.0216, 0.0227, 0.0221, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:29:37,016 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5123, 4.9907, 4.9464, 5.5091, 5.1254, 4.6624, 5.3990, 4.3651], device='cuda:0'), covar=tensor([0.0397, 0.1229, 0.0322, 0.0379, 0.0752, 0.0493, 0.0609, 0.0649], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0255, 0.0176, 0.0172, 0.0171, 0.0139, 0.0259, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 21:29:41,732 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:29:44,895 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4131, 3.5101, 3.2473, 3.1513, 2.4797, 3.7369, 3.1120, 1.5826], device='cuda:0'), covar=tensor([0.2535, 0.1313, 0.1677, 0.1674, 0.1249, 0.0940, 0.1647, 0.2953], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0072, 0.0058, 0.0061, 0.0088, 0.0069, 0.0094, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:0') 2022-12-07 21:29:44,939 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:29:45,571 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.288e+02 3.002e+02 3.760e+02 7.437e+02, threshold=6.005e+02, percent-clipped=6.0 2022-12-07 21:29:53,310 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:29:57,184 INFO [train.py:873] (0/4) Epoch 10, batch 700, loss[loss=0.2048, simple_loss=0.1896, pruned_loss=0.11, over 1193.00 frames. ], tot_loss[loss=0.1394, simple_loss=0.1651, pruned_loss=0.05682, over 1875667.24 frames. ], batch size: 100, lr: 8.11e-03, grad_scale: 8.0 2022-12-07 21:30:27,493 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:30:48,145 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-07 21:31:12,431 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8058, 0.7761, 0.6203, 0.8246, 0.8163, 0.1862, 0.7165, 0.7726], device='cuda:0'), covar=tensor([0.0214, 0.0264, 0.0255, 0.0211, 0.0267, 0.0147, 0.0461, 0.0436], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0025, 0.0023, 0.0023, 0.0034, 0.0024, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 21:31:20,804 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.251e+02 2.713e+02 3.475e+02 5.744e+02, threshold=5.426e+02, percent-clipped=0.0 2022-12-07 21:31:32,391 INFO [train.py:873] (0/4) Epoch 10, batch 800, loss[loss=0.17, simple_loss=0.1628, pruned_loss=0.08859, over 1232.00 frames. ], tot_loss[loss=0.1402, simple_loss=0.1652, pruned_loss=0.0576, over 1813015.39 frames. ], batch size: 100, lr: 8.10e-03, grad_scale: 8.0 2022-12-07 21:31:58,881 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7359, 1.4399, 1.4799, 1.4822, 1.4168, 0.8931, 1.5125, 1.7652], device='cuda:0'), covar=tensor([0.0687, 0.0737, 0.0676, 0.0909, 0.0945, 0.0676, 0.0768, 0.1522], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0023, 0.0025, 0.0022, 0.0023, 0.0034, 0.0024, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 21:32:43,289 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:32:43,948 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2022-12-07 21:32:55,164 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.204e+02 2.859e+02 3.964e+02 7.134e+02, threshold=5.717e+02, percent-clipped=4.0 2022-12-07 21:33:06,466 INFO [train.py:873] (0/4) Epoch 10, batch 900, loss[loss=0.1141, simple_loss=0.1503, pruned_loss=0.03898, over 6949.00 frames. ], tot_loss[loss=0.1396, simple_loss=0.1648, pruned_loss=0.05718, over 1918596.75 frames. ], batch size: 100, lr: 8.09e-03, grad_scale: 8.0 2022-12-07 21:33:50,482 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3622, 2.2322, 4.9567, 4.4888, 4.4703, 5.0850, 4.9622, 5.1597], device='cuda:0'), covar=tensor([0.1127, 0.1160, 0.0075, 0.0148, 0.0142, 0.0079, 0.0073, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0156, 0.0121, 0.0163, 0.0141, 0.0134, 0.0115, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:34:12,276 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6929, 1.8375, 1.9439, 1.3005, 1.2784, 1.8145, 1.0230, 1.7139], device='cuda:0'), covar=tensor([0.1383, 0.1602, 0.0709, 0.2033, 0.2834, 0.0809, 0.3380, 0.0852], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0093, 0.0086, 0.0093, 0.0113, 0.0080, 0.0127, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:0') 2022-12-07 21:34:24,751 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:26,757 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:30,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 2.392e+02 2.888e+02 3.362e+02 7.513e+02, threshold=5.776e+02, percent-clipped=4.0 2022-12-07 21:34:37,733 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:34:41,173 INFO [train.py:873] (0/4) Epoch 10, batch 1000, loss[loss=0.1324, simple_loss=0.1678, pruned_loss=0.04853, over 14201.00 frames. ], tot_loss[loss=0.1391, simple_loss=0.1646, pruned_loss=0.05684, over 1891656.17 frames. ], batch size: 46, lr: 8.09e-03, grad_scale: 8.0 2022-12-07 21:34:47,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5294, 2.4601, 4.6797, 3.0653, 4.2960, 2.2728, 3.3694, 4.4107], device='cuda:0'), covar=tensor([0.0498, 0.3823, 0.0261, 0.6937, 0.0582, 0.3218, 0.1207, 0.0353], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0220, 0.0194, 0.0300, 0.0214, 0.0225, 0.0218, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:35:04,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0168, 2.0524, 4.0976, 2.7685, 3.8546, 1.9992, 2.9448, 3.8459], device='cuda:0'), covar=tensor([0.0563, 0.4409, 0.0377, 0.6317, 0.0532, 0.3658, 0.1423, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0220, 0.0194, 0.0300, 0.0214, 0.0225, 0.0218, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:35:22,730 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:35:23,631 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:35:26,646 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2486, 2.2989, 5.1031, 4.7418, 4.5467, 5.1501, 5.0015, 5.3138], device='cuda:0'), covar=tensor([0.1243, 0.1182, 0.0090, 0.0138, 0.0165, 0.0089, 0.0100, 0.0079], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0157, 0.0122, 0.0165, 0.0142, 0.0136, 0.0116, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:35:46,529 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6067, 1.7138, 2.8714, 2.1182, 2.7564, 1.7932, 2.2704, 2.5998], device='cuda:0'), covar=tensor([0.1344, 0.4106, 0.0551, 0.4395, 0.0822, 0.3237, 0.1241, 0.0603], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0218, 0.0193, 0.0299, 0.0213, 0.0225, 0.0217, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:36:04,000 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.114e+02 2.200e+02 2.692e+02 3.189e+02 5.615e+02, threshold=5.384e+02, percent-clipped=0.0 2022-12-07 21:36:10,738 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-07 21:36:15,496 INFO [train.py:873] (0/4) Epoch 10, batch 1100, loss[loss=0.1071, simple_loss=0.1443, pruned_loss=0.03499, over 14337.00 frames. ], tot_loss[loss=0.1395, simple_loss=0.1645, pruned_loss=0.05731, over 1887289.24 frames. ], batch size: 18, lr: 8.08e-03, grad_scale: 8.0 2022-12-07 21:36:18,775 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:36:24,353 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:36:31,205 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2974, 3.7381, 2.8866, 3.1976, 2.6124, 3.7437, 3.5497, 1.4778], device='cuda:0'), covar=tensor([0.2408, 0.0597, 0.3441, 0.0994, 0.1019, 0.0347, 0.1113, 0.2915], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0071, 0.0058, 0.0060, 0.0087, 0.0069, 0.0092, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 21:36:37,510 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5806, 1.2384, 1.9912, 1.9056, 1.9096, 2.0419, 1.4684, 1.9592], device='cuda:0'), covar=tensor([0.0687, 0.1035, 0.0206, 0.0370, 0.0528, 0.0246, 0.0505, 0.0306], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0158, 0.0123, 0.0165, 0.0142, 0.0136, 0.0116, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:37:09,570 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2208, 1.5605, 1.8215, 1.6624, 1.5733, 1.7045, 1.3527, 1.2379], device='cuda:0'), covar=tensor([0.1950, 0.0821, 0.0367, 0.0432, 0.1140, 0.0864, 0.2247, 0.1601], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0072, 0.0057, 0.0061, 0.0088, 0.0068, 0.0094, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2022-12-07 21:37:17,015 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:18,934 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:22,791 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:27,450 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:37:30,289 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8146, 1.3146, 2.4360, 2.2515, 2.3798, 2.4952, 1.5962, 2.4265], device='cuda:0'), covar=tensor([0.0771, 0.1134, 0.0192, 0.0383, 0.0400, 0.0164, 0.0603, 0.0263], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0157, 0.0122, 0.0164, 0.0142, 0.0135, 0.0115, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:37:36,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9540, 4.7452, 4.4341, 4.5772, 4.5938, 4.8559, 4.9154, 4.9195], device='cuda:0'), covar=tensor([0.0728, 0.0372, 0.2113, 0.2739, 0.0784, 0.0692, 0.0852, 0.0742], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0244, 0.0409, 0.0527, 0.0305, 0.0400, 0.0387, 0.0342], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:37:37,842 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:37:39,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.182e+02 2.689e+02 3.231e+02 7.876e+02, threshold=5.378e+02, percent-clipped=5.0 2022-12-07 21:37:51,109 INFO [train.py:873] (0/4) Epoch 10, batch 1200, loss[loss=0.1284, simple_loss=0.1515, pruned_loss=0.0526, over 4926.00 frames. ], tot_loss[loss=0.1382, simple_loss=0.1639, pruned_loss=0.05625, over 1924473.53 frames. ], batch size: 100, lr: 8.08e-03, grad_scale: 8.0 2022-12-07 21:38:04,684 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:13,076 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:16,915 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:38:36,050 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:38:42,850 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6479, 1.9281, 2.0260, 2.1010, 1.7341, 2.0753, 1.6668, 0.9796], device='cuda:0'), covar=tensor([0.1856, 0.1506, 0.1113, 0.0776, 0.1568, 0.0912, 0.2410, 0.3762], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0073, 0.0058, 0.0062, 0.0089, 0.0069, 0.0095, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0006], device='cuda:0') 2022-12-07 21:38:49,279 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:02,185 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:39:08,982 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:14,524 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 2.336e+02 2.846e+02 3.621e+02 9.526e+02, threshold=5.693e+02, percent-clipped=8.0 2022-12-07 21:39:25,826 INFO [train.py:873] (0/4) Epoch 10, batch 1300, loss[loss=0.1345, simple_loss=0.1336, pruned_loss=0.06764, over 2625.00 frames. ], tot_loss[loss=0.1384, simple_loss=0.1643, pruned_loss=0.0563, over 1948480.87 frames. ], batch size: 100, lr: 8.07e-03, grad_scale: 8.0 2022-12-07 21:39:34,561 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:47,006 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:54,596 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:39:59,650 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:03,953 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:13,149 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3218, 2.2761, 3.0849, 2.3585, 3.1698, 3.0582, 2.8960, 2.4000], device='cuda:0'), covar=tensor([0.1215, 0.3346, 0.1188, 0.2495, 0.1123, 0.1051, 0.1376, 0.2303], device='cuda:0'), in_proj_covar=tensor([0.0341, 0.0323, 0.0396, 0.0309, 0.0378, 0.0314, 0.0363, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:40:17,296 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 21:40:32,789 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:40:49,782 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.414e+02 2.899e+02 3.828e+02 8.841e+02, threshold=5.798e+02, percent-clipped=2.0 2022-12-07 21:40:57,630 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:41:01,161 INFO [train.py:873] (0/4) Epoch 10, batch 1400, loss[loss=0.1369, simple_loss=0.167, pruned_loss=0.05339, over 14270.00 frames. ], tot_loss[loss=0.1383, simple_loss=0.1647, pruned_loss=0.05597, over 1993160.62 frames. ], batch size: 76, lr: 8.07e-03, grad_scale: 8.0 2022-12-07 21:41:22,211 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8393, 1.9591, 2.8669, 2.9376, 2.8534, 2.1270, 2.8777, 2.2977], device='cuda:0'), covar=tensor([0.0305, 0.0691, 0.0477, 0.0313, 0.0320, 0.0885, 0.0275, 0.0660], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0241, 0.0360, 0.0303, 0.0247, 0.0290, 0.0276, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:41:57,169 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:42:02,574 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:42:24,558 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.449e+02 2.299e+02 2.933e+02 3.622e+02 8.157e+02, threshold=5.865e+02, percent-clipped=2.0 2022-12-07 21:42:26,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-07 21:42:35,973 INFO [train.py:873] (0/4) Epoch 10, batch 1500, loss[loss=0.1396, simple_loss=0.1721, pruned_loss=0.05353, over 14094.00 frames. ], tot_loss[loss=0.1379, simple_loss=0.1642, pruned_loss=0.05578, over 1980804.42 frames. ], batch size: 29, lr: 8.06e-03, grad_scale: 8.0 2022-12-07 21:42:53,628 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1611, 1.2714, 1.3759, 0.9365, 0.9596, 1.2125, 0.7601, 1.2292], device='cuda:0'), covar=tensor([0.1654, 0.2988, 0.1036, 0.2575, 0.3153, 0.1026, 0.2754, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0093, 0.0088, 0.0093, 0.0114, 0.0080, 0.0130, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:0') 2022-12-07 21:42:57,158 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:43:01,234 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-07 21:43:15,901 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 21:43:35,245 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-07 21:43:42,193 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:43:59,277 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.142e+02 2.757e+02 3.564e+02 6.285e+02, threshold=5.515e+02, percent-clipped=1.0 2022-12-07 21:44:10,745 INFO [train.py:873] (0/4) Epoch 10, batch 1600, loss[loss=0.1275, simple_loss=0.1352, pruned_loss=0.05988, over 3865.00 frames. ], tot_loss[loss=0.1385, simple_loss=0.1645, pruned_loss=0.05623, over 2016170.40 frames. ], batch size: 100, lr: 8.05e-03, grad_scale: 8.0 2022-12-07 21:44:24,851 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:44:26,621 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:44:48,034 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:11,243 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:21,874 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:45:33,107 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.146e+01 2.291e+02 2.751e+02 3.642e+02 1.145e+03, threshold=5.503e+02, percent-clipped=2.0 2022-12-07 21:45:33,214 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:36,292 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:45:45,008 INFO [train.py:873] (0/4) Epoch 10, batch 1700, loss[loss=0.1043, simple_loss=0.1368, pruned_loss=0.03586, over 13664.00 frames. ], tot_loss[loss=0.1381, simple_loss=0.1643, pruned_loss=0.05596, over 2023954.90 frames. ], batch size: 17, lr: 8.05e-03, grad_scale: 8.0 2022-12-07 21:45:48,617 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9784, 4.7199, 4.4018, 4.5109, 4.6283, 4.8457, 4.9357, 4.9128], device='cuda:0'), covar=tensor([0.0714, 0.0492, 0.2068, 0.3020, 0.0670, 0.0741, 0.0968, 0.0765], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0248, 0.0413, 0.0532, 0.0311, 0.0407, 0.0383, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:46:40,948 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:46:44,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([6.1746, 5.7084, 5.6088, 6.1063, 5.9057, 5.0592, 6.1704, 4.9843], device='cuda:0'), covar=tensor([0.0315, 0.0890, 0.0270, 0.0456, 0.0644, 0.0303, 0.0455, 0.0595], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0253, 0.0176, 0.0173, 0.0168, 0.0139, 0.0260, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 21:46:46,133 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:08,313 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.221e+02 2.761e+02 3.390e+02 7.818e+02, threshold=5.522e+02, percent-clipped=1.0 2022-12-07 21:47:15,959 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0984, 1.8439, 3.3624, 2.3407, 3.1509, 1.7819, 2.5499, 3.0674], device='cuda:0'), covar=tensor([0.0880, 0.4360, 0.0475, 0.5738, 0.0906, 0.3750, 0.1317, 0.0891], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0222, 0.0191, 0.0300, 0.0216, 0.0227, 0.0219, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:47:19,440 INFO [train.py:873] (0/4) Epoch 10, batch 1800, loss[loss=0.1396, simple_loss=0.1705, pruned_loss=0.05431, over 14380.00 frames. ], tot_loss[loss=0.138, simple_loss=0.1644, pruned_loss=0.05576, over 2007212.82 frames. ], batch size: 73, lr: 8.04e-03, grad_scale: 8.0 2022-12-07 21:47:26,412 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:32,301 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:40,893 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:47:53,174 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6429, 3.7377, 3.9608, 3.5722, 3.7902, 3.7906, 1.6068, 3.5472], device='cuda:0'), covar=tensor([0.0300, 0.0342, 0.0301, 0.0407, 0.0330, 0.0407, 0.3008, 0.0301], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0130, 0.0131, 0.0188, 0.0129, 0.0152, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:47:59,745 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:48:15,078 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.16 vs. limit=5.0 2022-12-07 21:48:26,734 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:48:26,800 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:48:27,225 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2022-12-07 21:48:37,278 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8115, 1.4935, 1.7738, 1.9782, 1.4215, 1.6793, 1.8002, 1.8400], device='cuda:0'), covar=tensor([0.0113, 0.0225, 0.0120, 0.0104, 0.0231, 0.0269, 0.0142, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0241, 0.0361, 0.0304, 0.0248, 0.0293, 0.0277, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:48:43,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.361e+02 2.916e+02 3.771e+02 6.970e+02, threshold=5.833e+02, percent-clipped=3.0 2022-12-07 21:48:45,615 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:48:47,607 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6871, 1.7156, 1.8039, 1.8334, 1.6192, 1.4138, 1.2207, 1.1577], device='cuda:0'), covar=tensor([0.0474, 0.1158, 0.0401, 0.0335, 0.0478, 0.0366, 0.0347, 0.0555], device='cuda:0'), in_proj_covar=tensor([0.0013, 0.0015, 0.0012, 0.0013, 0.0013, 0.0021, 0.0017, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.0123e-04, 1.1016e-04, 9.6687e-05, 1.0482e-04, 1.0191e-04, 1.5092e-04, 1.2831e-04, 1.4675e-04], device='cuda:0') 2022-12-07 21:48:55,138 INFO [train.py:873] (0/4) Epoch 10, batch 1900, loss[loss=0.14, simple_loss=0.1687, pruned_loss=0.05568, over 14281.00 frames. ], tot_loss[loss=0.1377, simple_loss=0.1639, pruned_loss=0.05578, over 1940399.94 frames. ], batch size: 60, lr: 8.04e-03, grad_scale: 8.0 2022-12-07 21:49:11,045 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:49:11,853 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:49:34,719 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-70000.pt 2022-12-07 21:50:02,114 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:02,179 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:08,123 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 21:50:08,448 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.56 vs. limit=2.0 2022-12-07 21:50:24,070 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.053e+02 2.126e+02 2.650e+02 3.300e+02 6.481e+02, threshold=5.300e+02, percent-clipped=2.0 2022-12-07 21:50:27,048 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:50:35,176 INFO [train.py:873] (0/4) Epoch 10, batch 2000, loss[loss=0.2025, simple_loss=0.1874, pruned_loss=0.1088, over 1286.00 frames. ], tot_loss[loss=0.1383, simple_loss=0.1643, pruned_loss=0.0562, over 1915549.65 frames. ], batch size: 100, lr: 8.03e-03, grad_scale: 8.0 2022-12-07 21:50:38,196 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2307, 4.9560, 4.8333, 5.2706, 4.7809, 4.4897, 5.2413, 5.1861], device='cuda:0'), covar=tensor([0.0467, 0.0662, 0.0536, 0.0362, 0.0590, 0.0480, 0.0468, 0.0413], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0121, 0.0131, 0.0137, 0.0134, 0.0110, 0.0152, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 21:50:43,874 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6155, 2.4609, 2.1979, 2.3047, 2.5120, 2.5112, 2.5679, 2.5509], device='cuda:0'), covar=tensor([0.1044, 0.0749, 0.2207, 0.2692, 0.0909, 0.1100, 0.1241, 0.0953], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0249, 0.0417, 0.0531, 0.0313, 0.0406, 0.0386, 0.0352], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:50:48,019 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:12,859 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:24,477 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:51:59,346 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.415e+02 2.238e+02 2.810e+02 3.599e+02 8.591e+02, threshold=5.620e+02, percent-clipped=6.0 2022-12-07 21:52:10,762 INFO [train.py:873] (0/4) Epoch 10, batch 2100, loss[loss=0.1464, simple_loss=0.1649, pruned_loss=0.064, over 3827.00 frames. ], tot_loss[loss=0.1378, simple_loss=0.1638, pruned_loss=0.05595, over 1934253.44 frames. ], batch size: 100, lr: 8.03e-03, grad_scale: 16.0 2022-12-07 21:52:20,520 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8041, 1.5605, 2.0941, 1.6509, 1.9116, 1.4770, 1.6094, 1.8496], device='cuda:0'), covar=tensor([0.1970, 0.1999, 0.0331, 0.1928, 0.1047, 0.1321, 0.0965, 0.0634], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0222, 0.0192, 0.0301, 0.0215, 0.0227, 0.0220, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:52:22,408 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:53:17,303 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5864, 4.0928, 3.1573, 4.9462, 4.2061, 4.5085, 3.8010, 3.5595], device='cuda:0'), covar=tensor([0.0650, 0.1233, 0.3984, 0.0435, 0.1195, 0.1511, 0.1509, 0.3230], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0299, 0.0272, 0.0243, 0.0298, 0.0291, 0.0259, 0.0261], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 21:53:34,938 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.579e+01 2.218e+02 2.863e+02 3.700e+02 6.629e+02, threshold=5.727e+02, percent-clipped=4.0 2022-12-07 21:53:45,295 INFO [train.py:873] (0/4) Epoch 10, batch 2200, loss[loss=0.1273, simple_loss=0.1536, pruned_loss=0.05052, over 5997.00 frames. ], tot_loss[loss=0.1381, simple_loss=0.1641, pruned_loss=0.05604, over 1951926.48 frames. ], batch size: 100, lr: 8.02e-03, grad_scale: 8.0 2022-12-07 21:53:46,779 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:54:13,416 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2022-12-07 21:54:13,745 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0254, 2.9538, 2.1091, 3.0557, 2.8668, 2.9886, 2.7398, 2.2393], device='cuda:0'), covar=tensor([0.1033, 0.1254, 0.3519, 0.0645, 0.0933, 0.1038, 0.1321, 0.3626], device='cuda:0'), in_proj_covar=tensor([0.0262, 0.0296, 0.0270, 0.0242, 0.0296, 0.0290, 0.0257, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 21:54:14,653 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6038, 2.4602, 2.1354, 2.2917, 2.5294, 2.4746, 2.5861, 2.5614], device='cuda:0'), covar=tensor([0.1131, 0.0785, 0.2406, 0.2603, 0.0956, 0.1299, 0.1267, 0.1049], device='cuda:0'), in_proj_covar=tensor([0.0361, 0.0249, 0.0417, 0.0531, 0.0314, 0.0411, 0.0389, 0.0353], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:54:17,348 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7304, 3.7711, 4.0453, 3.4532, 3.8360, 3.8616, 1.4350, 3.5999], device='cuda:0'), covar=tensor([0.0268, 0.0319, 0.0317, 0.0534, 0.0340, 0.0400, 0.3291, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0132, 0.0133, 0.0190, 0.0129, 0.0153, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 21:54:43,868 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:54:52,796 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 21:55:09,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.038e+02 2.253e+02 2.794e+02 3.378e+02 9.827e+02, threshold=5.588e+02, percent-clipped=3.0 2022-12-07 21:55:19,407 INFO [train.py:873] (0/4) Epoch 10, batch 2300, loss[loss=0.1407, simple_loss=0.1584, pruned_loss=0.06154, over 6908.00 frames. ], tot_loss[loss=0.1381, simple_loss=0.1642, pruned_loss=0.05604, over 1934699.86 frames. ], batch size: 100, lr: 8.01e-03, grad_scale: 8.0 2022-12-07 21:55:29,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-07 21:55:37,468 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 21:56:43,386 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.243e+01 2.133e+02 2.592e+02 3.520e+02 8.171e+02, threshold=5.183e+02, percent-clipped=4.0 2022-12-07 21:56:47,290 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8364, 1.8304, 2.0886, 1.3416, 1.4515, 1.8968, 1.3237, 1.8354], device='cuda:0'), covar=tensor([0.1232, 0.1850, 0.0825, 0.2640, 0.3181, 0.0789, 0.3865, 0.1403], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0092, 0.0086, 0.0093, 0.0111, 0.0078, 0.0126, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:0') 2022-12-07 21:56:53,786 INFO [train.py:873] (0/4) Epoch 10, batch 2400, loss[loss=0.1635, simple_loss=0.176, pruned_loss=0.0755, over 7763.00 frames. ], tot_loss[loss=0.1384, simple_loss=0.1643, pruned_loss=0.05629, over 1928303.11 frames. ], batch size: 100, lr: 8.01e-03, grad_scale: 8.0 2022-12-07 21:57:00,296 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-07 21:57:00,707 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:57:25,669 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6525, 2.3367, 3.0480, 1.6325, 1.9314, 2.5456, 1.4852, 2.4372], device='cuda:0'), covar=tensor([0.1030, 0.1658, 0.0570, 0.3442, 0.2844, 0.1009, 0.4389, 0.1378], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0093, 0.0086, 0.0094, 0.0113, 0.0079, 0.0127, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0003], device='cuda:0') 2022-12-07 21:57:34,975 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4356, 2.4339, 1.8590, 2.4757, 2.2798, 2.3748, 2.1335, 2.0767], device='cuda:0'), covar=tensor([0.0633, 0.0716, 0.2193, 0.0532, 0.0966, 0.0634, 0.1407, 0.1268], device='cuda:0'), in_proj_covar=tensor([0.0262, 0.0299, 0.0271, 0.0242, 0.0299, 0.0290, 0.0258, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 21:58:07,875 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8231, 1.5167, 2.0906, 1.5757, 1.9153, 1.4856, 1.6243, 1.9143], device='cuda:0'), covar=tensor([0.2389, 0.2619, 0.0337, 0.1414, 0.0767, 0.1201, 0.1200, 0.0580], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0223, 0.0194, 0.0301, 0.0215, 0.0225, 0.0220, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 21:58:09,113 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-07 21:58:18,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.395e+02 3.014e+02 3.583e+02 8.568e+02, threshold=6.028e+02, percent-clipped=4.0 2022-12-07 21:58:25,283 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6448, 5.2216, 5.1225, 5.7220, 5.3917, 4.6592, 5.7404, 4.7433], device='cuda:0'), covar=tensor([0.0450, 0.1025, 0.0355, 0.0395, 0.0738, 0.0356, 0.0435, 0.0501], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0255, 0.0176, 0.0173, 0.0166, 0.0137, 0.0259, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 21:58:28,703 INFO [train.py:873] (0/4) Epoch 10, batch 2500, loss[loss=0.164, simple_loss=0.1514, pruned_loss=0.08831, over 1304.00 frames. ], tot_loss[loss=0.1369, simple_loss=0.1633, pruned_loss=0.05522, over 1895482.86 frames. ], batch size: 100, lr: 8.00e-03, grad_scale: 8.0 2022-12-07 21:58:31,436 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.36 vs. limit=5.0 2022-12-07 21:58:45,176 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7280, 1.3622, 2.8961, 1.6277, 2.9602, 2.8525, 2.2292, 3.0365], device='cuda:0'), covar=tensor([0.0305, 0.2853, 0.0351, 0.1971, 0.0362, 0.0479, 0.0855, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0158, 0.0154, 0.0167, 0.0168, 0.0167, 0.0134, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 21:58:50,153 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-07 21:58:56,137 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-07 21:58:57,098 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2022-12-07 21:59:22,847 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 21:59:53,710 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.220e+02 2.772e+02 3.507e+02 5.637e+02, threshold=5.543e+02, percent-clipped=0.0 2022-12-07 22:00:03,811 INFO [train.py:873] (0/4) Epoch 10, batch 2600, loss[loss=0.1532, simple_loss=0.1708, pruned_loss=0.06783, over 6030.00 frames. ], tot_loss[loss=0.139, simple_loss=0.1647, pruned_loss=0.05664, over 1897340.54 frames. ], batch size: 100, lr: 8.00e-03, grad_scale: 8.0 2022-12-07 22:00:25,346 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-07 22:01:16,045 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-07 22:01:27,325 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.257e+02 2.909e+02 3.862e+02 1.406e+03, threshold=5.817e+02, percent-clipped=9.0 2022-12-07 22:01:38,028 INFO [train.py:873] (0/4) Epoch 10, batch 2700, loss[loss=0.1078, simple_loss=0.144, pruned_loss=0.03576, over 14232.00 frames. ], tot_loss[loss=0.1377, simple_loss=0.164, pruned_loss=0.05571, over 1910843.30 frames. ], batch size: 37, lr: 7.99e-03, grad_scale: 8.0 2022-12-07 22:01:43,164 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2022-12-07 22:01:44,452 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.39 vs. limit=5.0 2022-12-07 22:01:45,102 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:02:30,607 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:03:03,174 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.053e+02 2.251e+02 2.633e+02 3.197e+02 4.846e+02, threshold=5.267e+02, percent-clipped=0.0 2022-12-07 22:03:13,059 INFO [train.py:873] (0/4) Epoch 10, batch 2800, loss[loss=0.1483, simple_loss=0.1728, pruned_loss=0.06193, over 12739.00 frames. ], tot_loss[loss=0.1358, simple_loss=0.1628, pruned_loss=0.05437, over 1984866.51 frames. ], batch size: 100, lr: 7.99e-03, grad_scale: 8.0 2022-12-07 22:04:04,845 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2022-12-07 22:04:06,274 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2050, 2.0258, 2.1248, 2.2030, 2.1378, 2.0952, 2.2872, 1.9182], device='cuda:0'), covar=tensor([0.0692, 0.1258, 0.0664, 0.0789, 0.0901, 0.0604, 0.0894, 0.0738], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0256, 0.0177, 0.0174, 0.0166, 0.0138, 0.0262, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:04:08,166 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:04:32,060 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-07 22:04:38,138 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.119e+02 2.647e+02 3.407e+02 7.312e+02, threshold=5.294e+02, percent-clipped=4.0 2022-12-07 22:04:48,438 INFO [train.py:873] (0/4) Epoch 10, batch 2900, loss[loss=0.1434, simple_loss=0.1355, pruned_loss=0.07565, over 1296.00 frames. ], tot_loss[loss=0.1365, simple_loss=0.1631, pruned_loss=0.05499, over 1919003.00 frames. ], batch size: 100, lr: 7.98e-03, grad_scale: 8.0 2022-12-07 22:04:52,940 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:05:19,679 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:05:55,178 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:06:12,367 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.365e+02 2.871e+02 3.545e+02 4.870e+02, threshold=5.742e+02, percent-clipped=0.0 2022-12-07 22:06:17,093 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:06:22,388 INFO [train.py:873] (0/4) Epoch 10, batch 3000, loss[loss=0.1285, simple_loss=0.1617, pruned_loss=0.0477, over 14269.00 frames. ], tot_loss[loss=0.1387, simple_loss=0.1645, pruned_loss=0.05649, over 1936499.23 frames. ], batch size: 60, lr: 7.98e-03, grad_scale: 8.0 2022-12-07 22:06:22,389 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 22:06:39,846 INFO [train.py:905] (0/4) Epoch 10, validation: loss=0.1251, simple_loss=0.1669, pruned_loss=0.04162, over 857387.00 frames. 2022-12-07 22:06:39,846 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 22:07:07,967 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2625, 1.8687, 2.2514, 1.9113, 2.3751, 2.0613, 2.0474, 2.0965], device='cuda:0'), covar=tensor([0.0543, 0.1727, 0.0321, 0.0775, 0.0313, 0.0717, 0.0351, 0.0673], device='cuda:0'), in_proj_covar=tensor([0.0344, 0.0323, 0.0400, 0.0307, 0.0378, 0.0316, 0.0365, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:07:10,031 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:07:43,142 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4312, 2.3961, 1.8671, 2.4391, 2.2867, 2.3618, 2.1274, 2.0817], device='cuda:0'), covar=tensor([0.0678, 0.0719, 0.1872, 0.0529, 0.0777, 0.0590, 0.1489, 0.1299], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0295, 0.0267, 0.0239, 0.0298, 0.0287, 0.0258, 0.0253], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:07:50,116 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4985, 3.1465, 2.3357, 3.5757, 3.4144, 3.3802, 2.9580, 2.5061], device='cuda:0'), covar=tensor([0.0729, 0.1483, 0.3720, 0.0538, 0.0886, 0.1240, 0.1432, 0.3630], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0296, 0.0267, 0.0239, 0.0298, 0.0288, 0.0258, 0.0253], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:07:50,396 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-07 22:08:04,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.301e+01 2.256e+02 2.878e+02 3.587e+02 6.087e+02, threshold=5.757e+02, percent-clipped=3.0 2022-12-07 22:08:14,752 INFO [train.py:873] (0/4) Epoch 10, batch 3100, loss[loss=0.1288, simple_loss=0.1312, pruned_loss=0.0632, over 2629.00 frames. ], tot_loss[loss=0.1366, simple_loss=0.1632, pruned_loss=0.05497, over 1909091.37 frames. ], batch size: 100, lr: 7.97e-03, grad_scale: 8.0 2022-12-07 22:08:37,777 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1494, 2.2922, 4.0055, 4.2422, 4.1216, 2.5264, 4.1173, 3.2518], device='cuda:0'), covar=tensor([0.0280, 0.0827, 0.0667, 0.0301, 0.0253, 0.1175, 0.0309, 0.0720], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0246, 0.0367, 0.0314, 0.0254, 0.0296, 0.0280, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:09:18,651 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2022-12-07 22:09:23,929 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.95 vs. limit=5.0 2022-12-07 22:09:25,489 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2542, 3.6862, 2.7690, 4.3944, 4.1855, 4.1614, 3.7336, 3.1673], device='cuda:0'), covar=tensor([0.0590, 0.1240, 0.3753, 0.0503, 0.0771, 0.1437, 0.1066, 0.2954], device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0301, 0.0273, 0.0246, 0.0303, 0.0294, 0.0263, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:09:33,458 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8867, 1.5817, 1.6193, 1.6105, 1.6157, 0.9292, 1.8382, 1.6327], device='cuda:0'), covar=tensor([0.0850, 0.0851, 0.0933, 0.1283, 0.1347, 0.0703, 0.0599, 0.1374], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0026, 0.0023, 0.0025, 0.0036, 0.0024, 0.0026], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 22:09:38,773 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.264e+02 3.028e+02 3.832e+02 6.668e+02, threshold=6.056e+02, percent-clipped=4.0 2022-12-07 22:09:49,302 INFO [train.py:873] (0/4) Epoch 10, batch 3200, loss[loss=0.1311, simple_loss=0.1665, pruned_loss=0.04785, over 14597.00 frames. ], tot_loss[loss=0.137, simple_loss=0.1635, pruned_loss=0.05525, over 1926153.25 frames. ], batch size: 30, lr: 7.96e-03, grad_scale: 8.0 2022-12-07 22:11:13,451 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.362e+02 2.792e+02 3.327e+02 7.945e+02, threshold=5.585e+02, percent-clipped=1.0 2022-12-07 22:11:13,584 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:11:24,235 INFO [train.py:873] (0/4) Epoch 10, batch 3300, loss[loss=0.1231, simple_loss=0.1439, pruned_loss=0.05117, over 3898.00 frames. ], tot_loss[loss=0.1358, simple_loss=0.1626, pruned_loss=0.05452, over 1915041.37 frames. ], batch size: 100, lr: 7.96e-03, grad_scale: 8.0 2022-12-07 22:11:49,860 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:12:28,544 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1284, 4.9002, 4.4601, 4.6861, 4.6780, 5.0120, 5.0853, 5.1013], device='cuda:0'), covar=tensor([0.0779, 0.0424, 0.2250, 0.2663, 0.0763, 0.0752, 0.0810, 0.0767], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0245, 0.0414, 0.0534, 0.0314, 0.0401, 0.0383, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:12:49,402 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.167e+02 2.744e+02 3.343e+02 6.323e+02, threshold=5.487e+02, percent-clipped=2.0 2022-12-07 22:12:59,544 INFO [train.py:873] (0/4) Epoch 10, batch 3400, loss[loss=0.1179, simple_loss=0.1536, pruned_loss=0.04105, over 14272.00 frames. ], tot_loss[loss=0.1342, simple_loss=0.1615, pruned_loss=0.05339, over 1961132.37 frames. ], batch size: 28, lr: 7.95e-03, grad_scale: 8.0 2022-12-07 22:13:03,860 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-07 22:13:43,455 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-07 22:14:22,585 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 2.276e+02 2.825e+02 3.448e+02 6.629e+02, threshold=5.650e+02, percent-clipped=5.0 2022-12-07 22:14:27,514 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3807, 2.2186, 4.4481, 3.0600, 4.2348, 2.2037, 3.2728, 4.3064], device='cuda:0'), covar=tensor([0.0497, 0.4495, 0.0339, 0.6773, 0.0463, 0.3474, 0.1303, 0.0301], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0223, 0.0194, 0.0297, 0.0214, 0.0224, 0.0216, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:14:33,334 INFO [train.py:873] (0/4) Epoch 10, batch 3500, loss[loss=0.1515, simple_loss=0.1608, pruned_loss=0.0711, over 6020.00 frames. ], tot_loss[loss=0.1348, simple_loss=0.1619, pruned_loss=0.05383, over 1963178.95 frames. ], batch size: 100, lr: 7.95e-03, grad_scale: 8.0 2022-12-07 22:15:07,928 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2022-12-07 22:15:27,725 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:15:49,475 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4228, 2.4274, 2.4851, 2.5219, 2.4775, 2.1290, 1.4019, 2.1824], device='cuda:0'), covar=tensor([0.0474, 0.0514, 0.0528, 0.0350, 0.0455, 0.1112, 0.2450, 0.0390], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0161, 0.0134, 0.0137, 0.0194, 0.0131, 0.0156, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:15:57,136 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.664e+02 3.016e+02 4.035e+02 8.090e+02, threshold=6.032e+02, percent-clipped=4.0 2022-12-07 22:15:57,296 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:15:57,752 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-07 22:16:07,710 INFO [train.py:873] (0/4) Epoch 10, batch 3600, loss[loss=0.152, simple_loss=0.1367, pruned_loss=0.08367, over 2641.00 frames. ], tot_loss[loss=0.1357, simple_loss=0.1625, pruned_loss=0.05439, over 1951136.15 frames. ], batch size: 100, lr: 7.94e-03, grad_scale: 8.0 2022-12-07 22:16:25,160 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:16:33,286 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:16:38,400 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7145, 2.0006, 2.8695, 2.9020, 2.7084, 1.9737, 2.8274, 2.1560], device='cuda:0'), covar=tensor([0.0312, 0.0808, 0.0468, 0.0316, 0.0355, 0.1174, 0.0270, 0.0787], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0246, 0.0365, 0.0312, 0.0255, 0.0298, 0.0281, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:16:42,636 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:17:18,939 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:17:31,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.595e+01 2.418e+02 2.819e+02 3.742e+02 8.382e+02, threshold=5.637e+02, percent-clipped=6.0 2022-12-07 22:17:42,532 INFO [train.py:873] (0/4) Epoch 10, batch 3700, loss[loss=0.1303, simple_loss=0.1609, pruned_loss=0.04983, over 14145.00 frames. ], tot_loss[loss=0.1367, simple_loss=0.163, pruned_loss=0.05518, over 1933106.87 frames. ], batch size: 84, lr: 7.94e-03, grad_scale: 8.0 2022-12-07 22:18:21,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9259, 2.5544, 3.9927, 4.0683, 4.0727, 2.2712, 4.0120, 3.3271], device='cuda:0'), covar=tensor([0.0329, 0.0728, 0.0771, 0.0360, 0.0259, 0.1203, 0.0322, 0.0666], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0246, 0.0366, 0.0313, 0.0256, 0.0298, 0.0283, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:19:05,971 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 2.175e+02 2.639e+02 3.434e+02 5.313e+02, threshold=5.277e+02, percent-clipped=0.0 2022-12-07 22:19:16,069 INFO [train.py:873] (0/4) Epoch 10, batch 3800, loss[loss=0.1286, simple_loss=0.164, pruned_loss=0.04658, over 14222.00 frames. ], tot_loss[loss=0.1364, simple_loss=0.1631, pruned_loss=0.05491, over 1967040.04 frames. ], batch size: 94, lr: 7.93e-03, grad_scale: 8.0 2022-12-07 22:19:56,523 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7875, 3.1514, 3.1013, 3.1172, 2.2917, 3.1645, 2.8948, 1.3973], device='cuda:0'), covar=tensor([0.2305, 0.0904, 0.1257, 0.0740, 0.1119, 0.0595, 0.1297, 0.2963], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0073, 0.0059, 0.0062, 0.0089, 0.0071, 0.0092, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:20:12,799 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:20:17,209 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8353, 0.8225, 0.6987, 0.6900, 0.6477, 0.3538, 0.4373, 0.6638], device='cuda:0'), covar=tensor([0.0138, 0.0132, 0.0119, 0.0123, 0.0137, 0.0345, 0.0233, 0.0303], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0015, 0.0013, 0.0014, 0.0014, 0.0023, 0.0018, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-07 22:20:32,617 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-07 22:20:39,494 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6829, 1.9936, 2.0642, 2.1372, 1.8736, 2.1179, 1.7959, 1.2991], device='cuda:0'), covar=tensor([0.1591, 0.1565, 0.0592, 0.0566, 0.1152, 0.0695, 0.1588, 0.2780], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0073, 0.0059, 0.0062, 0.0089, 0.0071, 0.0092, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:20:40,197 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.292e+02 2.855e+02 3.626e+02 5.894e+02, threshold=5.709e+02, percent-clipped=2.0 2022-12-07 22:20:49,330 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-07 22:20:49,793 INFO [train.py:873] (0/4) Epoch 10, batch 3900, loss[loss=0.1252, simple_loss=0.1567, pruned_loss=0.04685, over 14050.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1624, pruned_loss=0.05422, over 1962584.14 frames. ], batch size: 22, lr: 7.93e-03, grad_scale: 4.0 2022-12-07 22:21:00,692 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5009, 1.1941, 2.0607, 1.8743, 1.9854, 2.0762, 1.4487, 2.0486], device='cuda:0'), covar=tensor([0.0558, 0.0913, 0.0176, 0.0315, 0.0304, 0.0151, 0.0427, 0.0235], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0122, 0.0165, 0.0142, 0.0133, 0.0114, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:21:01,560 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:21:09,393 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:21:15,739 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9367, 4.7373, 4.4513, 4.9919, 4.4206, 4.0268, 4.9948, 4.8263], device='cuda:0'), covar=tensor([0.0566, 0.0740, 0.0781, 0.0442, 0.0734, 0.0543, 0.0540, 0.0554], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0120, 0.0131, 0.0138, 0.0133, 0.0110, 0.0154, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 22:21:25,346 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9598, 3.5643, 2.8818, 4.1700, 4.0329, 4.0224, 3.6329, 2.7449], device='cuda:0'), covar=tensor([0.0761, 0.1296, 0.3471, 0.0475, 0.0837, 0.1077, 0.1042, 0.4072], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0295, 0.0269, 0.0245, 0.0299, 0.0288, 0.0256, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 22:22:01,458 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-07 22:22:14,706 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.367e+02 2.840e+02 3.454e+02 7.877e+02, threshold=5.680e+02, percent-clipped=2.0 2022-12-07 22:22:23,932 INFO [train.py:873] (0/4) Epoch 10, batch 4000, loss[loss=0.1176, simple_loss=0.1546, pruned_loss=0.04032, over 14140.00 frames. ], tot_loss[loss=0.1353, simple_loss=0.1623, pruned_loss=0.05421, over 1935017.71 frames. ], batch size: 99, lr: 7.92e-03, grad_scale: 8.0 2022-12-07 22:23:16,706 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4147, 2.7115, 2.5764, 2.7577, 2.1693, 2.7592, 2.4462, 1.2713], device='cuda:0'), covar=tensor([0.1754, 0.0933, 0.0863, 0.0551, 0.1031, 0.0686, 0.1378, 0.2980], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0073, 0.0060, 0.0062, 0.0089, 0.0071, 0.0092, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:23:29,100 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:23:48,079 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.453e+01 2.051e+02 2.497e+02 3.120e+02 6.082e+02, threshold=4.994e+02, percent-clipped=1.0 2022-12-07 22:23:57,371 INFO [train.py:873] (0/4) Epoch 10, batch 4100, loss[loss=0.118, simple_loss=0.1554, pruned_loss=0.04034, over 14534.00 frames. ], tot_loss[loss=0.1352, simple_loss=0.162, pruned_loss=0.05421, over 1862573.74 frames. ], batch size: 34, lr: 7.91e-03, grad_scale: 4.0 2022-12-07 22:24:26,151 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:25:22,726 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 2.564e+02 3.159e+02 3.830e+02 6.918e+02, threshold=6.317e+02, percent-clipped=10.0 2022-12-07 22:25:31,387 INFO [train.py:873] (0/4) Epoch 10, batch 4200, loss[loss=0.1279, simple_loss=0.1621, pruned_loss=0.04688, over 14303.00 frames. ], tot_loss[loss=0.137, simple_loss=0.1628, pruned_loss=0.05555, over 1875990.43 frames. ], batch size: 76, lr: 7.91e-03, grad_scale: 4.0 2022-12-07 22:25:44,007 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:25:46,919 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:26:27,856 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:26:33,226 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:26:44,136 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3435, 1.9680, 2.3723, 2.3773, 2.2017, 1.8959, 2.3800, 2.1597], device='cuda:0'), covar=tensor([0.0249, 0.0533, 0.0260, 0.0260, 0.0313, 0.0705, 0.0295, 0.0383], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0244, 0.0363, 0.0310, 0.0253, 0.0296, 0.0283, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:26:55,037 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.250e+02 2.962e+02 3.836e+02 8.638e+02, threshold=5.923e+02, percent-clipped=4.0 2022-12-07 22:27:03,697 INFO [train.py:873] (0/4) Epoch 10, batch 4300, loss[loss=0.1299, simple_loss=0.1618, pruned_loss=0.04897, over 14550.00 frames. ], tot_loss[loss=0.1362, simple_loss=0.1632, pruned_loss=0.0546, over 1991942.42 frames. ], batch size: 43, lr: 7.90e-03, grad_scale: 4.0 2022-12-07 22:27:18,527 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2022-12-07 22:27:29,509 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:28:05,986 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-07 22:28:26,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.139e+02 2.736e+02 3.399e+02 6.791e+02, threshold=5.473e+02, percent-clipped=1.0 2022-12-07 22:28:34,601 INFO [train.py:873] (0/4) Epoch 10, batch 4400, loss[loss=0.1304, simple_loss=0.1621, pruned_loss=0.04937, over 14497.00 frames. ], tot_loss[loss=0.1355, simple_loss=0.1625, pruned_loss=0.05422, over 2015169.48 frames. ], batch size: 51, lr: 7.90e-03, grad_scale: 8.0 2022-12-07 22:28:58,337 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:29:45,402 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6691, 1.6818, 1.9370, 1.6194, 1.7380, 1.4071, 1.3120, 1.1074], device='cuda:0'), covar=tensor([0.0413, 0.0555, 0.0395, 0.0453, 0.0389, 0.0643, 0.0411, 0.0706], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0015, 0.0013, 0.0013, 0.0014, 0.0022, 0.0018, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.0542e-04, 1.1432e-04, 9.9101e-05, 1.0655e-04, 1.0460e-04, 1.5734e-04, 1.3284e-04, 1.5234e-04], device='cuda:0') 2022-12-07 22:29:52,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7543, 1.3693, 2.4632, 2.2946, 2.4192, 2.4958, 1.8005, 2.5055], device='cuda:0'), covar=tensor([0.0854, 0.1111, 0.0158, 0.0363, 0.0366, 0.0151, 0.0524, 0.0232], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0160, 0.0123, 0.0167, 0.0143, 0.0134, 0.0116, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:29:56,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.111e+02 2.675e+02 3.314e+02 5.959e+02, threshold=5.350e+02, percent-clipped=1.0 2022-12-07 22:30:05,238 INFO [train.py:873] (0/4) Epoch 10, batch 4500, loss[loss=0.1496, simple_loss=0.1658, pruned_loss=0.0667, over 5970.00 frames. ], tot_loss[loss=0.1343, simple_loss=0.1617, pruned_loss=0.05345, over 1986363.88 frames. ], batch size: 100, lr: 7.89e-03, grad_scale: 8.0 2022-12-07 22:30:19,394 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:30:20,274 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9290, 2.7102, 2.6704, 2.9110, 2.8069, 2.7828, 2.9922, 2.3795], device='cuda:0'), covar=tensor([0.0551, 0.1095, 0.0598, 0.0645, 0.0738, 0.0449, 0.0736, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0254, 0.0175, 0.0173, 0.0168, 0.0139, 0.0263, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:30:56,174 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2199, 1.7439, 2.2446, 1.8972, 2.3423, 2.0807, 2.0260, 2.0982], device='cuda:0'), covar=tensor([0.0430, 0.1635, 0.0341, 0.0691, 0.0334, 0.0698, 0.0397, 0.0580], device='cuda:0'), in_proj_covar=tensor([0.0342, 0.0316, 0.0396, 0.0302, 0.0377, 0.0315, 0.0361, 0.0316], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:31:03,233 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:31:25,185 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2022-12-07 22:31:27,223 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.124e+02 2.389e+02 3.143e+02 3.960e+02 1.878e+03, threshold=6.285e+02, percent-clipped=10.0 2022-12-07 22:31:35,245 INFO [train.py:873] (0/4) Epoch 10, batch 4600, loss[loss=0.13, simple_loss=0.1638, pruned_loss=0.04811, over 14028.00 frames. ], tot_loss[loss=0.1355, simple_loss=0.1626, pruned_loss=0.05418, over 1960448.67 frames. ], batch size: 26, lr: 7.89e-03, grad_scale: 8.0 2022-12-07 22:31:56,308 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:32:44,271 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8324, 1.5572, 2.0637, 1.6529, 1.8979, 1.5201, 1.6249, 1.8593], device='cuda:0'), covar=tensor([0.2286, 0.2447, 0.0284, 0.1010, 0.0968, 0.1342, 0.1041, 0.0642], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0221, 0.0196, 0.0299, 0.0213, 0.0223, 0.0220, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:32:51,192 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.23 vs. limit=2.0 2022-12-07 22:32:57,785 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.006e+02 2.571e+02 3.192e+02 6.224e+02, threshold=5.142e+02, percent-clipped=0.0 2022-12-07 22:32:58,810 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2820, 4.0970, 3.9654, 4.3708, 3.9428, 3.6539, 4.3861, 4.2408], device='cuda:0'), covar=tensor([0.0663, 0.0732, 0.0785, 0.0606, 0.0777, 0.0723, 0.0621, 0.0747], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0121, 0.0132, 0.0138, 0.0133, 0.0109, 0.0154, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 22:33:06,364 INFO [train.py:873] (0/4) Epoch 10, batch 4700, loss[loss=0.1267, simple_loss=0.1574, pruned_loss=0.04804, over 14311.00 frames. ], tot_loss[loss=0.1344, simple_loss=0.162, pruned_loss=0.05337, over 1968573.94 frames. ], batch size: 31, lr: 7.88e-03, grad_scale: 8.0 2022-12-07 22:33:10,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7178, 3.5862, 3.4293, 3.7879, 3.3070, 3.1854, 3.7814, 3.6673], device='cuda:0'), covar=tensor([0.0763, 0.0884, 0.0952, 0.0701, 0.1150, 0.0903, 0.0764, 0.0833], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0122, 0.0133, 0.0139, 0.0134, 0.0110, 0.0155, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 22:33:29,684 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:34:10,897 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2200, 4.2860, 4.5494, 3.8070, 4.3854, 4.6421, 1.6375, 4.1404], device='cuda:0'), covar=tensor([0.0262, 0.0287, 0.0366, 0.0504, 0.0355, 0.0231, 0.3337, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0163, 0.0136, 0.0136, 0.0196, 0.0131, 0.0157, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 22:34:11,922 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3882, 2.8493, 4.1799, 3.3566, 4.2534, 4.1402, 3.9883, 3.6449], device='cuda:0'), covar=tensor([0.0589, 0.2960, 0.0904, 0.1627, 0.0830, 0.0738, 0.1764, 0.1867], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0315, 0.0396, 0.0302, 0.0374, 0.0316, 0.0363, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:34:12,587 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:34:23,848 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.71 vs. limit=5.0 2022-12-07 22:34:27,384 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.194e+02 2.219e+02 2.922e+02 3.522e+02 5.753e+02, threshold=5.843e+02, percent-clipped=6.0 2022-12-07 22:34:35,285 INFO [train.py:873] (0/4) Epoch 10, batch 4800, loss[loss=0.1495, simple_loss=0.1474, pruned_loss=0.07576, over 2636.00 frames. ], tot_loss[loss=0.1334, simple_loss=0.1612, pruned_loss=0.05279, over 1967723.75 frames. ], batch size: 100, lr: 7.88e-03, grad_scale: 8.0 2022-12-07 22:34:48,175 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7325, 1.9722, 2.7218, 2.0589, 2.6619, 2.6134, 2.4217, 2.2528], device='cuda:0'), covar=tensor([0.0628, 0.2533, 0.0850, 0.1835, 0.0613, 0.0947, 0.0891, 0.1475], device='cuda:0'), in_proj_covar=tensor([0.0339, 0.0315, 0.0394, 0.0301, 0.0373, 0.0315, 0.0360, 0.0314], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:34:59,065 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7215, 1.7339, 1.7712, 1.9858, 1.9052, 1.0223, 1.9350, 2.1377], device='cuda:0'), covar=tensor([0.1134, 0.1260, 0.0621, 0.1303, 0.0922, 0.1039, 0.1456, 0.0593], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0025, 0.0027, 0.0024, 0.0025, 0.0037, 0.0025, 0.0026], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 22:35:06,557 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:35:56,441 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 2.330e+02 2.761e+02 3.317e+02 8.665e+02, threshold=5.521e+02, percent-clipped=2.0 2022-12-07 22:35:56,645 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9831, 1.6951, 4.1507, 3.8912, 3.9455, 4.1986, 3.4517, 4.2125], device='cuda:0'), covar=tensor([0.1424, 0.1445, 0.0096, 0.0223, 0.0195, 0.0128, 0.0264, 0.0114], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0161, 0.0123, 0.0167, 0.0142, 0.0134, 0.0116, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:36:01,292 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:36:04,982 INFO [train.py:873] (0/4) Epoch 10, batch 4900, loss[loss=0.1362, simple_loss=0.1609, pruned_loss=0.0558, over 13867.00 frames. ], tot_loss[loss=0.1336, simple_loss=0.1613, pruned_loss=0.05301, over 1936634.92 frames. ], batch size: 23, lr: 7.87e-03, grad_scale: 8.0 2022-12-07 22:36:09,386 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 22:36:24,925 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:37:03,898 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 22:37:05,512 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-07 22:37:08,118 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:37:26,027 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.275e+02 2.705e+02 3.390e+02 8.104e+02, threshold=5.410e+02, percent-clipped=1.0 2022-12-07 22:37:33,743 INFO [train.py:873] (0/4) Epoch 10, batch 5000, loss[loss=0.1473, simple_loss=0.1672, pruned_loss=0.06368, over 9502.00 frames. ], tot_loss[loss=0.1348, simple_loss=0.1622, pruned_loss=0.05374, over 1923553.52 frames. ], batch size: 100, lr: 7.87e-03, grad_scale: 8.0 2022-12-07 22:38:03,255 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9494, 4.0090, 3.5329, 2.7632, 3.4482, 3.7312, 4.1114, 3.4027], device='cuda:0'), covar=tensor([0.0472, 0.1115, 0.0843, 0.1624, 0.0751, 0.0604, 0.0742, 0.1013], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0185, 0.0135, 0.0126, 0.0130, 0.0138, 0.0114, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:38:10,812 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9199, 2.6279, 2.6623, 1.7934, 2.4881, 2.7265, 3.0592, 2.3784], device='cuda:0'), covar=tensor([0.0627, 0.1290, 0.1090, 0.1825, 0.0964, 0.0598, 0.0516, 0.1427], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0185, 0.0135, 0.0126, 0.0130, 0.0138, 0.0113, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:38:54,056 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.200e+01 2.088e+02 2.701e+02 3.421e+02 5.931e+02, threshold=5.402e+02, percent-clipped=1.0 2022-12-07 22:39:01,815 INFO [train.py:873] (0/4) Epoch 10, batch 5100, loss[loss=0.1136, simple_loss=0.1193, pruned_loss=0.05402, over 1257.00 frames. ], tot_loss[loss=0.1345, simple_loss=0.1618, pruned_loss=0.05358, over 1878122.96 frames. ], batch size: 100, lr: 7.86e-03, grad_scale: 4.0 2022-12-07 22:39:41,015 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0894, 2.8270, 3.7612, 2.7564, 2.3067, 3.0654, 1.4606, 2.9994], device='cuda:0'), covar=tensor([0.1197, 0.1310, 0.0786, 0.1984, 0.2343, 0.1129, 0.4518, 0.1353], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0095, 0.0088, 0.0093, 0.0112, 0.0078, 0.0124, 0.0086], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-07 22:39:55,369 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-07 22:40:21,588 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.92 vs. limit=5.0 2022-12-07 22:40:21,984 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:40:22,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.094e+02 2.487e+02 3.025e+02 3.884e+02 1.180e+03, threshold=6.049e+02, percent-clipped=5.0 2022-12-07 22:40:29,836 INFO [train.py:873] (0/4) Epoch 10, batch 5200, loss[loss=0.1134, simple_loss=0.1427, pruned_loss=0.04207, over 6958.00 frames. ], tot_loss[loss=0.1343, simple_loss=0.1618, pruned_loss=0.05342, over 1910410.18 frames. ], batch size: 100, lr: 7.85e-03, grad_scale: 8.0 2022-12-07 22:41:00,502 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:41:13,982 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2662, 4.6979, 4.7058, 5.2071, 4.9065, 4.4712, 5.2018, 4.2971], device='cuda:0'), covar=tensor([0.0325, 0.1273, 0.0330, 0.0464, 0.0708, 0.0427, 0.0522, 0.0531], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0254, 0.0177, 0.0173, 0.0168, 0.0140, 0.0262, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:41:25,168 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 22:41:52,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.362e+02 2.946e+02 3.664e+02 8.798e+02, threshold=5.892e+02, percent-clipped=2.0 2022-12-07 22:41:54,685 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:41:59,540 INFO [train.py:873] (0/4) Epoch 10, batch 5300, loss[loss=0.1074, simple_loss=0.137, pruned_loss=0.03887, over 13912.00 frames. ], tot_loss[loss=0.1357, simple_loss=0.1625, pruned_loss=0.05444, over 1891992.06 frames. ], batch size: 20, lr: 7.85e-03, grad_scale: 4.0 2022-12-07 22:42:15,276 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6019, 1.4889, 1.5781, 1.5074, 1.5280, 0.8656, 1.3674, 1.4458], device='cuda:0'), covar=tensor([0.0786, 0.0588, 0.0635, 0.1086, 0.0768, 0.1059, 0.0981, 0.1042], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0026, 0.0024, 0.0025, 0.0036, 0.0025, 0.0026], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 22:42:24,105 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2329, 4.3019, 4.5617, 3.7983, 4.3470, 4.5862, 1.6242, 4.1121], device='cuda:0'), covar=tensor([0.0276, 0.0277, 0.0304, 0.0400, 0.0324, 0.0199, 0.3177, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0161, 0.0135, 0.0135, 0.0195, 0.0133, 0.0156, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 22:43:21,802 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 2.215e+02 2.822e+02 3.427e+02 6.404e+02, threshold=5.643e+02, percent-clipped=2.0 2022-12-07 22:43:27,863 INFO [train.py:873] (0/4) Epoch 10, batch 5400, loss[loss=0.1712, simple_loss=0.1738, pruned_loss=0.08425, over 3876.00 frames. ], tot_loss[loss=0.135, simple_loss=0.1619, pruned_loss=0.05404, over 1868125.72 frames. ], batch size: 100, lr: 7.84e-03, grad_scale: 4.0 2022-12-07 22:43:27,956 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3291, 4.0733, 3.9765, 4.3766, 4.0886, 3.8679, 4.4051, 3.7259], device='cuda:0'), covar=tensor([0.0452, 0.0940, 0.0390, 0.0463, 0.0718, 0.1062, 0.0526, 0.0479], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0259, 0.0181, 0.0177, 0.0172, 0.0143, 0.0267, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:43:31,662 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6171, 2.0215, 3.7499, 2.5696, 3.5481, 1.9088, 2.8112, 3.5365], device='cuda:0'), covar=tensor([0.0631, 0.4553, 0.0401, 0.6079, 0.0637, 0.4018, 0.1469, 0.0559], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0217, 0.0193, 0.0294, 0.0213, 0.0223, 0.0219, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:43:58,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3912, 3.1265, 2.3516, 3.5141, 3.3827, 3.3699, 2.8908, 2.3008], device='cuda:0'), covar=tensor([0.1185, 0.1587, 0.4302, 0.0648, 0.0918, 0.1172, 0.1443, 0.4433], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0295, 0.0269, 0.0244, 0.0301, 0.0291, 0.0255, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 22:44:48,881 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:44:50,661 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.129e+02 2.539e+02 3.119e+02 4.865e+02, threshold=5.078e+02, percent-clipped=0.0 2022-12-07 22:44:57,303 INFO [train.py:873] (0/4) Epoch 10, batch 5500, loss[loss=0.1322, simple_loss=0.1643, pruned_loss=0.0501, over 14356.00 frames. ], tot_loss[loss=0.1341, simple_loss=0.1616, pruned_loss=0.05336, over 1962330.60 frames. ], batch size: 55, lr: 7.84e-03, grad_scale: 4.0 2022-12-07 22:45:23,816 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-07 22:45:24,704 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:45:31,972 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:45:40,156 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8180, 1.8650, 3.9024, 2.6823, 3.7775, 1.9286, 2.9287, 3.7430], device='cuda:0'), covar=tensor([0.0651, 0.5756, 0.0524, 0.8162, 0.0678, 0.4470, 0.1581, 0.0499], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0216, 0.0192, 0.0294, 0.0213, 0.0220, 0.0216, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:45:50,897 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 22:46:17,371 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:46:19,067 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:46:19,731 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.890e+01 2.291e+02 2.969e+02 3.792e+02 1.060e+03, threshold=5.938e+02, percent-clipped=6.0 2022-12-07 22:46:25,680 INFO [train.py:873] (0/4) Epoch 10, batch 5600, loss[loss=0.1121, simple_loss=0.1481, pruned_loss=0.03806, over 14307.00 frames. ], tot_loss[loss=0.1349, simple_loss=0.1623, pruned_loss=0.05373, over 1956349.80 frames. ], batch size: 39, lr: 7.83e-03, grad_scale: 8.0 2022-12-07 22:46:33,871 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 22:46:52,814 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2263, 1.5967, 1.7523, 1.7080, 1.5783, 1.7088, 1.4421, 1.2395], device='cuda:0'), covar=tensor([0.1429, 0.0872, 0.0469, 0.0391, 0.1397, 0.0678, 0.1726, 0.1968], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0075, 0.0060, 0.0063, 0.0090, 0.0072, 0.0093, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:47:17,538 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1760, 1.2616, 3.2144, 1.5345, 2.9742, 3.2629, 2.2855, 3.4279], device='cuda:0'), covar=tensor([0.0245, 0.3032, 0.0389, 0.2247, 0.1118, 0.0382, 0.0877, 0.0236], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0157, 0.0156, 0.0168, 0.0170, 0.0170, 0.0132, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 22:47:24,675 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0821, 1.2890, 1.4282, 1.0085, 0.8166, 1.2402, 0.8772, 1.2920], device='cuda:0'), covar=tensor([0.1802, 0.2538, 0.0834, 0.2440, 0.3262, 0.1062, 0.2125, 0.1031], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0093, 0.0087, 0.0094, 0.0113, 0.0077, 0.0125, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-07 22:47:30,671 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:47:43,131 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8425, 1.1653, 2.0388, 1.3032, 2.0047, 2.0472, 1.7414, 2.0614], device='cuda:0'), covar=tensor([0.0356, 0.2220, 0.0446, 0.1737, 0.0546, 0.0559, 0.1028, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0156, 0.0156, 0.0168, 0.0169, 0.0169, 0.0131, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 22:47:48,144 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.591e+02 3.035e+02 3.423e+02 1.031e+03, threshold=6.069e+02, percent-clipped=1.0 2022-12-07 22:47:55,242 INFO [train.py:873] (0/4) Epoch 10, batch 5700, loss[loss=0.1089, simple_loss=0.1439, pruned_loss=0.03691, over 13937.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1623, pruned_loss=0.05423, over 1923505.79 frames. ], batch size: 20, lr: 7.83e-03, grad_scale: 8.0 2022-12-07 22:48:24,146 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-07 22:48:24,531 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:48:29,883 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.48 vs. limit=2.0 2022-12-07 22:49:17,092 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.156e+02 2.743e+02 3.331e+02 7.800e+02, threshold=5.486e+02, percent-clipped=2.0 2022-12-07 22:49:23,292 INFO [train.py:873] (0/4) Epoch 10, batch 5800, loss[loss=0.1404, simple_loss=0.1622, pruned_loss=0.05928, over 11144.00 frames. ], tot_loss[loss=0.1353, simple_loss=0.1624, pruned_loss=0.05413, over 1921849.71 frames. ], batch size: 100, lr: 7.82e-03, grad_scale: 8.0 2022-12-07 22:49:24,323 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9547, 4.6683, 4.3463, 4.5878, 4.5984, 4.8543, 5.0047, 4.8814], device='cuda:0'), covar=tensor([0.0917, 0.0477, 0.2261, 0.3114, 0.0764, 0.0849, 0.0819, 0.1050], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0253, 0.0423, 0.0555, 0.0318, 0.0410, 0.0383, 0.0357], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:50:09,151 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3107, 1.8300, 3.4484, 2.3956, 3.3496, 1.8326, 2.6057, 3.3105], device='cuda:0'), covar=tensor([0.0695, 0.4477, 0.0610, 0.6565, 0.0638, 0.3705, 0.1468, 0.0535], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0216, 0.0192, 0.0293, 0.0213, 0.0220, 0.0214, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:50:16,295 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1974, 4.8206, 4.6255, 5.1872, 4.8662, 4.5303, 5.1614, 4.4646], device='cuda:0'), covar=tensor([0.0314, 0.0850, 0.0339, 0.0390, 0.0696, 0.0455, 0.0444, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0258, 0.0182, 0.0178, 0.0172, 0.0144, 0.0269, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:50:21,890 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5473, 2.8338, 4.3068, 3.1623, 4.3424, 4.3318, 4.1991, 3.6999], device='cuda:0'), covar=tensor([0.0613, 0.3458, 0.0857, 0.1909, 0.0794, 0.0698, 0.1519, 0.2268], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0320, 0.0401, 0.0306, 0.0378, 0.0317, 0.0363, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:50:40,904 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:50:43,991 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:50:46,443 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 2.192e+02 2.728e+02 3.266e+02 1.115e+03, threshold=5.456e+02, percent-clipped=1.0 2022-12-07 22:50:52,925 INFO [train.py:873] (0/4) Epoch 10, batch 5900, loss[loss=0.1265, simple_loss=0.1638, pruned_loss=0.04461, over 14346.00 frames. ], tot_loss[loss=0.1346, simple_loss=0.1622, pruned_loss=0.05354, over 1960392.42 frames. ], batch size: 31, lr: 7.82e-03, grad_scale: 8.0 2022-12-07 22:51:01,936 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7273, 3.2700, 2.6562, 3.9076, 3.7250, 3.7369, 3.3190, 2.6022], device='cuda:0'), covar=tensor([0.0986, 0.1555, 0.3601, 0.0608, 0.0959, 0.1227, 0.1225, 0.3718], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0295, 0.0268, 0.0245, 0.0302, 0.0292, 0.0253, 0.0256], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 22:51:25,808 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8846, 2.7072, 2.7115, 2.8861, 2.8013, 2.8755, 2.9812, 2.4164], device='cuda:0'), covar=tensor([0.0606, 0.1153, 0.0624, 0.0713, 0.0766, 0.0498, 0.0751, 0.0704], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0259, 0.0182, 0.0178, 0.0173, 0.0144, 0.0268, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:51:26,779 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:51:32,624 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.01 vs. limit=5.0 2022-12-07 22:51:41,227 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3381, 2.3957, 4.2366, 4.4031, 4.2634, 2.5294, 4.2057, 3.4340], device='cuda:0'), covar=tensor([0.0259, 0.0823, 0.0529, 0.0250, 0.0265, 0.1226, 0.0368, 0.0625], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0245, 0.0363, 0.0311, 0.0252, 0.0294, 0.0285, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 22:52:14,955 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.132e+02 2.894e+02 3.484e+02 7.158e+02, threshold=5.787e+02, percent-clipped=1.0 2022-12-07 22:52:20,976 INFO [train.py:873] (0/4) Epoch 10, batch 6000, loss[loss=0.2062, simple_loss=0.1798, pruned_loss=0.1163, over 1183.00 frames. ], tot_loss[loss=0.1338, simple_loss=0.1613, pruned_loss=0.0532, over 1919496.37 frames. ], batch size: 100, lr: 7.81e-03, grad_scale: 8.0 2022-12-07 22:52:20,977 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 22:52:36,794 INFO [train.py:905] (0/4) Epoch 10, validation: loss=0.1251, simple_loss=0.167, pruned_loss=0.04163, over 857387.00 frames. 2022-12-07 22:52:36,795 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 22:52:41,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:52:48,229 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0449, 3.3436, 3.0323, 3.1632, 2.4858, 3.3273, 3.0440, 1.6302], device='cuda:0'), covar=tensor([0.2332, 0.0774, 0.1319, 0.0918, 0.1127, 0.0575, 0.1353, 0.2893], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0076, 0.0061, 0.0064, 0.0090, 0.0072, 0.0094, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-07 22:52:48,258 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7745, 1.5428, 1.6488, 1.8411, 1.6693, 0.8418, 1.6918, 1.8449], device='cuda:0'), covar=tensor([0.1332, 0.0992, 0.0866, 0.0691, 0.1071, 0.0894, 0.0635, 0.0806], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0025, 0.0026, 0.0024, 0.0025, 0.0037, 0.0025, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 22:53:02,933 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:53:35,987 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:54:00,210 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.199e+02 2.430e+02 2.883e+02 3.808e+02 1.343e+03, threshold=5.765e+02, percent-clipped=6.0 2022-12-07 22:54:06,423 INFO [train.py:873] (0/4) Epoch 10, batch 6100, loss[loss=0.1727, simple_loss=0.1867, pruned_loss=0.07933, over 14169.00 frames. ], tot_loss[loss=0.1344, simple_loss=0.1614, pruned_loss=0.05368, over 1923567.48 frames. ], batch size: 84, lr: 7.81e-03, grad_scale: 8.0 2022-12-07 22:55:19,001 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:55:19,208 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-07 22:55:23,989 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:55:25,376 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-07 22:55:29,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.250e+02 2.648e+02 3.376e+02 5.030e+02, threshold=5.296e+02, percent-clipped=0.0 2022-12-07 22:55:35,296 INFO [train.py:873] (0/4) Epoch 10, batch 6200, loss[loss=0.1314, simple_loss=0.1319, pruned_loss=0.06552, over 2582.00 frames. ], tot_loss[loss=0.1342, simple_loss=0.1615, pruned_loss=0.05345, over 1915434.49 frames. ], batch size: 100, lr: 7.80e-03, grad_scale: 8.0 2022-12-07 22:56:06,622 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:56:09,289 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3462, 4.0402, 4.0075, 4.3963, 4.1218, 3.8682, 4.3655, 3.7024], device='cuda:0'), covar=tensor([0.0493, 0.1044, 0.0384, 0.0549, 0.0879, 0.0981, 0.0641, 0.0598], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0257, 0.0180, 0.0177, 0.0173, 0.0144, 0.0266, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:56:13,667 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:56:41,973 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3111, 2.7641, 2.7749, 1.5310, 2.8847, 3.0639, 3.2948, 2.4438], device='cuda:0'), covar=tensor([0.0752, 0.1916, 0.1300, 0.2844, 0.1212, 0.0731, 0.1045, 0.1800], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0189, 0.0137, 0.0128, 0.0135, 0.0142, 0.0116, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-07 22:56:58,219 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.014e+02 2.284e+02 2.758e+02 3.409e+02 5.517e+02, threshold=5.515e+02, percent-clipped=2.0 2022-12-07 22:57:04,251 INFO [train.py:873] (0/4) Epoch 10, batch 6300, loss[loss=0.1572, simple_loss=0.1808, pruned_loss=0.0668, over 12752.00 frames. ], tot_loss[loss=0.1335, simple_loss=0.1613, pruned_loss=0.05287, over 1950597.46 frames. ], batch size: 100, lr: 7.80e-03, grad_scale: 8.0 2022-12-07 22:57:12,155 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2022-12-07 22:57:30,141 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:57:35,004 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3288, 3.9797, 3.9190, 4.3286, 4.0641, 3.7950, 4.3061, 3.6229], device='cuda:0'), covar=tensor([0.0370, 0.0971, 0.0371, 0.0482, 0.0747, 0.1074, 0.0583, 0.0514], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0255, 0.0177, 0.0176, 0.0170, 0.0143, 0.0263, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 22:57:55,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9363, 3.6919, 3.3693, 3.6083, 3.8210, 3.8516, 3.9176, 3.9043], device='cuda:0'), covar=tensor([0.0834, 0.0633, 0.2190, 0.2545, 0.0770, 0.0794, 0.1015, 0.0815], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0252, 0.0423, 0.0546, 0.0317, 0.0407, 0.0390, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:57:57,823 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:58:11,974 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 22:58:12,138 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5307, 1.8505, 3.6627, 2.4987, 3.5228, 1.8647, 2.7766, 3.4476], device='cuda:0'), covar=tensor([0.0923, 0.5549, 0.0649, 0.7144, 0.0668, 0.4652, 0.1604, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0219, 0.0196, 0.0293, 0.0214, 0.0223, 0.0218, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:58:25,309 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.328e+02 2.936e+02 3.672e+02 8.689e+02, threshold=5.873e+02, percent-clipped=4.0 2022-12-07 22:58:28,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4940, 4.2537, 3.9400, 4.1302, 4.2703, 4.4210, 4.4911, 4.4510], device='cuda:0'), covar=tensor([0.0753, 0.0480, 0.2082, 0.2466, 0.0732, 0.0699, 0.0902, 0.0771], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0253, 0.0423, 0.0545, 0.0317, 0.0405, 0.0389, 0.0352], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:58:32,158 INFO [train.py:873] (0/4) Epoch 10, batch 6400, loss[loss=0.158, simple_loss=0.1756, pruned_loss=0.07026, over 14607.00 frames. ], tot_loss[loss=0.1334, simple_loss=0.1615, pruned_loss=0.05264, over 2019563.42 frames. ], batch size: 22, lr: 7.79e-03, grad_scale: 8.0 2022-12-07 22:58:36,150 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2657, 2.2479, 3.1069, 2.4651, 3.1770, 3.0691, 2.9733, 2.6158], device='cuda:0'), covar=tensor([0.0522, 0.2442, 0.0717, 0.1540, 0.0506, 0.0712, 0.0953, 0.1570], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0318, 0.0397, 0.0305, 0.0374, 0.0316, 0.0359, 0.0318], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 22:59:09,246 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5927, 1.6157, 1.6586, 1.5913, 1.5224, 1.1959, 1.0003, 1.0320], device='cuda:0'), covar=tensor([0.0218, 0.0406, 0.0413, 0.0229, 0.0271, 0.0329, 0.0327, 0.0636], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0015, 0.0013, 0.0014, 0.0014, 0.0023, 0.0018, 0.0023], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-07 22:59:53,824 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.247e+02 2.848e+02 3.548e+02 7.995e+02, threshold=5.696e+02, percent-clipped=2.0 2022-12-07 22:59:59,600 INFO [train.py:873] (0/4) Epoch 10, batch 6500, loss[loss=0.1446, simple_loss=0.1617, pruned_loss=0.06376, over 4997.00 frames. ], tot_loss[loss=0.1337, simple_loss=0.1618, pruned_loss=0.05279, over 1991179.69 frames. ], batch size: 100, lr: 7.79e-03, grad_scale: 8.0 2022-12-07 23:00:20,096 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9403, 3.6870, 3.4567, 3.5777, 3.8191, 3.8477, 3.9261, 3.9097], device='cuda:0'), covar=tensor([0.0929, 0.0633, 0.2265, 0.2950, 0.0800, 0.0884, 0.0979, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0249, 0.0419, 0.0543, 0.0316, 0.0404, 0.0389, 0.0348], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:00:30,727 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2936, 4.9999, 4.5588, 4.9003, 4.7840, 5.1957, 5.2448, 5.2031], device='cuda:0'), covar=tensor([0.0744, 0.0402, 0.2230, 0.2723, 0.0785, 0.0689, 0.0820, 0.0945], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0249, 0.0419, 0.0543, 0.0317, 0.0405, 0.0388, 0.0348], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:00:31,688 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:00:32,430 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:01:09,387 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0722, 1.9157, 1.9994, 2.1027, 2.0139, 2.0040, 2.1534, 1.8552], device='cuda:0'), covar=tensor([0.0893, 0.1289, 0.0735, 0.0816, 0.0957, 0.0735, 0.0944, 0.0745], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0255, 0.0178, 0.0175, 0.0171, 0.0141, 0.0263, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 23:01:22,223 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 2.214e+02 2.798e+02 3.905e+02 7.921e+02, threshold=5.597e+02, percent-clipped=3.0 2022-12-07 23:01:25,270 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:01:28,053 INFO [train.py:873] (0/4) Epoch 10, batch 6600, loss[loss=0.1672, simple_loss=0.1538, pruned_loss=0.09031, over 1254.00 frames. ], tot_loss[loss=0.1333, simple_loss=0.1616, pruned_loss=0.05246, over 2036566.70 frames. ], batch size: 100, lr: 7.78e-03, grad_scale: 4.0 2022-12-07 23:01:40,115 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0852, 4.8381, 4.3825, 4.6769, 4.6999, 4.9764, 5.0344, 5.0324], device='cuda:0'), covar=tensor([0.0686, 0.0382, 0.1992, 0.2540, 0.0645, 0.0718, 0.0763, 0.0751], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0245, 0.0416, 0.0537, 0.0313, 0.0403, 0.0387, 0.0347], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:02:22,362 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:02:51,954 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.183e+02 2.052e+02 2.533e+02 3.147e+02 6.445e+02, threshold=5.065e+02, percent-clipped=3.0 2022-12-07 23:02:57,100 INFO [train.py:873] (0/4) Epoch 10, batch 6700, loss[loss=0.1541, simple_loss=0.1495, pruned_loss=0.07933, over 2621.00 frames. ], tot_loss[loss=0.1338, simple_loss=0.1615, pruned_loss=0.05307, over 2003657.10 frames. ], batch size: 100, lr: 7.78e-03, grad_scale: 4.0 2022-12-07 23:03:04,996 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:04:19,048 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.393e+02 2.874e+02 3.943e+02 8.045e+02, threshold=5.748e+02, percent-clipped=8.0 2022-12-07 23:04:20,343 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5821, 2.8362, 4.3412, 3.3002, 4.3957, 4.3475, 4.0631, 3.8188], device='cuda:0'), covar=tensor([0.0598, 0.3009, 0.1125, 0.1779, 0.0714, 0.0766, 0.1826, 0.1702], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0317, 0.0395, 0.0307, 0.0371, 0.0318, 0.0361, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:04:24,812 INFO [train.py:873] (0/4) Epoch 10, batch 6800, loss[loss=0.1464, simple_loss=0.1715, pruned_loss=0.06068, over 14197.00 frames. ], tot_loss[loss=0.1355, simple_loss=0.1625, pruned_loss=0.05424, over 1992535.94 frames. ], batch size: 69, lr: 7.77e-03, grad_scale: 8.0 2022-12-07 23:04:56,625 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:08,601 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:38,775 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:45,384 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:05:47,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 2.283e+02 2.713e+02 3.632e+02 6.989e+02, threshold=5.426e+02, percent-clipped=3.0 2022-12-07 23:05:52,206 INFO [train.py:873] (0/4) Epoch 10, batch 6900, loss[loss=0.1485, simple_loss=0.1749, pruned_loss=0.06106, over 14347.00 frames. ], tot_loss[loss=0.136, simple_loss=0.1631, pruned_loss=0.05442, over 2014401.18 frames. ], batch size: 55, lr: 7.77e-03, grad_scale: 4.0 2022-12-07 23:05:56,628 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:06:01,554 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:06:27,590 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-75000.pt 2022-12-07 23:06:32,112 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.97 vs. limit=5.0 2022-12-07 23:06:52,479 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:07:16,832 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 2.278e+02 2.897e+02 3.473e+02 8.232e+02, threshold=5.794e+02, percent-clipped=1.0 2022-12-07 23:07:21,370 INFO [train.py:873] (0/4) Epoch 10, batch 7000, loss[loss=0.1494, simple_loss=0.1798, pruned_loss=0.05948, over 14197.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1626, pruned_loss=0.05409, over 1997411.41 frames. ], batch size: 80, lr: 7.76e-03, grad_scale: 4.0 2022-12-07 23:07:36,521 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1537, 1.5623, 1.7253, 1.6528, 1.6130, 1.7111, 1.3519, 1.2810], device='cuda:0'), covar=tensor([0.1645, 0.1168, 0.0390, 0.0468, 0.0994, 0.0733, 0.1921, 0.1546], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0074, 0.0058, 0.0063, 0.0089, 0.0072, 0.0093, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0005, 0.0006], device='cuda:0') 2022-12-07 23:07:55,690 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2022-12-07 23:08:29,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-07 23:08:45,730 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.185e+02 2.832e+02 3.460e+02 7.150e+02, threshold=5.664e+02, percent-clipped=2.0 2022-12-07 23:08:48,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-07 23:08:49,892 INFO [train.py:873] (0/4) Epoch 10, batch 7100, loss[loss=0.1478, simple_loss=0.1658, pruned_loss=0.06495, over 7774.00 frames. ], tot_loss[loss=0.1354, simple_loss=0.1622, pruned_loss=0.05425, over 1945336.34 frames. ], batch size: 100, lr: 7.76e-03, grad_scale: 4.0 2022-12-07 23:09:16,317 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5818, 1.5578, 1.4861, 1.4756, 1.6492, 0.9025, 1.6994, 1.4454], device='cuda:0'), covar=tensor([0.1062, 0.1011, 0.1296, 0.0739, 0.1194, 0.0939, 0.0824, 0.1391], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0027, 0.0024, 0.0025, 0.0037, 0.0025, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 23:09:51,409 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:09:57,239 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2022-12-07 23:10:11,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:13,566 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.322e+02 2.903e+02 3.521e+02 5.847e+02, threshold=5.807e+02, percent-clipped=2.0 2022-12-07 23:10:18,336 INFO [train.py:873] (0/4) Epoch 10, batch 7200, loss[loss=0.134, simple_loss=0.1633, pruned_loss=0.05239, over 14219.00 frames. ], tot_loss[loss=0.1353, simple_loss=0.1622, pruned_loss=0.05418, over 2009455.78 frames. ], batch size: 69, lr: 7.75e-03, grad_scale: 8.0 2022-12-07 23:10:23,618 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:28,166 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:44,820 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:10:53,490 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:11:11,940 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:11:21,372 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:11:39,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2892, 3.3644, 3.4137, 3.3070, 3.3897, 3.2076, 1.4699, 3.2230], device='cuda:0'), covar=tensor([0.0366, 0.0380, 0.0560, 0.0468, 0.0401, 0.0556, 0.3424, 0.0357], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0164, 0.0137, 0.0136, 0.0195, 0.0132, 0.0157, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 23:11:41,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 2.194e+02 2.882e+02 3.919e+02 8.679e+02, threshold=5.764e+02, percent-clipped=4.0 2022-12-07 23:11:46,065 INFO [train.py:873] (0/4) Epoch 10, batch 7300, loss[loss=0.1246, simple_loss=0.1544, pruned_loss=0.04742, over 5991.00 frames. ], tot_loss[loss=0.135, simple_loss=0.162, pruned_loss=0.05399, over 1962429.17 frames. ], batch size: 100, lr: 7.75e-03, grad_scale: 8.0 2022-12-07 23:12:34,377 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7726, 4.4191, 4.2239, 4.7869, 4.4859, 4.2393, 4.8088, 3.9369], device='cuda:0'), covar=tensor([0.0331, 0.0991, 0.0385, 0.0404, 0.0777, 0.0593, 0.0494, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0257, 0.0181, 0.0175, 0.0175, 0.0141, 0.0264, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 23:13:09,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.421e+02 2.835e+02 3.319e+02 6.698e+02, threshold=5.670e+02, percent-clipped=3.0 2022-12-07 23:13:14,008 INFO [train.py:873] (0/4) Epoch 10, batch 7400, loss[loss=0.133, simple_loss=0.147, pruned_loss=0.05952, over 4986.00 frames. ], tot_loss[loss=0.1337, simple_loss=0.1614, pruned_loss=0.05299, over 1994605.39 frames. ], batch size: 100, lr: 7.74e-03, grad_scale: 8.0 2022-12-07 23:13:34,748 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5584, 2.1419, 2.9855, 1.7879, 1.8450, 2.6759, 1.5114, 2.5330], device='cuda:0'), covar=tensor([0.0929, 0.1774, 0.0798, 0.2644, 0.2729, 0.0784, 0.4282, 0.1043], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0093, 0.0088, 0.0095, 0.0112, 0.0079, 0.0124, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-07 23:14:38,882 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.214e+02 2.894e+02 3.789e+02 9.059e+02, threshold=5.789e+02, percent-clipped=5.0 2022-12-07 23:14:42,300 INFO [train.py:873] (0/4) Epoch 10, batch 7500, loss[loss=0.1467, simple_loss=0.1688, pruned_loss=0.06228, over 11158.00 frames. ], tot_loss[loss=0.1352, simple_loss=0.1623, pruned_loss=0.05409, over 1968196.28 frames. ], batch size: 100, lr: 7.73e-03, grad_scale: 4.0 2022-12-07 23:14:47,592 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:15:04,134 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:15:21,410 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 23:15:24,290 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:15:29,504 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-10.pt 2022-12-07 23:16:08,910 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:16:09,626 INFO [train.py:873] (0/4) Epoch 11, batch 0, loss[loss=0.1396, simple_loss=0.1725, pruned_loss=0.05338, over 14209.00 frames. ], tot_loss[loss=0.1396, simple_loss=0.1725, pruned_loss=0.05338, over 14209.00 frames. ], batch size: 35, lr: 7.38e-03, grad_scale: 8.0 2022-12-07 23:16:09,627 INFO [train.py:896] (0/4) Computing validation loss 2022-12-07 23:16:16,872 INFO [train.py:905] (0/4) Epoch 11, validation: loss=0.1341, simple_loss=0.1756, pruned_loss=0.0463, over 857387.00 frames. 2022-12-07 23:16:16,873 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-07 23:16:21,384 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:16:27,155 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-07 23:16:47,137 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.744e+01 1.774e+02 2.985e+02 4.009e+02 1.076e+03, threshold=5.970e+02, percent-clipped=9.0 2022-12-07 23:16:58,371 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:17:45,591 INFO [train.py:873] (0/4) Epoch 11, batch 100, loss[loss=0.1327, simple_loss=0.1643, pruned_loss=0.05057, over 14295.00 frames. ], tot_loss[loss=0.1351, simple_loss=0.1625, pruned_loss=0.05389, over 840694.76 frames. ], batch size: 39, lr: 7.38e-03, grad_scale: 8.0 2022-12-07 23:18:14,580 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.145e+02 2.643e+02 3.199e+02 5.049e+02, threshold=5.286e+02, percent-clipped=0.0 2022-12-07 23:19:13,452 INFO [train.py:873] (0/4) Epoch 11, batch 200, loss[loss=0.1449, simple_loss=0.1389, pruned_loss=0.07541, over 1271.00 frames. ], tot_loss[loss=0.1355, simple_loss=0.1621, pruned_loss=0.05447, over 1211547.44 frames. ], batch size: 100, lr: 7.37e-03, grad_scale: 8.0 2022-12-07 23:19:14,134 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-07 23:19:14,445 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6250, 2.4719, 2.2024, 2.3343, 2.5262, 2.5554, 2.6019, 2.5800], device='cuda:0'), covar=tensor([0.0963, 0.0800, 0.2531, 0.2627, 0.1077, 0.1160, 0.1363, 0.0889], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0242, 0.0417, 0.0536, 0.0311, 0.0398, 0.0378, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:19:43,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.275e+02 2.829e+02 3.715e+02 6.696e+02, threshold=5.659e+02, percent-clipped=6.0 2022-12-07 23:20:08,858 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:20:41,220 INFO [train.py:873] (0/4) Epoch 11, batch 300, loss[loss=0.1311, simple_loss=0.1635, pruned_loss=0.04932, over 14227.00 frames. ], tot_loss[loss=0.133, simple_loss=0.1601, pruned_loss=0.05294, over 1452005.54 frames. ], batch size: 60, lr: 7.37e-03, grad_scale: 8.0 2022-12-07 23:20:46,315 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:20:51,522 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:21:02,092 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-07 23:21:02,693 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0885, 1.4976, 4.0571, 1.7812, 3.9172, 4.1316, 3.2491, 4.4522], device='cuda:0'), covar=tensor([0.0206, 0.3020, 0.0343, 0.2248, 0.0403, 0.0389, 0.0660, 0.0126], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0156, 0.0156, 0.0167, 0.0171, 0.0174, 0.0132, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 23:21:11,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.182e+02 2.664e+02 3.344e+02 7.501e+02, threshold=5.329e+02, percent-clipped=3.0 2022-12-07 23:21:24,523 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1283, 3.6298, 3.1112, 4.4744, 4.1251, 4.2279, 3.7366, 3.1509], device='cuda:0'), covar=tensor([0.0866, 0.1453, 0.3388, 0.0495, 0.1221, 0.1324, 0.1191, 0.3090], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0298, 0.0269, 0.0247, 0.0306, 0.0292, 0.0254, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 23:21:28,761 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:22:10,197 INFO [train.py:873] (0/4) Epoch 11, batch 400, loss[loss=0.1241, simple_loss=0.1573, pruned_loss=0.04545, over 14230.00 frames. ], tot_loss[loss=0.1328, simple_loss=0.16, pruned_loss=0.0528, over 1639270.36 frames. ], batch size: 69, lr: 7.36e-03, grad_scale: 8.0 2022-12-07 23:22:40,270 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.357e+02 2.915e+02 3.546e+02 6.707e+02, threshold=5.830e+02, percent-clipped=5.0 2022-12-07 23:23:38,043 INFO [train.py:873] (0/4) Epoch 11, batch 500, loss[loss=0.1406, simple_loss=0.1631, pruned_loss=0.05904, over 11972.00 frames. ], tot_loss[loss=0.1343, simple_loss=0.1612, pruned_loss=0.05375, over 1729236.35 frames. ], batch size: 100, lr: 7.36e-03, grad_scale: 8.0 2022-12-07 23:24:07,886 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.432e+01 2.214e+02 2.601e+02 3.343e+02 5.170e+02, threshold=5.201e+02, percent-clipped=0.0 2022-12-07 23:25:05,640 INFO [train.py:873] (0/4) Epoch 11, batch 600, loss[loss=0.1197, simple_loss=0.1572, pruned_loss=0.04109, over 14568.00 frames. ], tot_loss[loss=0.1323, simple_loss=0.1601, pruned_loss=0.0522, over 1841028.91 frames. ], batch size: 22, lr: 7.35e-03, grad_scale: 8.0 2022-12-07 23:25:28,351 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9056, 4.5609, 4.4355, 4.9661, 4.5004, 4.1523, 4.9034, 4.8449], device='cuda:0'), covar=tensor([0.0547, 0.0618, 0.0701, 0.0433, 0.0567, 0.0571, 0.0563, 0.0519], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0123, 0.0130, 0.0141, 0.0132, 0.0111, 0.0156, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:25:35,309 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 2.199e+02 2.747e+02 3.310e+02 7.404e+02, threshold=5.494e+02, percent-clipped=4.0 2022-12-07 23:25:45,120 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-07 23:25:47,481 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:26:05,095 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6288, 1.6691, 4.4368, 1.9314, 4.2488, 4.5671, 4.1219, 5.0028], device='cuda:0'), covar=tensor([0.0187, 0.2836, 0.0290, 0.2216, 0.0334, 0.0312, 0.0296, 0.0135], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0158, 0.0158, 0.0168, 0.0172, 0.0175, 0.0133, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 23:26:08,216 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2022-12-07 23:26:09,591 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:26:32,273 INFO [train.py:873] (0/4) Epoch 11, batch 700, loss[loss=0.1106, simple_loss=0.1466, pruned_loss=0.03728, over 13931.00 frames. ], tot_loss[loss=0.1319, simple_loss=0.16, pruned_loss=0.0519, over 1896223.85 frames. ], batch size: 20, lr: 7.35e-03, grad_scale: 8.0 2022-12-07 23:26:34,473 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:26:39,082 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2022-12-07 23:26:40,826 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:27:01,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.063e+02 2.502e+02 2.955e+02 5.307e+02, threshold=5.004e+02, percent-clipped=0.0 2022-12-07 23:27:02,091 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:27:27,272 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:27:58,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7433, 2.3105, 3.1492, 2.1622, 2.0568, 2.7644, 1.6659, 2.6268], device='cuda:0'), covar=tensor([0.1001, 0.1476, 0.0724, 0.2044, 0.2599, 0.0787, 0.3906, 0.1065], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0094, 0.0090, 0.0095, 0.0114, 0.0081, 0.0125, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-07 23:27:59,959 INFO [train.py:873] (0/4) Epoch 11, batch 800, loss[loss=0.1111, simple_loss=0.151, pruned_loss=0.03563, over 14281.00 frames. ], tot_loss[loss=0.1307, simple_loss=0.1593, pruned_loss=0.05108, over 1885943.49 frames. ], batch size: 31, lr: 7.34e-03, grad_scale: 8.0 2022-12-07 23:28:12,228 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0553, 2.5134, 3.4716, 2.4962, 2.1244, 2.9512, 1.6697, 2.9181], device='cuda:0'), covar=tensor([0.1110, 0.1384, 0.0767, 0.1658, 0.2545, 0.0953, 0.4183, 0.1176], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0094, 0.0090, 0.0095, 0.0114, 0.0081, 0.0125, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-07 23:28:29,451 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.357e+02 2.806e+02 3.598e+02 6.382e+02, threshold=5.612e+02, percent-clipped=5.0 2022-12-07 23:28:37,179 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8693, 4.4357, 4.2664, 4.8362, 4.5642, 4.2535, 4.8216, 3.9473], device='cuda:0'), covar=tensor([0.0372, 0.1288, 0.0411, 0.0475, 0.0812, 0.0670, 0.0593, 0.0632], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0262, 0.0186, 0.0179, 0.0176, 0.0145, 0.0268, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 23:29:26,690 INFO [train.py:873] (0/4) Epoch 11, batch 900, loss[loss=0.1422, simple_loss=0.1685, pruned_loss=0.05794, over 8611.00 frames. ], tot_loss[loss=0.1329, simple_loss=0.1606, pruned_loss=0.05261, over 1888348.21 frames. ], batch size: 100, lr: 7.34e-03, grad_scale: 8.0 2022-12-07 23:29:47,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1521, 1.9770, 2.0768, 2.1882, 2.0770, 2.0497, 2.2394, 1.9021], device='cuda:0'), covar=tensor([0.0941, 0.1415, 0.0736, 0.0769, 0.0982, 0.0789, 0.0921, 0.0786], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0262, 0.0185, 0.0179, 0.0176, 0.0145, 0.0267, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-07 23:29:56,493 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.233e+02 2.846e+02 3.366e+02 5.399e+02, threshold=5.691e+02, percent-clipped=0.0 2022-12-07 23:30:16,903 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9152, 2.4725, 3.6378, 2.7424, 3.8358, 3.6215, 3.4911, 3.0553], device='cuda:0'), covar=tensor([0.0742, 0.2971, 0.1205, 0.2023, 0.0879, 0.0869, 0.1431, 0.2014], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0323, 0.0400, 0.0305, 0.0381, 0.0323, 0.0363, 0.0316], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:30:54,355 INFO [train.py:873] (0/4) Epoch 11, batch 1000, loss[loss=0.1841, simple_loss=0.1647, pruned_loss=0.1017, over 1266.00 frames. ], tot_loss[loss=0.133, simple_loss=0.1608, pruned_loss=0.0526, over 1902172.30 frames. ], batch size: 100, lr: 7.33e-03, grad_scale: 8.0 2022-12-07 23:30:57,850 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:31:10,941 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2105, 1.7354, 2.5046, 2.0402, 2.2911, 1.7100, 2.0425, 2.2861], device='cuda:0'), covar=tensor([0.1904, 0.2642, 0.0344, 0.2041, 0.0889, 0.1788, 0.0981, 0.0589], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0218, 0.0198, 0.0292, 0.0215, 0.0218, 0.0214, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:31:19,955 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:31:24,209 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.175e+02 2.720e+02 3.257e+02 6.419e+02, threshold=5.440e+02, percent-clipped=2.0 2022-12-07 23:31:44,632 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:31:49,009 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7444, 1.2617, 3.0111, 2.7315, 2.8979, 3.0131, 2.2303, 2.9836], device='cuda:0'), covar=tensor([0.1417, 0.1569, 0.0168, 0.0359, 0.0339, 0.0204, 0.0441, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0124, 0.0168, 0.0142, 0.0135, 0.0116, 0.0117], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:31:59,760 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-07 23:32:21,331 INFO [train.py:873] (0/4) Epoch 11, batch 1100, loss[loss=0.1254, simple_loss=0.1599, pruned_loss=0.0454, over 14307.00 frames. ], tot_loss[loss=0.1325, simple_loss=0.1603, pruned_loss=0.0523, over 1903323.53 frames. ], batch size: 46, lr: 7.33e-03, grad_scale: 8.0 2022-12-07 23:32:40,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2022-12-07 23:32:50,793 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.125e+02 2.272e+02 2.937e+02 3.615e+02 8.026e+02, threshold=5.874e+02, percent-clipped=3.0 2022-12-07 23:32:57,366 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 23:33:49,946 INFO [train.py:873] (0/4) Epoch 11, batch 1200, loss[loss=0.1154, simple_loss=0.1533, pruned_loss=0.03879, over 14294.00 frames. ], tot_loss[loss=0.1317, simple_loss=0.1595, pruned_loss=0.05189, over 1845966.93 frames. ], batch size: 63, lr: 7.32e-03, grad_scale: 8.0 2022-12-07 23:33:58,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 23:34:05,557 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:34:12,532 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2286, 4.9672, 4.6886, 5.2645, 4.7727, 4.5262, 5.2826, 5.0672], device='cuda:0'), covar=tensor([0.0659, 0.0633, 0.0733, 0.0560, 0.0644, 0.0451, 0.0562, 0.0725], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0121, 0.0130, 0.0138, 0.0131, 0.0107, 0.0152, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:34:19,397 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.391e+02 2.914e+02 3.659e+02 7.427e+02, threshold=5.827e+02, percent-clipped=2.0 2022-12-07 23:34:46,593 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0284, 2.0971, 1.8577, 2.1585, 1.6806, 1.9366, 2.0669, 2.0686], device='cuda:0'), covar=tensor([0.0966, 0.1029, 0.1241, 0.0815, 0.1664, 0.0978, 0.1086, 0.0896], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0120, 0.0129, 0.0137, 0.0131, 0.0107, 0.0150, 0.0130], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:34:57,505 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1705, 4.2281, 4.4696, 3.7120, 4.3177, 4.4817, 1.5587, 4.0810], device='cuda:0'), covar=tensor([0.0233, 0.0331, 0.0352, 0.0620, 0.0285, 0.0223, 0.3161, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0164, 0.0137, 0.0136, 0.0193, 0.0131, 0.0156, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:34:59,334 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:35:17,347 INFO [train.py:873] (0/4) Epoch 11, batch 1300, loss[loss=0.1494, simple_loss=0.1492, pruned_loss=0.07484, over 2659.00 frames. ], tot_loss[loss=0.1316, simple_loss=0.1596, pruned_loss=0.05181, over 1828831.91 frames. ], batch size: 100, lr: 7.32e-03, grad_scale: 8.0 2022-12-07 23:35:20,706 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={2} 2022-12-07 23:35:42,602 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:35:44,649 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2816, 1.3708, 3.3996, 1.4866, 3.0942, 3.3706, 2.3957, 3.6114], device='cuda:0'), covar=tensor([0.0249, 0.3037, 0.0354, 0.2289, 0.0926, 0.0448, 0.0860, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0157, 0.0157, 0.0168, 0.0170, 0.0173, 0.0133, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 23:35:47,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.104e+02 2.573e+02 3.185e+02 4.865e+02, threshold=5.146e+02, percent-clipped=0.0 2022-12-07 23:36:02,816 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:36:08,340 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:36:24,625 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:36:35,236 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-07 23:36:44,064 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-07 23:36:44,709 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3202, 1.4368, 1.5866, 1.1758, 1.3139, 1.1042, 0.9036, 0.9356], device='cuda:0'), covar=tensor([0.0191, 0.0322, 0.0161, 0.0183, 0.0188, 0.0331, 0.0242, 0.0449], device='cuda:0'), in_proj_covar=tensor([0.0014, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-07 23:36:45,317 INFO [train.py:873] (0/4) Epoch 11, batch 1400, loss[loss=0.1673, simple_loss=0.18, pruned_loss=0.07734, over 7746.00 frames. ], tot_loss[loss=0.1322, simple_loss=0.1601, pruned_loss=0.05213, over 1842289.16 frames. ], batch size: 100, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:36:50,504 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={1} 2022-12-07 23:36:51,393 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9461, 1.8719, 3.2431, 2.3832, 3.1133, 1.7902, 2.4026, 3.0399], device='cuda:0'), covar=tensor([0.0885, 0.4325, 0.0541, 0.5872, 0.0850, 0.3473, 0.1363, 0.0676], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0217, 0.0199, 0.0291, 0.0217, 0.0218, 0.0214, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:37:14,871 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.332e+02 2.864e+02 3.537e+02 8.782e+02, threshold=5.729e+02, percent-clipped=8.0 2022-12-07 23:37:38,396 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2022-12-07 23:37:47,341 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-07 23:37:49,620 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:38:05,909 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-07 23:38:12,197 INFO [train.py:873] (0/4) Epoch 11, batch 1500, loss[loss=0.147, simple_loss=0.1514, pruned_loss=0.07127, over 2582.00 frames. ], tot_loss[loss=0.1316, simple_loss=0.1601, pruned_loss=0.05157, over 1929881.17 frames. ], batch size: 100, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:38:17,570 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7965, 1.3279, 3.7089, 1.5188, 3.6626, 3.8672, 2.8211, 4.1072], device='cuda:0'), covar=tensor([0.0214, 0.3247, 0.0420, 0.2504, 0.0479, 0.0343, 0.0779, 0.0159], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0155, 0.0155, 0.0166, 0.0168, 0.0173, 0.0131, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 23:38:41,966 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.474e+02 2.983e+02 4.107e+02 8.785e+02, threshold=5.967e+02, percent-clipped=9.0 2022-12-07 23:38:43,044 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:39:17,244 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:39:30,257 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2704, 4.0015, 3.9641, 4.3395, 3.8449, 3.6504, 4.3226, 4.2152], device='cuda:0'), covar=tensor([0.0597, 0.0831, 0.0745, 0.0531, 0.0816, 0.0654, 0.0582, 0.0646], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0124, 0.0133, 0.0142, 0.0134, 0.0110, 0.0155, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:39:32,277 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:39:40,746 INFO [train.py:873] (0/4) Epoch 11, batch 1600, loss[loss=0.1404, simple_loss=0.1685, pruned_loss=0.05612, over 14179.00 frames. ], tot_loss[loss=0.1314, simple_loss=0.1599, pruned_loss=0.05145, over 1952429.67 frames. ], batch size: 99, lr: 7.31e-03, grad_scale: 8.0 2022-12-07 23:40:10,413 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.174e+02 1.982e+02 2.587e+02 3.263e+02 7.799e+02, threshold=5.175e+02, percent-clipped=3.0 2022-12-07 23:40:17,232 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4302, 2.1972, 3.3879, 3.4686, 3.4251, 2.1999, 3.4111, 2.6641], device='cuda:0'), covar=tensor([0.0358, 0.0807, 0.0566, 0.0420, 0.0347, 0.1232, 0.0331, 0.0793], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0250, 0.0369, 0.0317, 0.0261, 0.0298, 0.0293, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:40:25,853 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:26,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:30,469 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:40:53,069 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-07 23:41:08,995 INFO [train.py:873] (0/4) Epoch 11, batch 1700, loss[loss=0.1175, simple_loss=0.1168, pruned_loss=0.05912, over 1329.00 frames. ], tot_loss[loss=0.1319, simple_loss=0.16, pruned_loss=0.05194, over 1927376.24 frames. ], batch size: 100, lr: 7.30e-03, grad_scale: 8.0 2022-12-07 23:41:11,718 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3338, 2.3956, 1.9138, 2.4732, 2.2827, 2.3556, 2.1069, 2.0091], device='cuda:0'), covar=tensor([0.0794, 0.0799, 0.2219, 0.0567, 0.0737, 0.0509, 0.1277, 0.1348], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0299, 0.0271, 0.0250, 0.0306, 0.0290, 0.0257, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 23:41:20,844 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:41:24,009 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:41:39,032 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.095e+02 2.641e+02 3.320e+02 6.995e+02, threshold=5.282e+02, percent-clipped=0.0 2022-12-07 23:41:42,656 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9850, 3.0637, 3.1792, 3.0683, 3.1004, 2.9551, 1.3936, 2.8975], device='cuda:0'), covar=tensor([0.0383, 0.0430, 0.0422, 0.0414, 0.0415, 0.0675, 0.3028, 0.0339], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0166, 0.0138, 0.0137, 0.0196, 0.0134, 0.0157, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-07 23:41:45,261 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1194, 2.1676, 4.9520, 4.5115, 4.4109, 5.0369, 4.8579, 5.0609], device='cuda:0'), covar=tensor([0.1322, 0.1317, 0.0076, 0.0159, 0.0171, 0.0097, 0.0096, 0.0089], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0160, 0.0125, 0.0169, 0.0142, 0.0136, 0.0117, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:42:37,340 INFO [train.py:873] (0/4) Epoch 11, batch 1800, loss[loss=0.1457, simple_loss=0.16, pruned_loss=0.06569, over 4991.00 frames. ], tot_loss[loss=0.1321, simple_loss=0.1605, pruned_loss=0.05183, over 1976572.54 frames. ], batch size: 100, lr: 7.30e-03, grad_scale: 8.0 2022-12-07 23:43:04,200 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={3} 2022-12-07 23:43:07,257 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.136e+02 2.704e+02 3.283e+02 4.688e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-07 23:43:42,022 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2022-12-07 23:43:42,411 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:44:00,909 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8799, 1.8250, 1.5811, 1.9754, 1.8489, 1.8784, 1.7670, 1.7004], device='cuda:0'), covar=tensor([0.1190, 0.0977, 0.2064, 0.0576, 0.0868, 0.0589, 0.1508, 0.0840], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0299, 0.0271, 0.0249, 0.0306, 0.0290, 0.0257, 0.0255], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-07 23:44:05,123 INFO [train.py:873] (0/4) Epoch 11, batch 1900, loss[loss=0.1227, simple_loss=0.1701, pruned_loss=0.03768, over 13968.00 frames. ], tot_loss[loss=0.1332, simple_loss=0.1608, pruned_loss=0.05282, over 1914560.94 frames. ], batch size: 23, lr: 7.29e-03, grad_scale: 8.0 2022-12-07 23:44:24,710 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:44:36,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.774e+01 2.327e+02 2.859e+02 3.630e+02 7.780e+02, threshold=5.718e+02, percent-clipped=3.0 2022-12-07 23:44:45,944 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:44:52,396 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4390, 1.9637, 2.3576, 2.0554, 2.4138, 2.3180, 2.1762, 2.2015], device='cuda:0'), covar=tensor([0.0557, 0.2227, 0.0491, 0.1209, 0.0423, 0.0877, 0.0556, 0.1049], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0316, 0.0398, 0.0304, 0.0376, 0.0318, 0.0361, 0.0315], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:45:29,326 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:33,420 INFO [train.py:873] (0/4) Epoch 11, batch 2000, loss[loss=0.1191, simple_loss=0.1512, pruned_loss=0.0435, over 12760.00 frames. ], tot_loss[loss=0.1332, simple_loss=0.1607, pruned_loss=0.05284, over 1887255.09 frames. ], batch size: 100, lr: 7.29e-03, grad_scale: 8.0 2022-12-07 23:45:40,502 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:44,035 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:45:50,741 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-07 23:45:59,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9606, 1.1151, 1.1807, 0.7422, 0.7815, 0.7433, 0.7243, 0.7614], device='cuda:0'), covar=tensor([0.0223, 0.0210, 0.0197, 0.0239, 0.0270, 0.0540, 0.0345, 0.0554], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0015, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-07 23:46:04,192 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.333e+02 2.909e+02 3.438e+02 9.067e+02, threshold=5.818e+02, percent-clipped=5.0 2022-12-07 23:46:21,827 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:46:21,847 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:46:59,455 INFO [train.py:873] (0/4) Epoch 11, batch 2100, loss[loss=0.1191, simple_loss=0.1508, pruned_loss=0.04372, over 14109.00 frames. ], tot_loss[loss=0.133, simple_loss=0.1606, pruned_loss=0.05273, over 1902444.40 frames. ], batch size: 29, lr: 7.28e-03, grad_scale: 8.0 2022-12-07 23:47:14,268 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:47:25,621 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:47:29,642 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 2.408e+02 3.077e+02 4.325e+02 1.206e+03, threshold=6.153e+02, percent-clipped=10.0 2022-12-07 23:48:07,052 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:48:26,366 INFO [train.py:873] (0/4) Epoch 11, batch 2200, loss[loss=0.134, simple_loss=0.1589, pruned_loss=0.05456, over 11133.00 frames. ], tot_loss[loss=0.133, simple_loss=0.1604, pruned_loss=0.05282, over 1900510.12 frames. ], batch size: 100, lr: 7.28e-03, grad_scale: 8.0 2022-12-07 23:48:57,072 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.100e+02 2.823e+02 3.472e+02 8.203e+02, threshold=5.646e+02, percent-clipped=4.0 2022-12-07 23:49:06,546 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:49:13,455 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0278, 3.1650, 3.2467, 3.3157, 2.4380, 3.3178, 3.2275, 1.5794], device='cuda:0'), covar=tensor([0.2139, 0.1275, 0.1072, 0.0810, 0.1120, 0.0907, 0.1092, 0.2801], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0078, 0.0061, 0.0065, 0.0093, 0.0075, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-07 23:49:21,718 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-07 23:49:46,595 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8689, 3.5048, 3.2237, 2.9931, 3.2292, 3.4745, 3.8958, 3.1813], device='cuda:0'), covar=tensor([0.0592, 0.1674, 0.1033, 0.1419, 0.1054, 0.0794, 0.1045, 0.1271], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0181, 0.0134, 0.0125, 0.0133, 0.0139, 0.0115, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-07 23:49:48,079 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:49:53,346 INFO [train.py:873] (0/4) Epoch 11, batch 2300, loss[loss=0.1158, simple_loss=0.1555, pruned_loss=0.03811, over 14287.00 frames. ], tot_loss[loss=0.1316, simple_loss=0.1595, pruned_loss=0.05183, over 1915991.46 frames. ], batch size: 44, lr: 7.27e-03, grad_scale: 8.0 2022-12-07 23:50:00,478 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:04,132 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:23,859 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.313e+02 2.906e+02 3.798e+02 8.346e+02, threshold=5.811e+02, percent-clipped=6.0 2022-12-07 23:50:26,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6850, 1.4003, 3.6230, 1.4023, 3.5559, 3.7195, 2.4752, 3.9885], device='cuda:0'), covar=tensor([0.0211, 0.3040, 0.0404, 0.2395, 0.0589, 0.0358, 0.0879, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0157, 0.0158, 0.0168, 0.0171, 0.0174, 0.0132, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-07 23:50:38,036 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1626, 3.9829, 3.8484, 4.2051, 3.8041, 3.4584, 4.2046, 4.0439], device='cuda:0'), covar=tensor([0.0708, 0.0818, 0.0810, 0.0602, 0.0861, 0.0769, 0.0627, 0.0798], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0125, 0.0133, 0.0142, 0.0135, 0.0111, 0.0156, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:50:38,063 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:42,324 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:50:45,713 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:51:11,488 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0628, 1.8854, 4.6158, 4.2403, 4.2061, 4.7559, 4.3069, 4.7123], device='cuda:0'), covar=tensor([0.1327, 0.1311, 0.0084, 0.0178, 0.0164, 0.0097, 0.0115, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0125, 0.0168, 0.0143, 0.0136, 0.0119, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:51:21,249 INFO [train.py:873] (0/4) Epoch 11, batch 2400, loss[loss=0.1264, simple_loss=0.1479, pruned_loss=0.05245, over 5960.00 frames. ], tot_loss[loss=0.1316, simple_loss=0.16, pruned_loss=0.05165, over 1939947.98 frames. ], batch size: 100, lr: 7.27e-03, grad_scale: 8.0 2022-12-07 23:51:32,077 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:51:51,986 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.203e+02 2.586e+02 3.154e+02 7.524e+02, threshold=5.173e+02, percent-clipped=2.0 2022-12-07 23:52:35,137 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-07 23:52:49,055 INFO [train.py:873] (0/4) Epoch 11, batch 2500, loss[loss=0.1508, simple_loss=0.158, pruned_loss=0.07183, over 5020.00 frames. ], tot_loss[loss=0.1306, simple_loss=0.1596, pruned_loss=0.05074, over 2010076.29 frames. ], batch size: 100, lr: 7.26e-03, grad_scale: 8.0 2022-12-07 23:53:19,543 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.052e+02 2.774e+02 3.543e+02 6.822e+02, threshold=5.549e+02, percent-clipped=2.0 2022-12-07 23:54:09,068 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0718, 1.0477, 0.9436, 0.9757, 1.0809, 0.5910, 0.9304, 1.0116], device='cuda:0'), covar=tensor([0.0564, 0.1103, 0.0614, 0.0615, 0.0479, 0.0681, 0.0799, 0.0601], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0026, 0.0028, 0.0025, 0.0027, 0.0038, 0.0026, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-07 23:54:16,599 INFO [train.py:873] (0/4) Epoch 11, batch 2600, loss[loss=0.1504, simple_loss=0.1712, pruned_loss=0.06477, over 6936.00 frames. ], tot_loss[loss=0.1307, simple_loss=0.1598, pruned_loss=0.05083, over 1981896.55 frames. ], batch size: 100, lr: 7.26e-03, grad_scale: 8.0 2022-12-07 23:54:26,554 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:54:28,154 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9617, 1.9340, 1.6952, 2.0503, 1.7970, 1.9664, 1.8495, 1.7355], device='cuda:0'), covar=tensor([0.0902, 0.0952, 0.1963, 0.0548, 0.0895, 0.0655, 0.1501, 0.0864], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0301, 0.0276, 0.0253, 0.0312, 0.0294, 0.0259, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:54:37,524 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8497, 2.1382, 2.7201, 2.3366, 2.7408, 2.8058, 2.6317, 2.3596], device='cuda:0'), covar=tensor([0.0773, 0.2692, 0.0838, 0.1640, 0.0577, 0.0833, 0.1012, 0.1708], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0320, 0.0399, 0.0305, 0.0384, 0.0324, 0.0368, 0.0319], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-07 23:54:47,973 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.197e+02 2.874e+02 3.504e+02 7.274e+02, threshold=5.748e+02, percent-clipped=5.0 2022-12-07 23:55:00,962 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:55:13,351 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-07 23:55:19,349 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={0} 2022-12-07 23:55:42,558 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:55:43,477 INFO [train.py:873] (0/4) Epoch 11, batch 2700, loss[loss=0.1531, simple_loss=0.1428, pruned_loss=0.08171, over 1176.00 frames. ], tot_loss[loss=0.1307, simple_loss=0.1597, pruned_loss=0.05085, over 2004734.34 frames. ], batch size: 100, lr: 7.25e-03, grad_scale: 4.0 2022-12-07 23:55:54,592 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:56:15,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.898e+01 2.288e+02 2.843e+02 3.409e+02 7.759e+02, threshold=5.687e+02, percent-clipped=1.0 2022-12-07 23:56:23,454 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-07 23:56:29,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-07 23:56:36,322 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:57:03,991 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7798, 2.0189, 2.8274, 2.8747, 2.7061, 1.9961, 2.8078, 2.2047], device='cuda:0'), covar=tensor([0.0346, 0.0831, 0.0433, 0.0326, 0.0399, 0.1048, 0.0280, 0.0741], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0245, 0.0362, 0.0310, 0.0253, 0.0293, 0.0285, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-07 23:57:11,629 INFO [train.py:873] (0/4) Epoch 11, batch 2800, loss[loss=0.1674, simple_loss=0.1758, pruned_loss=0.07951, over 9510.00 frames. ], tot_loss[loss=0.1294, simple_loss=0.1588, pruned_loss=0.05002, over 2011012.11 frames. ], batch size: 100, lr: 7.25e-03, grad_scale: 8.0 2022-12-07 23:57:42,813 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0487, 2.4011, 4.0266, 4.1354, 4.0453, 2.4847, 4.0300, 3.2370], device='cuda:0'), covar=tensor([0.0298, 0.0800, 0.0679, 0.0337, 0.0303, 0.1258, 0.0314, 0.0745], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0246, 0.0362, 0.0310, 0.0253, 0.0294, 0.0285, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:57:43,400 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.311e+02 2.686e+02 3.123e+02 7.325e+02, threshold=5.372e+02, percent-clipped=4.0 2022-12-07 23:57:46,014 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:58:39,403 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8176, 0.7189, 0.7697, 0.7068, 0.7224, 0.4664, 0.4954, 0.7202], device='cuda:0'), covar=tensor([0.0117, 0.0144, 0.0094, 0.0097, 0.0147, 0.0287, 0.0173, 0.0223], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0014, 0.0014, 0.0024, 0.0019, 0.0024], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-07 23:58:40,122 INFO [train.py:873] (0/4) Epoch 11, batch 2900, loss[loss=0.1555, simple_loss=0.1476, pruned_loss=0.08166, over 1284.00 frames. ], tot_loss[loss=0.1299, simple_loss=0.159, pruned_loss=0.05042, over 1982715.24 frames. ], batch size: 100, lr: 7.24e-03, grad_scale: 8.0 2022-12-07 23:58:40,334 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:59:00,637 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2022-12-07 23:59:10,552 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2022-12-07 23:59:11,184 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 2.237e+02 2.916e+02 3.680e+02 6.575e+02, threshold=5.831e+02, percent-clipped=6.0 2022-12-07 23:59:16,920 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8363, 3.9615, 4.1743, 3.7150, 4.0238, 4.1289, 1.5091, 3.7688], device='cuda:0'), covar=tensor([0.0300, 0.0326, 0.0324, 0.0451, 0.0284, 0.0258, 0.3179, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0165, 0.0138, 0.0137, 0.0194, 0.0133, 0.0154, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-07 23:59:38,736 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:00:05,474 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:00:08,934 INFO [train.py:873] (0/4) Epoch 11, batch 3000, loss[loss=0.1296, simple_loss=0.1579, pruned_loss=0.05059, over 12761.00 frames. ], tot_loss[loss=0.1308, simple_loss=0.1592, pruned_loss=0.05114, over 1973083.47 frames. ], batch size: 100, lr: 7.24e-03, grad_scale: 8.0 2022-12-08 00:00:08,934 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 00:00:17,381 INFO [train.py:905] (0/4) Epoch 11, validation: loss=0.1282, simple_loss=0.1681, pruned_loss=0.04413, over 857387.00 frames. 2022-12-08 00:00:17,382 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 00:00:17,573 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7517, 2.4950, 2.5911, 1.7182, 2.3241, 2.5884, 2.8041, 2.3638], device='cuda:0'), covar=tensor([0.0674, 0.1044, 0.1093, 0.1627, 0.1215, 0.0746, 0.0533, 0.1312], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0183, 0.0136, 0.0125, 0.0134, 0.0141, 0.0117, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:00:48,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.287e+02 2.909e+02 3.866e+02 8.652e+02, threshold=5.818e+02, percent-clipped=3.0 2022-12-08 00:00:59,153 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8824, 2.6430, 2.7339, 1.7595, 2.4422, 2.7010, 2.9599, 2.4330], device='cuda:0'), covar=tensor([0.0701, 0.1138, 0.1002, 0.1791, 0.0992, 0.0732, 0.0656, 0.1427], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0184, 0.0136, 0.0126, 0.0134, 0.0142, 0.0118, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:01:05,051 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7052, 3.4983, 3.3654, 2.5013, 3.2750, 3.5659, 4.0317, 3.1509], device='cuda:0'), covar=tensor([0.0656, 0.1940, 0.0889, 0.1698, 0.0998, 0.0698, 0.0536, 0.1221], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0184, 0.0136, 0.0126, 0.0134, 0.0142, 0.0118, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:01:32,933 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9092, 5.3960, 5.3246, 5.9199, 5.4925, 4.8329, 5.8467, 4.7532], device='cuda:0'), covar=tensor([0.0313, 0.0828, 0.0293, 0.0343, 0.0649, 0.0307, 0.0431, 0.0538], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0259, 0.0181, 0.0176, 0.0173, 0.0140, 0.0264, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:01:46,386 INFO [train.py:873] (0/4) Epoch 11, batch 3100, loss[loss=0.1029, simple_loss=0.1418, pruned_loss=0.03197, over 14240.00 frames. ], tot_loss[loss=0.1305, simple_loss=0.1589, pruned_loss=0.05102, over 2010978.75 frames. ], batch size: 57, lr: 7.24e-03, grad_scale: 8.0 2022-12-08 00:02:18,621 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.078e+02 2.524e+02 3.137e+02 7.110e+02, threshold=5.049e+02, percent-clipped=1.0 2022-12-08 00:03:11,321 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:03:15,284 INFO [train.py:873] (0/4) Epoch 11, batch 3200, loss[loss=0.1228, simple_loss=0.1597, pruned_loss=0.04295, over 14215.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1585, pruned_loss=0.04997, over 2044671.72 frames. ], batch size: 35, lr: 7.23e-03, grad_scale: 8.0 2022-12-08 00:03:21,844 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1431, 2.9067, 3.8114, 2.6893, 2.1953, 3.1153, 1.6623, 3.2014], device='cuda:0'), covar=tensor([0.1200, 0.1371, 0.0624, 0.2231, 0.2520, 0.1064, 0.4354, 0.0992], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0094, 0.0090, 0.0094, 0.0112, 0.0082, 0.0124, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:03:46,996 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.001e+02 2.170e+02 2.756e+02 3.335e+02 6.522e+02, threshold=5.512e+02, percent-clipped=3.0 2022-12-08 00:04:14,664 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:04:35,330 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:04:43,622 INFO [train.py:873] (0/4) Epoch 11, batch 3300, loss[loss=0.1175, simple_loss=0.1537, pruned_loss=0.04068, over 14561.00 frames. ], tot_loss[loss=0.1295, simple_loss=0.1583, pruned_loss=0.0503, over 1987537.28 frames. ], batch size: 43, lr: 7.23e-03, grad_scale: 8.0 2022-12-08 00:04:56,443 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:05:15,456 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.245e+02 2.789e+02 3.616e+02 5.775e+02, threshold=5.578e+02, percent-clipped=1.0 2022-12-08 00:05:22,607 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:06:10,656 INFO [train.py:873] (0/4) Epoch 11, batch 3400, loss[loss=0.1173, simple_loss=0.1534, pruned_loss=0.04066, over 14224.00 frames. ], tot_loss[loss=0.1306, simple_loss=0.1589, pruned_loss=0.05119, over 1948741.01 frames. ], batch size: 69, lr: 7.22e-03, grad_scale: 4.0 2022-12-08 00:06:15,702 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:06:43,352 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.208e+02 2.821e+02 3.585e+02 7.488e+02, threshold=5.641e+02, percent-clipped=3.0 2022-12-08 00:06:45,697 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3809, 2.0145, 2.6794, 1.8189, 1.6821, 2.3910, 1.2675, 2.3513], device='cuda:0'), covar=tensor([0.1173, 0.1956, 0.0907, 0.2257, 0.3082, 0.1216, 0.5184, 0.1106], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0095, 0.0091, 0.0095, 0.0114, 0.0083, 0.0126, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:06:46,206 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-08 00:07:34,202 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:07:38,504 INFO [train.py:873] (0/4) Epoch 11, batch 3500, loss[loss=0.1373, simple_loss=0.1508, pruned_loss=0.0619, over 4979.00 frames. ], tot_loss[loss=0.132, simple_loss=0.1593, pruned_loss=0.05239, over 1896414.09 frames. ], batch size: 100, lr: 7.22e-03, grad_scale: 4.0 2022-12-08 00:08:10,356 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.398e+02 2.883e+02 3.497e+02 6.738e+02, threshold=5.767e+02, percent-clipped=3.0 2022-12-08 00:08:15,415 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:08:57,410 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:09:04,924 INFO [train.py:873] (0/4) Epoch 11, batch 3600, loss[loss=0.1225, simple_loss=0.1531, pruned_loss=0.04596, over 14266.00 frames. ], tot_loss[loss=0.1318, simple_loss=0.1594, pruned_loss=0.05209, over 1948412.83 frames. ], batch size: 25, lr: 7.21e-03, grad_scale: 8.0 2022-12-08 00:09:37,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.170e+02 2.534e+02 3.229e+02 7.657e+02, threshold=5.069e+02, percent-clipped=4.0 2022-12-08 00:09:39,211 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:09:45,893 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1753, 4.4746, 4.8137, 5.1823, 4.7254, 4.3428, 5.0459, 4.2238], device='cuda:0'), covar=tensor([0.0836, 0.1628, 0.0722, 0.0817, 0.1224, 0.0647, 0.0937, 0.0888], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0262, 0.0182, 0.0180, 0.0175, 0.0142, 0.0266, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:10:01,598 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:10:16,285 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 00:10:16,763 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4966, 3.2193, 3.2074, 3.4790, 3.2929, 3.4634, 3.5126, 2.9109], device='cuda:0'), covar=tensor([0.0517, 0.1059, 0.0498, 0.0557, 0.0844, 0.0392, 0.0616, 0.0648], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0263, 0.0183, 0.0181, 0.0176, 0.0143, 0.0267, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:10:32,863 INFO [train.py:873] (0/4) Epoch 11, batch 3700, loss[loss=0.1031, simple_loss=0.1471, pruned_loss=0.0295, over 14250.00 frames. ], tot_loss[loss=0.1318, simple_loss=0.1595, pruned_loss=0.05205, over 1978893.49 frames. ], batch size: 44, lr: 7.21e-03, grad_scale: 8.0 2022-12-08 00:10:32,985 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:10:33,919 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3307, 1.9482, 2.2524, 1.5200, 1.9701, 2.2991, 2.3070, 2.0308], device='cuda:0'), covar=tensor([0.0845, 0.0770, 0.1132, 0.1489, 0.1319, 0.0755, 0.0595, 0.1406], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0180, 0.0135, 0.0124, 0.0131, 0.0139, 0.0117, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:10:43,189 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:10:54,292 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:11:04,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.300e+02 2.713e+02 3.691e+02 7.855e+02, threshold=5.426e+02, percent-clipped=7.0 2022-12-08 00:11:09,832 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:11:19,320 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:11:35,743 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:11:58,465 INFO [train.py:873] (0/4) Epoch 11, batch 3800, loss[loss=0.1298, simple_loss=0.1594, pruned_loss=0.05005, over 14222.00 frames. ], tot_loss[loss=0.1314, simple_loss=0.1593, pruned_loss=0.05179, over 2027983.82 frames. ], batch size: 94, lr: 7.20e-03, grad_scale: 8.0 2022-12-08 00:12:02,448 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:12:11,297 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 00:12:11,637 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:12:30,563 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.086e+02 2.258e+02 2.922e+02 3.679e+02 8.434e+02, threshold=5.844e+02, percent-clipped=6.0 2022-12-08 00:12:42,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.09 vs. limit=2.0 2022-12-08 00:13:00,014 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1877, 2.1246, 4.2468, 2.8881, 4.0879, 1.9930, 3.1823, 4.0010], device='cuda:0'), covar=tensor([0.0737, 0.4491, 0.0552, 0.7183, 0.0644, 0.3910, 0.1462, 0.0567], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0214, 0.0197, 0.0292, 0.0217, 0.0218, 0.0213, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:13:21,310 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7454, 1.4002, 2.8599, 2.6179, 2.7986, 2.9117, 2.1242, 2.8514], device='cuda:0'), covar=tensor([0.1407, 0.1492, 0.0227, 0.0414, 0.0389, 0.0196, 0.0608, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0164, 0.0142, 0.0134, 0.0116, 0.0116], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 00:13:26,570 INFO [train.py:873] (0/4) Epoch 11, batch 3900, loss[loss=0.1156, simple_loss=0.1523, pruned_loss=0.0395, over 14642.00 frames. ], tot_loss[loss=0.1299, simple_loss=0.1587, pruned_loss=0.05055, over 2041609.73 frames. ], batch size: 33, lr: 7.20e-03, grad_scale: 8.0 2022-12-08 00:13:33,538 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7025, 3.8152, 3.9728, 3.4717, 3.8842, 3.8651, 1.4479, 3.6700], device='cuda:0'), covar=tensor([0.0305, 0.0308, 0.0375, 0.0534, 0.0334, 0.0365, 0.3251, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0162, 0.0135, 0.0135, 0.0192, 0.0130, 0.0153, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 00:13:48,682 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8538, 1.1892, 1.3151, 1.2784, 0.9917, 1.3437, 1.0299, 0.8119], device='cuda:0'), covar=tensor([0.1932, 0.1066, 0.0443, 0.0440, 0.1919, 0.0649, 0.1728, 0.1379], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0077, 0.0061, 0.0064, 0.0092, 0.0074, 0.0094, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 00:13:59,193 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.132e+02 2.623e+02 3.165e+02 6.507e+02, threshold=5.247e+02, percent-clipped=1.0 2022-12-08 00:14:20,189 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0766, 2.9062, 2.3093, 3.1243, 2.9336, 2.9832, 2.5658, 2.2353], device='cuda:0'), covar=tensor([0.0905, 0.1438, 0.3327, 0.0748, 0.1033, 0.1144, 0.1646, 0.3123], device='cuda:0'), in_proj_covar=tensor([0.0267, 0.0296, 0.0270, 0.0250, 0.0307, 0.0291, 0.0253, 0.0255], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 00:14:22,703 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:14:54,999 INFO [train.py:873] (0/4) Epoch 11, batch 4000, loss[loss=0.1449, simple_loss=0.1546, pruned_loss=0.06762, over 3883.00 frames. ], tot_loss[loss=0.1294, simple_loss=0.1582, pruned_loss=0.0503, over 1963818.17 frames. ], batch size: 100, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:14:55,169 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:14:57,425 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2022-12-08 00:15:01,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 00:15:12,968 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:15:17,558 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:26,974 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:27,677 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.890e+01 2.190e+02 2.797e+02 3.644e+02 7.365e+02, threshold=5.595e+02, percent-clipped=3.0 2022-12-08 00:15:37,268 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:15:41,944 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5774, 1.4087, 1.5440, 1.4600, 1.5385, 0.9296, 1.3390, 1.4398], device='cuda:0'), covar=tensor([0.0897, 0.0997, 0.0713, 0.0874, 0.0860, 0.0930, 0.0903, 0.0888], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0026, 0.0028, 0.0025, 0.0026, 0.0038, 0.0026, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-08 00:15:55,364 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:21,096 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:23,000 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:16:23,755 INFO [train.py:873] (0/4) Epoch 11, batch 4100, loss[loss=0.1084, simple_loss=0.1324, pruned_loss=0.0422, over 3848.00 frames. ], tot_loss[loss=0.13, simple_loss=0.1585, pruned_loss=0.05074, over 1977973.96 frames. ], batch size: 100, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:16:32,680 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:16:34,502 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7316, 1.4298, 2.9619, 1.4685, 3.0022, 2.8668, 2.0980, 3.0784], device='cuda:0'), covar=tensor([0.0263, 0.2563, 0.0320, 0.1925, 0.0338, 0.0431, 0.0916, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0159, 0.0159, 0.0170, 0.0172, 0.0175, 0.0135, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:16:56,454 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.218e+02 2.886e+02 3.615e+02 6.897e+02, threshold=5.771e+02, percent-clipped=4.0 2022-12-08 00:17:00,592 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:17:20,369 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7071, 2.0868, 2.2077, 1.9253, 1.9878, 1.1438, 2.1821, 2.2404], device='cuda:0'), covar=tensor([0.1404, 0.0776, 0.0612, 0.1917, 0.1718, 0.0957, 0.0802, 0.0586], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0026, 0.0028, 0.0025, 0.0026, 0.0038, 0.0026, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2022-12-08 00:17:52,398 INFO [train.py:873] (0/4) Epoch 11, batch 4200, loss[loss=0.1219, simple_loss=0.1413, pruned_loss=0.05127, over 2648.00 frames. ], tot_loss[loss=0.1311, simple_loss=0.1591, pruned_loss=0.05156, over 1927091.97 frames. ], batch size: 100, lr: 7.19e-03, grad_scale: 8.0 2022-12-08 00:17:54,383 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:18:25,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.219e+02 2.651e+02 3.356e+02 1.038e+03, threshold=5.303e+02, percent-clipped=1.0 2022-12-08 00:18:52,935 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1667, 1.2942, 1.4329, 0.9066, 0.7200, 1.1778, 0.8135, 1.2065], device='cuda:0'), covar=tensor([0.1654, 0.2374, 0.0907, 0.2502, 0.3564, 0.0907, 0.2356, 0.1218], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0096, 0.0092, 0.0095, 0.0116, 0.0083, 0.0126, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:18:54,638 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6760, 1.3991, 2.9268, 1.3619, 2.9574, 2.8391, 2.1736, 3.0289], device='cuda:0'), covar=tensor([0.0310, 0.2750, 0.0364, 0.2130, 0.0336, 0.0499, 0.0975, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0159, 0.0159, 0.0168, 0.0172, 0.0174, 0.0134, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:19:05,827 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8887, 2.5491, 3.6477, 2.5330, 1.9582, 3.1098, 1.6852, 3.1663], device='cuda:0'), covar=tensor([0.1871, 0.1602, 0.0743, 0.2283, 0.3189, 0.1097, 0.5102, 0.0974], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0096, 0.0092, 0.0095, 0.0117, 0.0084, 0.0126, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:19:11,903 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6258, 3.9443, 3.7202, 3.3145, 2.7637, 3.6898, 3.5828, 2.0224], device='cuda:0'), covar=tensor([0.1756, 0.0568, 0.1050, 0.1853, 0.0984, 0.0703, 0.0946, 0.2190], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0075, 0.0060, 0.0063, 0.0091, 0.0072, 0.0093, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 00:19:21,010 INFO [train.py:873] (0/4) Epoch 11, batch 4300, loss[loss=0.1036, simple_loss=0.1326, pruned_loss=0.03733, over 4967.00 frames. ], tot_loss[loss=0.1297, simple_loss=0.159, pruned_loss=0.0502, over 2005078.26 frames. ], batch size: 100, lr: 7.18e-03, grad_scale: 8.0 2022-12-08 00:19:38,655 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:19:38,711 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:19:53,915 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.309e+02 2.745e+02 3.460e+02 6.233e+02, threshold=5.491e+02, percent-clipped=1.0 2022-12-08 00:19:57,636 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9546, 1.9616, 1.8928, 1.9983, 1.8925, 1.5359, 1.2824, 1.6892], device='cuda:0'), covar=tensor([0.0672, 0.0562, 0.0729, 0.0483, 0.0747, 0.1618, 0.2395, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0165, 0.0138, 0.0136, 0.0196, 0.0133, 0.0156, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 00:19:59,194 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 00:20:21,811 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:20:21,859 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:32,245 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-80000.pt 2022-12-08 00:20:36,931 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7063, 5.1293, 5.1458, 5.7056, 5.2463, 4.6350, 5.6299, 4.6597], device='cuda:0'), covar=tensor([0.0364, 0.1437, 0.0345, 0.0449, 0.0772, 0.0469, 0.0580, 0.0583], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0257, 0.0179, 0.0178, 0.0172, 0.0142, 0.0262, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:20:41,716 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 00:20:47,243 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:53,272 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:20:53,975 INFO [train.py:873] (0/4) Epoch 11, batch 4400, loss[loss=0.1296, simple_loss=0.1614, pruned_loss=0.04886, over 13966.00 frames. ], tot_loss[loss=0.1297, simple_loss=0.1588, pruned_loss=0.05028, over 2011792.20 frames. ], batch size: 26, lr: 7.18e-03, grad_scale: 8.0 2022-12-08 00:21:03,157 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:21:05,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2022-12-08 00:21:08,527 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:21:23,718 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.14 vs. limit=2.0 2022-12-08 00:21:26,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.889e+01 2.388e+02 2.902e+02 3.759e+02 8.737e+02, threshold=5.804e+02, percent-clipped=8.0 2022-12-08 00:21:34,899 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-08 00:21:35,078 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:21:44,980 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:21:59,470 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7165, 1.5622, 1.7448, 1.6040, 1.6486, 0.8516, 1.3579, 1.5457], device='cuda:0'), covar=tensor([0.0650, 0.0767, 0.0633, 0.1174, 0.0792, 0.0859, 0.0869, 0.0733], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0026, 0.0028, 0.0025, 0.0027, 0.0039, 0.0027, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2022-12-08 00:22:02,847 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9768, 4.6371, 4.4922, 5.0252, 4.6285, 4.3017, 5.0101, 4.2249], device='cuda:0'), covar=tensor([0.0373, 0.0842, 0.0362, 0.0394, 0.0744, 0.0513, 0.0454, 0.0534], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0257, 0.0178, 0.0177, 0.0171, 0.0142, 0.0261, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:22:07,995 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5175, 1.1894, 3.7211, 1.4368, 3.5303, 3.8348, 3.1000, 3.8065], device='cuda:0'), covar=tensor([0.0534, 0.5069, 0.0745, 0.3923, 0.0977, 0.0621, 0.0778, 0.0592], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0158, 0.0161, 0.0169, 0.0173, 0.0175, 0.0135, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:22:19,560 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:22:21,958 INFO [train.py:873] (0/4) Epoch 11, batch 4500, loss[loss=0.1036, simple_loss=0.145, pruned_loss=0.03109, over 13888.00 frames. ], tot_loss[loss=0.1293, simple_loss=0.1583, pruned_loss=0.05017, over 1943802.10 frames. ], batch size: 20, lr: 7.17e-03, grad_scale: 8.0 2022-12-08 00:22:33,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7353, 2.8370, 4.4715, 3.4974, 4.4979, 4.2930, 4.2137, 3.8126], device='cuda:0'), covar=tensor([0.0602, 0.3409, 0.0889, 0.1784, 0.0762, 0.0961, 0.2013, 0.1906], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0318, 0.0400, 0.0308, 0.0381, 0.0323, 0.0367, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:22:54,366 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.254e+02 2.921e+02 3.671e+02 7.258e+02, threshold=5.842e+02, percent-clipped=4.0 2022-12-08 00:22:55,306 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8737, 1.2776, 2.0702, 1.1877, 1.9896, 2.0794, 1.6153, 2.1488], device='cuda:0'), covar=tensor([0.0293, 0.1682, 0.0370, 0.1589, 0.0438, 0.0405, 0.1109, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0158, 0.0160, 0.0168, 0.0172, 0.0174, 0.0135, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:23:11,740 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-08 00:23:48,489 INFO [train.py:873] (0/4) Epoch 11, batch 4600, loss[loss=0.1662, simple_loss=0.1849, pruned_loss=0.07381, over 10340.00 frames. ], tot_loss[loss=0.1309, simple_loss=0.1597, pruned_loss=0.05101, over 1967604.23 frames. ], batch size: 100, lr: 7.17e-03, grad_scale: 8.0 2022-12-08 00:24:06,646 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:24:21,385 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.172e+02 2.626e+02 3.163e+02 7.453e+02, threshold=5.252e+02, percent-clipped=2.0 2022-12-08 00:24:42,718 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:24:48,267 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:09,166 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:16,060 INFO [train.py:873] (0/4) Epoch 11, batch 4700, loss[loss=0.1376, simple_loss=0.159, pruned_loss=0.05813, over 5025.00 frames. ], tot_loss[loss=0.1301, simple_loss=0.1594, pruned_loss=0.05043, over 2029184.76 frames. ], batch size: 100, lr: 7.16e-03, grad_scale: 8.0 2022-12-08 00:25:36,792 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:25:49,170 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 2.137e+02 2.805e+02 3.644e+02 8.749e+02, threshold=5.610e+02, percent-clipped=7.0 2022-12-08 00:25:51,864 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:20,331 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:42,748 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:26:45,211 INFO [train.py:873] (0/4) Epoch 11, batch 4800, loss[loss=0.1442, simple_loss=0.1606, pruned_loss=0.06388, over 5998.00 frames. ], tot_loss[loss=0.1288, simple_loss=0.1587, pruned_loss=0.04946, over 2045413.88 frames. ], batch size: 100, lr: 7.16e-03, grad_scale: 8.0 2022-12-08 00:26:58,829 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 00:27:14,048 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:17,205 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.171e+02 2.910e+02 3.468e+02 6.443e+02, threshold=5.821e+02, percent-clipped=1.0 2022-12-08 00:27:21,579 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:23,922 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:27:49,339 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:06,001 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5976, 1.3378, 2.7769, 1.3455, 2.8001, 2.7365, 1.9656, 2.9132], device='cuda:0'), covar=tensor([0.0272, 0.2542, 0.0370, 0.2028, 0.0361, 0.0451, 0.1112, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0157, 0.0159, 0.0169, 0.0171, 0.0174, 0.0135, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:28:10,722 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8580, 4.8897, 4.2909, 4.2907, 4.6445, 4.8378, 5.0384, 4.9707], device='cuda:0'), covar=tensor([0.1292, 0.0532, 0.3193, 0.3808, 0.1068, 0.1217, 0.1070, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0367, 0.0253, 0.0430, 0.0540, 0.0327, 0.0412, 0.0391, 0.0355], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:28:12,374 INFO [train.py:873] (0/4) Epoch 11, batch 4900, loss[loss=0.1428, simple_loss=0.1439, pruned_loss=0.0708, over 2561.00 frames. ], tot_loss[loss=0.1294, simple_loss=0.159, pruned_loss=0.0499, over 2061414.55 frames. ], batch size: 100, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:28:15,001 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:27,715 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8808, 1.6132, 2.0958, 1.6459, 1.9515, 1.5197, 1.6996, 1.9983], device='cuda:0'), covar=tensor([0.1927, 0.2531, 0.0379, 0.1632, 0.0941, 0.1327, 0.0959, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0215, 0.0199, 0.0291, 0.0215, 0.0219, 0.0212, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:28:42,413 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:28:44,856 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.165e+02 2.757e+02 3.267e+02 6.600e+02, threshold=5.514e+02, percent-clipped=1.0 2022-12-08 00:29:28,423 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 00:29:38,362 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3558, 3.6625, 3.5656, 3.3998, 2.7305, 3.4703, 3.4569, 1.8734], device='cuda:0'), covar=tensor([0.2033, 0.0733, 0.0934, 0.0907, 0.1018, 0.0485, 0.1099, 0.2599], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0077, 0.0060, 0.0064, 0.0092, 0.0074, 0.0095, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 00:29:41,614 INFO [train.py:873] (0/4) Epoch 11, batch 5000, loss[loss=0.175, simple_loss=0.1927, pruned_loss=0.07865, over 10342.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1591, pruned_loss=0.04967, over 2085412.52 frames. ], batch size: 100, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:29:46,348 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.35 vs. limit=2.0 2022-12-08 00:29:57,600 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:30:14,651 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.220e+02 2.746e+02 3.480e+02 5.243e+02, threshold=5.491e+02, percent-clipped=0.0 2022-12-08 00:30:43,099 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 00:31:10,693 INFO [train.py:873] (0/4) Epoch 11, batch 5100, loss[loss=0.1073, simple_loss=0.1469, pruned_loss=0.03388, over 14371.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1585, pruned_loss=0.04997, over 2030833.59 frames. ], batch size: 18, lr: 7.15e-03, grad_scale: 8.0 2022-12-08 00:31:24,339 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:31:35,061 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:31:43,012 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.114e+02 2.657e+02 3.455e+02 5.035e+02, threshold=5.315e+02, percent-clipped=0.0 2022-12-08 00:31:56,753 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:17,369 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:36,958 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:32:38,683 INFO [train.py:873] (0/4) Epoch 11, batch 5200, loss[loss=0.1412, simple_loss=0.1618, pruned_loss=0.06026, over 4979.00 frames. ], tot_loss[loss=0.1298, simple_loss=0.159, pruned_loss=0.05033, over 2019446.59 frames. ], batch size: 100, lr: 7.14e-03, grad_scale: 8.0 2022-12-08 00:32:50,392 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:33:04,567 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:33:11,236 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.222e+02 2.775e+02 3.597e+02 5.956e+02, threshold=5.550e+02, percent-clipped=3.0 2022-12-08 00:33:27,866 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.33 vs. limit=5.0 2022-12-08 00:33:56,410 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2022-12-08 00:34:04,553 INFO [train.py:873] (0/4) Epoch 11, batch 5300, loss[loss=0.1242, simple_loss=0.1544, pruned_loss=0.04697, over 14139.00 frames. ], tot_loss[loss=0.1297, simple_loss=0.1587, pruned_loss=0.05032, over 1953852.68 frames. ], batch size: 84, lr: 7.14e-03, grad_scale: 8.0 2022-12-08 00:34:20,770 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:34:21,569 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1794, 1.2597, 1.3981, 0.9072, 0.8455, 1.1748, 0.9180, 1.2165], device='cuda:0'), covar=tensor([0.1751, 0.2168, 0.0988, 0.2380, 0.3149, 0.0918, 0.2235, 0.1172], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0095, 0.0090, 0.0094, 0.0113, 0.0083, 0.0124, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:34:37,142 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.266e+02 2.683e+02 3.445e+02 6.231e+02, threshold=5.367e+02, percent-clipped=3.0 2022-12-08 00:35:02,401 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:35:29,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7463, 4.5210, 4.2712, 4.8029, 4.3891, 4.1015, 4.7747, 4.6472], device='cuda:0'), covar=tensor([0.0610, 0.0808, 0.0770, 0.0573, 0.0695, 0.0753, 0.0620, 0.0680], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0128, 0.0133, 0.0144, 0.0136, 0.0112, 0.0155, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 00:35:32,523 INFO [train.py:873] (0/4) Epoch 11, batch 5400, loss[loss=0.132, simple_loss=0.163, pruned_loss=0.05052, over 12726.00 frames. ], tot_loss[loss=0.1287, simple_loss=0.158, pruned_loss=0.04967, over 1904845.60 frames. ], batch size: 100, lr: 7.13e-03, grad_scale: 16.0 2022-12-08 00:35:57,936 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81049.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:00,463 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 00:36:05,567 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.933e+01 1.955e+02 2.740e+02 3.056e+02 5.675e+02, threshold=5.479e+02, percent-clipped=1.0 2022-12-08 00:36:35,727 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:39,872 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:46,828 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1327, 2.0552, 1.7674, 1.7982, 2.0515, 2.0930, 2.0792, 2.0645], device='cuda:0'), covar=tensor([0.1123, 0.0951, 0.3027, 0.2862, 0.1408, 0.1093, 0.1633, 0.1011], device='cuda:0'), in_proj_covar=tensor([0.0367, 0.0258, 0.0435, 0.0543, 0.0325, 0.0418, 0.0387, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:36:56,325 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:36:58,928 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81119.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:00,419 INFO [train.py:873] (0/4) Epoch 11, batch 5500, loss[loss=0.11, simple_loss=0.1495, pruned_loss=0.03528, over 14393.00 frames. ], tot_loss[loss=0.1295, simple_loss=0.1583, pruned_loss=0.05038, over 1897597.69 frames. ], batch size: 44, lr: 7.13e-03, grad_scale: 8.0 2022-12-08 00:37:07,685 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81129.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:25,875 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81150.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:33,796 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.115e+02 2.510e+02 3.295e+02 5.828e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 00:37:37,162 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.72 vs. limit=5.0 2022-12-08 00:37:40,966 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81167.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:37:47,729 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6155, 2.4529, 2.1931, 2.3387, 2.5398, 2.5183, 2.5593, 2.5526], device='cuda:0'), covar=tensor([0.1198, 0.0850, 0.2928, 0.2589, 0.1144, 0.1123, 0.1560, 0.1013], device='cuda:0'), in_proj_covar=tensor([0.0369, 0.0257, 0.0434, 0.0539, 0.0323, 0.0417, 0.0386, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:37:49,663 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81177.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:38:02,340 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8633, 1.7629, 1.9811, 1.7655, 2.0606, 1.8208, 1.7460, 1.8766], device='cuda:0'), covar=tensor([0.0661, 0.1338, 0.0312, 0.0465, 0.0419, 0.0730, 0.0284, 0.0420], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0320, 0.0404, 0.0311, 0.0384, 0.0326, 0.0370, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:38:07,659 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81198.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:38:28,043 INFO [train.py:873] (0/4) Epoch 11, batch 5600, loss[loss=0.1247, simple_loss=0.142, pruned_loss=0.05373, over 3849.00 frames. ], tot_loss[loss=0.1304, simple_loss=0.1587, pruned_loss=0.05104, over 1862993.07 frames. ], batch size: 100, lr: 7.12e-03, grad_scale: 8.0 2022-12-08 00:38:59,957 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2868, 1.6431, 1.8735, 1.7589, 1.6400, 1.7504, 1.3949, 1.2636], device='cuda:0'), covar=tensor([0.1399, 0.1699, 0.0497, 0.0599, 0.1424, 0.0814, 0.1878, 0.2028], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0077, 0.0061, 0.0065, 0.0093, 0.0074, 0.0093, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 00:39:02,412 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.253e+02 2.769e+02 3.640e+02 8.621e+02, threshold=5.537e+02, percent-clipped=7.0 2022-12-08 00:39:25,951 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 00:39:36,105 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 00:39:37,327 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:39:38,999 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81301.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:39:55,988 INFO [train.py:873] (0/4) Epoch 11, batch 5700, loss[loss=0.1386, simple_loss=0.1705, pruned_loss=0.05334, over 14114.00 frames. ], tot_loss[loss=0.1293, simple_loss=0.158, pruned_loss=0.05026, over 1892537.24 frames. ], batch size: 29, lr: 7.12e-03, grad_scale: 8.0 2022-12-08 00:40:10,694 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 00:40:21,461 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2022-12-08 00:40:28,144 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1334, 2.8697, 2.9212, 1.9428, 2.6438, 2.9165, 3.1737, 2.5968], device='cuda:0'), covar=tensor([0.0781, 0.1350, 0.0942, 0.1764, 0.1056, 0.0925, 0.0920, 0.1430], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0183, 0.0136, 0.0126, 0.0134, 0.0144, 0.0119, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:40:28,773 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.149e+01 2.189e+02 2.731e+02 3.467e+02 4.923e+02, threshold=5.462e+02, percent-clipped=0.0 2022-12-08 00:40:29,840 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81360.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:40:31,267 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81362.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:40:38,970 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1285, 2.0106, 1.7556, 1.8325, 2.0450, 2.0802, 2.0551, 2.0586], device='cuda:0'), covar=tensor([0.1565, 0.1060, 0.3590, 0.3426, 0.1528, 0.1302, 0.2027, 0.1258], device='cuda:0'), in_proj_covar=tensor([0.0366, 0.0257, 0.0429, 0.0536, 0.0320, 0.0417, 0.0385, 0.0358], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:40:44,085 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1281, 2.0131, 2.0636, 2.1725, 2.0347, 2.0112, 2.1953, 1.8873], device='cuda:0'), covar=tensor([0.0721, 0.1120, 0.0696, 0.0718, 0.1004, 0.0669, 0.0793, 0.0726], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0258, 0.0181, 0.0179, 0.0174, 0.0143, 0.0263, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:40:57,889 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81392.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:23,788 INFO [train.py:873] (0/4) Epoch 11, batch 5800, loss[loss=0.1235, simple_loss=0.1569, pruned_loss=0.045, over 14363.00 frames. ], tot_loss[loss=0.1292, simple_loss=0.1582, pruned_loss=0.05007, over 1931722.99 frames. ], batch size: 41, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:41:30,629 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:40,079 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:41:56,860 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.147e+02 2.711e+02 3.541e+02 7.030e+02, threshold=5.421e+02, percent-clipped=5.0 2022-12-08 00:42:08,761 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:42:12,677 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81477.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:42:17,837 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7737, 3.6657, 3.3017, 3.3588, 3.8424, 3.7753, 3.9462, 3.8417], device='cuda:0'), covar=tensor([0.1527, 0.0899, 0.2819, 0.3912, 0.1094, 0.1263, 0.1318, 0.1402], device='cuda:0'), in_proj_covar=tensor([0.0368, 0.0261, 0.0432, 0.0542, 0.0325, 0.0418, 0.0391, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:42:36,786 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3706, 3.0941, 4.0072, 2.6532, 2.3452, 3.4941, 1.6579, 3.4495], device='cuda:0'), covar=tensor([0.1263, 0.1315, 0.0594, 0.2337, 0.2293, 0.0860, 0.4096, 0.1100], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0096, 0.0091, 0.0095, 0.0114, 0.0083, 0.0125, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:42:45,817 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.36 vs. limit=5.0 2022-12-08 00:42:51,270 INFO [train.py:873] (0/4) Epoch 11, batch 5900, loss[loss=0.1255, simple_loss=0.1555, pruned_loss=0.04773, over 14231.00 frames. ], tot_loss[loss=0.1282, simple_loss=0.1577, pruned_loss=0.04936, over 2005016.94 frames. ], batch size: 89, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:43:24,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.246e+02 2.674e+02 3.531e+02 7.780e+02, threshold=5.349e+02, percent-clipped=2.0 2022-12-08 00:43:57,933 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9088, 2.1364, 2.7858, 2.9035, 2.8371, 2.1168, 2.8085, 2.3755], device='cuda:0'), covar=tensor([0.0350, 0.0735, 0.0548, 0.0386, 0.0403, 0.1085, 0.0357, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0251, 0.0367, 0.0316, 0.0258, 0.0298, 0.0295, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 00:44:19,735 INFO [train.py:873] (0/4) Epoch 11, batch 6000, loss[loss=0.1485, simple_loss=0.1681, pruned_loss=0.06446, over 10310.00 frames. ], tot_loss[loss=0.1285, simple_loss=0.158, pruned_loss=0.0495, over 1975765.76 frames. ], batch size: 100, lr: 7.11e-03, grad_scale: 8.0 2022-12-08 00:44:19,736 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 00:44:28,213 INFO [train.py:905] (0/4) Epoch 11, validation: loss=0.1286, simple_loss=0.169, pruned_loss=0.04409, over 857387.00 frames. 2022-12-08 00:44:28,213 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 00:44:37,529 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-08 00:44:38,746 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8793, 2.5499, 3.6865, 2.8200, 3.7937, 3.6003, 3.5732, 3.1306], device='cuda:0'), covar=tensor([0.1061, 0.2927, 0.1119, 0.2110, 0.1028, 0.1031, 0.1667, 0.2132], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0314, 0.0399, 0.0307, 0.0376, 0.0321, 0.0365, 0.0312], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:44:45,515 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8873, 1.7017, 1.9719, 1.7506, 2.0556, 1.8538, 1.7035, 1.9329], device='cuda:0'), covar=tensor([0.0485, 0.1128, 0.0211, 0.0344, 0.0303, 0.0519, 0.0219, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0315, 0.0400, 0.0307, 0.0377, 0.0321, 0.0366, 0.0313], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:44:58,267 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81655.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:00,138 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81657.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:01,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.152e+02 2.716e+02 3.284e+02 6.077e+02, threshold=5.431e+02, percent-clipped=1.0 2022-12-08 00:45:11,717 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81670.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:45:24,650 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1646, 4.0472, 3.5978, 3.6567, 4.0885, 4.1182, 4.2720, 4.1605], device='cuda:0'), covar=tensor([0.1442, 0.0768, 0.2540, 0.3967, 0.1010, 0.1169, 0.1238, 0.1360], device='cuda:0'), in_proj_covar=tensor([0.0371, 0.0261, 0.0432, 0.0545, 0.0325, 0.0418, 0.0393, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:45:25,080 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 00:45:55,906 INFO [train.py:873] (0/4) Epoch 11, batch 6100, loss[loss=0.1541, simple_loss=0.1694, pruned_loss=0.06938, over 7803.00 frames. ], tot_loss[loss=0.1287, simple_loss=0.1579, pruned_loss=0.04974, over 1947198.86 frames. ], batch size: 100, lr: 7.10e-03, grad_scale: 8.0 2022-12-08 00:46:04,825 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81731.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:46:29,253 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.111e+02 2.193e+02 2.528e+02 3.178e+02 6.202e+02, threshold=5.055e+02, percent-clipped=2.0 2022-12-08 00:46:40,533 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:46:44,184 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6745, 3.3281, 2.5873, 3.8437, 3.6423, 3.5881, 3.1675, 2.5774], device='cuda:0'), covar=tensor([0.0795, 0.1375, 0.3578, 0.0520, 0.0898, 0.1260, 0.1287, 0.3801], device='cuda:0'), in_proj_covar=tensor([0.0271, 0.0298, 0.0271, 0.0256, 0.0310, 0.0299, 0.0254, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 00:47:00,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1983, 2.0559, 2.1326, 2.2258, 2.1624, 2.1170, 2.2649, 1.9605], device='cuda:0'), covar=tensor([0.0998, 0.1375, 0.0718, 0.0783, 0.0928, 0.0808, 0.0849, 0.0687], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0261, 0.0182, 0.0182, 0.0175, 0.0145, 0.0268, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:47:23,437 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:47:24,228 INFO [train.py:873] (0/4) Epoch 11, batch 6200, loss[loss=0.1633, simple_loss=0.1607, pruned_loss=0.08294, over 3843.00 frames. ], tot_loss[loss=0.1288, simple_loss=0.1577, pruned_loss=0.05, over 1942377.72 frames. ], batch size: 100, lr: 7.10e-03, grad_scale: 8.0 2022-12-08 00:47:58,181 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.324e+02 2.708e+02 3.471e+02 1.552e+03, threshold=5.417e+02, percent-clipped=7.0 2022-12-08 00:48:52,506 INFO [train.py:873] (0/4) Epoch 11, batch 6300, loss[loss=0.1307, simple_loss=0.1633, pruned_loss=0.04905, over 14269.00 frames. ], tot_loss[loss=0.1284, simple_loss=0.1575, pruned_loss=0.04967, over 1883317.85 frames. ], batch size: 76, lr: 7.09e-03, grad_scale: 8.0 2022-12-08 00:49:22,579 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81955.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:49:24,326 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81957.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:49:25,808 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.360e+02 2.936e+02 3.429e+02 7.128e+02, threshold=5.872e+02, percent-clipped=5.0 2022-12-08 00:49:50,185 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:04,802 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:06,460 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82005.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:20,250 INFO [train.py:873] (0/4) Epoch 11, batch 6400, loss[loss=0.1243, simple_loss=0.1579, pruned_loss=0.04533, over 14297.00 frames. ], tot_loss[loss=0.1287, simple_loss=0.1577, pruned_loss=0.0498, over 1888938.18 frames. ], batch size: 63, lr: 7.09e-03, grad_scale: 8.0 2022-12-08 00:50:24,711 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82026.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 00:50:33,396 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2022-12-08 00:50:43,921 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82048.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:49,093 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4203, 5.2197, 4.8293, 5.0315, 5.0860, 5.3368, 5.4764, 5.4167], device='cuda:0'), covar=tensor([0.0582, 0.0360, 0.1964, 0.2353, 0.0581, 0.0493, 0.0641, 0.0620], device='cuda:0'), in_proj_covar=tensor([0.0373, 0.0259, 0.0431, 0.0545, 0.0324, 0.0416, 0.0389, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:50:50,081 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:50:53,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.188e+02 2.645e+02 3.155e+02 5.407e+02, threshold=5.290e+02, percent-clipped=0.0 2022-12-08 00:51:43,595 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:51:47,737 INFO [train.py:873] (0/4) Epoch 11, batch 6500, loss[loss=0.1047, simple_loss=0.1471, pruned_loss=0.03117, over 14609.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1579, pruned_loss=0.04992, over 1876741.08 frames. ], batch size: 22, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:52:12,641 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7833, 4.4481, 4.2574, 4.8163, 4.4540, 4.1873, 4.8136, 4.0731], device='cuda:0'), covar=tensor([0.0381, 0.1107, 0.0386, 0.0399, 0.0790, 0.0542, 0.0548, 0.0493], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0255, 0.0178, 0.0179, 0.0171, 0.0143, 0.0264, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 00:52:13,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.43 vs. limit=5.0 2022-12-08 00:52:20,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.664e+01 2.301e+02 3.022e+02 3.856e+02 8.544e+02, threshold=6.044e+02, percent-clipped=9.0 2022-12-08 00:52:40,860 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5261, 3.1481, 3.0135, 2.2602, 3.0274, 3.2552, 3.5784, 2.7847], device='cuda:0'), covar=tensor([0.0587, 0.1655, 0.1009, 0.1674, 0.0799, 0.0613, 0.0561, 0.1317], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0183, 0.0136, 0.0126, 0.0134, 0.0141, 0.0119, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 00:52:44,392 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 00:53:14,953 INFO [train.py:873] (0/4) Epoch 11, batch 6600, loss[loss=0.138, simple_loss=0.1676, pruned_loss=0.05424, over 13986.00 frames. ], tot_loss[loss=0.129, simple_loss=0.1586, pruned_loss=0.04969, over 1997701.54 frames. ], batch size: 26, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:53:40,546 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0264, 4.8677, 4.4590, 4.6319, 4.6880, 4.9774, 5.0605, 5.0321], device='cuda:0'), covar=tensor([0.0853, 0.0382, 0.2156, 0.2787, 0.0670, 0.0708, 0.0836, 0.0773], device='cuda:0'), in_proj_covar=tensor([0.0372, 0.0258, 0.0429, 0.0547, 0.0324, 0.0416, 0.0389, 0.0365], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:53:48,293 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.109e+02 2.550e+02 3.112e+02 5.386e+02, threshold=5.100e+02, percent-clipped=0.0 2022-12-08 00:54:40,165 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3507, 2.1420, 2.5695, 1.4927, 1.6773, 2.3519, 1.2632, 2.3666], device='cuda:0'), covar=tensor([0.0815, 0.1525, 0.0687, 0.2530, 0.2561, 0.0769, 0.3963, 0.0675], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0096, 0.0090, 0.0095, 0.0113, 0.0082, 0.0123, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:54:43,679 INFO [train.py:873] (0/4) Epoch 11, batch 6700, loss[loss=0.1258, simple_loss=0.1533, pruned_loss=0.04911, over 14313.00 frames. ], tot_loss[loss=0.1274, simple_loss=0.1572, pruned_loss=0.04882, over 1935197.03 frames. ], batch size: 46, lr: 7.08e-03, grad_scale: 8.0 2022-12-08 00:54:48,043 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82326.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:02,373 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82343.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:05,062 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:07,216 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 00:55:10,490 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1982, 1.5711, 2.4560, 1.9524, 2.2010, 1.6705, 1.9495, 2.0815], device='cuda:0'), covar=tensor([0.1873, 0.3128, 0.0389, 0.1911, 0.1005, 0.1724, 0.0915, 0.0788], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0215, 0.0200, 0.0291, 0.0218, 0.0218, 0.0218, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:55:16,416 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.173e+02 2.670e+02 3.578e+02 7.367e+02, threshold=5.339e+02, percent-clipped=2.0 2022-12-08 00:55:16,991 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2022-12-08 00:55:29,465 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:55:56,361 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 00:55:59,160 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:56:02,589 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82411.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:56:11,098 INFO [train.py:873] (0/4) Epoch 11, batch 6800, loss[loss=0.1061, simple_loss=0.1466, pruned_loss=0.03285, over 14349.00 frames. ], tot_loss[loss=0.1275, simple_loss=0.1574, pruned_loss=0.04884, over 1960547.78 frames. ], batch size: 44, lr: 7.07e-03, grad_scale: 8.0 2022-12-08 00:56:43,724 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.471e+02 3.089e+02 4.263e+02 8.393e+02, threshold=6.178e+02, percent-clipped=10.0 2022-12-08 00:56:50,890 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 00:57:14,020 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82493.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 00:57:21,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8330, 1.4060, 3.8368, 1.7126, 3.7956, 3.9494, 2.7800, 4.2114], device='cuda:0'), covar=tensor([0.0242, 0.3333, 0.0413, 0.2248, 0.0426, 0.0335, 0.0703, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0158, 0.0161, 0.0170, 0.0171, 0.0173, 0.0132, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 00:57:22,416 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4489, 3.3240, 4.0514, 2.8034, 2.2528, 3.4541, 1.8695, 3.6185], device='cuda:0'), covar=tensor([0.0810, 0.0981, 0.0518, 0.1854, 0.2273, 0.0811, 0.3593, 0.1148], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0094, 0.0088, 0.0093, 0.0111, 0.0081, 0.0120, 0.0085], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0005, 0.0004], device='cuda:0') 2022-12-08 00:57:37,892 INFO [train.py:873] (0/4) Epoch 11, batch 6900, loss[loss=0.1161, simple_loss=0.1556, pruned_loss=0.03828, over 14571.00 frames. ], tot_loss[loss=0.1277, simple_loss=0.1569, pruned_loss=0.04923, over 1922064.53 frames. ], batch size: 34, lr: 7.07e-03, grad_scale: 8.0 2022-12-08 00:57:53,968 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8669, 2.1030, 3.9356, 2.6823, 3.7976, 2.0934, 2.9682, 3.7821], device='cuda:0'), covar=tensor([0.0708, 0.4088, 0.0445, 0.6553, 0.0589, 0.3569, 0.1344, 0.0492], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0216, 0.0202, 0.0294, 0.0220, 0.0220, 0.0219, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 00:58:04,808 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 00:58:06,348 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82554.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 00:58:11,136 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.298e+02 3.049e+02 3.803e+02 8.603e+02, threshold=6.098e+02, percent-clipped=2.0 2022-12-08 00:58:20,044 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82569.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:58:58,502 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 00:59:06,536 INFO [train.py:873] (0/4) Epoch 11, batch 7000, loss[loss=0.117, simple_loss=0.1457, pruned_loss=0.04419, over 6950.00 frames. ], tot_loss[loss=0.1278, simple_loss=0.157, pruned_loss=0.04931, over 1925221.31 frames. ], batch size: 100, lr: 7.06e-03, grad_scale: 8.0 2022-12-08 00:59:14,700 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:59:26,134 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 00:59:26,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 00:59:39,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.154e+02 2.641e+02 3.288e+02 6.152e+02, threshold=5.282e+02, percent-clipped=1.0 2022-12-08 01:00:03,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9124, 1.8671, 2.0483, 1.6033, 2.0428, 1.0384, 2.0860, 2.2189], device='cuda:0'), covar=tensor([0.0994, 0.1007, 0.0659, 0.2261, 0.0913, 0.0840, 0.1031, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:00:07,991 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:17,853 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82702.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:26,001 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82711.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:00:34,803 INFO [train.py:873] (0/4) Epoch 11, batch 7100, loss[loss=0.1335, simple_loss=0.1524, pruned_loss=0.05728, over 4964.00 frames. ], tot_loss[loss=0.1264, simple_loss=0.1563, pruned_loss=0.04828, over 1964692.48 frames. ], batch size: 100, lr: 7.06e-03, grad_scale: 8.0 2022-12-08 01:00:40,176 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.94 vs. limit=5.0 2022-12-08 01:00:48,188 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 01:01:07,645 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.263e+02 2.729e+02 3.602e+02 7.969e+02, threshold=5.458e+02, percent-clipped=3.0 2022-12-08 01:01:07,747 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82759.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:02:00,838 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5283, 1.1305, 2.0230, 1.8362, 1.8673, 2.0338, 1.2814, 2.0146], device='cuda:0'), covar=tensor([0.0913, 0.1287, 0.0236, 0.0423, 0.0529, 0.0262, 0.0710, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0160, 0.0129, 0.0167, 0.0144, 0.0138, 0.0119, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:02:03,362 INFO [train.py:873] (0/4) Epoch 11, batch 7200, loss[loss=0.1149, simple_loss=0.1516, pruned_loss=0.03912, over 14293.00 frames. ], tot_loss[loss=0.1279, simple_loss=0.1576, pruned_loss=0.04909, over 1997384.90 frames. ], batch size: 46, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:02:28,450 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82849.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:02:36,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.101e+02 2.276e+02 3.049e+02 3.669e+02 5.112e+02, threshold=6.097e+02, percent-clipped=0.0 2022-12-08 01:03:06,169 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3065, 1.8117, 2.5217, 2.0632, 2.3247, 1.6555, 1.9877, 2.3282], device='cuda:0'), covar=tensor([0.1751, 0.3033, 0.0453, 0.2303, 0.0887, 0.2158, 0.1210, 0.0652], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0214, 0.0199, 0.0289, 0.0216, 0.0217, 0.0216, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:03:31,180 INFO [train.py:873] (0/4) Epoch 11, batch 7300, loss[loss=0.157, simple_loss=0.1569, pruned_loss=0.07853, over 3918.00 frames. ], tot_loss[loss=0.1277, simple_loss=0.157, pruned_loss=0.04921, over 1965867.75 frames. ], batch size: 100, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:03:34,879 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:03:51,476 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6574, 3.5129, 3.1510, 2.4512, 3.1394, 3.4051, 3.6624, 2.8694], device='cuda:0'), covar=tensor([0.0581, 0.1169, 0.0995, 0.1550, 0.0900, 0.0682, 0.0662, 0.1313], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0181, 0.0136, 0.0126, 0.0133, 0.0140, 0.0118, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 01:04:04,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.016e+02 2.470e+02 3.084e+02 3.689e+02 1.024e+03, threshold=6.168e+02, percent-clipped=3.0 2022-12-08 01:04:08,549 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82963.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:04:43,169 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:04:50,795 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6909, 1.7833, 2.9525, 1.9826, 2.8260, 1.7889, 2.3017, 2.7842], device='cuda:0'), covar=tensor([0.0975, 0.4022, 0.0603, 0.5106, 0.0874, 0.3225, 0.1229, 0.0617], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0215, 0.0200, 0.0288, 0.0218, 0.0217, 0.0216, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:04:59,089 INFO [train.py:873] (0/4) Epoch 11, batch 7400, loss[loss=0.1363, simple_loss=0.1381, pruned_loss=0.06722, over 1288.00 frames. ], tot_loss[loss=0.1285, simple_loss=0.1574, pruned_loss=0.04979, over 1888239.01 frames. ], batch size: 100, lr: 7.05e-03, grad_scale: 8.0 2022-12-08 01:05:02,195 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83024.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:05:25,167 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83050.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:05:32,577 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 2.140e+02 2.552e+02 3.210e+02 1.054e+03, threshold=5.104e+02, percent-clipped=1.0 2022-12-08 01:06:26,609 INFO [train.py:873] (0/4) Epoch 11, batch 7500, loss[loss=0.1291, simple_loss=0.1306, pruned_loss=0.06379, over 1165.00 frames. ], tot_loss[loss=0.1296, simple_loss=0.1581, pruned_loss=0.05057, over 1900120.54 frames. ], batch size: 100, lr: 7.04e-03, grad_scale: 16.0 2022-12-08 01:06:50,682 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83149.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:06:58,814 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.439e+02 2.915e+02 3.570e+02 7.862e+02, threshold=5.829e+02, percent-clipped=6.0 2022-12-08 01:07:02,527 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2022-12-08 01:07:13,079 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-11.pt 2022-12-08 01:07:53,459 INFO [train.py:873] (0/4) Epoch 12, batch 0, loss[loss=0.1275, simple_loss=0.167, pruned_loss=0.04397, over 14522.00 frames. ], tot_loss[loss=0.1275, simple_loss=0.167, pruned_loss=0.04397, over 14522.00 frames. ], batch size: 51, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:07:53,460 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 01:07:56,163 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2115, 3.6044, 3.0521, 3.4704, 2.4187, 3.4724, 3.2449, 1.9029], device='cuda:0'), covar=tensor([0.1832, 0.0480, 0.1171, 0.0537, 0.1065, 0.0482, 0.0907, 0.2214], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0076, 0.0062, 0.0064, 0.0093, 0.0076, 0.0093, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 01:08:00,688 INFO [train.py:905] (0/4) Epoch 12, validation: loss=0.1326, simple_loss=0.1738, pruned_loss=0.04568, over 857387.00 frames. 2022-12-08 01:08:00,688 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 01:08:12,326 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83196.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:08:13,194 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83197.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:08:38,643 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:08:44,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8855, 2.6844, 2.7454, 2.9049, 2.8028, 2.8056, 2.9392, 2.4591], device='cuda:0'), covar=tensor([0.0712, 0.1026, 0.0535, 0.0553, 0.0808, 0.0549, 0.0646, 0.0624], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0256, 0.0179, 0.0179, 0.0174, 0.0143, 0.0266, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 01:09:07,348 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83257.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:08,850 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.351e+01 1.849e+02 2.933e+02 3.942e+02 9.490e+02, threshold=5.866e+02, percent-clipped=6.0 2022-12-08 01:09:21,071 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:30,010 INFO [train.py:873] (0/4) Epoch 12, batch 100, loss[loss=0.1628, simple_loss=0.1817, pruned_loss=0.07192, over 10354.00 frames. ], tot_loss[loss=0.1239, simple_loss=0.1561, pruned_loss=0.04591, over 888713.17 frames. ], batch size: 100, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:09:30,962 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:09:49,985 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7790, 3.5810, 3.5654, 3.8476, 3.4306, 3.2541, 3.8362, 3.6986], device='cuda:0'), covar=tensor([0.0740, 0.0872, 0.0821, 0.0635, 0.0915, 0.0725, 0.0679, 0.0758], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0129, 0.0136, 0.0145, 0.0137, 0.0114, 0.0156, 0.0133], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:10:00,916 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:10:24,285 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83345.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:10:31,137 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3871, 2.0871, 2.6750, 1.5632, 1.7774, 2.3979, 1.3891, 2.4000], device='cuda:0'), covar=tensor([0.1009, 0.1819, 0.0867, 0.2780, 0.2781, 0.0996, 0.4202, 0.1017], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0095, 0.0089, 0.0095, 0.0111, 0.0082, 0.0122, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:10:36,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.243e+02 2.760e+02 3.223e+02 5.001e+02, threshold=5.521e+02, percent-clipped=0.0 2022-12-08 01:10:37,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.72 vs. limit=5.0 2022-12-08 01:10:46,424 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4482, 1.0163, 1.2800, 0.8375, 1.1263, 1.4006, 1.1294, 1.1281], device='cuda:0'), covar=tensor([0.0370, 0.0958, 0.0709, 0.0525, 0.0913, 0.0758, 0.0418, 0.1145], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0183, 0.0138, 0.0127, 0.0135, 0.0142, 0.0121, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 01:10:53,259 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9867, 5.3744, 5.3285, 5.9466, 5.5662, 4.7752, 5.8720, 4.8312], device='cuda:0'), covar=tensor([0.0303, 0.1072, 0.0301, 0.0389, 0.0719, 0.0404, 0.0443, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0255, 0.0180, 0.0177, 0.0173, 0.0143, 0.0266, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 01:10:57,432 INFO [train.py:873] (0/4) Epoch 12, batch 200, loss[loss=0.129, simple_loss=0.1619, pruned_loss=0.04806, over 14138.00 frames. ], tot_loss[loss=0.1278, simple_loss=0.158, pruned_loss=0.0488, over 1280145.20 frames. ], batch size: 84, lr: 6.74e-03, grad_scale: 16.0 2022-12-08 01:11:22,193 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3209, 1.3773, 3.5138, 1.5196, 3.1983, 3.4234, 2.3590, 3.6352], device='cuda:0'), covar=tensor([0.0217, 0.2978, 0.0293, 0.2185, 0.0874, 0.0377, 0.0980, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0158, 0.0168, 0.0169, 0.0172, 0.0133, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 01:11:50,462 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4952, 4.1991, 4.1849, 4.4818, 4.0998, 3.8536, 4.4995, 4.4153], device='cuda:0'), covar=tensor([0.0615, 0.0733, 0.0721, 0.0601, 0.0700, 0.0612, 0.0606, 0.0668], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0129, 0.0136, 0.0145, 0.0137, 0.0113, 0.0156, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:12:04,537 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.214e+02 2.674e+02 3.340e+02 6.097e+02, threshold=5.348e+02, percent-clipped=3.0 2022-12-08 01:12:13,216 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3642, 4.8250, 4.8142, 5.3718, 4.9575, 4.5458, 5.2570, 4.4519], device='cuda:0'), covar=tensor([0.0283, 0.0978, 0.0309, 0.0372, 0.0705, 0.0451, 0.0479, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0254, 0.0180, 0.0177, 0.0173, 0.0144, 0.0266, 0.0158], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 01:12:21,422 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-08 01:12:24,713 INFO [train.py:873] (0/4) Epoch 12, batch 300, loss[loss=0.1709, simple_loss=0.1874, pruned_loss=0.07718, over 9499.00 frames. ], tot_loss[loss=0.1295, simple_loss=0.1586, pruned_loss=0.05024, over 1572901.84 frames. ], batch size: 100, lr: 6.73e-03, grad_scale: 8.0 2022-12-08 01:13:02,683 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 01:13:25,757 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83552.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:13:29,292 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6999, 1.9927, 1.7489, 1.6352, 1.7175, 1.7639, 1.5598, 1.4862], device='cuda:0'), covar=tensor([0.0399, 0.0417, 0.0469, 0.0548, 0.0595, 0.0465, 0.0478, 0.0588], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0016, 0.0014, 0.0015, 0.0015, 0.0025, 0.0020, 0.0026], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:13:32,202 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.254e+02 2.674e+02 3.399e+02 5.874e+02, threshold=5.349e+02, percent-clipped=4.0 2022-12-08 01:13:35,595 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9753, 1.8529, 3.9824, 3.6955, 3.8258, 4.0233, 3.3705, 4.0989], device='cuda:0'), covar=tensor([0.1353, 0.1341, 0.0104, 0.0206, 0.0183, 0.0114, 0.0204, 0.0098], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0128, 0.0166, 0.0142, 0.0138, 0.0119, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:13:53,021 INFO [train.py:873] (0/4) Epoch 12, batch 400, loss[loss=0.1372, simple_loss=0.1655, pruned_loss=0.05446, over 14196.00 frames. ], tot_loss[loss=0.1285, simple_loss=0.1575, pruned_loss=0.04978, over 1656262.81 frames. ], batch size: 84, lr: 6.73e-03, grad_scale: 8.0 2022-12-08 01:13:55,435 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2022-12-08 01:14:25,356 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83619.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:14:43,832 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83640.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:15:01,084 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.765e+01 2.109e+02 2.628e+02 3.451e+02 6.835e+02, threshold=5.257e+02, percent-clipped=4.0 2022-12-08 01:15:01,318 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4027, 1.7551, 2.3182, 2.0175, 2.4056, 2.2390, 2.1849, 2.1657], device='cuda:0'), covar=tensor([0.0506, 0.2381, 0.0588, 0.1330, 0.0392, 0.0901, 0.0561, 0.1104], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0313, 0.0394, 0.0301, 0.0372, 0.0319, 0.0364, 0.0310], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:15:07,446 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83667.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:15:19,186 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0745, 2.1154, 1.9393, 2.1725, 1.8596, 2.0161, 2.1042, 2.0724], device='cuda:0'), covar=tensor([0.0907, 0.1116, 0.1122, 0.0929, 0.1409, 0.0871, 0.1068, 0.0989], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0128, 0.0135, 0.0143, 0.0136, 0.0112, 0.0155, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:15:21,383 INFO [train.py:873] (0/4) Epoch 12, batch 500, loss[loss=0.09062, simple_loss=0.1381, pruned_loss=0.02157, over 14023.00 frames. ], tot_loss[loss=0.1279, simple_loss=0.1574, pruned_loss=0.04924, over 1782361.21 frames. ], batch size: 22, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:15:47,846 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.49 vs. limit=5.0 2022-12-08 01:16:01,532 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8327, 3.1917, 3.9957, 2.6657, 2.3658, 3.5038, 1.9977, 3.6159], device='cuda:0'), covar=tensor([0.1370, 0.1132, 0.0547, 0.3532, 0.2291, 0.0650, 0.3447, 0.0797], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0095, 0.0089, 0.0096, 0.0113, 0.0083, 0.0124, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:16:19,767 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0030, 4.7453, 4.5665, 5.0887, 4.5909, 4.2259, 5.0755, 4.9024], device='cuda:0'), covar=tensor([0.0574, 0.0574, 0.0743, 0.0467, 0.0636, 0.0568, 0.0492, 0.0566], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0129, 0.0136, 0.0144, 0.0137, 0.0114, 0.0156, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:16:28,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.083e+02 2.759e+02 3.343e+02 7.485e+02, threshold=5.518e+02, percent-clipped=2.0 2022-12-08 01:16:29,185 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83760.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:16:49,376 INFO [train.py:873] (0/4) Epoch 12, batch 600, loss[loss=0.1385, simple_loss=0.1576, pruned_loss=0.05967, over 6905.00 frames. ], tot_loss[loss=0.1283, simple_loss=0.158, pruned_loss=0.04936, over 1891592.55 frames. ], batch size: 100, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:17:06,849 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3446, 2.3710, 1.8835, 2.4008, 2.3105, 2.3525, 2.1517, 1.9947], device='cuda:0'), covar=tensor([0.1024, 0.0838, 0.2577, 0.0731, 0.0945, 0.0673, 0.1287, 0.1244], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0294, 0.0266, 0.0255, 0.0307, 0.0294, 0.0252, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 01:17:22,765 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:17:50,247 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:17:57,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.241e+02 2.685e+02 3.305e+02 1.014e+03, threshold=5.370e+02, percent-clipped=1.0 2022-12-08 01:18:17,198 INFO [train.py:873] (0/4) Epoch 12, batch 700, loss[loss=0.1271, simple_loss=0.1612, pruned_loss=0.04653, over 14248.00 frames. ], tot_loss[loss=0.1288, simple_loss=0.1583, pruned_loss=0.04971, over 1931453.57 frames. ], batch size: 46, lr: 6.72e-03, grad_scale: 8.0 2022-12-08 01:18:32,214 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83900.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:05,292 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2670, 2.0599, 3.0815, 3.2492, 3.1975, 2.1995, 3.1087, 2.5040], device='cuda:0'), covar=tensor([0.0319, 0.0863, 0.0581, 0.0413, 0.0370, 0.1202, 0.0316, 0.0771], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0247, 0.0365, 0.0315, 0.0256, 0.0294, 0.0294, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:19:06,832 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83940.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:14,777 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83949.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:23,841 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.241e+02 2.717e+02 3.335e+02 9.146e+02, threshold=5.433e+02, percent-clipped=9.0 2022-12-08 01:19:44,262 INFO [train.py:873] (0/4) Epoch 12, batch 800, loss[loss=0.1292, simple_loss=0.1613, pruned_loss=0.04858, over 13988.00 frames. ], tot_loss[loss=0.1283, simple_loss=0.1579, pruned_loss=0.04934, over 1934955.08 frames. ], batch size: 22, lr: 6.71e-03, grad_scale: 8.0 2022-12-08 01:19:46,274 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:19:48,765 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83988.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:20:08,121 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84010.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:20:32,837 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4477, 1.0246, 1.3096, 0.8154, 1.1060, 1.4417, 1.1351, 1.1015], device='cuda:0'), covar=tensor([0.0321, 0.0815, 0.0583, 0.0475, 0.1043, 0.0562, 0.0464, 0.1248], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0182, 0.0137, 0.0127, 0.0136, 0.0143, 0.0121, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 01:20:40,375 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:20:52,176 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.165e+02 2.773e+02 3.425e+02 7.521e+02, threshold=5.546e+02, percent-clipped=4.0 2022-12-08 01:21:11,644 INFO [train.py:873] (0/4) Epoch 12, batch 900, loss[loss=0.1406, simple_loss=0.1644, pruned_loss=0.05842, over 12738.00 frames. ], tot_loss[loss=0.1281, simple_loss=0.1576, pruned_loss=0.04934, over 1883137.78 frames. ], batch size: 100, lr: 6.71e-03, grad_scale: 8.0 2022-12-08 01:21:24,314 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:21:30,547 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.83 vs. limit=5.0 2022-12-08 01:21:40,408 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:22:16,967 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:22:18,574 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.043e+02 2.201e+02 2.633e+02 3.303e+02 7.770e+02, threshold=5.265e+02, percent-clipped=5.0 2022-12-08 01:22:38,827 INFO [train.py:873] (0/4) Epoch 12, batch 1000, loss[loss=0.2152, simple_loss=0.1808, pruned_loss=0.1248, over 1251.00 frames. ], tot_loss[loss=0.1276, simple_loss=0.1571, pruned_loss=0.04905, over 1893737.63 frames. ], batch size: 100, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:22:52,814 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84199.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:23:14,872 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2022-12-08 01:23:25,696 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84236.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:23:46,259 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.176e+02 2.741e+02 3.221e+02 6.313e+02, threshold=5.481e+02, percent-clipped=1.0 2022-12-08 01:23:46,512 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:06,154 INFO [train.py:873] (0/4) Epoch 12, batch 1100, loss[loss=0.1321, simple_loss=0.1633, pruned_loss=0.05042, over 8618.00 frames. ], tot_loss[loss=0.1275, simple_loss=0.1568, pruned_loss=0.04909, over 1926680.76 frames. ], batch size: 100, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:24:16,245 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2509, 1.5083, 4.3040, 1.9886, 4.0444, 4.2767, 3.4858, 4.5523], device='cuda:0'), covar=tensor([0.0234, 0.3558, 0.0340, 0.2421, 0.0462, 0.0388, 0.0633, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0158, 0.0160, 0.0170, 0.0172, 0.0174, 0.0135, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:24:18,668 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84297.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:25,609 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84305.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:24:52,831 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:24:56,726 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84341.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:25:02,546 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 01:25:13,032 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.219e+02 2.734e+02 3.210e+02 6.975e+02, threshold=5.467e+02, percent-clipped=2.0 2022-12-08 01:25:25,201 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9582, 4.1006, 4.1636, 3.7127, 4.0754, 4.3024, 1.5718, 3.8741], device='cuda:0'), covar=tensor([0.0343, 0.0441, 0.0493, 0.0630, 0.0398, 0.0279, 0.3475, 0.0345], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0166, 0.0138, 0.0135, 0.0199, 0.0134, 0.0154, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:25:33,227 INFO [train.py:873] (0/4) Epoch 12, batch 1200, loss[loss=0.1188, simple_loss=0.1514, pruned_loss=0.04308, over 14468.00 frames. ], tot_loss[loss=0.1273, simple_loss=0.157, pruned_loss=0.04884, over 1970741.44 frames. ], batch size: 51, lr: 6.70e-03, grad_scale: 8.0 2022-12-08 01:25:35,866 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4693, 4.6293, 4.8433, 4.1714, 4.6552, 4.9920, 1.7278, 4.4100], device='cuda:0'), covar=tensor([0.0207, 0.0246, 0.0298, 0.0343, 0.0235, 0.0122, 0.2899, 0.0233], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0166, 0.0138, 0.0135, 0.0199, 0.0134, 0.0154, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:25:37,315 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 01:25:45,475 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84397.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:02,551 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84416.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:34,089 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84453.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:26:36,714 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 01:26:40,233 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.298e+02 2.755e+02 3.792e+02 6.830e+02, threshold=5.510e+02, percent-clipped=5.0 2022-12-08 01:26:43,683 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84464.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:27:00,263 INFO [train.py:873] (0/4) Epoch 12, batch 1300, loss[loss=0.1248, simple_loss=0.1555, pruned_loss=0.04705, over 14030.00 frames. ], tot_loss[loss=0.1279, simple_loss=0.1572, pruned_loss=0.04925, over 1956396.92 frames. ], batch size: 22, lr: 6.69e-03, grad_scale: 8.0 2022-12-08 01:27:36,891 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84525.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:27:53,595 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-08 01:28:03,449 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:07,761 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 2.228e+02 2.625e+02 3.303e+02 8.483e+02, threshold=5.249e+02, percent-clipped=5.0 2022-12-08 01:28:28,445 INFO [train.py:873] (0/4) Epoch 12, batch 1400, loss[loss=0.1178, simple_loss=0.156, pruned_loss=0.0398, over 14593.00 frames. ], tot_loss[loss=0.128, simple_loss=0.1575, pruned_loss=0.04925, over 1994880.99 frames. ], batch size: 34, lr: 6.69e-03, grad_scale: 8.0 2022-12-08 01:28:31,160 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84586.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:36,457 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84592.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:28:37,361 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2308, 2.9354, 2.2397, 3.3164, 3.1415, 3.1668, 2.7515, 2.2207], device='cuda:0'), covar=tensor([0.0881, 0.1586, 0.4218, 0.0661, 0.1047, 0.1012, 0.1745, 0.4180], device='cuda:0'), in_proj_covar=tensor([0.0267, 0.0293, 0.0266, 0.0256, 0.0308, 0.0293, 0.0253, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 01:28:47,801 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:12,514 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84633.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:19,404 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:29,689 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:29:35,866 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.687e+01 2.089e+02 2.664e+02 3.503e+02 6.178e+02, threshold=5.327e+02, percent-clipped=6.0 2022-12-08 01:29:55,407 INFO [train.py:873] (0/4) Epoch 12, batch 1500, loss[loss=0.1332, simple_loss=0.1388, pruned_loss=0.06383, over 2650.00 frames. ], tot_loss[loss=0.1275, simple_loss=0.157, pruned_loss=0.04907, over 1986909.90 frames. ], batch size: 100, lr: 6.68e-03, grad_scale: 8.0 2022-12-08 01:30:00,979 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84689.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:03,541 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:05,386 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84694.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:30:26,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5652, 2.2349, 3.6426, 3.7038, 3.5445, 2.1734, 3.6295, 2.6721], device='cuda:0'), covar=tensor([0.0406, 0.0972, 0.0652, 0.0441, 0.0436, 0.1503, 0.0392, 0.0956], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0251, 0.0369, 0.0320, 0.0260, 0.0297, 0.0299, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:30:27,739 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4545, 3.5175, 3.6270, 3.3162, 3.5521, 3.5516, 1.5359, 3.3869], device='cuda:0'), covar=tensor([0.0430, 0.0468, 0.0612, 0.0568, 0.0497, 0.0448, 0.3434, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0170, 0.0142, 0.0137, 0.0202, 0.0137, 0.0157, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 01:30:35,267 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 01:30:56,716 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84753.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:02,108 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.424e+02 2.855e+02 3.432e+02 8.447e+02, threshold=5.711e+02, percent-clipped=4.0 2022-12-08 01:31:22,262 INFO [train.py:873] (0/4) Epoch 12, batch 1600, loss[loss=0.1178, simple_loss=0.1323, pruned_loss=0.05165, over 2615.00 frames. ], tot_loss[loss=0.1273, simple_loss=0.1566, pruned_loss=0.04897, over 1925838.73 frames. ], batch size: 100, lr: 6.68e-03, grad_scale: 8.0 2022-12-08 01:31:36,904 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84800.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:37,610 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84801.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:31:49,142 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2022-12-08 01:31:52,203 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7788, 1.5930, 3.8535, 3.5151, 3.6613, 3.9214, 3.2890, 3.8666], device='cuda:0'), covar=tensor([0.1509, 0.1440, 0.0111, 0.0222, 0.0217, 0.0112, 0.0248, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0156, 0.0127, 0.0164, 0.0143, 0.0136, 0.0118, 0.0118], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:32:25,431 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84855.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:29,445 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.720e+01 2.080e+02 2.547e+02 3.159e+02 6.486e+02, threshold=5.095e+02, percent-clipped=3.0 2022-12-08 01:32:30,461 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84861.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:42,192 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0702, 1.3321, 1.4375, 0.8730, 0.9106, 1.0863, 0.8238, 1.2477], device='cuda:0'), covar=tensor([0.1678, 0.2860, 0.0914, 0.2370, 0.3257, 0.1105, 0.2336, 0.1257], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0098, 0.0090, 0.0096, 0.0114, 0.0083, 0.0124, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:32:47,191 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:32:48,864 INFO [train.py:873] (0/4) Epoch 12, batch 1700, loss[loss=0.123, simple_loss=0.1496, pruned_loss=0.04827, over 5970.00 frames. ], tot_loss[loss=0.1272, simple_loss=0.1567, pruned_loss=0.04882, over 1924791.96 frames. ], batch size: 100, lr: 6.68e-03, grad_scale: 4.0 2022-12-08 01:32:57,343 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:06,787 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84903.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:21,991 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4998, 4.2127, 4.1603, 4.5773, 4.0477, 3.7952, 4.5913, 4.3822], device='cuda:0'), covar=tensor([0.0648, 0.0877, 0.0778, 0.0500, 0.0848, 0.0670, 0.0598, 0.0671], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0131, 0.0138, 0.0149, 0.0141, 0.0115, 0.0158, 0.0137], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:33:39,690 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84940.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:49,406 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84951.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:33:57,762 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.143e+02 2.620e+02 3.286e+02 6.897e+02, threshold=5.240e+02, percent-clipped=6.0 2022-12-08 01:34:17,443 INFO [train.py:873] (0/4) Epoch 12, batch 1800, loss[loss=0.09566, simple_loss=0.126, pruned_loss=0.03267, over 10776.00 frames. ], tot_loss[loss=0.1267, simple_loss=0.1563, pruned_loss=0.04856, over 1950279.93 frames. ], batch size: 13, lr: 6.67e-03, grad_scale: 4.0 2022-12-08 01:34:22,814 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:34:25,415 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:34:32,928 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-85000.pt 2022-12-08 01:34:47,069 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85012.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:34:48,837 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0787, 2.1237, 4.1287, 2.9508, 3.9408, 2.0729, 3.1934, 3.9546], device='cuda:0'), covar=tensor([0.0548, 0.4198, 0.0452, 0.5701, 0.0543, 0.3636, 0.1231, 0.0456], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0213, 0.0200, 0.0287, 0.0222, 0.0217, 0.0215, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:35:10,612 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:35:29,534 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.196e+02 2.693e+02 3.379e+02 9.036e+02, threshold=5.386e+02, percent-clipped=1.0 2022-12-08 01:35:48,290 INFO [train.py:873] (0/4) Epoch 12, batch 1900, loss[loss=0.1347, simple_loss=0.1622, pruned_loss=0.05357, over 14566.00 frames. ], tot_loss[loss=0.1262, simple_loss=0.1563, pruned_loss=0.04807, over 1967215.00 frames. ], batch size: 23, lr: 6.67e-03, grad_scale: 4.0 2022-12-08 01:35:51,492 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.13 vs. limit=5.0 2022-12-08 01:36:09,443 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85107.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:16,277 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:52,390 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85156.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:36:56,466 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.007e+02 2.116e+02 2.633e+02 3.281e+02 1.062e+03, threshold=5.267e+02, percent-clipped=7.0 2022-12-08 01:37:02,868 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85168.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:37:10,096 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9086, 2.9773, 3.0977, 2.9814, 3.0205, 2.8030, 1.4691, 2.8036], device='cuda:0'), covar=tensor([0.0410, 0.0398, 0.0415, 0.0344, 0.0402, 0.0912, 0.2771, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0169, 0.0140, 0.0137, 0.0201, 0.0135, 0.0157, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:37:10,172 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:37:14,026 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85181.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:37:14,587 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 01:37:15,662 INFO [train.py:873] (0/4) Epoch 12, batch 2000, loss[loss=0.1331, simple_loss=0.163, pruned_loss=0.05163, over 14280.00 frames. ], tot_loss[loss=0.1268, simple_loss=0.1567, pruned_loss=0.04844, over 1904285.56 frames. ], batch size: 80, lr: 6.66e-03, grad_scale: 8.0 2022-12-08 01:37:56,102 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:38:25,282 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.573e+01 2.203e+02 2.703e+02 3.626e+02 1.921e+03, threshold=5.405e+02, percent-clipped=8.0 2022-12-08 01:38:28,902 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0856, 2.1637, 2.2863, 2.3482, 2.0000, 2.3460, 2.0282, 1.3792], device='cuda:0'), covar=tensor([0.1027, 0.0805, 0.0719, 0.0470, 0.0878, 0.0695, 0.1416, 0.2115], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0078, 0.0062, 0.0065, 0.0092, 0.0076, 0.0094, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 01:38:43,505 INFO [train.py:873] (0/4) Epoch 12, batch 2100, loss[loss=0.1299, simple_loss=0.1281, pruned_loss=0.0658, over 2595.00 frames. ], tot_loss[loss=0.1278, simple_loss=0.1569, pruned_loss=0.04929, over 1878408.50 frames. ], batch size: 100, lr: 6.66e-03, grad_scale: 4.0 2022-12-08 01:38:48,644 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:38:50,644 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4937, 1.1155, 2.0747, 1.8570, 1.9133, 2.0909, 1.4237, 2.0734], device='cuda:0'), covar=tensor([0.0658, 0.1143, 0.0206, 0.0389, 0.0449, 0.0212, 0.0569, 0.0241], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0157, 0.0127, 0.0165, 0.0143, 0.0138, 0.0119, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:39:04,429 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:39:05,313 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8709, 1.8635, 2.1875, 1.3896, 1.5965, 1.9342, 1.2369, 1.9868], device='cuda:0'), covar=tensor([0.1479, 0.1902, 0.0916, 0.3027, 0.3136, 0.1122, 0.4187, 0.1130], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0097, 0.0089, 0.0095, 0.0113, 0.0083, 0.0122, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:39:23,324 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7590, 1.6105, 1.6758, 1.7440, 1.6522, 0.8900, 1.5756, 1.5610], device='cuda:0'), covar=tensor([0.0984, 0.0835, 0.0529, 0.0999, 0.1058, 0.0934, 0.1059, 0.0719], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0026, 0.0029, 0.0026, 0.0028, 0.0039, 0.0027, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:39:30,441 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:39:43,928 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.56 vs. limit=5.0 2022-12-08 01:39:51,892 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.818e+01 2.339e+02 2.945e+02 3.643e+02 8.230e+02, threshold=5.891e+02, percent-clipped=7.0 2022-12-08 01:40:10,082 INFO [train.py:873] (0/4) Epoch 12, batch 2200, loss[loss=0.1209, simple_loss=0.1318, pruned_loss=0.05505, over 1284.00 frames. ], tot_loss[loss=0.1276, simple_loss=0.1568, pruned_loss=0.04919, over 1902862.68 frames. ], batch size: 100, lr: 6.66e-03, grad_scale: 4.0 2022-12-08 01:40:19,650 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 01:40:37,858 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 01:40:49,607 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:40:54,731 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5124, 2.2359, 2.9144, 1.7009, 1.8610, 2.4060, 1.4265, 2.5326], device='cuda:0'), covar=tensor([0.0878, 0.1384, 0.0677, 0.2336, 0.2390, 0.1046, 0.3660, 0.0866], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0095, 0.0089, 0.0094, 0.0112, 0.0082, 0.0120, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:41:14,575 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85456.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:19,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.150e+02 2.815e+02 3.417e+02 5.996e+02, threshold=5.631e+02, percent-clipped=2.0 2022-12-08 01:41:20,080 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85463.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 01:41:23,100 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 01:41:26,821 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:36,989 INFO [train.py:873] (0/4) Epoch 12, batch 2300, loss[loss=0.1309, simple_loss=0.1541, pruned_loss=0.05382, over 6908.00 frames. ], tot_loss[loss=0.1286, simple_loss=0.1574, pruned_loss=0.04992, over 1901152.46 frames. ], batch size: 100, lr: 6.65e-03, grad_scale: 4.0 2022-12-08 01:41:37,186 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:42,376 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:41:55,413 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85504.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:18,113 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85530.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:30,248 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85544.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:42:39,981 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0310, 2.1600, 2.0152, 2.1578, 1.8768, 2.0408, 2.1375, 2.0949], device='cuda:0'), covar=tensor([0.0968, 0.1124, 0.1180, 0.0908, 0.1463, 0.0918, 0.0943, 0.0861], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0128, 0.0135, 0.0146, 0.0137, 0.0112, 0.0154, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:42:45,171 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.375e+02 2.779e+02 3.534e+02 5.701e+02, threshold=5.557e+02, percent-clipped=1.0 2022-12-08 01:43:03,730 INFO [train.py:873] (0/4) Epoch 12, batch 2400, loss[loss=0.1009, simple_loss=0.143, pruned_loss=0.0294, over 13971.00 frames. ], tot_loss[loss=0.129, simple_loss=0.1579, pruned_loss=0.05002, over 1903049.65 frames. ], batch size: 19, lr: 6.65e-03, grad_scale: 8.0 2022-12-08 01:43:08,449 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2022-12-08 01:43:10,467 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:43:24,430 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85607.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:43:40,286 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:43:47,700 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1307, 2.0180, 1.7422, 1.7992, 2.0615, 2.0561, 2.0376, 2.0378], device='cuda:0'), covar=tensor([0.1151, 0.1027, 0.3183, 0.3084, 0.1238, 0.1208, 0.1646, 0.1288], device='cuda:0'), in_proj_covar=tensor([0.0371, 0.0257, 0.0432, 0.0552, 0.0326, 0.0421, 0.0392, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:43:50,481 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.61 vs. limit=5.0 2022-12-08 01:44:06,765 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85655.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:44:12,664 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 2.042e+02 2.557e+02 3.044e+02 5.063e+02, threshold=5.114e+02, percent-clipped=0.0 2022-12-08 01:44:30,324 INFO [train.py:873] (0/4) Epoch 12, batch 2500, loss[loss=0.1423, simple_loss=0.1684, pruned_loss=0.05811, over 14449.00 frames. ], tot_loss[loss=0.1277, simple_loss=0.1571, pruned_loss=0.04917, over 1883096.57 frames. ], batch size: 51, lr: 6.65e-03, grad_scale: 8.0 2022-12-08 01:44:33,077 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85686.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:45:11,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6397, 1.5267, 1.6080, 1.5176, 1.5496, 1.0288, 1.5489, 1.4760], device='cuda:0'), covar=tensor([0.0821, 0.0700, 0.0640, 0.1147, 0.0870, 0.0755, 0.0689, 0.0807], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0028, 0.0030, 0.0027, 0.0029, 0.0040, 0.0028, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:45:15,224 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0427, 2.7335, 2.7994, 1.8855, 2.4861, 2.7541, 3.1084, 2.4495], device='cuda:0'), covar=tensor([0.0676, 0.1194, 0.1024, 0.1818, 0.0991, 0.0991, 0.0592, 0.1395], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0181, 0.0139, 0.0126, 0.0136, 0.0143, 0.0121, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 01:45:40,095 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.190e+02 2.751e+02 3.272e+02 6.413e+02, threshold=5.502e+02, percent-clipped=4.0 2022-12-08 01:45:40,267 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:45:47,415 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:45:58,010 INFO [train.py:873] (0/4) Epoch 12, batch 2600, loss[loss=0.1311, simple_loss=0.1612, pruned_loss=0.05046, over 11181.00 frames. ], tot_loss[loss=0.1289, simple_loss=0.1576, pruned_loss=0.05013, over 1877998.28 frames. ], batch size: 100, lr: 6.64e-03, grad_scale: 4.0 2022-12-08 01:45:58,942 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:22,740 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85811.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:29,916 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:46:35,733 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0213, 2.0396, 5.0351, 4.6149, 4.4470, 5.1800, 4.9542, 5.2656], device='cuda:0'), covar=tensor([0.1424, 0.1352, 0.0086, 0.0143, 0.0172, 0.0105, 0.0087, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0157, 0.0127, 0.0166, 0.0143, 0.0138, 0.0118, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 01:46:46,864 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85839.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:47:08,283 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.414e+01 2.205e+02 2.825e+02 3.419e+02 7.121e+02, threshold=5.651e+02, percent-clipped=1.0 2022-12-08 01:47:18,514 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:47:22,943 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 01:47:25,017 INFO [train.py:873] (0/4) Epoch 12, batch 2700, loss[loss=0.1096, simple_loss=0.1469, pruned_loss=0.03612, over 14653.00 frames. ], tot_loss[loss=0.1274, simple_loss=0.157, pruned_loss=0.04883, over 1972935.54 frames. ], batch size: 33, lr: 6.64e-03, grad_scale: 4.0 2022-12-08 01:47:27,750 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85886.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:48:06,104 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85930.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:48:11,815 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:48:29,703 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85957.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:48:34,529 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.895e+01 2.358e+02 3.006e+02 3.476e+02 6.598e+02, threshold=6.011e+02, percent-clipped=3.0 2022-12-08 01:48:35,911 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 01:48:51,220 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85981.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:48:52,815 INFO [train.py:873] (0/4) Epoch 12, batch 2800, loss[loss=0.1133, simple_loss=0.1263, pruned_loss=0.05012, over 2686.00 frames. ], tot_loss[loss=0.1268, simple_loss=0.1567, pruned_loss=0.04842, over 1975257.12 frames. ], batch size: 100, lr: 6.63e-03, grad_scale: 8.0 2022-12-08 01:48:59,865 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3289, 2.2576, 2.5709, 1.7354, 1.7953, 2.3720, 1.3302, 2.3235], device='cuda:0'), covar=tensor([0.1105, 0.1657, 0.0938, 0.2373, 0.2953, 0.1002, 0.4644, 0.1293], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0096, 0.0089, 0.0093, 0.0113, 0.0082, 0.0121, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 01:48:59,914 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85991.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:49:00,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3977, 1.3429, 1.4595, 1.1563, 1.1802, 1.0539, 1.0860, 1.0008], device='cuda:0'), covar=tensor([0.0184, 0.0345, 0.0133, 0.0182, 0.0191, 0.0382, 0.0214, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0016, 0.0017, 0.0015, 0.0016, 0.0015, 0.0026, 0.0021, 0.0026], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:49:09,741 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:49:21,416 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 01:49:24,365 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:49:40,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9360, 1.6043, 1.8981, 1.6377, 2.0525, 1.8053, 1.6563, 1.8683], device='cuda:0'), covar=tensor([0.0514, 0.1383, 0.0277, 0.0460, 0.0378, 0.0648, 0.0279, 0.0371], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0316, 0.0399, 0.0305, 0.0377, 0.0321, 0.0371, 0.0311], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:49:53,842 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86052.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:03,560 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.107e+02 2.317e+02 2.627e+02 3.139e+02 9.999e+02, threshold=5.255e+02, percent-clipped=3.0 2022-12-08 01:50:03,799 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:08,734 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86069.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:20,497 INFO [train.py:873] (0/4) Epoch 12, batch 2900, loss[loss=0.09794, simple_loss=0.1311, pruned_loss=0.03241, over 5987.00 frames. ], tot_loss[loss=0.1262, simple_loss=0.1562, pruned_loss=0.04813, over 1965075.54 frames. ], batch size: 100, lr: 6.63e-03, grad_scale: 8.0 2022-12-08 01:50:21,285 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:46,773 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86113.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:50:46,807 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2560, 2.2207, 3.0128, 2.4317, 3.0753, 3.0715, 2.8682, 2.6099], device='cuda:0'), covar=tensor([0.0748, 0.2642, 0.0799, 0.1791, 0.0641, 0.0845, 0.1114, 0.1676], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0315, 0.0399, 0.0305, 0.0376, 0.0321, 0.0371, 0.0310], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:51:01,355 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:03,158 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:09,866 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86139.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:26,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3412, 2.2738, 4.3203, 2.9360, 4.1137, 2.0664, 3.2144, 4.0966], device='cuda:0'), covar=tensor([0.0486, 0.3922, 0.0411, 0.6185, 0.0507, 0.3630, 0.1311, 0.0367], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0216, 0.0205, 0.0289, 0.0226, 0.0219, 0.0218, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:51:30,509 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.360e+01 2.186e+02 2.706e+02 3.386e+02 6.755e+02, threshold=5.412e+02, percent-clipped=5.0 2022-12-08 01:51:35,871 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 01:51:47,631 INFO [train.py:873] (0/4) Epoch 12, batch 3000, loss[loss=0.117, simple_loss=0.1448, pruned_loss=0.04456, over 5944.00 frames. ], tot_loss[loss=0.1269, simple_loss=0.1569, pruned_loss=0.04841, over 1986183.71 frames. ], batch size: 100, lr: 6.63e-03, grad_scale: 4.0 2022-12-08 01:51:47,632 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 01:51:56,040 INFO [train.py:905] (0/4) Epoch 12, validation: loss=0.1299, simple_loss=0.1698, pruned_loss=0.04501, over 857387.00 frames. 2022-12-08 01:51:56,041 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 01:51:58,741 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86186.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:51:59,512 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86187.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:03,227 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1081, 1.1092, 1.1479, 0.8499, 0.7989, 0.7231, 0.8686, 0.8109], device='cuda:0'), covar=tensor([0.0145, 0.0157, 0.0152, 0.0165, 0.0185, 0.0359, 0.0228, 0.0350], device='cuda:0'), in_proj_covar=tensor([0.0016, 0.0017, 0.0015, 0.0016, 0.0016, 0.0027, 0.0021, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:52:38,310 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86231.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:41,112 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:52:42,886 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 01:52:51,122 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9186, 5.0025, 5.3607, 4.5091, 5.1389, 5.5154, 1.8545, 4.8144], device='cuda:0'), covar=tensor([0.0206, 0.0256, 0.0276, 0.0503, 0.0255, 0.0086, 0.3149, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0170, 0.0140, 0.0138, 0.0202, 0.0135, 0.0157, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 01:52:52,148 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86246.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:07,352 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.699e+01 2.315e+02 2.814e+02 3.910e+02 8.994e+02, threshold=5.629e+02, percent-clipped=7.0 2022-12-08 01:53:22,998 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:24,629 INFO [train.py:873] (0/4) Epoch 12, batch 3100, loss[loss=0.1231, simple_loss=0.161, pruned_loss=0.04259, over 14234.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1556, pruned_loss=0.04728, over 1986318.35 frames. ], batch size: 25, lr: 6.62e-03, grad_scale: 4.0 2022-12-08 01:53:27,164 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86286.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 01:53:45,097 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:53:50,621 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86313.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:04,957 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:31,027 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86358.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:54:36,315 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.157e+02 2.639e+02 3.347e+02 7.885e+02, threshold=5.278e+02, percent-clipped=4.0 2022-12-08 01:54:50,766 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 01:54:51,747 INFO [train.py:873] (0/4) Epoch 12, batch 3200, loss[loss=0.1875, simple_loss=0.1954, pruned_loss=0.08976, over 7802.00 frames. ], tot_loss[loss=0.1255, simple_loss=0.156, pruned_loss=0.04747, over 1976643.12 frames. ], batch size: 100, lr: 6.62e-03, grad_scale: 8.0 2022-12-08 01:55:01,731 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5368, 2.8494, 4.2328, 3.2472, 4.3085, 4.0957, 4.0121, 3.6586], device='cuda:0'), covar=tensor([0.0591, 0.2597, 0.0832, 0.1612, 0.0728, 0.0835, 0.1583, 0.1639], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0313, 0.0398, 0.0304, 0.0375, 0.0319, 0.0368, 0.0310], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:55:14,618 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:28,827 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:42,188 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:55:46,286 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:56:02,446 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.245e+02 2.734e+02 3.789e+02 7.990e+02, threshold=5.469e+02, percent-clipped=5.0 2022-12-08 01:56:05,228 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6453, 2.6154, 2.0003, 2.7102, 2.5233, 2.5712, 2.3257, 2.1051], device='cuda:0'), covar=tensor([0.0984, 0.1072, 0.2703, 0.0869, 0.1142, 0.1013, 0.1567, 0.2011], device='cuda:0'), in_proj_covar=tensor([0.0269, 0.0294, 0.0268, 0.0263, 0.0313, 0.0298, 0.0258, 0.0252], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 01:56:20,043 INFO [train.py:873] (0/4) Epoch 12, batch 3300, loss[loss=0.1245, simple_loss=0.1435, pruned_loss=0.05277, over 4952.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1555, pruned_loss=0.04732, over 1986635.31 frames. ], batch size: 100, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:56:35,431 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86501.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:56:39,510 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86506.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 01:57:01,777 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86531.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:30,031 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.093e+02 2.507e+02 3.209e+02 6.718e+02, threshold=5.013e+02, percent-clipped=2.0 2022-12-08 01:57:42,902 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86579.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:43,919 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86580.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:57:46,319 INFO [train.py:873] (0/4) Epoch 12, batch 3400, loss[loss=0.1456, simple_loss=0.1738, pruned_loss=0.05873, over 14247.00 frames. ], tot_loss[loss=0.1248, simple_loss=0.1555, pruned_loss=0.04704, over 1999157.62 frames. ], batch size: 80, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:57:48,944 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86586.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 01:58:03,133 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2022-12-08 01:58:03,560 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86602.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:04,114 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-08 01:58:13,123 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86613.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:31,337 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86634.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 01:58:38,069 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:52,723 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86658.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:55,348 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86661.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:58:57,326 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5420, 2.5677, 4.3155, 4.4150, 4.3794, 2.5301, 4.4031, 3.5225], device='cuda:0'), covar=tensor([0.0294, 0.0946, 0.0782, 0.0351, 0.0359, 0.1572, 0.0346, 0.0801], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0251, 0.0368, 0.0321, 0.0262, 0.0299, 0.0298, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 01:58:57,881 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.267e+02 2.858e+02 3.502e+02 8.320e+02, threshold=5.717e+02, percent-clipped=7.0 2022-12-08 01:59:14,481 INFO [train.py:873] (0/4) Epoch 12, batch 3500, loss[loss=0.1496, simple_loss=0.1672, pruned_loss=0.06594, over 12789.00 frames. ], tot_loss[loss=0.1244, simple_loss=0.1546, pruned_loss=0.04711, over 1975362.03 frames. ], batch size: 100, lr: 6.61e-03, grad_scale: 8.0 2022-12-08 01:59:17,063 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9659, 3.0772, 3.2056, 3.0799, 3.0892, 2.9351, 1.4792, 2.8737], device='cuda:0'), covar=tensor([0.0379, 0.0352, 0.0360, 0.0372, 0.0340, 0.0737, 0.2787, 0.0305], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0170, 0.0140, 0.0138, 0.0203, 0.0135, 0.0158, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 01:59:34,026 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86706.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:59:35,793 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86708.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:59:41,128 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3363, 1.4124, 1.4115, 1.4990, 1.4972, 0.9381, 1.2514, 1.3528], device='cuda:0'), covar=tensor([0.0662, 0.0707, 0.0578, 0.0567, 0.0447, 0.0880, 0.0858, 0.0596], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0027, 0.0030, 0.0026, 0.0029, 0.0040, 0.0028, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 01:59:51,310 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86725.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 01:59:52,151 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9893, 1.9658, 2.0583, 2.0892, 1.9603, 1.5634, 1.3373, 1.7522], device='cuda:0'), covar=tensor([0.0551, 0.0534, 0.0457, 0.0340, 0.0435, 0.1307, 0.2101, 0.0442], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0170, 0.0140, 0.0138, 0.0202, 0.0134, 0.0158, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 02:00:02,466 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2022-12-08 02:00:17,807 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86756.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:24,901 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.023e+02 2.697e+02 3.316e+02 6.067e+02, threshold=5.393e+02, percent-clipped=1.0 2022-12-08 02:00:32,574 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86773.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:41,038 INFO [train.py:873] (0/4) Epoch 12, batch 3600, loss[loss=0.1122, simple_loss=0.1484, pruned_loss=0.03801, over 14340.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1552, pruned_loss=0.04691, over 1967279.76 frames. ], batch size: 28, lr: 6.60e-03, grad_scale: 8.0 2022-12-08 02:00:52,488 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86796.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:00:57,574 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86801.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:01:08,648 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:01:15,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8901, 1.9971, 2.1791, 1.5436, 1.5157, 2.0381, 1.2340, 1.9437], device='cuda:0'), covar=tensor([0.1539, 0.1920, 0.0995, 0.2445, 0.3463, 0.1137, 0.4670, 0.1157], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0097, 0.0090, 0.0096, 0.0114, 0.0085, 0.0123, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 02:01:51,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 2.201e+02 2.694e+02 3.354e+02 5.968e+02, threshold=5.388e+02, percent-clipped=2.0 2022-12-08 02:02:01,399 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:01,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-08 02:02:04,399 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86878.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:08,669 INFO [train.py:873] (0/4) Epoch 12, batch 3700, loss[loss=0.1099, simple_loss=0.1483, pruned_loss=0.03571, over 13539.00 frames. ], tot_loss[loss=0.1243, simple_loss=0.1549, pruned_loss=0.04681, over 1962159.06 frames. ], batch size: 100, lr: 6.60e-03, grad_scale: 8.0 2022-12-08 02:02:24,511 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86902.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:32,462 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.77 vs. limit=5.0 2022-12-08 02:02:49,085 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2022-12-08 02:02:49,273 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2022-12-08 02:02:54,832 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:02:57,541 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:03:02,231 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7574, 1.8078, 1.8129, 1.8995, 1.9098, 1.1127, 1.5880, 1.8810], device='cuda:0'), covar=tensor([0.1592, 0.0535, 0.1222, 0.1221, 0.1094, 0.0905, 0.0764, 0.0528], device='cuda:0'), in_proj_covar=tensor([0.0027, 0.0027, 0.0029, 0.0026, 0.0028, 0.0039, 0.0028, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 02:03:06,351 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:03:19,132 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.304e+02 2.133e+02 2.735e+02 3.535e+02 1.030e+03, threshold=5.470e+02, percent-clipped=7.0 2022-12-08 02:03:35,277 INFO [train.py:873] (0/4) Epoch 12, batch 3800, loss[loss=0.1071, simple_loss=0.1496, pruned_loss=0.03231, over 14379.00 frames. ], tot_loss[loss=0.1241, simple_loss=0.155, pruned_loss=0.04661, over 1942308.40 frames. ], batch size: 41, lr: 6.60e-03, grad_scale: 4.0 2022-12-08 02:03:52,744 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87002.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:04:45,917 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87063.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:04:47,286 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.236e+02 2.550e+02 3.017e+02 4.626e+02, threshold=5.100e+02, percent-clipped=0.0 2022-12-08 02:05:03,579 INFO [train.py:873] (0/4) Epoch 12, batch 3900, loss[loss=0.1381, simple_loss=0.1494, pruned_loss=0.06342, over 3899.00 frames. ], tot_loss[loss=0.1245, simple_loss=0.1553, pruned_loss=0.04686, over 1975030.21 frames. ], batch size: 100, lr: 6.59e-03, grad_scale: 4.0 2022-12-08 02:05:14,059 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87096.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:05:14,082 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7213, 3.0309, 2.9401, 2.9143, 2.2050, 3.0824, 2.7558, 1.5398], device='cuda:0'), covar=tensor([0.1684, 0.0711, 0.0957, 0.0776, 0.1173, 0.0547, 0.1244, 0.2370], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0078, 0.0062, 0.0067, 0.0091, 0.0077, 0.0094, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 02:05:18,341 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87101.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:05:48,145 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0949, 2.0028, 4.8360, 4.4010, 4.2453, 4.9216, 4.6906, 4.9556], device='cuda:0'), covar=tensor([0.1482, 0.1435, 0.0087, 0.0166, 0.0185, 0.0093, 0.0091, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0159, 0.0127, 0.0166, 0.0144, 0.0140, 0.0119, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:05:56,018 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87144.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:00,188 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87149.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:13,579 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9762, 1.4832, 3.8764, 1.8914, 3.8727, 4.0140, 3.2158, 4.3609], device='cuda:0'), covar=tensor([0.0233, 0.3225, 0.0448, 0.2233, 0.0444, 0.0486, 0.0634, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0160, 0.0169, 0.0170, 0.0177, 0.0133, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 02:06:14,161 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.280e+02 2.775e+02 3.261e+02 6.407e+02, threshold=5.550e+02, percent-clipped=4.0 2022-12-08 02:06:18,667 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:06:29,720 INFO [train.py:873] (0/4) Epoch 12, batch 4000, loss[loss=0.1377, simple_loss=0.162, pruned_loss=0.05665, over 11989.00 frames. ], tot_loss[loss=0.1246, simple_loss=0.1554, pruned_loss=0.04694, over 1999571.17 frames. ], batch size: 100, lr: 6.59e-03, grad_scale: 8.0 2022-12-08 02:07:10,541 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87230.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:14,032 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:15,980 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87236.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:07:33,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-08 02:07:40,897 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.153e+02 2.853e+02 3.504e+02 7.130e+02, threshold=5.705e+02, percent-clipped=5.0 2022-12-08 02:07:48,313 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-08 02:07:57,456 INFO [train.py:873] (0/4) Epoch 12, batch 4100, loss[loss=0.1264, simple_loss=0.1528, pruned_loss=0.05003, over 14394.00 frames. ], tot_loss[loss=0.1255, simple_loss=0.1556, pruned_loss=0.04773, over 1935141.57 frames. ], batch size: 53, lr: 6.58e-03, grad_scale: 8.0 2022-12-08 02:07:58,283 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:08:04,338 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87291.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:08:15,045 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-08 02:09:02,397 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87358.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:09:06,572 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87362.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:09:09,785 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.165e+02 2.894e+02 3.729e+02 1.916e+03, threshold=5.789e+02, percent-clipped=7.0 2022-12-08 02:09:24,060 INFO [train.py:873] (0/4) Epoch 12, batch 4200, loss[loss=0.1056, simple_loss=0.1473, pruned_loss=0.03191, over 13836.00 frames. ], tot_loss[loss=0.1258, simple_loss=0.1559, pruned_loss=0.04781, over 1929220.73 frames. ], batch size: 20, lr: 6.58e-03, grad_scale: 4.0 2022-12-08 02:09:58,920 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87423.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:32,215 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87461.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:36,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.802e+01 2.101e+02 2.786e+02 3.317e+02 5.801e+02, threshold=5.572e+02, percent-clipped=2.0 2022-12-08 02:10:39,540 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87470.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:10:51,434 INFO [train.py:873] (0/4) Epoch 12, batch 4300, loss[loss=0.1431, simple_loss=0.1393, pruned_loss=0.07346, over 1221.00 frames. ], tot_loss[loss=0.1262, simple_loss=0.1563, pruned_loss=0.04808, over 1933189.82 frames. ], batch size: 100, lr: 6.58e-03, grad_scale: 4.0 2022-12-08 02:11:10,466 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:21,405 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87518.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:25,961 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87522.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:11:35,390 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5084, 5.3136, 4.9049, 5.5411, 5.0548, 4.7805, 5.5569, 5.4173], device='cuda:0'), covar=tensor([0.0596, 0.0632, 0.0808, 0.0463, 0.0742, 0.0450, 0.0592, 0.0535], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0131, 0.0138, 0.0147, 0.0140, 0.0113, 0.0157, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 02:11:36,352 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87534.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:04,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.182e+02 2.709e+02 3.348e+02 1.136e+03, threshold=5.418e+02, percent-clipped=3.0 2022-12-08 02:12:04,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87566.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:12:18,387 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87582.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:19,183 INFO [train.py:873] (0/4) Epoch 12, batch 4400, loss[loss=0.1253, simple_loss=0.1503, pruned_loss=0.05018, over 6006.00 frames. ], tot_loss[loss=0.126, simple_loss=0.1559, pruned_loss=0.04804, over 1917299.30 frames. ], batch size: 100, lr: 6.57e-03, grad_scale: 8.0 2022-12-08 02:12:19,958 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87584.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:21,533 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87586.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:24,098 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87589.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:38,858 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:12:47,304 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1466, 2.0724, 1.8209, 1.8635, 2.0807, 2.1080, 2.0715, 2.0853], device='cuda:0'), covar=tensor([0.1006, 0.0847, 0.2325, 0.2501, 0.0996, 0.0991, 0.1376, 0.1006], device='cuda:0'), in_proj_covar=tensor([0.0375, 0.0259, 0.0436, 0.0555, 0.0330, 0.0423, 0.0387, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:13:13,669 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87645.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:13:18,221 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:13:24,946 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87658.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:13:29,722 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 02:13:31,918 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87666.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:13:32,488 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.847e+01 2.130e+02 2.616e+02 3.250e+02 5.248e+02, threshold=5.232e+02, percent-clipped=0.0 2022-12-08 02:13:46,587 INFO [train.py:873] (0/4) Epoch 12, batch 4500, loss[loss=0.1315, simple_loss=0.1595, pruned_loss=0.05175, over 14212.00 frames. ], tot_loss[loss=0.1243, simple_loss=0.1549, pruned_loss=0.04687, over 1927235.35 frames. ], batch size: 94, lr: 6.57e-03, grad_scale: 4.0 2022-12-08 02:14:06,776 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87706.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:14:17,065 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:14:35,555 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0828, 1.6747, 4.5263, 4.2142, 4.0756, 4.5881, 4.0737, 4.5909], device='cuda:0'), covar=tensor([0.1424, 0.1518, 0.0095, 0.0178, 0.0196, 0.0092, 0.0174, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0160, 0.0128, 0.0168, 0.0145, 0.0142, 0.0121, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:14:59,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.389e+02 2.965e+02 3.391e+02 5.507e+02, threshold=5.931e+02, percent-clipped=1.0 2022-12-08 02:15:13,018 INFO [train.py:873] (0/4) Epoch 12, batch 4600, loss[loss=0.1453, simple_loss=0.1669, pruned_loss=0.06179, over 14179.00 frames. ], tot_loss[loss=0.1252, simple_loss=0.1554, pruned_loss=0.04747, over 1910963.70 frames. ], batch size: 99, lr: 6.57e-03, grad_scale: 4.0 2022-12-08 02:15:20,935 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2633, 3.3145, 3.2744, 3.4138, 2.4448, 3.3752, 3.1777, 1.8969], device='cuda:0'), covar=tensor([0.2017, 0.1214, 0.1145, 0.0786, 0.1219, 0.0610, 0.1307, 0.2498], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0081, 0.0065, 0.0068, 0.0094, 0.0079, 0.0095, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 02:15:42,228 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 02:15:43,402 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87817.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:16:21,463 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87861.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:16:26,308 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 2.249e+02 2.775e+02 3.473e+02 5.522e+02, threshold=5.551e+02, percent-clipped=0.0 2022-12-08 02:16:40,648 INFO [train.py:873] (0/4) Epoch 12, batch 4700, loss[loss=0.1403, simple_loss=0.1627, pruned_loss=0.059, over 9501.00 frames. ], tot_loss[loss=0.1243, simple_loss=0.1549, pruned_loss=0.04681, over 1918968.94 frames. ], batch size: 100, lr: 6.56e-03, grad_scale: 4.0 2022-12-08 02:16:43,352 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87886.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:25,805 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87934.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:31,128 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87940.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:17:35,298 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:17:48,995 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87961.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:17:54,230 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.984e+01 2.201e+02 2.736e+02 3.426e+02 5.956e+02, threshold=5.473e+02, percent-clipped=2.0 2022-12-08 02:17:58,821 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 02:18:08,683 INFO [train.py:873] (0/4) Epoch 12, batch 4800, loss[loss=0.1493, simple_loss=0.1407, pruned_loss=0.07893, over 1277.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1551, pruned_loss=0.04754, over 1895342.41 frames. ], batch size: 100, lr: 6.56e-03, grad_scale: 8.0 2022-12-08 02:18:15,753 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0620, 3.1471, 3.0600, 3.2653, 2.3954, 3.3120, 3.2010, 1.5797], device='cuda:0'), covar=tensor([0.1731, 0.0833, 0.1262, 0.0650, 0.1106, 0.0608, 0.0889, 0.2592], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0080, 0.0064, 0.0068, 0.0093, 0.0078, 0.0095, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 02:18:39,753 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:19:07,589 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 02:19:21,827 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:19:22,466 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.264e+02 2.796e+02 3.655e+02 7.619e+02, threshold=5.592e+02, percent-clipped=2.0 2022-12-08 02:19:26,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4411, 3.2531, 2.9458, 3.1669, 3.3451, 3.3336, 3.3843, 3.3931], device='cuda:0'), covar=tensor([0.0922, 0.0707, 0.2408, 0.2436, 0.0832, 0.0981, 0.1277, 0.0878], device='cuda:0'), in_proj_covar=tensor([0.0374, 0.0260, 0.0431, 0.0553, 0.0327, 0.0422, 0.0388, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:19:36,608 INFO [train.py:873] (0/4) Epoch 12, batch 4900, loss[loss=0.1134, simple_loss=0.1476, pruned_loss=0.03958, over 14279.00 frames. ], tot_loss[loss=0.1258, simple_loss=0.1559, pruned_loss=0.04787, over 1905094.29 frames. ], batch size: 31, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:19:52,012 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88100.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:06,819 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:45,831 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88161.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:20:45,864 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88161.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:49,480 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:20:51,096 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.168e+02 2.607e+02 3.248e+02 6.908e+02, threshold=5.213e+02, percent-clipped=2.0 2022-12-08 02:20:54,371 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-08 02:21:05,439 INFO [train.py:873] (0/4) Epoch 12, batch 5000, loss[loss=0.1407, simple_loss=0.1593, pruned_loss=0.06106, over 7742.00 frames. ], tot_loss[loss=0.1264, simple_loss=0.156, pruned_loss=0.04838, over 1834436.34 frames. ], batch size: 100, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:21:28,451 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88209.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:21:55,850 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88240.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:00,492 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88245.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:07,641 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8659, 1.4188, 3.6826, 1.8731, 3.7326, 3.9067, 2.9145, 4.2457], device='cuda:0'), covar=tensor([0.0231, 0.3107, 0.0527, 0.2041, 0.0445, 0.0378, 0.0789, 0.0160], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0161, 0.0168, 0.0170, 0.0177, 0.0132, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 02:22:14,890 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88261.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:22:19,851 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.015e+02 2.163e+02 2.571e+02 3.278e+02 8.920e+02, threshold=5.143e+02, percent-clipped=2.0 2022-12-08 02:22:23,555 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7138, 1.9235, 3.8035, 2.6082, 3.5557, 1.8418, 2.7943, 3.5644], device='cuda:0'), covar=tensor([0.0786, 0.4533, 0.0540, 0.5910, 0.0815, 0.3751, 0.1473, 0.0607], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0213, 0.0203, 0.0285, 0.0226, 0.0213, 0.0213, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:22:33,614 INFO [train.py:873] (0/4) Epoch 12, batch 5100, loss[loss=0.1147, simple_loss=0.1187, pruned_loss=0.05533, over 1257.00 frames. ], tot_loss[loss=0.1268, simple_loss=0.1559, pruned_loss=0.0488, over 1812702.71 frames. ], batch size: 100, lr: 6.55e-03, grad_scale: 8.0 2022-12-08 02:22:38,197 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88288.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:41,170 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88291.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:22:42,664 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:22:56,533 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88309.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:23:34,075 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88352.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:23:46,914 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.245e+02 2.755e+02 3.354e+02 6.298e+02, threshold=5.511e+02, percent-clipped=3.0 2022-12-08 02:23:52,394 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-08 02:24:01,890 INFO [train.py:873] (0/4) Epoch 12, batch 5200, loss[loss=0.1238, simple_loss=0.1599, pruned_loss=0.04386, over 14523.00 frames. ], tot_loss[loss=0.1265, simple_loss=0.156, pruned_loss=0.04853, over 1859715.83 frames. ], batch size: 34, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:25:05,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 02:25:06,265 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88456.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:25:15,464 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.113e+02 2.685e+02 3.247e+02 4.954e+02, threshold=5.369e+02, percent-clipped=0.0 2022-12-08 02:25:29,579 INFO [train.py:873] (0/4) Epoch 12, batch 5300, loss[loss=0.119, simple_loss=0.1504, pruned_loss=0.0438, over 14139.00 frames. ], tot_loss[loss=0.1263, simple_loss=0.1563, pruned_loss=0.04818, over 1926010.89 frames. ], batch size: 84, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:26:43,275 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.234e+01 2.154e+02 2.570e+02 3.365e+02 7.012e+02, threshold=5.140e+02, percent-clipped=5.0 2022-12-08 02:26:57,899 INFO [train.py:873] (0/4) Epoch 12, batch 5400, loss[loss=0.1308, simple_loss=0.1587, pruned_loss=0.05148, over 14228.00 frames. ], tot_loss[loss=0.1244, simple_loss=0.1551, pruned_loss=0.04686, over 1925947.28 frames. ], batch size: 80, lr: 6.54e-03, grad_scale: 8.0 2022-12-08 02:27:45,512 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.39 vs. limit=5.0 2022-12-08 02:27:53,977 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88647.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:28:11,936 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.088e+01 2.108e+02 2.673e+02 3.283e+02 5.238e+02, threshold=5.346e+02, percent-clipped=1.0 2022-12-08 02:28:26,023 INFO [train.py:873] (0/4) Epoch 12, batch 5500, loss[loss=0.145, simple_loss=0.1688, pruned_loss=0.06062, over 11188.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1548, pruned_loss=0.0463, over 1953649.39 frames. ], batch size: 100, lr: 6.53e-03, grad_scale: 8.0 2022-12-08 02:29:15,078 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3551, 4.0956, 3.7352, 4.0048, 4.1603, 4.2473, 4.3117, 4.2668], device='cuda:0'), covar=tensor([0.0885, 0.0527, 0.2056, 0.2577, 0.0737, 0.0776, 0.0970, 0.1017], device='cuda:0'), in_proj_covar=tensor([0.0377, 0.0262, 0.0435, 0.0556, 0.0328, 0.0426, 0.0386, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:29:29,709 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88756.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:29:38,829 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.096e+02 2.766e+02 3.641e+02 8.752e+02, threshold=5.532e+02, percent-clipped=4.0 2022-12-08 02:29:53,654 INFO [train.py:873] (0/4) Epoch 12, batch 5600, loss[loss=0.1063, simple_loss=0.1475, pruned_loss=0.03254, over 14219.00 frames. ], tot_loss[loss=0.124, simple_loss=0.1549, pruned_loss=0.04651, over 1954840.49 frames. ], batch size: 35, lr: 6.53e-03, grad_scale: 8.0 2022-12-08 02:30:12,307 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88804.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:30:26,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2022-12-08 02:31:06,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.217e+02 2.679e+02 3.309e+02 6.646e+02, threshold=5.358e+02, percent-clipped=2.0 2022-12-08 02:31:20,847 INFO [train.py:873] (0/4) Epoch 12, batch 5700, loss[loss=0.1281, simple_loss=0.1526, pruned_loss=0.05181, over 10360.00 frames. ], tot_loss[loss=0.1226, simple_loss=0.154, pruned_loss=0.04555, over 1979085.97 frames. ], batch size: 100, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:31:46,848 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0998, 2.9061, 2.7024, 2.7956, 3.0468, 3.0312, 3.0413, 3.0335], device='cuda:0'), covar=tensor([0.0963, 0.0774, 0.2095, 0.2527, 0.0798, 0.0927, 0.1231, 0.0938], device='cuda:0'), in_proj_covar=tensor([0.0375, 0.0259, 0.0430, 0.0550, 0.0324, 0.0423, 0.0382, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:32:16,507 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88947.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:32:32,857 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 2.272e+02 2.731e+02 3.384e+02 1.131e+03, threshold=5.463e+02, percent-clipped=4.0 2022-12-08 02:32:46,743 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3740, 4.4145, 4.1442, 4.1301, 4.1301, 4.7135, 1.6637, 3.9512], device='cuda:0'), covar=tensor([0.0583, 0.0632, 0.0982, 0.0646, 0.0940, 0.0290, 0.4986, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0168, 0.0139, 0.0137, 0.0199, 0.0136, 0.0157, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 02:32:47,397 INFO [train.py:873] (0/4) Epoch 12, batch 5800, loss[loss=0.1348, simple_loss=0.1685, pruned_loss=0.05053, over 14285.00 frames. ], tot_loss[loss=0.1239, simple_loss=0.1548, pruned_loss=0.04651, over 1948757.05 frames. ], batch size: 31, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:32:50,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6089, 4.1271, 3.2287, 4.8736, 4.2294, 4.6579, 4.0198, 3.3918], device='cuda:0'), covar=tensor([0.0607, 0.1212, 0.3850, 0.0561, 0.1213, 0.1309, 0.1254, 0.3391], device='cuda:0'), in_proj_covar=tensor([0.0263, 0.0294, 0.0265, 0.0259, 0.0310, 0.0292, 0.0252, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 02:32:57,690 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88995.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:33:26,344 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:33:35,741 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7562, 1.3248, 2.5297, 2.2864, 2.3500, 2.5179, 1.7576, 2.5388], device='cuda:0'), covar=tensor([0.0915, 0.1299, 0.0175, 0.0406, 0.0491, 0.0205, 0.0609, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0164, 0.0143, 0.0140, 0.0120, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:34:01,515 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.293e+02 2.084e+02 2.490e+02 3.041e+02 5.969e+02, threshold=4.980e+02, percent-clipped=1.0 2022-12-08 02:34:15,417 INFO [train.py:873] (0/4) Epoch 12, batch 5900, loss[loss=0.1215, simple_loss=0.1447, pruned_loss=0.0492, over 6915.00 frames. ], tot_loss[loss=0.1229, simple_loss=0.1542, pruned_loss=0.04587, over 1932968.06 frames. ], batch size: 100, lr: 6.52e-03, grad_scale: 8.0 2022-12-08 02:34:17,640 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1333, 1.9420, 5.1502, 4.5801, 4.4280, 5.1860, 4.9963, 5.2364], device='cuda:0'), covar=tensor([0.1418, 0.1425, 0.0069, 0.0162, 0.0188, 0.0081, 0.0076, 0.0089], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0163, 0.0143, 0.0140, 0.0120, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:34:20,399 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:35:00,706 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3595, 4.1229, 3.9606, 4.4239, 3.8879, 3.6356, 4.3780, 4.2831], device='cuda:0'), covar=tensor([0.0572, 0.0684, 0.0794, 0.0528, 0.0819, 0.0694, 0.0609, 0.0651], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0131, 0.0138, 0.0147, 0.0138, 0.0115, 0.0158, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 02:35:29,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.297e+02 2.664e+02 3.393e+02 7.245e+02, threshold=5.328e+02, percent-clipped=3.0 2022-12-08 02:35:31,816 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89169.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:35:40,311 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6969, 2.5489, 3.1500, 2.1030, 2.0859, 2.6350, 1.5494, 2.7572], device='cuda:0'), covar=tensor([0.0914, 0.1245, 0.0694, 0.1925, 0.2413, 0.1037, 0.3825, 0.0876], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0096, 0.0090, 0.0096, 0.0114, 0.0083, 0.0121, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 02:35:44,587 INFO [train.py:873] (0/4) Epoch 12, batch 6000, loss[loss=0.145, simple_loss=0.1666, pruned_loss=0.06175, over 9499.00 frames. ], tot_loss[loss=0.1242, simple_loss=0.1552, pruned_loss=0.0466, over 2011589.65 frames. ], batch size: 100, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:35:44,587 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 02:35:52,918 INFO [train.py:905] (0/4) Epoch 12, validation: loss=0.1296, simple_loss=0.1695, pruned_loss=0.04492, over 857387.00 frames. 2022-12-08 02:35:52,919 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 02:35:57,930 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-08 02:36:34,402 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89230.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:37:06,542 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.059e+02 2.275e+02 2.670e+02 3.684e+02 9.566e+02, threshold=5.341e+02, percent-clipped=7.0 2022-12-08 02:37:21,418 INFO [train.py:873] (0/4) Epoch 12, batch 6100, loss[loss=0.163, simple_loss=0.1788, pruned_loss=0.07358, over 14169.00 frames. ], tot_loss[loss=0.1249, simple_loss=0.1554, pruned_loss=0.04723, over 2001478.01 frames. ], batch size: 84, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:37:52,465 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2825, 2.5249, 4.2581, 4.4805, 4.3797, 2.4407, 4.4960, 3.1948], device='cuda:0'), covar=tensor([0.0354, 0.1068, 0.0852, 0.0339, 0.0331, 0.1668, 0.0295, 0.1047], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0256, 0.0372, 0.0325, 0.0266, 0.0302, 0.0304, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:38:27,779 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6439, 1.4136, 3.3466, 3.1296, 3.2553, 3.3576, 2.4423, 3.3691], device='cuda:0'), covar=tensor([0.2073, 0.2188, 0.0228, 0.0401, 0.0381, 0.0265, 0.0643, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0163, 0.0144, 0.0139, 0.0120, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:38:34,355 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 2.240e+02 2.644e+02 3.331e+02 6.090e+02, threshold=5.289e+02, percent-clipped=8.0 2022-12-08 02:38:40,603 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:38:48,056 INFO [train.py:873] (0/4) Epoch 12, batch 6200, loss[loss=0.1481, simple_loss=0.1376, pruned_loss=0.07928, over 2582.00 frames. ], tot_loss[loss=0.1244, simple_loss=0.155, pruned_loss=0.04692, over 2002537.10 frames. ], batch size: 100, lr: 6.51e-03, grad_scale: 8.0 2022-12-08 02:38:48,167 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:38:58,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-08 02:39:02,067 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2022-12-08 02:39:11,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.53 vs. limit=5.0 2022-12-08 02:39:24,537 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:39:33,550 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:39:37,335 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89439.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:39:38,945 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5419, 3.2084, 3.0828, 2.0550, 2.9432, 3.3112, 3.5957, 2.8906], device='cuda:0'), covar=tensor([0.0602, 0.1156, 0.1050, 0.1823, 0.0843, 0.0663, 0.0581, 0.1216], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0179, 0.0139, 0.0126, 0.0136, 0.0146, 0.0124, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 02:39:41,486 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1830, 1.8680, 2.1896, 1.4872, 1.8562, 2.2659, 2.1635, 1.9534], device='cuda:0'), covar=tensor([0.0839, 0.0703, 0.0861, 0.1594, 0.1311, 0.0719, 0.0629, 0.1536], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0179, 0.0139, 0.0126, 0.0136, 0.0146, 0.0124, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 02:40:00,789 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.324e+02 2.890e+02 3.661e+02 8.443e+02, threshold=5.779e+02, percent-clipped=4.0 2022-12-08 02:40:15,101 INFO [train.py:873] (0/4) Epoch 12, batch 6300, loss[loss=0.1557, simple_loss=0.1711, pruned_loss=0.07008, over 9426.00 frames. ], tot_loss[loss=0.1234, simple_loss=0.1545, pruned_loss=0.0461, over 1986238.08 frames. ], batch size: 100, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:40:17,844 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:40:29,981 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89500.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 02:40:37,606 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89509.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:40:51,637 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89525.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:41:09,662 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0896, 1.7084, 4.4737, 4.0266, 4.0586, 4.5148, 4.0388, 4.5475], device='cuda:0'), covar=tensor([0.1360, 0.1469, 0.0099, 0.0193, 0.0198, 0.0108, 0.0215, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0156, 0.0126, 0.0163, 0.0143, 0.0138, 0.0119, 0.0119], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:41:27,938 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.271e+01 2.078e+02 2.565e+02 3.125e+02 6.537e+02, threshold=5.131e+02, percent-clipped=2.0 2022-12-08 02:41:30,762 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89570.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:41:41,866 INFO [train.py:873] (0/4) Epoch 12, batch 6400, loss[loss=0.1494, simple_loss=0.1378, pruned_loss=0.08053, over 1208.00 frames. ], tot_loss[loss=0.1236, simple_loss=0.1546, pruned_loss=0.04631, over 2000312.61 frames. ], batch size: 100, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:42:06,832 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6809, 2.0966, 2.6760, 2.7886, 2.6551, 2.1250, 2.7715, 2.3168], device='cuda:0'), covar=tensor([0.0373, 0.0818, 0.0465, 0.0454, 0.0413, 0.1021, 0.0355, 0.0658], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0253, 0.0369, 0.0323, 0.0264, 0.0301, 0.0303, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 02:42:11,324 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3809, 1.3826, 3.5327, 1.5321, 3.2997, 3.5404, 2.3920, 3.6910], device='cuda:0'), covar=tensor([0.0305, 0.3540, 0.0460, 0.2531, 0.0827, 0.0450, 0.1071, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0157, 0.0160, 0.0168, 0.0167, 0.0177, 0.0132, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 02:42:56,147 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 2.320e+02 2.969e+02 3.886e+02 7.379e+02, threshold=5.938e+02, percent-clipped=5.0 2022-12-08 02:43:06,409 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2368, 3.5540, 3.0695, 3.3929, 2.5849, 3.5603, 3.3023, 1.8392], device='cuda:0'), covar=tensor([0.1685, 0.0598, 0.1793, 0.0716, 0.0889, 0.0487, 0.1083, 0.2062], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0079, 0.0064, 0.0067, 0.0093, 0.0079, 0.0095, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 02:43:07,306 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:09,688 INFO [train.py:873] (0/4) Epoch 12, batch 6500, loss[loss=0.1268, simple_loss=0.1576, pruned_loss=0.04806, over 14190.00 frames. ], tot_loss[loss=0.1248, simple_loss=0.1554, pruned_loss=0.04706, over 1950471.21 frames. ], batch size: 37, lr: 6.50e-03, grad_scale: 8.0 2022-12-08 02:43:09,809 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89683.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:21,022 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.60 vs. limit=5.0 2022-12-08 02:43:32,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.08 vs. limit=5.0 2022-12-08 02:43:50,391 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89730.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:43:51,206 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:00,240 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89741.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:23,989 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.343e+02 2.899e+02 3.659e+02 1.529e+03, threshold=5.799e+02, percent-clipped=6.0 2022-12-08 02:44:32,235 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 02:44:35,224 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:44:36,900 INFO [train.py:873] (0/4) Epoch 12, batch 6600, loss[loss=0.1463, simple_loss=0.1658, pruned_loss=0.06339, over 4951.00 frames. ], tot_loss[loss=0.1253, simple_loss=0.1555, pruned_loss=0.04752, over 1912361.35 frames. ], batch size: 100, lr: 6.49e-03, grad_scale: 8.0 2022-12-08 02:44:47,412 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89795.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:45:13,782 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89825.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:48,169 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:48,249 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:50,625 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.462e+01 2.267e+02 2.755e+02 3.423e+02 7.380e+02, threshold=5.509e+02, percent-clipped=4.0 2022-12-08 02:45:55,153 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89873.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:45:55,302 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9996, 1.9721, 1.9046, 1.9580, 1.9781, 1.1225, 1.8859, 1.9397], device='cuda:0'), covar=tensor([0.0802, 0.0781, 0.0562, 0.1470, 0.0934, 0.0968, 0.0720, 0.1020], device='cuda:0'), in_proj_covar=tensor([0.0028, 0.0028, 0.0030, 0.0027, 0.0028, 0.0040, 0.0028, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 02:46:04,380 INFO [train.py:873] (0/4) Epoch 12, batch 6700, loss[loss=0.127, simple_loss=0.1579, pruned_loss=0.04809, over 14320.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1555, pruned_loss=0.04739, over 1904043.24 frames. ], batch size: 25, lr: 6.49e-03, grad_scale: 8.0 2022-12-08 02:46:07,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 02:46:08,954 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-08 02:46:41,490 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89926.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:47:17,826 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 2.198e+02 2.718e+02 3.192e+02 1.007e+03, threshold=5.437e+02, percent-clipped=1.0 2022-12-08 02:47:30,899 INFO [train.py:873] (0/4) Epoch 12, batch 6800, loss[loss=0.1257, simple_loss=0.1516, pruned_loss=0.04992, over 4942.00 frames. ], tot_loss[loss=0.1261, simple_loss=0.1558, pruned_loss=0.04825, over 1903317.36 frames. ], batch size: 100, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:47:43,377 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1190, 3.2049, 3.3915, 3.1678, 3.2952, 2.8089, 1.4386, 3.0915], device='cuda:0'), covar=tensor([0.0426, 0.0432, 0.0432, 0.0470, 0.0366, 0.0923, 0.3242, 0.0341], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0170, 0.0141, 0.0140, 0.0202, 0.0135, 0.0157, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 02:47:46,392 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-90000.pt 2022-12-08 02:47:59,942 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5822, 1.2869, 2.0488, 1.8739, 1.8374, 2.0199, 1.3186, 2.0317], device='cuda:0'), covar=tensor([0.0804, 0.1096, 0.0237, 0.0414, 0.0547, 0.0257, 0.0613, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0127, 0.0164, 0.0144, 0.0139, 0.0120, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:48:01,957 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.78 vs. limit=5.0 2022-12-08 02:48:07,798 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7289, 2.4264, 3.5052, 2.5843, 3.5780, 3.4674, 3.3807, 2.8171], device='cuda:0'), covar=tensor([0.0799, 0.2941, 0.1056, 0.2215, 0.0885, 0.0978, 0.1338, 0.2227], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0313, 0.0396, 0.0304, 0.0373, 0.0320, 0.0363, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:48:15,338 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90030.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:20,544 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:21,730 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9349, 3.7147, 3.4103, 3.5566, 3.8219, 3.8537, 3.8883, 3.8922], device='cuda:0'), covar=tensor([0.0905, 0.0652, 0.2290, 0.3017, 0.0850, 0.0930, 0.1248, 0.0888], device='cuda:0'), in_proj_covar=tensor([0.0370, 0.0258, 0.0429, 0.0544, 0.0321, 0.0417, 0.0381, 0.0358], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:48:47,746 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.301e+02 2.958e+02 4.111e+02 8.301e+02, threshold=5.916e+02, percent-clipped=6.0 2022-12-08 02:48:56,870 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:48:59,856 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90081.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:01,447 INFO [train.py:873] (0/4) Epoch 12, batch 6900, loss[loss=0.1209, simple_loss=0.1575, pruned_loss=0.04212, over 14003.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1551, pruned_loss=0.04757, over 1919201.11 frames. ], batch size: 22, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:49:11,852 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90095.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 02:49:18,252 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5575, 1.6380, 4.3624, 2.1072, 4.3281, 4.5696, 4.0680, 4.9601], device='cuda:0'), covar=tensor([0.0193, 0.3101, 0.0379, 0.2096, 0.0350, 0.0379, 0.0363, 0.0136], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0158, 0.0161, 0.0169, 0.0169, 0.0178, 0.0132, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:49:41,776 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90129.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:42,818 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:49:53,771 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90143.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:50:13,156 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:50:15,735 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.354e+02 2.821e+02 3.688e+02 2.263e+03, threshold=5.641e+02, percent-clipped=6.0 2022-12-08 02:50:28,847 INFO [train.py:873] (0/4) Epoch 12, batch 7000, loss[loss=0.102, simple_loss=0.1432, pruned_loss=0.03041, over 14045.00 frames. ], tot_loss[loss=0.1237, simple_loss=0.1543, pruned_loss=0.04658, over 1909747.54 frames. ], batch size: 19, lr: 6.48e-03, grad_scale: 8.0 2022-12-08 02:50:35,989 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90191.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:50:49,150 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1334, 2.0121, 2.1842, 2.3621, 1.9829, 1.9003, 2.0943, 2.2112], device='cuda:0'), covar=tensor([0.0256, 0.0418, 0.0222, 0.0262, 0.0341, 0.0557, 0.0316, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0287, 0.0253, 0.0369, 0.0321, 0.0262, 0.0300, 0.0300, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 02:50:55,447 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:51:02,565 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90221.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 02:51:12,011 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90232.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:51:24,339 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2065, 2.1477, 4.1740, 2.8008, 3.9993, 2.0427, 3.1300, 4.0306], device='cuda:0'), covar=tensor([0.0596, 0.4447, 0.0467, 0.7015, 0.0675, 0.3470, 0.1321, 0.0448], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0215, 0.0208, 0.0291, 0.0228, 0.0215, 0.0213, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 02:51:44,201 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 1.997e+02 2.595e+02 3.353e+02 6.185e+02, threshold=5.189e+02, percent-clipped=2.0 2022-12-08 02:51:51,056 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 02:51:53,736 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8589, 1.4094, 2.5510, 2.2611, 2.4026, 2.5689, 1.6667, 2.5224], device='cuda:0'), covar=tensor([0.0768, 0.1136, 0.0165, 0.0383, 0.0379, 0.0177, 0.0639, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0128, 0.0166, 0.0145, 0.0140, 0.0121, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 02:51:57,578 INFO [train.py:873] (0/4) Epoch 12, batch 7100, loss[loss=0.1029, simple_loss=0.1429, pruned_loss=0.03148, over 14508.00 frames. ], tot_loss[loss=0.1229, simple_loss=0.1539, pruned_loss=0.04595, over 1878531.61 frames. ], batch size: 34, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:52:06,214 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:52:32,977 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7783, 1.3681, 1.7405, 1.1583, 1.3742, 1.8307, 1.5080, 1.4710], device='cuda:0'), covar=tensor([0.0747, 0.0821, 0.0651, 0.0956, 0.1613, 0.0778, 0.0780, 0.1691], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0178, 0.0136, 0.0124, 0.0134, 0.0146, 0.0123, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 02:52:39,783 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.8870, 5.2625, 5.2569, 5.8116, 5.2615, 4.7456, 5.7944, 4.6059], device='cuda:0'), covar=tensor([0.0307, 0.0999, 0.0311, 0.0444, 0.0908, 0.0399, 0.0472, 0.0581], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0260, 0.0186, 0.0181, 0.0176, 0.0147, 0.0268, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 02:52:44,240 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:52:57,298 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90351.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:53:06,052 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6133, 1.6132, 1.8737, 1.6261, 1.7289, 1.5183, 1.2122, 1.1435], device='cuda:0'), covar=tensor([0.0238, 0.0422, 0.0260, 0.0300, 0.0268, 0.0296, 0.0325, 0.0570], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0015, 0.0016, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 02:53:12,653 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 2.346e+02 2.734e+02 3.324e+02 7.118e+02, threshold=5.469e+02, percent-clipped=5.0 2022-12-08 02:53:26,144 INFO [train.py:873] (0/4) Epoch 12, batch 7200, loss[loss=0.135, simple_loss=0.1626, pruned_loss=0.0537, over 12745.00 frames. ], tot_loss[loss=0.1238, simple_loss=0.1544, pruned_loss=0.04657, over 1945827.86 frames. ], batch size: 100, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:53:26,994 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:53:51,997 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90412.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:54:25,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2022-12-08 02:54:33,027 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90459.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:54:40,950 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.297e+02 2.801e+02 3.599e+02 6.534e+02, threshold=5.602e+02, percent-clipped=3.0 2022-12-08 02:54:54,141 INFO [train.py:873] (0/4) Epoch 12, batch 7300, loss[loss=0.1386, simple_loss=0.1689, pruned_loss=0.05413, over 11155.00 frames. ], tot_loss[loss=0.1231, simple_loss=0.1541, pruned_loss=0.0461, over 1973365.86 frames. ], batch size: 100, lr: 6.47e-03, grad_scale: 8.0 2022-12-08 02:54:56,704 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:05,326 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90496.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:26,482 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:55:27,613 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90521.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 02:55:58,934 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90557.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:56:08,321 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.439e+02 2.820e+02 3.761e+02 1.640e+03, threshold=5.641e+02, percent-clipped=6.0 2022-12-08 02:56:09,293 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90569.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:56:21,034 INFO [train.py:873] (0/4) Epoch 12, batch 7400, loss[loss=0.1212, simple_loss=0.1315, pruned_loss=0.05546, over 2639.00 frames. ], tot_loss[loss=0.1234, simple_loss=0.1543, pruned_loss=0.04631, over 1942306.50 frames. ], batch size: 100, lr: 6.46e-03, grad_scale: 8.0 2022-12-08 02:56:25,651 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90588.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:57:36,484 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.056e+02 2.705e+02 3.388e+02 5.902e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-08 02:57:50,343 INFO [train.py:873] (0/4) Epoch 12, batch 7500, loss[loss=0.1043, simple_loss=0.1416, pruned_loss=0.03349, over 14286.00 frames. ], tot_loss[loss=0.1223, simple_loss=0.1538, pruned_loss=0.04543, over 1988979.70 frames. ], batch size: 31, lr: 6.46e-03, grad_scale: 8.0 2022-12-08 02:58:06,065 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 02:58:10,889 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 02:58:28,869 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=10.05 vs. limit=5.0 2022-12-08 02:58:29,717 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8668, 0.7929, 0.7829, 0.6679, 0.7270, 0.5483, 0.6089, 0.7852], device='cuda:0'), covar=tensor([0.0160, 0.0144, 0.0136, 0.0144, 0.0154, 0.0306, 0.0227, 0.0247], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0016, 0.0017, 0.0028, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 02:58:36,968 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-12.pt 2022-12-08 02:59:13,080 INFO [train.py:873] (0/4) Epoch 13, batch 0, loss[loss=0.163, simple_loss=0.1849, pruned_loss=0.07054, over 14253.00 frames. ], tot_loss[loss=0.163, simple_loss=0.1849, pruned_loss=0.07054, over 14253.00 frames. ], batch size: 46, lr: 6.21e-03, grad_scale: 8.0 2022-12-08 02:59:13,081 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 02:59:18,103 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1589, 2.0251, 1.9464, 2.3974, 1.9783, 2.2310, 1.9245, 1.7044], device='cuda:0'), covar=tensor([0.0421, 0.0713, 0.0279, 0.0231, 0.0496, 0.0429, 0.0671, 0.0778], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0080, 0.0064, 0.0067, 0.0093, 0.0078, 0.0094, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 02:59:20,362 INFO [train.py:905] (0/4) Epoch 13, validation: loss=0.1364, simple_loss=0.1777, pruned_loss=0.04756, over 857387.00 frames. 2022-12-08 02:59:20,363 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 02:59:41,454 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 6.584e+01 1.659e+02 2.772e+02 3.641e+02 1.065e+03, threshold=5.544e+02, percent-clipped=8.0 2022-12-08 02:59:48,113 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3529, 1.4740, 4.1244, 1.8607, 4.1551, 4.3006, 3.6456, 4.7659], device='cuda:0'), covar=tensor([0.0187, 0.3091, 0.0420, 0.2224, 0.0354, 0.0300, 0.0453, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0156, 0.0160, 0.0168, 0.0169, 0.0177, 0.0133, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:0') 2022-12-08 02:59:57,584 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90786.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:24,291 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90815.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:38,292 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5120, 1.4584, 1.5191, 1.3048, 1.3209, 1.2171, 1.2288, 1.0517], device='cuda:0'), covar=tensor([0.0216, 0.0203, 0.0229, 0.0236, 0.0250, 0.0343, 0.0270, 0.0467], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0015, 0.0016, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:00:40,590 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90834.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:50,553 INFO [train.py:873] (0/4) Epoch 13, batch 100, loss[loss=0.1154, simple_loss=0.1513, pruned_loss=0.03974, over 14257.00 frames. ], tot_loss[loss=0.1261, simple_loss=0.1562, pruned_loss=0.04803, over 845388.28 frames. ], batch size: 80, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:00:54,341 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90849.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:54,638 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 03:00:56,671 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:00:58,794 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-08 03:01:09,979 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.335e+02 2.702e+02 3.282e+02 9.825e+02, threshold=5.404e+02, percent-clipped=3.0 2022-12-08 03:01:27,468 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:01:46,436 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90910.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:02:09,087 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90936.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:02:16,740 INFO [train.py:873] (0/4) Epoch 13, batch 200, loss[loss=0.1139, simple_loss=0.1536, pruned_loss=0.03711, over 14138.00 frames. ], tot_loss[loss=0.1251, simple_loss=0.1557, pruned_loss=0.04727, over 1321101.23 frames. ], batch size: 84, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:02:36,077 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.982e+01 2.220e+02 2.675e+02 3.390e+02 6.533e+02, threshold=5.349e+02, percent-clipped=6.0 2022-12-08 03:02:59,234 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0203, 1.1203, 0.9286, 1.1333, 1.1229, 0.6560, 0.9802, 1.0651], device='cuda:0'), covar=tensor([0.0445, 0.0654, 0.0440, 0.0364, 0.0324, 0.0354, 0.0948, 0.0734], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0030, 0.0027, 0.0029, 0.0041, 0.0028, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:03:11,346 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:03:18,823 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5067, 1.3322, 1.4389, 1.4553, 1.5519, 0.8967, 1.3383, 1.3775], device='cuda:0'), covar=tensor([0.0617, 0.0912, 0.0616, 0.0619, 0.0606, 0.0868, 0.0857, 0.0657], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0030, 0.0028, 0.0029, 0.0041, 0.0029, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:03:41,571 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4782, 1.8322, 2.3329, 2.0785, 2.4964, 2.2708, 2.2023, 2.2154], device='cuda:0'), covar=tensor([0.0434, 0.1950, 0.0446, 0.1072, 0.0427, 0.0837, 0.0553, 0.0968], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0313, 0.0396, 0.0302, 0.0373, 0.0318, 0.0360, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:03:44,490 INFO [train.py:873] (0/4) Epoch 13, batch 300, loss[loss=0.131, simple_loss=0.1329, pruned_loss=0.06452, over 1251.00 frames. ], tot_loss[loss=0.1241, simple_loss=0.1543, pruned_loss=0.04698, over 1509881.95 frames. ], batch size: 100, lr: 6.20e-03, grad_scale: 8.0 2022-12-08 03:03:53,659 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:04:04,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.005e+02 2.425e+02 3.069e+02 6.147e+02, threshold=4.849e+02, percent-clipped=1.0 2022-12-08 03:04:39,518 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.40 vs. limit=5.0 2022-12-08 03:04:46,078 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:13,072 INFO [train.py:873] (0/4) Epoch 13, batch 400, loss[loss=0.1305, simple_loss=0.1638, pruned_loss=0.04863, over 12746.00 frames. ], tot_loss[loss=0.1224, simple_loss=0.1533, pruned_loss=0.04576, over 1681522.09 frames. ], batch size: 100, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:05:19,353 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91152.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:28,724 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91163.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:05:33,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.284e+02 2.800e+02 3.397e+02 6.782e+02, threshold=5.601e+02, percent-clipped=4.0 2022-12-08 03:06:00,539 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3725, 1.9044, 3.4775, 2.5836, 3.3207, 1.9354, 2.7397, 3.3132], device='cuda:0'), covar=tensor([0.0787, 0.4254, 0.0499, 0.5201, 0.0818, 0.3235, 0.1256, 0.0591], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0212, 0.0206, 0.0290, 0.0229, 0.0213, 0.0211, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:06:01,543 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91200.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:06:06,394 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91205.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:06:26,870 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2022-12-08 03:06:41,227 INFO [train.py:873] (0/4) Epoch 13, batch 500, loss[loss=0.1404, simple_loss=0.1708, pruned_loss=0.05503, over 14295.00 frames. ], tot_loss[loss=0.1229, simple_loss=0.1538, pruned_loss=0.04601, over 1733572.22 frames. ], batch size: 80, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:07:01,832 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.197e+02 2.790e+02 3.555e+02 6.703e+02, threshold=5.580e+02, percent-clipped=3.0 2022-12-08 03:07:16,631 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91285.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:07:40,385 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1858, 2.0327, 2.0992, 2.1836, 2.0959, 2.0855, 2.2561, 1.8984], device='cuda:0'), covar=tensor([0.0783, 0.1182, 0.0670, 0.0738, 0.0962, 0.0670, 0.0800, 0.0733], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0264, 0.0188, 0.0184, 0.0179, 0.0149, 0.0270, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 03:08:03,441 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5393, 1.4084, 1.5022, 1.3304, 1.3013, 1.1750, 1.1460, 1.0750], device='cuda:0'), covar=tensor([0.0161, 0.0224, 0.0217, 0.0191, 0.0207, 0.0338, 0.0236, 0.0400], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0015, 0.0016, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:08:09,276 INFO [train.py:873] (0/4) Epoch 13, batch 600, loss[loss=0.1103, simple_loss=0.1512, pruned_loss=0.03469, over 14220.00 frames. ], tot_loss[loss=0.1224, simple_loss=0.1537, pruned_loss=0.04556, over 1888648.16 frames. ], batch size: 80, lr: 6.19e-03, grad_scale: 8.0 2022-12-08 03:08:09,362 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9157, 2.7803, 2.5053, 2.6114, 2.8380, 2.8556, 2.8328, 2.8228], device='cuda:0'), covar=tensor([0.0888, 0.0653, 0.2099, 0.2504, 0.0929, 0.0986, 0.1201, 0.0979], device='cuda:0'), in_proj_covar=tensor([0.0376, 0.0261, 0.0440, 0.0552, 0.0327, 0.0427, 0.0386, 0.0365], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:08:10,259 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:08:29,164 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 2.257e+02 2.792e+02 3.414e+02 7.089e+02, threshold=5.584e+02, percent-clipped=8.0 2022-12-08 03:09:36,011 INFO [train.py:873] (0/4) Epoch 13, batch 700, loss[loss=0.1117, simple_loss=0.1461, pruned_loss=0.0386, over 14233.00 frames. ], tot_loss[loss=0.1225, simple_loss=0.1539, pruned_loss=0.04555, over 1968241.61 frames. ], batch size: 69, lr: 6.18e-03, grad_scale: 8.0 2022-12-08 03:09:55,978 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.741e+01 2.087e+02 2.616e+02 3.337e+02 7.440e+02, threshold=5.233e+02, percent-clipped=3.0 2022-12-08 03:10:16,776 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-08 03:10:28,657 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:11:01,236 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91542.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:11:03,620 INFO [train.py:873] (0/4) Epoch 13, batch 800, loss[loss=0.11, simple_loss=0.1495, pruned_loss=0.03521, over 14281.00 frames. ], tot_loss[loss=0.1226, simple_loss=0.1538, pruned_loss=0.04568, over 1961079.70 frames. ], batch size: 31, lr: 6.18e-03, grad_scale: 8.0 2022-12-08 03:11:03,769 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4978, 4.6798, 4.9249, 3.9810, 4.7387, 4.9998, 1.9459, 4.3872], device='cuda:0'), covar=tensor([0.0268, 0.0270, 0.0341, 0.0483, 0.0277, 0.0141, 0.3049, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0167, 0.0140, 0.0139, 0.0198, 0.0132, 0.0156, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:11:10,454 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91553.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:11:23,656 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.089e+02 2.194e+02 2.551e+02 3.061e+02 6.051e+02, threshold=5.101e+02, percent-clipped=1.0 2022-12-08 03:11:54,251 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91603.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:12:27,334 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91641.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:12:30,940 INFO [train.py:873] (0/4) Epoch 13, batch 900, loss[loss=0.09974, simple_loss=0.1394, pruned_loss=0.03002, over 13975.00 frames. ], tot_loss[loss=0.1234, simple_loss=0.1544, pruned_loss=0.04619, over 1971934.52 frames. ], batch size: 26, lr: 6.18e-03, grad_scale: 16.0 2022-12-08 03:12:52,063 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.371e+02 2.875e+02 3.801e+02 8.432e+02, threshold=5.750e+02, percent-clipped=7.0 2022-12-08 03:13:43,446 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4666, 5.3142, 4.9837, 5.5641, 5.0610, 4.9105, 5.5441, 5.4052], device='cuda:0'), covar=tensor([0.0624, 0.0735, 0.0694, 0.0552, 0.0799, 0.0437, 0.0520, 0.0631], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0133, 0.0142, 0.0152, 0.0142, 0.0119, 0.0161, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:13:56,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9915, 1.7890, 1.9747, 2.0237, 1.9562, 1.0868, 1.6505, 1.9570], device='cuda:0'), covar=tensor([0.0709, 0.1220, 0.0643, 0.0857, 0.1249, 0.0938, 0.1306, 0.0981], device='cuda:0'), in_proj_covar=tensor([0.0028, 0.0029, 0.0030, 0.0027, 0.0029, 0.0041, 0.0029, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:13:58,929 INFO [train.py:873] (0/4) Epoch 13, batch 1000, loss[loss=0.1135, simple_loss=0.1399, pruned_loss=0.04349, over 5958.00 frames. ], tot_loss[loss=0.1222, simple_loss=0.1536, pruned_loss=0.0454, over 1975619.90 frames. ], batch size: 100, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:14:19,310 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.021e+02 2.131e+02 2.597e+02 3.397e+02 6.255e+02, threshold=5.195e+02, percent-clipped=2.0 2022-12-08 03:14:22,944 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4269, 2.4179, 1.9502, 2.0019, 2.3535, 2.4533, 2.4769, 2.4285], device='cuda:0'), covar=tensor([0.1603, 0.1005, 0.3629, 0.4151, 0.1793, 0.1409, 0.2085, 0.1482], device='cuda:0'), in_proj_covar=tensor([0.0376, 0.0263, 0.0445, 0.0556, 0.0331, 0.0430, 0.0391, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:14:48,103 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0578, 2.1026, 2.3489, 1.4980, 1.6310, 2.2042, 1.2900, 2.0951], device='cuda:0'), covar=tensor([0.1430, 0.1588, 0.0913, 0.3082, 0.2712, 0.0884, 0.3570, 0.0989], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0097, 0.0091, 0.0098, 0.0115, 0.0086, 0.0122, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 03:15:26,329 INFO [train.py:873] (0/4) Epoch 13, batch 1100, loss[loss=0.1199, simple_loss=0.1519, pruned_loss=0.04399, over 14207.00 frames. ], tot_loss[loss=0.1221, simple_loss=0.1533, pruned_loss=0.04546, over 1990771.24 frames. ], batch size: 25, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:15:39,563 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2022-12-08 03:15:47,467 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.763e+01 2.120e+02 2.631e+02 3.193e+02 5.598e+02, threshold=5.262e+02, percent-clipped=1.0 2022-12-08 03:16:07,026 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91891.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:16:13,623 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91898.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:16:33,954 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8821, 1.4327, 3.9057, 3.6772, 3.6573, 3.9513, 3.2719, 3.9383], device='cuda:0'), covar=tensor([0.1536, 0.1694, 0.0119, 0.0233, 0.0231, 0.0134, 0.0257, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0129, 0.0167, 0.0146, 0.0141, 0.0122, 0.0121], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:16:51,294 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91941.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:16:54,657 INFO [train.py:873] (0/4) Epoch 13, batch 1200, loss[loss=0.1192, simple_loss=0.1357, pruned_loss=0.05141, over 5939.00 frames. ], tot_loss[loss=0.1233, simple_loss=0.154, pruned_loss=0.04629, over 2000785.60 frames. ], batch size: 100, lr: 6.17e-03, grad_scale: 8.0 2022-12-08 03:17:00,844 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91952.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:03,188 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91955.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:15,486 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.334e+02 2.878e+02 3.521e+02 1.024e+03, threshold=5.756e+02, percent-clipped=7.0 2022-12-08 03:17:26,263 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91981.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:17:33,103 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:17:43,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9092, 2.5109, 4.8356, 3.3309, 4.6136, 1.9927, 3.6455, 4.5458], device='cuda:0'), covar=tensor([0.0446, 0.3927, 0.0484, 0.6354, 0.0434, 0.3825, 0.1244, 0.0315], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0213, 0.0210, 0.0291, 0.0231, 0.0215, 0.0215, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0006, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:17:56,987 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92016.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:18:06,895 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9202, 1.4588, 2.9830, 1.5219, 3.1323, 2.9633, 2.0703, 3.2460], device='cuda:0'), covar=tensor([0.0292, 0.2693, 0.0485, 0.2082, 0.0372, 0.0479, 0.1101, 0.0235], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0159, 0.0161, 0.0169, 0.0170, 0.0179, 0.0135, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:18:11,870 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2022-12-08 03:18:13,550 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.19 vs. limit=5.0 2022-12-08 03:18:19,773 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92042.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:18:22,473 INFO [train.py:873] (0/4) Epoch 13, batch 1300, loss[loss=0.1182, simple_loss=0.1542, pruned_loss=0.04111, over 14336.00 frames. ], tot_loss[loss=0.1231, simple_loss=0.1534, pruned_loss=0.04643, over 1886560.76 frames. ], batch size: 55, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:18:44,122 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.023e+02 2.102e+02 2.497e+02 3.213e+02 6.107e+02, threshold=4.993e+02, percent-clipped=1.0 2022-12-08 03:19:52,207 INFO [train.py:873] (0/4) Epoch 13, batch 1400, loss[loss=0.1052, simple_loss=0.1468, pruned_loss=0.03177, over 14260.00 frames. ], tot_loss[loss=0.1218, simple_loss=0.1526, pruned_loss=0.04549, over 1877847.13 frames. ], batch size: 28, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:19:52,337 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0655, 2.1735, 2.4875, 1.5490, 1.6920, 2.3009, 1.3537, 2.2042], device='cuda:0'), covar=tensor([0.1318, 0.1528, 0.0840, 0.2386, 0.2658, 0.0927, 0.3779, 0.1104], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0098, 0.0090, 0.0098, 0.0115, 0.0085, 0.0122, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 03:20:13,182 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.362e+02 2.970e+02 3.546e+02 6.469e+02, threshold=5.940e+02, percent-clipped=3.0 2022-12-08 03:20:39,438 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92198.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:21:08,365 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5570, 2.6221, 2.7591, 2.6934, 2.6775, 2.3990, 1.5085, 2.4338], device='cuda:0'), covar=tensor([0.0479, 0.0467, 0.0401, 0.0397, 0.0415, 0.1113, 0.2515, 0.0362], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0167, 0.0141, 0.0139, 0.0197, 0.0133, 0.0156, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:21:21,063 INFO [train.py:873] (0/4) Epoch 13, batch 1500, loss[loss=0.1801, simple_loss=0.18, pruned_loss=0.09013, over 8602.00 frames. ], tot_loss[loss=0.1228, simple_loss=0.1532, pruned_loss=0.0462, over 1888655.87 frames. ], batch size: 100, lr: 6.16e-03, grad_scale: 8.0 2022-12-08 03:21:22,058 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92246.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:21:22,811 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:21:41,369 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.124e+02 2.104e+02 2.663e+02 3.435e+02 6.153e+02, threshold=5.325e+02, percent-clipped=2.0 2022-12-08 03:22:11,647 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6675, 1.6862, 1.8737, 1.5960, 1.5793, 1.2570, 0.6977, 1.0880], device='cuda:0'), covar=tensor([0.0232, 0.0382, 0.0216, 0.0315, 0.0298, 0.0381, 0.0362, 0.0543], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0016, 0.0016, 0.0027, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:22:18,314 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92311.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:22:40,836 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92337.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:22:48,256 INFO [train.py:873] (0/4) Epoch 13, batch 1600, loss[loss=0.1331, simple_loss=0.1626, pruned_loss=0.05181, over 14276.00 frames. ], tot_loss[loss=0.1227, simple_loss=0.1533, pruned_loss=0.04604, over 1885373.99 frames. ], batch size: 76, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:23:08,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.214e+02 2.735e+02 3.327e+02 6.728e+02, threshold=5.470e+02, percent-clipped=2.0 2022-12-08 03:23:57,879 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7801, 1.7349, 3.0093, 2.2127, 2.8499, 1.7496, 2.3071, 2.7852], device='cuda:0'), covar=tensor([0.1163, 0.4074, 0.0650, 0.3666, 0.1133, 0.3328, 0.1392, 0.0878], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0214, 0.0210, 0.0289, 0.0231, 0.0214, 0.0215, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:24:14,890 INFO [train.py:873] (0/4) Epoch 13, batch 1700, loss[loss=0.134, simple_loss=0.159, pruned_loss=0.0545, over 8648.00 frames. ], tot_loss[loss=0.1211, simple_loss=0.1528, pruned_loss=0.04465, over 2008020.81 frames. ], batch size: 100, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:24:36,007 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.177e+02 2.526e+02 3.126e+02 5.467e+02, threshold=5.052e+02, percent-clipped=0.0 2022-12-08 03:24:47,845 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5659, 1.5358, 1.4639, 1.5705, 1.7350, 0.8957, 1.3596, 1.4298], device='cuda:0'), covar=tensor([0.0591, 0.0690, 0.0615, 0.0633, 0.0608, 0.0895, 0.0783, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0030, 0.0027, 0.0029, 0.0041, 0.0029, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:25:22,694 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8549, 2.9328, 3.0962, 2.9929, 3.0501, 2.8838, 1.5279, 2.7870], device='cuda:0'), covar=tensor([0.0406, 0.0411, 0.0402, 0.0448, 0.0365, 0.0734, 0.2607, 0.0320], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0168, 0.0141, 0.0139, 0.0199, 0.0133, 0.0155, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:25:42,443 INFO [train.py:873] (0/4) Epoch 13, batch 1800, loss[loss=0.1196, simple_loss=0.1544, pruned_loss=0.0424, over 14380.00 frames. ], tot_loss[loss=0.1207, simple_loss=0.1526, pruned_loss=0.04438, over 1983554.18 frames. ], batch size: 73, lr: 6.15e-03, grad_scale: 8.0 2022-12-08 03:25:44,271 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92547.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:02,905 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.175e+02 2.080e+02 2.746e+02 3.507e+02 5.512e+02, threshold=5.493e+02, percent-clipped=3.0 2022-12-08 03:26:25,745 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92595.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:39,587 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92611.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:26:52,297 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:27:02,559 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92637.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:27:09,284 INFO [train.py:873] (0/4) Epoch 13, batch 1900, loss[loss=0.1035, simple_loss=0.14, pruned_loss=0.03355, over 14257.00 frames. ], tot_loss[loss=0.1226, simple_loss=0.1534, pruned_loss=0.04585, over 1924450.57 frames. ], batch size: 57, lr: 6.14e-03, grad_scale: 4.0 2022-12-08 03:27:20,542 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 03:27:21,758 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:27:24,768 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9137, 1.8549, 1.8253, 1.8614, 1.7577, 1.7234, 1.3817, 1.6317], device='cuda:0'), covar=tensor([0.0395, 0.0771, 0.0754, 0.0498, 0.0471, 0.0477, 0.0613, 0.0622], device='cuda:0'), in_proj_covar=tensor([0.0017, 0.0018, 0.0016, 0.0017, 0.0017, 0.0027, 0.0022, 0.0027], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:27:25,545 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3080, 4.0823, 3.7490, 3.9242, 4.1378, 4.2038, 4.2915, 4.2372], device='cuda:0'), covar=tensor([0.0848, 0.0552, 0.2271, 0.2654, 0.0723, 0.0893, 0.0965, 0.0827], device='cuda:0'), in_proj_covar=tensor([0.0384, 0.0266, 0.0453, 0.0565, 0.0334, 0.0438, 0.0393, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:27:31,531 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.202e+02 2.634e+02 3.218e+02 6.502e+02, threshold=5.268e+02, percent-clipped=3.0 2022-12-08 03:27:44,635 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92685.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:27:45,461 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:28:12,328 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7245, 3.1538, 2.8264, 3.1298, 2.4001, 3.0749, 2.9008, 1.6628], device='cuda:0'), covar=tensor([0.1937, 0.0620, 0.1283, 0.0539, 0.1117, 0.0717, 0.1099, 0.2597], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0081, 0.0065, 0.0067, 0.0096, 0.0081, 0.0096, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 03:28:15,776 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:28:37,930 INFO [train.py:873] (0/4) Epoch 13, batch 2000, loss[loss=0.1269, simple_loss=0.1538, pruned_loss=0.05001, over 12760.00 frames. ], tot_loss[loss=0.1232, simple_loss=0.154, pruned_loss=0.04619, over 1949676.95 frames. ], batch size: 100, lr: 6.14e-03, grad_scale: 8.0 2022-12-08 03:28:38,861 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5132, 5.0256, 5.0734, 5.4601, 5.1642, 4.6091, 5.5464, 4.5928], device='cuda:0'), covar=tensor([0.0347, 0.1052, 0.0338, 0.0475, 0.0665, 0.0422, 0.0420, 0.0493], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0269, 0.0191, 0.0189, 0.0182, 0.0151, 0.0276, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 03:28:59,687 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.366e+02 2.802e+02 3.411e+02 7.982e+02, threshold=5.603e+02, percent-clipped=8.0 2022-12-08 03:29:10,473 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:29:51,978 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2276, 2.3776, 4.0977, 4.2542, 4.0887, 2.5030, 4.1814, 3.2164], device='cuda:0'), covar=tensor([0.0391, 0.1084, 0.0904, 0.0421, 0.0403, 0.1558, 0.0458, 0.0952], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0258, 0.0377, 0.0329, 0.0268, 0.0305, 0.0307, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:29:54,484 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:30:06,190 INFO [train.py:873] (0/4) Epoch 13, batch 2100, loss[loss=0.1182, simple_loss=0.1544, pruned_loss=0.041, over 14476.00 frames. ], tot_loss[loss=0.1217, simple_loss=0.1533, pruned_loss=0.04507, over 2016364.00 frames. ], batch size: 51, lr: 6.14e-03, grad_scale: 8.0 2022-12-08 03:30:28,667 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.062e+02 2.147e+02 2.777e+02 3.486e+02 6.981e+02, threshold=5.554e+02, percent-clipped=1.0 2022-12-08 03:30:34,061 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2604, 1.2661, 1.2338, 1.3087, 1.4530, 0.8662, 1.1015, 1.1715], device='cuda:0'), covar=tensor([0.0989, 0.1205, 0.0665, 0.0666, 0.0583, 0.1103, 0.1191, 0.0756], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0030, 0.0031, 0.0028, 0.0029, 0.0041, 0.0029, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:30:37,404 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9162, 1.7763, 1.8644, 1.9862, 2.0657, 1.0783, 1.6563, 1.8714], device='cuda:0'), covar=tensor([0.0855, 0.1564, 0.0731, 0.0956, 0.0930, 0.0886, 0.1235, 0.0712], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0030, 0.0031, 0.0028, 0.0029, 0.0041, 0.0029, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:30:48,224 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:30:54,301 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:11,310 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9207, 3.6832, 3.6448, 3.9198, 3.5880, 3.3134, 3.9728, 3.8223], device='cuda:0'), covar=tensor([0.0599, 0.0909, 0.0772, 0.0657, 0.0879, 0.0851, 0.0630, 0.0716], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0139, 0.0150, 0.0139, 0.0117, 0.0158, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:31:19,211 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7624, 1.1630, 1.2958, 1.1902, 1.0785, 1.3002, 1.0663, 0.8611], device='cuda:0'), covar=tensor([0.1963, 0.0893, 0.0306, 0.0306, 0.1627, 0.0700, 0.1450, 0.1465], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0082, 0.0065, 0.0068, 0.0096, 0.0081, 0.0096, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 03:31:21,974 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92931.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:24,636 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3397, 5.1688, 4.9125, 5.4378, 4.9427, 4.7317, 5.4474, 5.2342], device='cuda:0'), covar=tensor([0.0538, 0.0608, 0.0648, 0.0399, 0.0697, 0.0418, 0.0481, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0130, 0.0138, 0.0149, 0.0138, 0.0117, 0.0158, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:31:34,898 INFO [train.py:873] (0/4) Epoch 13, batch 2200, loss[loss=0.1411, simple_loss=0.1424, pruned_loss=0.06992, over 1245.00 frames. ], tot_loss[loss=0.1233, simple_loss=0.154, pruned_loss=0.04632, over 1969571.67 frames. ], batch size: 100, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:31:37,317 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8391, 4.2123, 4.5763, 3.3850, 3.0906, 3.5400, 2.5333, 3.6436], device='cuda:0'), covar=tensor([0.0887, 0.0555, 0.0449, 0.1526, 0.1760, 0.1025, 0.2903, 0.1759], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0098, 0.0091, 0.0099, 0.0115, 0.0087, 0.0123, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 03:31:47,206 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 03:31:47,577 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92960.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:31:50,368 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 03:31:56,269 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.331e+02 2.968e+02 3.670e+02 7.983e+02, threshold=5.936e+02, percent-clipped=6.0 2022-12-08 03:32:01,005 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92975.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:06,288 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:12,647 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8838, 1.6380, 2.0503, 1.7044, 1.9276, 1.4944, 1.6421, 1.9661], device='cuda:0'), covar=tensor([0.2238, 0.2770, 0.0540, 0.1403, 0.1729, 0.1209, 0.1195, 0.0898], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0211, 0.0209, 0.0288, 0.0230, 0.0212, 0.0213, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:32:16,154 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:31,685 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93010.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:55,219 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:32:55,240 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:01,194 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8103, 3.6375, 3.5895, 3.8892, 3.7196, 3.5521, 3.9733, 3.3034], device='cuda:0'), covar=tensor([0.0743, 0.1001, 0.0471, 0.0506, 0.0703, 0.1167, 0.0571, 0.0578], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0268, 0.0191, 0.0188, 0.0182, 0.0151, 0.0274, 0.0167], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 03:33:02,671 INFO [train.py:873] (0/4) Epoch 13, batch 2300, loss[loss=0.1337, simple_loss=0.1522, pruned_loss=0.05756, over 5961.00 frames. ], tot_loss[loss=0.122, simple_loss=0.1529, pruned_loss=0.04557, over 1926603.36 frames. ], batch size: 100, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:33:25,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.107e+02 2.508e+02 3.191e+02 7.606e+02, threshold=5.016e+02, percent-clipped=1.0 2022-12-08 03:33:26,665 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93071.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:30,945 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93076.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:33:49,187 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93097.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:33:51,084 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4764, 4.3029, 4.1675, 4.5250, 4.0776, 3.9290, 4.5964, 4.4112], device='cuda:0'), covar=tensor([0.0558, 0.0698, 0.0651, 0.0476, 0.0707, 0.0562, 0.0448, 0.0530], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0129, 0.0137, 0.0148, 0.0137, 0.0116, 0.0156, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:34:05,090 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:34:12,514 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3004, 4.8244, 4.7622, 5.2856, 4.8995, 4.5608, 5.2938, 4.4368], device='cuda:0'), covar=tensor([0.0347, 0.1126, 0.0324, 0.0386, 0.0731, 0.0523, 0.0471, 0.0428], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0266, 0.0190, 0.0187, 0.0182, 0.0151, 0.0272, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 03:34:31,777 INFO [train.py:873] (0/4) Epoch 13, batch 2400, loss[loss=0.1831, simple_loss=0.1729, pruned_loss=0.0966, over 1267.00 frames. ], tot_loss[loss=0.123, simple_loss=0.1535, pruned_loss=0.04622, over 1882007.21 frames. ], batch size: 100, lr: 6.13e-03, grad_scale: 8.0 2022-12-08 03:34:53,124 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.312e+02 2.703e+02 3.635e+02 6.826e+02, threshold=5.406e+02, percent-clipped=5.0 2022-12-08 03:34:58,986 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:35:08,687 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8127, 1.6474, 1.8081, 2.0511, 1.5234, 1.7600, 1.8216, 1.9577], device='cuda:0'), covar=tensor([0.0135, 0.0254, 0.0121, 0.0101, 0.0198, 0.0306, 0.0156, 0.0116], device='cuda:0'), in_proj_covar=tensor([0.0286, 0.0255, 0.0369, 0.0324, 0.0264, 0.0302, 0.0304, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:35:09,389 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93187.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:35:18,504 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 03:35:59,849 INFO [train.py:873] (0/4) Epoch 13, batch 2500, loss[loss=0.1375, simple_loss=0.1394, pruned_loss=0.06781, over 1206.00 frames. ], tot_loss[loss=0.1221, simple_loss=0.1533, pruned_loss=0.04547, over 1929769.97 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:36:09,005 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93255.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:22,499 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.321e+02 2.910e+02 3.520e+02 9.929e+02, threshold=5.819e+02, percent-clipped=5.0 2022-12-08 03:36:23,807 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2022-12-08 03:36:32,065 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:37,172 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:36:49,084 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 03:37:15,090 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:16,484 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 03:37:16,863 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93331.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:28,915 INFO [train.py:873] (0/4) Epoch 13, batch 2600, loss[loss=0.1321, simple_loss=0.1619, pruned_loss=0.05114, over 9435.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1529, pruned_loss=0.045, over 1930544.06 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:37:47,477 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93366.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:37:50,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.180e+02 2.788e+02 3.347e+02 5.990e+02, threshold=5.575e+02, percent-clipped=1.0 2022-12-08 03:37:56,044 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:38:10,520 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93392.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:38:39,212 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93424.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:38:57,479 INFO [train.py:873] (0/4) Epoch 13, batch 2700, loss[loss=0.1467, simple_loss=0.1591, pruned_loss=0.06709, over 5035.00 frames. ], tot_loss[loss=0.121, simple_loss=0.1526, pruned_loss=0.04464, over 1939610.91 frames. ], batch size: 100, lr: 6.12e-03, grad_scale: 8.0 2022-12-08 03:39:19,857 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.738e+01 2.118e+02 2.554e+02 3.202e+02 5.751e+02, threshold=5.109e+02, percent-clipped=2.0 2022-12-08 03:39:20,915 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:39:34,819 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93487.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:40:16,912 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93535.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:40:25,981 INFO [train.py:873] (0/4) Epoch 13, batch 2800, loss[loss=0.1339, simple_loss=0.127, pruned_loss=0.07037, over 1261.00 frames. ], tot_loss[loss=0.1212, simple_loss=0.1527, pruned_loss=0.04488, over 1931366.90 frames. ], batch size: 100, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:40:35,135 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:40:47,337 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.714e+01 2.417e+02 3.010e+02 3.786e+02 1.160e+03, threshold=6.020e+02, percent-clipped=14.0 2022-12-08 03:41:02,960 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93587.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:16,933 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:29,930 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2022-12-08 03:41:41,566 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93631.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:44,846 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93635.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:41:53,318 INFO [train.py:873] (0/4) Epoch 13, batch 2900, loss[loss=0.1249, simple_loss=0.1587, pruned_loss=0.04554, over 14025.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1533, pruned_loss=0.04477, over 2005171.37 frames. ], batch size: 29, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:42:11,879 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93666.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:14,915 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2663, 1.8431, 2.2505, 1.9370, 2.3798, 2.1481, 1.9795, 2.1086], device='cuda:0'), covar=tensor([0.0462, 0.1599, 0.0425, 0.0715, 0.0412, 0.0726, 0.0382, 0.0470], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0310, 0.0394, 0.0297, 0.0376, 0.0317, 0.0356, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:42:15,457 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 2.101e+02 2.613e+02 3.235e+02 8.582e+02, threshold=5.226e+02, percent-clipped=1.0 2022-12-08 03:42:19,380 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2022-12-08 03:42:23,187 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:34,296 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:42:46,882 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7407, 3.0594, 4.4328, 3.4341, 4.4393, 4.5483, 4.2077, 3.8893], device='cuda:0'), covar=tensor([0.0592, 0.2634, 0.0982, 0.1601, 0.0933, 0.0684, 0.1526, 0.2244], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0310, 0.0395, 0.0299, 0.0376, 0.0317, 0.0357, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:42:53,844 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93714.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:43:06,706 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6980, 1.9135, 2.0793, 2.1329, 1.9396, 2.1517, 1.8141, 1.4034], device='cuda:0'), covar=tensor([0.1091, 0.1103, 0.0730, 0.0500, 0.1016, 0.0667, 0.1435, 0.2129], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0082, 0.0066, 0.0068, 0.0097, 0.0082, 0.0096, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 03:43:16,183 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:43:21,028 INFO [train.py:873] (0/4) Epoch 13, batch 3000, loss[loss=0.1426, simple_loss=0.1721, pruned_loss=0.05656, over 14259.00 frames. ], tot_loss[loss=0.1212, simple_loss=0.1531, pruned_loss=0.04462, over 1998070.28 frames. ], batch size: 44, lr: 6.11e-03, grad_scale: 8.0 2022-12-08 03:43:21,029 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 03:43:29,438 INFO [train.py:905] (0/4) Epoch 13, validation: loss=0.1304, simple_loss=0.1697, pruned_loss=0.04555, over 857387.00 frames. 2022-12-08 03:43:29,439 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 03:43:52,064 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.240e+02 2.864e+02 3.755e+02 7.950e+02, threshold=5.728e+02, percent-clipped=3.0 2022-12-08 03:43:52,313 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:43:53,152 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:14,628 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 03:44:15,297 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 03:44:35,621 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:46,169 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:44:58,704 INFO [train.py:873] (0/4) Epoch 13, batch 3100, loss[loss=0.1491, simple_loss=0.1668, pruned_loss=0.06566, over 8650.00 frames. ], tot_loss[loss=0.1206, simple_loss=0.1529, pruned_loss=0.04414, over 2053749.87 frames. ], batch size: 100, lr: 6.10e-03, grad_scale: 8.0 2022-12-08 03:45:12,337 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8262, 0.8559, 0.7086, 0.8534, 0.8416, 0.2200, 0.7470, 0.8679], device='cuda:0'), covar=tensor([0.0348, 0.0414, 0.0476, 0.0350, 0.0263, 0.0305, 0.0994, 0.0598], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0031, 0.0028, 0.0029, 0.0042, 0.0029, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:45:19,448 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.395e+02 2.887e+02 3.517e+02 6.635e+02, threshold=5.774e+02, percent-clipped=1.0 2022-12-08 03:45:42,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8079, 1.3563, 3.2306, 2.9431, 3.0902, 3.2489, 2.5764, 3.1945], device='cuda:0'), covar=tensor([0.1311, 0.1559, 0.0145, 0.0348, 0.0306, 0.0171, 0.0368, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0127, 0.0165, 0.0144, 0.0141, 0.0121, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:45:48,743 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2022-12-08 03:45:50,822 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5701, 2.6140, 2.7742, 2.7231, 2.7201, 2.4197, 1.5075, 2.4367], device='cuda:0'), covar=tensor([0.0530, 0.0484, 0.0446, 0.0418, 0.0424, 0.1040, 0.2437, 0.0412], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0166, 0.0139, 0.0138, 0.0198, 0.0133, 0.0154, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 03:46:08,347 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9179, 2.0396, 2.8627, 2.2469, 2.8697, 2.7890, 2.6101, 2.4566], device='cuda:0'), covar=tensor([0.0756, 0.3192, 0.0989, 0.1743, 0.0801, 0.0952, 0.1074, 0.1918], device='cuda:0'), in_proj_covar=tensor([0.0344, 0.0311, 0.0393, 0.0297, 0.0373, 0.0316, 0.0356, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:46:25,612 INFO [train.py:873] (0/4) Epoch 13, batch 3200, loss[loss=0.1166, simple_loss=0.1553, pruned_loss=0.03897, over 14405.00 frames. ], tot_loss[loss=0.1213, simple_loss=0.1532, pruned_loss=0.04474, over 1984752.27 frames. ], batch size: 53, lr: 6.10e-03, grad_scale: 8.0 2022-12-08 03:46:36,462 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0107, 2.2141, 1.6420, 2.0653, 2.3328, 1.8404, 1.8705, 1.9316], device='cuda:0'), covar=tensor([0.0322, 0.0776, 0.0406, 0.0389, 0.0392, 0.0420, 0.0397, 0.0435], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0018, 0.0016, 0.0017, 0.0017, 0.0028, 0.0023, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:46:48,614 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.002e+02 2.396e+02 3.046e+02 4.955e+02, threshold=4.792e+02, percent-clipped=0.0 2022-12-08 03:46:49,637 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4486, 3.3889, 4.1303, 2.9902, 2.5452, 3.5207, 1.9793, 3.5372], device='cuda:0'), covar=tensor([0.1003, 0.1301, 0.0497, 0.1565, 0.1996, 0.0650, 0.3065, 0.1477], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0097, 0.0090, 0.0098, 0.0115, 0.0085, 0.0122, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 03:46:57,608 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:47:23,542 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7088, 1.6360, 1.8990, 1.6918, 1.9086, 0.9526, 1.5457, 1.6245], device='cuda:0'), covar=tensor([0.1181, 0.0981, 0.0643, 0.0875, 0.1016, 0.0878, 0.0836, 0.0866], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0031, 0.0027, 0.0029, 0.0041, 0.0030, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 03:47:50,361 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94042.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:47:53,369 INFO [train.py:873] (0/4) Epoch 13, batch 3300, loss[loss=0.1447, simple_loss=0.1731, pruned_loss=0.05816, over 14127.00 frames. ], tot_loss[loss=0.1209, simple_loss=0.1527, pruned_loss=0.0446, over 1960234.43 frames. ], batch size: 99, lr: 6.10e-03, grad_scale: 4.0 2022-12-08 03:48:15,567 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.207e+02 2.819e+02 3.660e+02 9.376e+02, threshold=5.637e+02, percent-clipped=6.0 2022-12-08 03:49:03,570 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94126.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:49:19,899 INFO [train.py:873] (0/4) Epoch 13, batch 3400, loss[loss=0.1167, simple_loss=0.1573, pruned_loss=0.03805, over 13930.00 frames. ], tot_loss[loss=0.1204, simple_loss=0.1522, pruned_loss=0.04428, over 1956903.89 frames. ], batch size: 23, lr: 6.09e-03, grad_scale: 4.0 2022-12-08 03:49:42,649 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.362e+02 2.991e+02 3.918e+02 1.858e+03, threshold=5.982e+02, percent-clipped=6.0 2022-12-08 03:49:50,596 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-08 03:50:28,506 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94223.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 03:50:47,831 INFO [train.py:873] (0/4) Epoch 13, batch 3500, loss[loss=0.1255, simple_loss=0.1533, pruned_loss=0.04881, over 11987.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1522, pruned_loss=0.04466, over 1908016.78 frames. ], batch size: 100, lr: 6.09e-03, grad_scale: 4.0 2022-12-08 03:51:10,108 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.271e+02 2.788e+02 3.634e+02 1.227e+03, threshold=5.575e+02, percent-clipped=1.0 2022-12-08 03:51:11,254 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8865, 2.3778, 3.3079, 2.2883, 2.0838, 2.8428, 1.5402, 2.8097], device='cuda:0'), covar=tensor([0.1099, 0.1498, 0.0612, 0.1699, 0.2302, 0.1003, 0.3654, 0.1018], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0096, 0.0088, 0.0096, 0.0113, 0.0084, 0.0122, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 03:51:21,731 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94284.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:51:44,062 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 03:51:50,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4602, 4.0042, 3.1712, 4.7414, 4.3379, 4.5676, 4.0647, 3.2742], device='cuda:0'), covar=tensor([0.0620, 0.1037, 0.3133, 0.0419, 0.0675, 0.1297, 0.0983, 0.2882], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0293, 0.0264, 0.0266, 0.0314, 0.0296, 0.0256, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:52:08,028 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:52:14,716 INFO [train.py:873] (0/4) Epoch 13, batch 3600, loss[loss=0.1189, simple_loss=0.1498, pruned_loss=0.04407, over 14408.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.1518, pruned_loss=0.04393, over 1957450.73 frames. ], batch size: 55, lr: 6.09e-03, grad_scale: 8.0 2022-12-08 03:52:17,404 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94348.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:52:27,034 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7766, 1.4333, 3.3182, 3.0216, 3.1471, 3.3191, 2.4820, 3.2944], device='cuda:0'), covar=tensor([0.1613, 0.1724, 0.0144, 0.0323, 0.0312, 0.0170, 0.0391, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0127, 0.0166, 0.0144, 0.0140, 0.0122, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:52:30,900 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3134, 2.2740, 4.2767, 4.3714, 4.0941, 2.4699, 4.3310, 3.3433], device='cuda:0'), covar=tensor([0.0350, 0.1099, 0.0648, 0.0370, 0.0417, 0.1590, 0.0340, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0254, 0.0369, 0.0324, 0.0266, 0.0300, 0.0303, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:52:37,864 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.011e+02 2.604e+02 3.365e+02 5.868e+02, threshold=5.208e+02, percent-clipped=1.0 2022-12-08 03:53:01,429 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2022-12-08 03:53:10,921 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94409.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 03:53:25,268 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94426.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:53:38,766 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2022-12-08 03:53:41,901 INFO [train.py:873] (0/4) Epoch 13, batch 3700, loss[loss=0.1339, simple_loss=0.1624, pruned_loss=0.05273, over 14271.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1525, pruned_loss=0.04458, over 1949004.75 frames. ], batch size: 76, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:53:49,352 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4680, 2.7072, 2.7517, 2.7457, 2.3950, 2.7634, 2.4885, 1.4974], device='cuda:0'), covar=tensor([0.1163, 0.1034, 0.0716, 0.0699, 0.0912, 0.0622, 0.1199, 0.2265], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0083, 0.0066, 0.0070, 0.0097, 0.0083, 0.0097, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 03:54:04,972 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.353e+02 2.765e+02 3.627e+02 6.924e+02, threshold=5.531e+02, percent-clipped=2.0 2022-12-08 03:54:06,727 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:54:10,703 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 03:54:14,689 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0616, 2.1372, 1.9331, 2.1722, 1.7516, 2.0394, 2.1408, 2.0833], device='cuda:0'), covar=tensor([0.0868, 0.1077, 0.1125, 0.0815, 0.1447, 0.0792, 0.0951, 0.0890], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0131, 0.0142, 0.0150, 0.0139, 0.0116, 0.0158, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:55:09,007 INFO [train.py:873] (0/4) Epoch 13, batch 3800, loss[loss=0.1101, simple_loss=0.1493, pruned_loss=0.03546, over 14222.00 frames. ], tot_loss[loss=0.1209, simple_loss=0.1525, pruned_loss=0.04465, over 1981656.02 frames. ], batch size: 94, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:55:32,195 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.117e+02 2.466e+02 3.115e+02 4.944e+02, threshold=4.932e+02, percent-clipped=0.0 2022-12-08 03:55:38,518 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94579.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 03:55:54,293 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8673, 1.4424, 2.8067, 2.5210, 2.6692, 2.8408, 2.0480, 2.8026], device='cuda:0'), covar=tensor([0.0962, 0.1317, 0.0180, 0.0395, 0.0384, 0.0185, 0.0555, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0127, 0.0166, 0.0145, 0.0140, 0.0122, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 03:56:25,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0168, 2.0570, 4.0566, 2.8314, 3.8931, 1.9432, 2.9230, 3.8639], device='cuda:0'), covar=tensor([0.0598, 0.4199, 0.0529, 0.5789, 0.0663, 0.3453, 0.1418, 0.0536], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0210, 0.0208, 0.0284, 0.0226, 0.0211, 0.0212, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:56:29,438 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94637.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:56:36,277 INFO [train.py:873] (0/4) Epoch 13, batch 3900, loss[loss=0.1515, simple_loss=0.1632, pruned_loss=0.06989, over 7795.00 frames. ], tot_loss[loss=0.1213, simple_loss=0.1527, pruned_loss=0.04493, over 1906693.45 frames. ], batch size: 100, lr: 6.08e-03, grad_scale: 4.0 2022-12-08 03:56:47,271 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.8570, 5.6787, 5.2570, 5.9920, 5.4831, 5.2533, 5.9246, 5.8007], device='cuda:0'), covar=tensor([0.0527, 0.0479, 0.0802, 0.0374, 0.0558, 0.0352, 0.0468, 0.0449], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0132, 0.0142, 0.0150, 0.0139, 0.0116, 0.0160, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 03:56:59,461 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 2.050e+02 2.441e+02 3.382e+02 7.650e+02, threshold=4.881e+02, percent-clipped=6.0 2022-12-08 03:57:11,172 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 03:57:27,745 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94704.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 03:57:42,385 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 03:57:51,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7991, 1.7677, 3.0300, 2.2782, 2.8480, 1.7784, 2.3610, 2.8058], device='cuda:0'), covar=tensor([0.1259, 0.3768, 0.0568, 0.3937, 0.0836, 0.3050, 0.1308, 0.0754], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0208, 0.0206, 0.0281, 0.0225, 0.0209, 0.0210, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:57:58,881 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 03:58:02,597 INFO [train.py:873] (0/4) Epoch 13, batch 4000, loss[loss=0.1274, simple_loss=0.1261, pruned_loss=0.06431, over 1300.00 frames. ], tot_loss[loss=0.1228, simple_loss=0.1535, pruned_loss=0.0461, over 1869018.79 frames. ], batch size: 100, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 03:58:26,194 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.218e+02 2.831e+02 3.684e+02 6.338e+02, threshold=5.663e+02, percent-clipped=4.0 2022-12-08 03:58:26,506 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6936, 2.9547, 4.4113, 3.3754, 4.3699, 4.4765, 4.3790, 3.9958], device='cuda:0'), covar=tensor([0.0807, 0.2939, 0.0884, 0.1982, 0.0836, 0.0799, 0.1563, 0.1817], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0312, 0.0395, 0.0300, 0.0376, 0.0321, 0.0362, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:59:01,241 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6477, 2.0489, 3.7193, 2.6496, 3.6160, 1.9028, 2.7663, 3.5311], device='cuda:0'), covar=tensor([0.0691, 0.4155, 0.0616, 0.5436, 0.0724, 0.3468, 0.1386, 0.0619], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0208, 0.0206, 0.0282, 0.0225, 0.0210, 0.0211, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 03:59:29,380 INFO [train.py:873] (0/4) Epoch 13, batch 4100, loss[loss=0.1413, simple_loss=0.1675, pruned_loss=0.05753, over 14160.00 frames. ], tot_loss[loss=0.1223, simple_loss=0.1531, pruned_loss=0.04574, over 1887484.76 frames. ], batch size: 99, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 03:59:52,465 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 2.091e+02 2.510e+02 3.105e+02 7.353e+02, threshold=5.020e+02, percent-clipped=2.0 2022-12-08 03:59:58,825 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94879.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:00:02,653 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 04:00:40,982 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94927.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:00:56,135 INFO [train.py:873] (0/4) Epoch 13, batch 4200, loss[loss=0.1232, simple_loss=0.1486, pruned_loss=0.04887, over 9503.00 frames. ], tot_loss[loss=0.1221, simple_loss=0.1529, pruned_loss=0.04562, over 1884327.04 frames. ], batch size: 100, lr: 6.07e-03, grad_scale: 8.0 2022-12-08 04:01:19,978 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.402e+02 2.842e+02 3.518e+02 1.114e+03, threshold=5.683e+02, percent-clipped=6.0 2022-12-08 04:01:35,994 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0647, 1.8255, 3.2656, 2.3921, 3.0989, 1.8230, 2.4271, 3.1562], device='cuda:0'), covar=tensor([0.0771, 0.3821, 0.0527, 0.4238, 0.0716, 0.3132, 0.1321, 0.0497], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0208, 0.0206, 0.0282, 0.0227, 0.0208, 0.0211, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:01:43,977 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-95000.pt 2022-12-08 04:01:51,318 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95004.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:02:26,455 INFO [train.py:873] (0/4) Epoch 13, batch 4300, loss[loss=0.1225, simple_loss=0.1556, pruned_loss=0.04473, over 14550.00 frames. ], tot_loss[loss=0.1223, simple_loss=0.1534, pruned_loss=0.0456, over 1922755.80 frames. ], batch size: 43, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:02:30,783 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.15 vs. limit=5.0 2022-12-08 04:02:32,643 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95052.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:02:44,339 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2022-12-08 04:02:49,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.351e+02 2.797e+02 3.272e+02 6.145e+02, threshold=5.594e+02, percent-clipped=1.0 2022-12-08 04:03:09,984 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2577, 3.0895, 3.0783, 3.2916, 3.1618, 3.2450, 3.3669, 2.8138], device='cuda:0'), covar=tensor([0.0498, 0.1032, 0.0530, 0.0548, 0.0690, 0.0448, 0.0573, 0.0611], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0272, 0.0190, 0.0187, 0.0182, 0.0151, 0.0275, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 04:03:11,648 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4941, 3.2439, 3.1062, 2.0937, 2.9578, 3.2768, 3.5550, 2.7952], device='cuda:0'), covar=tensor([0.0594, 0.1104, 0.0920, 0.1577, 0.0933, 0.0687, 0.0624, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0178, 0.0139, 0.0127, 0.0139, 0.0150, 0.0126, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 04:03:11,718 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2164, 2.7547, 4.0990, 3.0329, 4.1101, 3.9304, 3.9047, 3.5328], device='cuda:0'), covar=tensor([0.0663, 0.2775, 0.0839, 0.1658, 0.0669, 0.0842, 0.1173, 0.1516], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0312, 0.0395, 0.0301, 0.0373, 0.0321, 0.0363, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:03:13,516 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.54 vs. limit=5.0 2022-12-08 04:03:18,403 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95105.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:03:44,330 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([6.2011, 5.6306, 5.6297, 6.1351, 5.6685, 4.9188, 6.1286, 5.0191], device='cuda:0'), covar=tensor([0.0228, 0.0837, 0.0264, 0.0352, 0.0676, 0.0303, 0.0344, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0272, 0.0191, 0.0188, 0.0183, 0.0152, 0.0276, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 04:03:53,023 INFO [train.py:873] (0/4) Epoch 13, batch 4400, loss[loss=0.1392, simple_loss=0.1583, pruned_loss=0.06008, over 7800.00 frames. ], tot_loss[loss=0.1212, simple_loss=0.1526, pruned_loss=0.0449, over 1943629.92 frames. ], batch size: 100, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:04:10,989 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95166.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:04:16,201 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.201e+02 2.204e+02 2.628e+02 3.192e+02 7.807e+02, threshold=5.256e+02, percent-clipped=2.0 2022-12-08 04:04:29,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2022-12-08 04:04:36,092 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3801, 3.5987, 3.6433, 3.4936, 2.6189, 3.6006, 3.2862, 1.8185], device='cuda:0'), covar=tensor([0.1703, 0.0809, 0.0697, 0.0576, 0.1152, 0.0455, 0.1241, 0.2340], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0083, 0.0066, 0.0069, 0.0097, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:05:20,305 INFO [train.py:873] (0/4) Epoch 13, batch 4500, loss[loss=0.1345, simple_loss=0.1366, pruned_loss=0.06614, over 1315.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.152, pruned_loss=0.04456, over 1939441.57 frames. ], batch size: 100, lr: 6.06e-03, grad_scale: 8.0 2022-12-08 04:05:33,308 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:05:40,416 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:05:43,635 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.015e+02 2.327e+02 3.103e+02 3.682e+02 7.295e+02, threshold=6.206e+02, percent-clipped=3.0 2022-12-08 04:06:20,723 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 04:06:26,032 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:33,082 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:41,651 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95339.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:06:46,981 INFO [train.py:873] (0/4) Epoch 13, batch 4600, loss[loss=0.1129, simple_loss=0.1383, pruned_loss=0.04376, over 3862.00 frames. ], tot_loss[loss=0.1211, simple_loss=0.1524, pruned_loss=0.0449, over 1934565.09 frames. ], batch size: 100, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:07:10,470 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.077e+02 2.550e+02 3.185e+02 6.474e+02, threshold=5.099e+02, percent-clipped=1.0 2022-12-08 04:07:35,098 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95400.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:08:14,530 INFO [train.py:873] (0/4) Epoch 13, batch 4700, loss[loss=0.1226, simple_loss=0.1548, pruned_loss=0.04519, over 14282.00 frames. ], tot_loss[loss=0.1201, simple_loss=0.1518, pruned_loss=0.04425, over 2020806.28 frames. ], batch size: 80, lr: 6.05e-03, grad_scale: 4.0 2022-12-08 04:08:28,700 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95461.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:08:39,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.071e+02 2.759e+02 3.394e+02 8.943e+02, threshold=5.519e+02, percent-clipped=4.0 2022-12-08 04:09:41,416 INFO [train.py:873] (0/4) Epoch 13, batch 4800, loss[loss=0.1254, simple_loss=0.1588, pruned_loss=0.04597, over 14298.00 frames. ], tot_loss[loss=0.1207, simple_loss=0.1526, pruned_loss=0.04437, over 2039940.34 frames. ], batch size: 60, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:09:51,422 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5465, 2.7037, 2.6320, 2.7019, 2.2964, 2.7737, 2.5200, 1.4881], device='cuda:0'), covar=tensor([0.1322, 0.0968, 0.0838, 0.0772, 0.1036, 0.0569, 0.1245, 0.2394], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0083, 0.0067, 0.0069, 0.0096, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:10:05,764 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.302e+02 2.972e+02 3.662e+02 7.055e+02, threshold=5.944e+02, percent-clipped=3.0 2022-12-08 04:10:17,131 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9607, 2.4707, 4.8902, 3.3447, 4.6146, 2.2937, 3.7982, 4.7003], device='cuda:0'), covar=tensor([0.0474, 0.4397, 0.0402, 0.6416, 0.0672, 0.3607, 0.1247, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0210, 0.0208, 0.0284, 0.0226, 0.0210, 0.0211, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:10:43,516 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:10:44,442 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:10:50,223 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:10:54,909 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 04:11:00,662 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2022-12-08 04:11:08,088 INFO [train.py:873] (0/4) Epoch 13, batch 4900, loss[loss=0.1419, simple_loss=0.1731, pruned_loss=0.05536, over 14376.00 frames. ], tot_loss[loss=0.1213, simple_loss=0.153, pruned_loss=0.04477, over 2035645.26 frames. ], batch size: 55, lr: 6.05e-03, grad_scale: 8.0 2022-12-08 04:11:25,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6827, 2.3597, 2.9291, 1.8883, 1.9460, 2.5582, 1.5017, 2.5784], device='cuda:0'), covar=tensor([0.1030, 0.1283, 0.0673, 0.2067, 0.2397, 0.0874, 0.3666, 0.1011], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0097, 0.0090, 0.0097, 0.0113, 0.0086, 0.0121, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 04:11:32,372 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.158e+02 2.610e+02 3.322e+02 1.003e+03, threshold=5.220e+02, percent-clipped=4.0 2022-12-08 04:11:36,534 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:11:51,525 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95695.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:12:02,805 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2550, 4.2751, 4.6142, 3.9224, 4.4499, 4.6327, 1.7996, 4.1590], device='cuda:0'), covar=tensor([0.0278, 0.0324, 0.0312, 0.0364, 0.0261, 0.0180, 0.3155, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0172, 0.0143, 0.0142, 0.0202, 0.0138, 0.0160, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:12:35,019 INFO [train.py:873] (0/4) Epoch 13, batch 5000, loss[loss=0.1529, simple_loss=0.1739, pruned_loss=0.06599, over 11940.00 frames. ], tot_loss[loss=0.1207, simple_loss=0.1524, pruned_loss=0.04448, over 1975676.65 frames. ], batch size: 100, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:12:44,997 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5896, 3.9058, 3.8113, 3.6274, 2.8643, 3.9364, 3.6503, 1.8917], device='cuda:0'), covar=tensor([0.1927, 0.0774, 0.0901, 0.1088, 0.0973, 0.0350, 0.1098, 0.2180], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0083, 0.0066, 0.0069, 0.0096, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:12:49,205 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95761.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:12:59,674 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.284e+02 2.864e+02 3.553e+02 7.569e+02, threshold=5.729e+02, percent-clipped=2.0 2022-12-08 04:13:06,360 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8615, 1.3440, 2.7831, 2.5515, 2.6779, 2.7395, 2.1008, 2.7995], device='cuda:0'), covar=tensor([0.1064, 0.1344, 0.0151, 0.0399, 0.0358, 0.0188, 0.0459, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0126, 0.0166, 0.0144, 0.0138, 0.0121, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 04:13:22,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1507, 4.7971, 4.5933, 5.1154, 4.7748, 4.4190, 5.1181, 4.2410], device='cuda:0'), covar=tensor([0.0303, 0.0907, 0.0365, 0.0385, 0.0775, 0.0540, 0.0505, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0270, 0.0190, 0.0187, 0.0183, 0.0151, 0.0276, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 04:13:31,718 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95809.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:13:34,716 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95812.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:14:03,053 INFO [train.py:873] (0/4) Epoch 13, batch 5100, loss[loss=0.1306, simple_loss=0.1602, pruned_loss=0.0505, over 14259.00 frames. ], tot_loss[loss=0.1198, simple_loss=0.1516, pruned_loss=0.04403, over 2003975.53 frames. ], batch size: 76, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:14:05,076 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95847.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:14:27,856 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.050e+02 2.508e+02 3.173e+02 5.976e+02, threshold=5.017e+02, percent-clipped=1.0 2022-12-08 04:14:28,136 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95873.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:14:59,256 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95908.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 04:15:06,299 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:13,276 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:31,126 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5357, 2.2710, 2.7578, 1.5989, 1.8057, 2.4758, 1.4574, 2.4635], device='cuda:0'), covar=tensor([0.1064, 0.1379, 0.0786, 0.2879, 0.2515, 0.1131, 0.3568, 0.0948], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0098, 0.0092, 0.0098, 0.0114, 0.0086, 0.0121, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 04:15:31,805 INFO [train.py:873] (0/4) Epoch 13, batch 5200, loss[loss=0.1387, simple_loss=0.167, pruned_loss=0.05514, over 11190.00 frames. ], tot_loss[loss=0.1202, simple_loss=0.1522, pruned_loss=0.04411, over 2047369.44 frames. ], batch size: 100, lr: 6.04e-03, grad_scale: 8.0 2022-12-08 04:15:48,347 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95964.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:52,955 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0596, 2.3955, 2.3563, 2.5082, 1.9896, 2.4863, 2.2175, 1.3122], device='cuda:0'), covar=tensor([0.1639, 0.0886, 0.0884, 0.0658, 0.1286, 0.0797, 0.1452, 0.2742], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0083, 0.0067, 0.0069, 0.0097, 0.0083, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:15:55,359 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:15:56,530 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 2.184e+02 2.660e+02 3.283e+02 4.676e+02, threshold=5.321e+02, percent-clipped=0.0 2022-12-08 04:15:56,625 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:03,022 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:03,894 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2999, 2.0028, 2.2511, 2.2337, 2.1610, 1.3873, 2.1971, 2.2806], device='cuda:0'), covar=tensor([0.1087, 0.0932, 0.0735, 0.1487, 0.2431, 0.0765, 0.1032, 0.1250], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0031, 0.0028, 0.0029, 0.0042, 0.0030, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:16:15,818 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95995.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:24,531 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9707, 2.1114, 1.9158, 2.0856, 1.7704, 1.9612, 2.0752, 1.9996], device='cuda:0'), covar=tensor([0.1054, 0.1118, 0.1166, 0.0896, 0.1473, 0.0843, 0.1134, 0.0997], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0135, 0.0144, 0.0155, 0.0143, 0.0119, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 04:16:34,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9943, 2.3882, 3.8761, 4.0663, 3.9011, 2.4731, 3.9221, 3.0764], device='cuda:0'), covar=tensor([0.0349, 0.0977, 0.0797, 0.0387, 0.0404, 0.1501, 0.0357, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0286, 0.0251, 0.0365, 0.0321, 0.0263, 0.0297, 0.0300, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 04:16:56,526 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96041.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:58,087 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:16:59,769 INFO [train.py:873] (0/4) Epoch 13, batch 5300, loss[loss=0.09624, simple_loss=0.1391, pruned_loss=0.02669, over 14250.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1515, pruned_loss=0.04321, over 2032498.70 frames. ], batch size: 39, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:17:23,907 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.223e+02 2.915e+02 3.617e+02 9.982e+02, threshold=5.829e+02, percent-clipped=6.0 2022-12-08 04:17:29,942 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:17:30,246 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 04:18:18,732 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6751, 1.7531, 1.7492, 1.6058, 1.7627, 1.5029, 1.4251, 1.0884], device='cuda:0'), covar=tensor([0.0220, 0.0217, 0.0235, 0.0286, 0.0202, 0.0292, 0.0262, 0.0467], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0016, 0.0017, 0.0017, 0.0028, 0.0023, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:18:23,865 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96141.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:18:27,058 INFO [train.py:873] (0/4) Epoch 13, batch 5400, loss[loss=0.1288, simple_loss=0.1619, pruned_loss=0.04787, over 14264.00 frames. ], tot_loss[loss=0.1198, simple_loss=0.152, pruned_loss=0.04375, over 2019421.59 frames. ], batch size: 57, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:18:44,252 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:18:46,927 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96168.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 04:18:51,342 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.106e+02 2.759e+02 3.696e+02 7.492e+02, threshold=5.517e+02, percent-clipped=3.0 2022-12-08 04:19:17,373 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96203.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:19:33,050 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6891, 3.3129, 2.6138, 3.8170, 3.7079, 3.6391, 3.1849, 2.5883], device='cuda:0'), covar=tensor([0.0784, 0.1420, 0.3288, 0.0546, 0.0710, 0.1371, 0.1258, 0.3639], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0294, 0.0269, 0.0270, 0.0316, 0.0300, 0.0260, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:19:36,529 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 04:19:37,234 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96226.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:19:38,882 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4074, 0.9911, 1.2287, 0.7965, 1.0859, 1.3930, 1.0231, 1.0479], device='cuda:0'), covar=tensor([0.0446, 0.0846, 0.0801, 0.0657, 0.1054, 0.0718, 0.0508, 0.1406], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0177, 0.0139, 0.0126, 0.0138, 0.0150, 0.0126, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 04:19:42,957 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 04:19:53,396 INFO [train.py:873] (0/4) Epoch 13, batch 5500, loss[loss=0.08841, simple_loss=0.1315, pruned_loss=0.02268, over 14058.00 frames. ], tot_loss[loss=0.1193, simple_loss=0.1516, pruned_loss=0.04354, over 1964163.16 frames. ], batch size: 22, lr: 6.03e-03, grad_scale: 8.0 2022-12-08 04:20:05,174 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7233, 2.3544, 3.0615, 1.9327, 1.8600, 2.4575, 1.4111, 2.5463], device='cuda:0'), covar=tensor([0.0917, 0.1372, 0.0658, 0.2572, 0.2711, 0.1253, 0.3908, 0.1271], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0098, 0.0091, 0.0098, 0.0116, 0.0087, 0.0123, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 04:20:17,918 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.945e+01 2.434e+02 2.937e+02 3.811e+02 8.980e+02, threshold=5.874e+02, percent-clipped=5.0 2022-12-08 04:20:18,449 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:20:54,583 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2476, 2.6699, 3.9827, 3.1568, 4.0080, 3.9012, 3.8147, 3.4136], device='cuda:0'), covar=tensor([0.0623, 0.2675, 0.0979, 0.1607, 0.0912, 0.0853, 0.1435, 0.1770], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0316, 0.0398, 0.0306, 0.0378, 0.0325, 0.0367, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:21:00,276 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:02,867 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96324.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:13,575 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:21:21,089 INFO [train.py:873] (0/4) Epoch 13, batch 5600, loss[loss=0.125, simple_loss=0.1584, pruned_loss=0.04575, over 14532.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1524, pruned_loss=0.04462, over 1963965.99 frames. ], batch size: 43, lr: 6.02e-03, grad_scale: 8.0 2022-12-08 04:21:22,872 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1972, 1.7149, 2.4519, 2.0171, 2.2924, 1.5814, 1.9509, 2.2181], device='cuda:0'), covar=tensor([0.1857, 0.3115, 0.0558, 0.1861, 0.1110, 0.2107, 0.0985, 0.0883], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0209, 0.0209, 0.0284, 0.0226, 0.0212, 0.0209, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:21:33,029 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8599, 1.4747, 3.3481, 3.0800, 3.1844, 3.3581, 2.6467, 3.3320], device='cuda:0'), covar=tensor([0.1348, 0.1577, 0.0134, 0.0312, 0.0288, 0.0145, 0.0363, 0.0162], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0127, 0.0167, 0.0145, 0.0139, 0.0123, 0.0120], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 04:21:45,613 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.301e+02 2.704e+02 3.458e+02 6.924e+02, threshold=5.409e+02, percent-clipped=1.0 2022-12-08 04:21:56,093 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:00,540 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:24,530 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2022-12-08 04:22:24,724 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5048, 3.4168, 3.2680, 3.6109, 3.1117, 3.1225, 3.6182, 3.4470], device='cuda:0'), covar=tensor([0.0799, 0.0923, 0.1037, 0.0688, 0.1208, 0.0878, 0.0731, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0137, 0.0146, 0.0156, 0.0144, 0.0121, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 04:22:27,159 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 04:22:41,373 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96436.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:48,795 INFO [train.py:873] (0/4) Epoch 13, batch 5700, loss[loss=0.1319, simple_loss=0.1578, pruned_loss=0.053, over 7783.00 frames. ], tot_loss[loss=0.1208, simple_loss=0.1525, pruned_loss=0.0446, over 1996940.08 frames. ], batch size: 100, lr: 6.02e-03, grad_scale: 8.0 2022-12-08 04:22:54,653 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96451.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:22:58,046 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8234, 1.1205, 1.3115, 1.2091, 1.0355, 1.2809, 1.0971, 0.8755], device='cuda:0'), covar=tensor([0.1934, 0.0952, 0.0394, 0.0437, 0.1850, 0.0876, 0.1457, 0.1485], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0083, 0.0065, 0.0068, 0.0097, 0.0081, 0.0097, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0005, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:23:09,332 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96468.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:23:10,618 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2022-12-08 04:23:13,579 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.382e+02 2.762e+02 3.323e+02 6.415e+02, threshold=5.524e+02, percent-clipped=3.0 2022-12-08 04:23:19,994 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1599, 4.2267, 4.5176, 4.0375, 4.3886, 4.5418, 1.7997, 4.1191], device='cuda:0'), covar=tensor([0.0319, 0.0301, 0.0338, 0.0413, 0.0286, 0.0221, 0.3116, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0172, 0.0144, 0.0142, 0.0203, 0.0138, 0.0159, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:23:40,469 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96503.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:23:51,381 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96516.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 04:23:55,874 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96521.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:24:17,060 INFO [train.py:873] (0/4) Epoch 13, batch 5800, loss[loss=0.09727, simple_loss=0.1439, pruned_loss=0.02534, over 14098.00 frames. ], tot_loss[loss=0.121, simple_loss=0.1527, pruned_loss=0.04466, over 2002065.59 frames. ], batch size: 29, lr: 6.02e-03, grad_scale: 4.0 2022-12-08 04:24:22,211 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:24:42,625 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.152e+02 2.229e+02 2.647e+02 3.440e+02 6.528e+02, threshold=5.293e+02, percent-clipped=3.0 2022-12-08 04:25:16,772 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-08 04:25:37,067 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96636.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:25:44,388 INFO [train.py:873] (0/4) Epoch 13, batch 5900, loss[loss=0.1245, simple_loss=0.1346, pruned_loss=0.05714, over 3863.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.1523, pruned_loss=0.04434, over 1946976.11 frames. ], batch size: 100, lr: 6.01e-03, grad_scale: 4.0 2022-12-08 04:26:10,182 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 2.158e+02 2.566e+02 3.278e+02 1.082e+03, threshold=5.131e+02, percent-clipped=5.0 2022-12-08 04:26:15,575 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:26:18,890 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96684.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:26:38,430 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5773, 3.9211, 3.5936, 3.6670, 2.6916, 3.7608, 3.6982, 2.0059], device='cuda:0'), covar=tensor([0.1505, 0.0841, 0.1277, 0.0762, 0.1003, 0.0410, 0.0764, 0.2086], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0084, 0.0066, 0.0069, 0.0098, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:27:02,604 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2022-12-08 04:27:03,981 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:11,548 INFO [train.py:873] (0/4) Epoch 13, batch 6000, loss[loss=0.09161, simple_loss=0.1289, pruned_loss=0.02716, over 13944.00 frames. ], tot_loss[loss=0.1205, simple_loss=0.1522, pruned_loss=0.04439, over 1921739.13 frames. ], batch size: 20, lr: 6.01e-03, grad_scale: 8.0 2022-12-08 04:27:11,549 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 04:27:19,805 INFO [train.py:905] (0/4) Epoch 13, validation: loss=0.132, simple_loss=0.1717, pruned_loss=0.04613, over 857387.00 frames. 2022-12-08 04:27:19,805 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 04:27:20,796 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96746.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:21,703 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1302, 3.5324, 2.6677, 4.3819, 4.0381, 4.1058, 3.5991, 2.9163], device='cuda:0'), covar=tensor([0.0708, 0.1447, 0.4261, 0.0446, 0.0896, 0.1482, 0.1327, 0.3697], device='cuda:0'), in_proj_covar=tensor([0.0271, 0.0290, 0.0265, 0.0265, 0.0311, 0.0295, 0.0255, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:27:45,010 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.016e+02 2.616e+02 3.028e+02 5.890e+02, threshold=5.232e+02, percent-clipped=1.0 2022-12-08 04:27:46,772 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96776.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:27:53,299 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:17,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.10 vs. limit=5.0 2022-12-08 04:28:25,749 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:35,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3668, 2.4440, 2.5312, 2.4783, 2.4466, 2.1303, 1.5845, 2.2370], device='cuda:0'), covar=tensor([0.0597, 0.0512, 0.0505, 0.0452, 0.0420, 0.1400, 0.2449, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0172, 0.0144, 0.0141, 0.0202, 0.0136, 0.0159, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:28:40,170 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96837.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:28:47,194 INFO [train.py:873] (0/4) Epoch 13, batch 6100, loss[loss=0.1237, simple_loss=0.157, pruned_loss=0.04526, over 14509.00 frames. ], tot_loss[loss=0.1214, simple_loss=0.1525, pruned_loss=0.04522, over 1820300.59 frames. ], batch size: 34, lr: 6.01e-03, grad_scale: 8.0 2022-12-08 04:29:01,848 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-08 04:29:07,672 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96869.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:29:12,500 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.137e+02 2.568e+02 3.326e+02 6.550e+02, threshold=5.137e+02, percent-clipped=4.0 2022-12-08 04:29:15,773 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1083, 3.8931, 3.5822, 3.7590, 3.9898, 4.0096, 4.0671, 4.0679], device='cuda:0'), covar=tensor([0.0811, 0.0512, 0.2130, 0.2613, 0.0747, 0.0829, 0.1027, 0.0842], device='cuda:0'), in_proj_covar=tensor([0.0376, 0.0266, 0.0440, 0.0552, 0.0330, 0.0430, 0.0382, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:30:13,974 INFO [train.py:873] (0/4) Epoch 13, batch 6200, loss[loss=0.1264, simple_loss=0.1577, pruned_loss=0.04757, over 13537.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.1515, pruned_loss=0.04452, over 1837642.29 frames. ], batch size: 100, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:30:39,951 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.243e+02 2.655e+02 3.164e+02 5.951e+02, threshold=5.310e+02, percent-clipped=2.0 2022-12-08 04:30:45,352 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:30:47,714 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2022-12-08 04:31:03,628 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97000.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:28,048 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97028.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:43,226 INFO [train.py:873] (0/4) Epoch 13, batch 6300, loss[loss=0.1469, simple_loss=0.1671, pruned_loss=0.06339, over 8638.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1514, pruned_loss=0.04331, over 1935001.59 frames. ], batch size: 100, lr: 6.00e-03, grad_scale: 4.0 2022-12-08 04:31:44,260 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:31:46,854 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8274, 2.0062, 2.0479, 1.8834, 2.0894, 1.8494, 1.5199, 1.4269], device='cuda:0'), covar=tensor([0.0391, 0.0418, 0.0233, 0.0310, 0.0270, 0.0392, 0.0358, 0.0548], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0016, 0.0017, 0.0017, 0.0028, 0.0023, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:31:56,982 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97061.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:32:09,641 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.227e+02 2.717e+02 3.666e+02 6.842e+02, threshold=5.435e+02, percent-clipped=6.0 2022-12-08 04:32:11,745 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.20 vs. limit=2.0 2022-12-08 04:32:25,893 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97094.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:32:58,837 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:33:04,433 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.28 vs. limit=5.0 2022-12-08 04:33:09,960 INFO [train.py:873] (0/4) Epoch 13, batch 6400, loss[loss=0.114, simple_loss=0.1585, pruned_loss=0.03478, over 13923.00 frames. ], tot_loss[loss=0.1193, simple_loss=0.1518, pruned_loss=0.04341, over 1969159.82 frames. ], batch size: 26, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:33:36,266 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.249e+02 2.635e+02 3.271e+02 6.214e+02, threshold=5.270e+02, percent-clipped=1.0 2022-12-08 04:34:37,201 INFO [train.py:873] (0/4) Epoch 13, batch 6500, loss[loss=0.1083, simple_loss=0.1526, pruned_loss=0.03203, over 14237.00 frames. ], tot_loss[loss=0.1203, simple_loss=0.1523, pruned_loss=0.04414, over 1955685.74 frames. ], batch size: 39, lr: 6.00e-03, grad_scale: 8.0 2022-12-08 04:34:48,682 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-08 04:34:50,074 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97260.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:35:03,585 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.327e+02 2.807e+02 3.571e+02 7.532e+02, threshold=5.613e+02, percent-clipped=5.0 2022-12-08 04:35:41,518 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3640, 3.4558, 3.5906, 3.3685, 3.4870, 3.3394, 1.5438, 3.3153], device='cuda:0'), covar=tensor([0.0372, 0.0372, 0.0394, 0.0451, 0.0303, 0.0578, 0.3215, 0.0303], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0171, 0.0143, 0.0140, 0.0200, 0.0135, 0.0157, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:35:44,043 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:02,230 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:04,578 INFO [train.py:873] (0/4) Epoch 13, batch 6600, loss[loss=0.1206, simple_loss=0.1433, pruned_loss=0.04895, over 5985.00 frames. ], tot_loss[loss=0.1188, simple_loss=0.1514, pruned_loss=0.0431, over 1983221.06 frames. ], batch size: 100, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:36:06,382 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:14,678 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:30,427 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97374.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:36:31,019 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.194e+02 2.777e+02 3.424e+02 5.452e+02, threshold=5.554e+02, percent-clipped=0.0 2022-12-08 04:36:55,836 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97403.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:37:00,355 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:20,868 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:23,916 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:37:26,461 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0867, 1.1697, 1.2114, 1.0249, 0.8939, 0.8618, 0.9466, 0.8810], device='cuda:0'), covar=tensor([0.0196, 0.0236, 0.0194, 0.0248, 0.0270, 0.0425, 0.0269, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0016, 0.0018, 0.0018, 0.0029, 0.0024, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:37:32,757 INFO [train.py:873] (0/4) Epoch 13, batch 6700, loss[loss=0.1126, simple_loss=0.1472, pruned_loss=0.03897, over 14279.00 frames. ], tot_loss[loss=0.1176, simple_loss=0.1509, pruned_loss=0.04212, over 1979932.59 frames. ], batch size: 60, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:37:40,503 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2235, 5.2382, 4.3528, 4.5604, 4.8739, 5.2306, 5.4040, 5.2533], device='cuda:0'), covar=tensor([0.1099, 0.0515, 0.2830, 0.3271, 0.1138, 0.0922, 0.0702, 0.1074], device='cuda:0'), in_proj_covar=tensor([0.0373, 0.0263, 0.0436, 0.0553, 0.0328, 0.0429, 0.0378, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:37:42,397 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7782, 3.0764, 4.5207, 3.4940, 4.5450, 4.4050, 4.2849, 3.8986], device='cuda:0'), covar=tensor([0.0889, 0.2924, 0.0885, 0.1731, 0.0639, 0.0945, 0.1561, 0.1817], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0315, 0.0396, 0.0304, 0.0376, 0.0319, 0.0365, 0.0307], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:37:59,104 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.128e+02 2.579e+02 3.037e+02 5.604e+02, threshold=5.159e+02, percent-clipped=1.0 2022-12-08 04:38:03,376 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97480.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:38:59,023 INFO [train.py:873] (0/4) Epoch 13, batch 6800, loss[loss=0.1035, simple_loss=0.1447, pruned_loss=0.03114, over 14349.00 frames. ], tot_loss[loss=0.1183, simple_loss=0.1511, pruned_loss=0.04274, over 1965821.53 frames. ], batch size: 44, lr: 5.99e-03, grad_scale: 8.0 2022-12-08 04:39:08,469 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:39:25,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.339e+02 3.043e+02 3.694e+02 1.051e+03, threshold=6.086e+02, percent-clipped=5.0 2022-12-08 04:39:32,445 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2107, 2.5533, 2.5452, 2.7369, 2.1330, 2.7129, 2.3956, 1.4409], device='cuda:0'), covar=tensor([0.1247, 0.0922, 0.0857, 0.0491, 0.1072, 0.0619, 0.1134, 0.2374], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0085, 0.0066, 0.0069, 0.0097, 0.0082, 0.0098, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:39:42,383 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6968, 2.0799, 1.9316, 2.0068, 2.1292, 1.8385, 1.7075, 1.3475], device='cuda:0'), covar=tensor([0.0444, 0.0567, 0.0345, 0.0256, 0.0256, 0.0408, 0.0384, 0.0682], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0017, 0.0018, 0.0017, 0.0029, 0.0023, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:39:53,316 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1536, 2.5056, 3.9024, 2.9397, 4.0137, 3.8345, 3.8075, 3.3420], device='cuda:0'), covar=tensor([0.0902, 0.3322, 0.1221, 0.1990, 0.0760, 0.0925, 0.1488, 0.1679], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0315, 0.0394, 0.0305, 0.0376, 0.0320, 0.0365, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:40:01,884 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:01,994 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:27,572 INFO [train.py:873] (0/4) Epoch 13, batch 6900, loss[loss=0.1111, simple_loss=0.1562, pruned_loss=0.03301, over 14466.00 frames. ], tot_loss[loss=0.1191, simple_loss=0.1517, pruned_loss=0.04325, over 1984561.95 frames. ], batch size: 51, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:40:36,924 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97656.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:40:52,694 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97674.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:40:53,268 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 2.182e+02 2.726e+02 3.211e+02 6.785e+02, threshold=5.451e+02, percent-clipped=1.0 2022-12-08 04:40:54,271 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 04:41:00,312 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 04:41:13,542 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97698.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 04:41:17,891 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97703.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:18,593 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:41,329 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97730.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:41:45,617 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97735.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:41:46,326 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0804, 2.9071, 2.6650, 2.7947, 3.0009, 2.9853, 3.0380, 3.0190], device='cuda:0'), covar=tensor([0.0980, 0.0772, 0.2505, 0.2698, 0.0856, 0.1161, 0.1403, 0.1001], device='cuda:0'), in_proj_covar=tensor([0.0380, 0.0270, 0.0444, 0.0561, 0.0335, 0.0438, 0.0388, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:41:53,810 INFO [train.py:873] (0/4) Epoch 13, batch 7000, loss[loss=0.1511, simple_loss=0.163, pruned_loss=0.06961, over 7798.00 frames. ], tot_loss[loss=0.1197, simple_loss=0.152, pruned_loss=0.04366, over 2041459.35 frames. ], batch size: 100, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:42:20,482 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.837e+01 2.234e+02 2.676e+02 3.367e+02 8.383e+02, threshold=5.352e+02, percent-clipped=3.0 2022-12-08 04:42:53,640 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0484, 1.9763, 2.0382, 2.1523, 2.0883, 1.9877, 2.1900, 1.7996], device='cuda:0'), covar=tensor([0.1084, 0.1297, 0.0777, 0.0725, 0.0950, 0.0759, 0.0828, 0.0791], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0270, 0.0193, 0.0188, 0.0186, 0.0154, 0.0276, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 04:43:12,210 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6344, 1.9548, 3.7405, 2.6730, 3.5784, 1.8257, 2.7250, 3.4412], device='cuda:0'), covar=tensor([0.0784, 0.4563, 0.0613, 0.5990, 0.0838, 0.3710, 0.1521, 0.0782], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0208, 0.0211, 0.0282, 0.0228, 0.0210, 0.0207, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:43:23,309 INFO [train.py:873] (0/4) Epoch 13, batch 7100, loss[loss=0.1351, simple_loss=0.1698, pruned_loss=0.05017, over 14487.00 frames. ], tot_loss[loss=0.1201, simple_loss=0.1523, pruned_loss=0.04393, over 1998203.89 frames. ], batch size: 49, lr: 5.98e-03, grad_scale: 8.0 2022-12-08 04:43:49,281 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.322e+02 2.890e+02 3.385e+02 9.839e+02, threshold=5.780e+02, percent-clipped=7.0 2022-12-08 04:44:01,127 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97888.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:21,121 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:26,011 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:36,866 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8076, 0.8647, 0.6665, 0.8807, 0.9243, 0.3754, 0.7569, 0.8444], device='cuda:0'), covar=tensor([0.0333, 0.0405, 0.0293, 0.0378, 0.0286, 0.0273, 0.0704, 0.0594], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0029, 0.0032, 0.0028, 0.0030, 0.0042, 0.0031, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:44:51,429 INFO [train.py:873] (0/4) Epoch 13, batch 7200, loss[loss=0.1459, simple_loss=0.1676, pruned_loss=0.0621, over 8603.00 frames. ], tot_loss[loss=0.1196, simple_loss=0.1516, pruned_loss=0.04378, over 1955269.50 frames. ], batch size: 100, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:44:54,790 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97949.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:44:55,830 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0630, 2.6148, 3.9431, 2.9525, 3.9187, 3.8190, 3.7354, 3.2784], device='cuda:0'), covar=tensor([0.0972, 0.2737, 0.1084, 0.1877, 0.0769, 0.0843, 0.1243, 0.1722], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0313, 0.0395, 0.0304, 0.0376, 0.0321, 0.0365, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:45:08,583 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97964.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:45:18,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.164e+02 2.816e+02 3.534e+02 1.223e+03, threshold=5.633e+02, percent-clipped=3.0 2022-12-08 04:45:29,648 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3863, 2.4535, 4.3691, 3.0360, 4.1707, 2.1815, 3.4276, 4.1531], device='cuda:0'), covar=tensor([0.0583, 0.3638, 0.0549, 0.6411, 0.0760, 0.3204, 0.1097, 0.0484], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0210, 0.0211, 0.0283, 0.0229, 0.0211, 0.0208, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:45:32,668 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.94 vs. limit=5.0 2022-12-08 04:45:38,771 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:45:43,480 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:07,204 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98030.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 04:46:07,236 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98030.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:20,635 INFO [train.py:873] (0/4) Epoch 13, batch 7300, loss[loss=0.1419, simple_loss=0.1482, pruned_loss=0.06778, over 3918.00 frames. ], tot_loss[loss=0.1186, simple_loss=0.1509, pruned_loss=0.04312, over 2022697.17 frames. ], batch size: 100, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:46:21,865 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98046.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:26,103 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:46:46,648 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.194e+02 2.668e+02 3.260e+02 4.997e+02, threshold=5.336e+02, percent-clipped=0.0 2022-12-08 04:46:49,730 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:47:14,639 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:47:46,638 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.08 vs. limit=5.0 2022-12-08 04:47:48,797 INFO [train.py:873] (0/4) Epoch 13, batch 7400, loss[loss=0.1103, simple_loss=0.133, pruned_loss=0.04377, over 4986.00 frames. ], tot_loss[loss=0.1185, simple_loss=0.1508, pruned_loss=0.0431, over 1978932.98 frames. ], batch size: 100, lr: 5.97e-03, grad_scale: 8.0 2022-12-08 04:48:08,641 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98167.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:48:15,367 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 1.931e+02 2.491e+02 3.118e+02 6.123e+02, threshold=4.982e+02, percent-clipped=1.0 2022-12-08 04:48:15,549 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0872, 2.9432, 3.5858, 2.5405, 2.0630, 3.0882, 1.4611, 3.3248], device='cuda:0'), covar=tensor([0.1205, 0.1119, 0.0765, 0.2330, 0.2765, 0.1159, 0.4508, 0.1071], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0091, 0.0100, 0.0117, 0.0087, 0.0123, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 04:48:28,459 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4763, 1.5614, 4.2860, 1.8467, 4.2207, 4.4582, 3.8721, 4.8472], device='cuda:0'), covar=tensor([0.0200, 0.3200, 0.0310, 0.2390, 0.0359, 0.0391, 0.0469, 0.0128], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0157, 0.0169, 0.0167, 0.0176, 0.0133, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:48:39,322 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2022-12-08 04:48:48,151 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98211.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:08,945 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9190, 1.7544, 3.0922, 2.2629, 2.9412, 1.7904, 2.4931, 2.8979], device='cuda:0'), covar=tensor([0.0983, 0.3990, 0.0660, 0.4119, 0.1023, 0.3268, 0.1139, 0.0802], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0209, 0.0211, 0.0283, 0.0228, 0.0211, 0.0207, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:49:16,623 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98244.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:17,647 INFO [train.py:873] (0/4) Epoch 13, batch 7500, loss[loss=0.1391, simple_loss=0.159, pruned_loss=0.05965, over 6000.00 frames. ], tot_loss[loss=0.1196, simple_loss=0.1517, pruned_loss=0.0438, over 1928332.68 frames. ], batch size: 100, lr: 5.96e-03, grad_scale: 4.0 2022-12-08 04:49:29,607 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:49:44,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.297e+02 2.932e+02 3.500e+02 6.688e+02, threshold=5.864e+02, percent-clipped=4.0 2022-12-08 04:49:52,013 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-08 04:50:03,941 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-13.pt 2022-12-08 04:50:44,931 INFO [train.py:873] (0/4) Epoch 14, batch 0, loss[loss=0.1411, simple_loss=0.1774, pruned_loss=0.05242, over 14207.00 frames. ], tot_loss[loss=0.1411, simple_loss=0.1774, pruned_loss=0.05242, over 14207.00 frames. ], batch size: 89, lr: 5.75e-03, grad_scale: 8.0 2022-12-08 04:50:44,932 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 04:50:49,721 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7337, 4.8085, 4.9560, 4.1804, 4.7724, 5.0932, 2.0264, 4.6298], device='cuda:0'), covar=tensor([0.0128, 0.0164, 0.0253, 0.0273, 0.0195, 0.0088, 0.2812, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0170, 0.0142, 0.0140, 0.0201, 0.0135, 0.0158, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:50:52,246 INFO [train.py:905] (0/4) Epoch 14, validation: loss=0.1389, simple_loss=0.1805, pruned_loss=0.04866, over 857387.00 frames. 2022-12-08 04:50:52,246 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 04:50:59,552 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 04:51:13,036 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98330.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:51:33,432 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 04:51:43,410 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7128, 3.8817, 2.9917, 5.1336, 4.5374, 4.8005, 3.8897, 3.4250], device='cuda:0'), covar=tensor([0.0790, 0.1524, 0.4668, 0.0458, 0.1002, 0.1045, 0.1637, 0.3826], device='cuda:0'), in_proj_covar=tensor([0.0272, 0.0289, 0.0264, 0.0270, 0.0310, 0.0294, 0.0254, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:51:53,529 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.725e+01 2.112e+02 3.078e+02 4.479e+02 1.243e+03, threshold=6.157e+02, percent-clipped=15.0 2022-12-08 04:51:55,465 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98378.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 04:52:20,895 INFO [train.py:873] (0/4) Epoch 14, batch 100, loss[loss=0.1248, simple_loss=0.1315, pruned_loss=0.0591, over 2691.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1523, pruned_loss=0.04287, over 895859.82 frames. ], batch size: 100, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:52:24,335 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98410.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:52:38,931 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0960, 1.0941, 1.2227, 0.9316, 1.0036, 0.8657, 1.0212, 0.8467], device='cuda:0'), covar=tensor([0.0257, 0.0358, 0.0197, 0.0265, 0.0250, 0.0447, 0.0300, 0.0462], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0019, 0.0017, 0.0017, 0.0018, 0.0029, 0.0024, 0.0028], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:53:03,321 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1561, 2.2403, 2.0875, 2.1521, 2.2135, 1.6165, 1.9265, 2.3395], device='cuda:0'), covar=tensor([0.0976, 0.0981, 0.1143, 0.1221, 0.1567, 0.0741, 0.1369, 0.0779], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0032, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 04:53:10,401 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:53:18,362 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:53:22,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.180e+02 2.038e+02 2.657e+02 3.395e+02 7.955e+02, threshold=5.314e+02, percent-clipped=1.0 2022-12-08 04:53:49,904 INFO [train.py:873] (0/4) Epoch 14, batch 200, loss[loss=0.1384, simple_loss=0.1729, pruned_loss=0.05192, over 14520.00 frames. ], tot_loss[loss=0.1198, simple_loss=0.1524, pruned_loss=0.0436, over 1349435.79 frames. ], batch size: 34, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:54:05,281 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1246, 3.4300, 3.0876, 3.2956, 2.6097, 3.3117, 3.1311, 1.9857], device='cuda:0'), covar=tensor([0.1539, 0.0726, 0.1290, 0.0940, 0.1043, 0.1282, 0.1137, 0.2163], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0084, 0.0066, 0.0069, 0.0097, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:54:22,226 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98543.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:54:22,907 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98544.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:54:50,566 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.243e+02 2.774e+02 3.161e+02 6.652e+02, threshold=5.548e+02, percent-clipped=3.0 2022-12-08 04:55:00,683 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0327, 2.0013, 2.0743, 2.0563, 2.0373, 1.7104, 1.3317, 1.8090], device='cuda:0'), covar=tensor([0.0517, 0.0505, 0.0491, 0.0365, 0.0437, 0.1292, 0.2329, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0170, 0.0142, 0.0140, 0.0199, 0.0135, 0.0159, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 04:55:03,806 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98592.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:55:07,528 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4967, 3.9491, 2.9956, 4.9180, 4.2434, 4.5935, 3.9805, 3.2072], device='cuda:0'), covar=tensor([0.0620, 0.1233, 0.3592, 0.0388, 0.1025, 0.1375, 0.1119, 0.3361], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0292, 0.0267, 0.0272, 0.0314, 0.0298, 0.0257, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:55:14,601 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98604.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:55:16,907 INFO [train.py:873] (0/4) Epoch 14, batch 300, loss[loss=0.1515, simple_loss=0.1729, pruned_loss=0.06503, over 9499.00 frames. ], tot_loss[loss=0.1191, simple_loss=0.152, pruned_loss=0.0431, over 1657771.24 frames. ], batch size: 100, lr: 5.74e-03, grad_scale: 8.0 2022-12-08 04:55:28,333 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-08 04:56:01,283 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0268, 3.5709, 2.6329, 4.3661, 3.9288, 4.1996, 3.6392, 3.0162], device='cuda:0'), covar=tensor([0.0819, 0.1365, 0.3704, 0.0721, 0.1053, 0.1477, 0.1183, 0.2923], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0290, 0.0265, 0.0271, 0.0314, 0.0297, 0.0255, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:56:18,354 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.234e+02 2.680e+02 3.385e+02 6.957e+02, threshold=5.360e+02, percent-clipped=3.0 2022-12-08 04:56:45,717 INFO [train.py:873] (0/4) Epoch 14, batch 400, loss[loss=0.1103, simple_loss=0.1423, pruned_loss=0.03909, over 13505.00 frames. ], tot_loss[loss=0.1182, simple_loss=0.1512, pruned_loss=0.04261, over 1817252.55 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 04:56:59,763 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0895, 2.2561, 2.2622, 2.4193, 2.1071, 2.3529, 2.2214, 1.4219], device='cuda:0'), covar=tensor([0.1269, 0.0946, 0.0775, 0.0574, 0.0971, 0.0854, 0.1198, 0.2117], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0084, 0.0065, 0.0069, 0.0096, 0.0082, 0.0097, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 04:57:14,690 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2022-12-08 04:57:33,670 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98762.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:57:36,966 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:57:46,468 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.221e+02 2.679e+02 3.803e+02 8.740e+02, threshold=5.359e+02, percent-clipped=4.0 2022-12-08 04:58:06,929 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2050, 2.1527, 3.1801, 3.3200, 3.1834, 2.2406, 3.1835, 2.5254], device='cuda:0'), covar=tensor([0.0440, 0.1020, 0.0780, 0.0464, 0.0463, 0.1375, 0.0443, 0.0921], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0254, 0.0369, 0.0326, 0.0266, 0.0299, 0.0305, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 04:58:13,695 INFO [train.py:873] (0/4) Epoch 14, batch 500, loss[loss=0.1337, simple_loss=0.1649, pruned_loss=0.05123, over 11171.00 frames. ], tot_loss[loss=0.1195, simple_loss=0.1516, pruned_loss=0.04366, over 1849407.99 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 04:58:16,683 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98810.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:14,610 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-08 04:59:14,755 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 1.935e+02 2.381e+02 2.894e+02 6.356e+02, threshold=4.763e+02, percent-clipped=3.0 2022-12-08 04:59:26,664 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98890.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:31,792 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98895.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:35,163 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 04:59:36,128 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6070, 1.6524, 4.3729, 2.3935, 4.2192, 4.5659, 4.0714, 4.9983], device='cuda:0'), covar=tensor([0.0185, 0.2831, 0.0350, 0.1769, 0.0353, 0.0328, 0.0368, 0.0119], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0158, 0.0161, 0.0172, 0.0169, 0.0178, 0.0136, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 04:59:42,010 INFO [train.py:873] (0/4) Epoch 14, batch 600, loss[loss=0.1844, simple_loss=0.157, pruned_loss=0.1059, over 1245.00 frames. ], tot_loss[loss=0.1196, simple_loss=0.1516, pruned_loss=0.0438, over 1857526.99 frames. ], batch size: 100, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 05:00:20,946 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98951.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:25,411 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98956.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:38,438 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98971.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:00:43,055 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.204e+01 1.954e+02 2.310e+02 3.088e+02 7.406e+02, threshold=4.620e+02, percent-clipped=2.0 2022-12-08 05:01:09,651 INFO [train.py:873] (0/4) Epoch 14, batch 700, loss[loss=0.1105, simple_loss=0.1418, pruned_loss=0.03966, over 13931.00 frames. ], tot_loss[loss=0.1198, simple_loss=0.1517, pruned_loss=0.044, over 1904808.15 frames. ], batch size: 20, lr: 5.73e-03, grad_scale: 8.0 2022-12-08 05:01:32,188 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:01:43,806 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0703, 2.1570, 1.9933, 2.2135, 1.8484, 2.0492, 2.1290, 2.0711], device='cuda:0'), covar=tensor([0.1047, 0.1071, 0.1111, 0.0819, 0.1389, 0.0863, 0.1066, 0.0985], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0136, 0.0144, 0.0155, 0.0144, 0.0120, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:01:48,591 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:01,365 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:09,758 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.198e+02 2.677e+02 3.490e+02 1.171e+03, threshold=5.353e+02, percent-clipped=9.0 2022-12-08 05:02:37,294 INFO [train.py:873] (0/4) Epoch 14, batch 800, loss[loss=0.1285, simple_loss=0.1617, pruned_loss=0.04762, over 14434.00 frames. ], tot_loss[loss=0.1209, simple_loss=0.1525, pruned_loss=0.04463, over 1966200.13 frames. ], batch size: 41, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:02:41,480 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99112.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:43,049 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99114.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:02:46,143 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-08 05:03:16,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2022-12-08 05:03:38,207 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.181e+02 2.713e+02 3.365e+02 5.447e+02, threshold=5.425e+02, percent-clipped=1.0 2022-12-08 05:03:58,127 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99199.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:05,015 INFO [train.py:873] (0/4) Epoch 14, batch 900, loss[loss=0.1496, simple_loss=0.1655, pruned_loss=0.06685, over 7794.00 frames. ], tot_loss[loss=0.1211, simple_loss=0.1522, pruned_loss=0.04499, over 1922498.46 frames. ], batch size: 100, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:04:07,790 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8442, 2.5744, 2.5806, 1.7652, 2.3527, 2.6122, 2.7388, 2.3176], device='cuda:0'), covar=tensor([0.0809, 0.0920, 0.0947, 0.1516, 0.1163, 0.0834, 0.0862, 0.1306], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0177, 0.0139, 0.0126, 0.0139, 0.0150, 0.0127, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 05:04:12,459 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7668, 5.6085, 5.3037, 5.7461, 5.3891, 5.1424, 5.7822, 5.6208], device='cuda:0'), covar=tensor([0.0585, 0.0696, 0.0732, 0.0552, 0.0693, 0.0461, 0.0601, 0.0652], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0137, 0.0145, 0.0155, 0.0144, 0.0120, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:04:24,543 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:39,117 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99246.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:39,937 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:43,443 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99251.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:04:58,907 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:05,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.212e+02 2.795e+02 3.543e+02 9.051e+02, threshold=5.590e+02, percent-clipped=4.0 2022-12-08 05:05:17,587 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99290.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:32,837 INFO [train.py:873] (0/4) Epoch 14, batch 1000, loss[loss=0.1174, simple_loss=0.1547, pruned_loss=0.04004, over 14154.00 frames. ], tot_loss[loss=0.12, simple_loss=0.1518, pruned_loss=0.04408, over 1941877.80 frames. ], batch size: 99, lr: 5.72e-03, grad_scale: 8.0 2022-12-08 05:05:41,528 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1293, 2.0269, 1.8338, 1.9011, 2.0761, 2.0835, 2.1051, 2.0416], device='cuda:0'), covar=tensor([0.1283, 0.0884, 0.2734, 0.2632, 0.1169, 0.1314, 0.1641, 0.1332], device='cuda:0'), in_proj_covar=tensor([0.0377, 0.0264, 0.0437, 0.0553, 0.0333, 0.0437, 0.0385, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:05:47,431 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99324.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:49,880 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99327.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:05:51,625 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:06:32,477 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.112e+02 2.804e+02 3.540e+02 1.187e+03, threshold=5.607e+02, percent-clipped=4.0 2022-12-08 05:06:41,190 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:06:59,881 INFO [train.py:873] (0/4) Epoch 14, batch 1100, loss[loss=0.1156, simple_loss=0.1513, pruned_loss=0.03994, over 14152.00 frames. ], tot_loss[loss=0.1199, simple_loss=0.1518, pruned_loss=0.04401, over 1988337.14 frames. ], batch size: 84, lr: 5.71e-03, grad_scale: 8.0 2022-12-08 05:06:59,987 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:07:12,210 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:07:21,697 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3702, 1.0276, 1.2904, 0.8329, 1.0963, 1.3633, 1.0296, 1.0607], device='cuda:0'), covar=tensor([0.0502, 0.0814, 0.0643, 0.0589, 0.1048, 0.0873, 0.0523, 0.1439], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0175, 0.0137, 0.0125, 0.0139, 0.0148, 0.0126, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 05:07:53,803 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99468.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:08:00,247 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.195e+02 2.512e+02 3.214e+02 5.194e+02, threshold=5.024e+02, percent-clipped=0.0 2022-12-08 05:08:04,837 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99481.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:08:25,772 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 05:08:27,690 INFO [train.py:873] (0/4) Epoch 14, batch 1200, loss[loss=0.1042, simple_loss=0.1497, pruned_loss=0.02937, over 14399.00 frames. ], tot_loss[loss=0.1192, simple_loss=0.1516, pruned_loss=0.04342, over 1988470.67 frames. ], batch size: 41, lr: 5.71e-03, grad_scale: 8.0 2022-12-08 05:08:38,739 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 05:08:41,615 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6441, 2.4648, 2.2350, 2.3362, 2.5270, 2.5764, 2.5671, 2.5875], device='cuda:0'), covar=tensor([0.0999, 0.0856, 0.2515, 0.2576, 0.1101, 0.1154, 0.1363, 0.0896], device='cuda:0'), in_proj_covar=tensor([0.0377, 0.0267, 0.0439, 0.0557, 0.0337, 0.0438, 0.0387, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:08:46,777 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99529.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:08:49,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.44 vs. limit=2.0 2022-12-08 05:09:01,585 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99546.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:05,831 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:20,240 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99568.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:28,000 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.156e+02 2.823e+02 3.405e+02 6.610e+02, threshold=5.645e+02, percent-clipped=3.0 2022-12-08 05:09:35,412 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99585.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:38,971 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99589.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:39,004 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5962, 2.2929, 4.6400, 3.2818, 4.3613, 2.2538, 3.4502, 4.2563], device='cuda:0'), covar=tensor([0.0475, 0.3876, 0.0341, 0.5481, 0.0675, 0.3130, 0.1194, 0.0542], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0209, 0.0213, 0.0282, 0.0228, 0.0208, 0.0206, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:09:43,002 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:47,157 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99599.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:09:54,364 INFO [train.py:873] (0/4) Epoch 14, batch 1300, loss[loss=0.1364, simple_loss=0.1612, pruned_loss=0.05579, over 12006.00 frames. ], tot_loss[loss=0.1189, simple_loss=0.151, pruned_loss=0.04335, over 1939149.07 frames. ], batch size: 100, lr: 5.71e-03, grad_scale: 4.0 2022-12-08 05:10:10,475 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:12,936 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99627.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:14,784 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99629.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:32,914 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:43,807 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0231, 3.5261, 2.7198, 4.2127, 3.9313, 4.0852, 3.5000, 2.7595], device='cuda:0'), covar=tensor([0.0730, 0.1242, 0.3349, 0.0484, 0.0988, 0.1123, 0.1186, 0.3200], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0293, 0.0265, 0.0271, 0.0315, 0.0297, 0.0254, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:10:54,747 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99675.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:56,553 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.185e+02 2.869e+02 3.573e+02 7.828e+02, threshold=5.739e+02, percent-clipped=4.0 2022-12-08 05:10:59,254 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:10:59,317 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:11:23,268 INFO [train.py:873] (0/4) Epoch 14, batch 1400, loss[loss=0.1097, simple_loss=0.1482, pruned_loss=0.0356, over 14192.00 frames. ], tot_loss[loss=0.1195, simple_loss=0.1514, pruned_loss=0.04383, over 1950243.65 frames. ], batch size: 57, lr: 5.71e-03, grad_scale: 4.0 2022-12-08 05:11:23,424 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:11:52,849 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99741.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:12:05,058 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99755.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:12:08,593 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5210, 3.2608, 3.2034, 3.5038, 3.3378, 3.5256, 3.5531, 2.9613], device='cuda:0'), covar=tensor([0.0441, 0.0983, 0.0470, 0.0485, 0.0735, 0.0400, 0.0577, 0.0606], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0270, 0.0190, 0.0189, 0.0183, 0.0153, 0.0278, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 05:12:23,747 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99776.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:12:24,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.204e+02 2.049e+02 2.625e+02 3.154e+02 6.097e+02, threshold=5.251e+02, percent-clipped=2.0 2022-12-08 05:12:50,315 INFO [train.py:873] (0/4) Epoch 14, batch 1500, loss[loss=0.1255, simple_loss=0.1519, pruned_loss=0.04954, over 8602.00 frames. ], tot_loss[loss=0.1174, simple_loss=0.15, pruned_loss=0.0424, over 1970050.78 frames. ], batch size: 100, lr: 5.70e-03, grad_scale: 4.0 2022-12-08 05:13:06,023 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99824.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 05:13:37,710 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9375, 4.6935, 4.6115, 4.9669, 4.5768, 4.3454, 4.9918, 4.8356], device='cuda:0'), covar=tensor([0.0566, 0.0655, 0.0712, 0.0431, 0.0664, 0.0606, 0.0512, 0.0565], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0136, 0.0143, 0.0154, 0.0142, 0.0119, 0.0162, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:13:43,822 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2022-12-08 05:13:51,974 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.709e+01 2.163e+02 2.561e+02 3.336e+02 7.134e+02, threshold=5.122e+02, percent-clipped=2.0 2022-12-08 05:13:58,867 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99885.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:00,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3609, 2.3640, 1.9702, 2.4223, 2.2866, 2.2803, 2.1611, 2.0720], device='cuda:0'), covar=tensor([0.0809, 0.0910, 0.2156, 0.0748, 0.0961, 0.0708, 0.1389, 0.1406], device='cuda:0'), in_proj_covar=tensor([0.0270, 0.0287, 0.0261, 0.0267, 0.0311, 0.0293, 0.0251, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2022-12-08 05:14:03,878 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 05:14:14,576 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1312, 4.2603, 4.2523, 3.8061, 4.1104, 4.5299, 1.5002, 3.9583], device='cuda:0'), covar=tensor([0.0430, 0.0453, 0.0603, 0.0564, 0.0576, 0.0244, 0.3998, 0.0395], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0173, 0.0145, 0.0142, 0.0204, 0.0137, 0.0160, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 05:14:18,360 INFO [train.py:873] (0/4) Epoch 14, batch 1600, loss[loss=0.1369, simple_loss=0.1668, pruned_loss=0.05355, over 14403.00 frames. ], tot_loss[loss=0.1183, simple_loss=0.1506, pruned_loss=0.04306, over 1932295.41 frames. ], batch size: 53, lr: 5.70e-03, grad_scale: 8.0 2022-12-08 05:14:21,824 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1786, 4.8043, 4.6269, 5.1859, 4.8447, 4.3507, 5.1325, 4.3640], device='cuda:0'), covar=tensor([0.0344, 0.0829, 0.0393, 0.0357, 0.0694, 0.0639, 0.0487, 0.0467], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0274, 0.0193, 0.0193, 0.0185, 0.0156, 0.0282, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 05:14:32,771 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:32,816 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:40,500 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:45,442 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2905, 2.2332, 3.1083, 2.5128, 3.1665, 3.0642, 3.0096, 2.7009], device='cuda:0'), covar=tensor([0.0852, 0.2868, 0.1025, 0.1945, 0.0740, 0.0927, 0.1047, 0.1689], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0315, 0.0399, 0.0306, 0.0378, 0.0323, 0.0364, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:14:52,044 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:14:52,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7810, 2.7197, 2.6800, 2.9407, 2.4782, 2.5303, 2.8297, 2.7784], device='cuda:0'), covar=tensor([0.0842, 0.1183, 0.1032, 0.0684, 0.1386, 0.0838, 0.0953, 0.0921], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0136, 0.0142, 0.0154, 0.0142, 0.0119, 0.0162, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:15:08,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4829, 1.9566, 2.6228, 2.6475, 2.5142, 1.9197, 2.6516, 2.1547], device='cuda:0'), covar=tensor([0.0430, 0.1045, 0.0470, 0.0488, 0.0514, 0.1500, 0.0403, 0.0806], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0256, 0.0374, 0.0328, 0.0267, 0.0302, 0.0307, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:15:09,539 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3306, 1.9479, 2.6073, 1.6719, 1.6807, 2.3160, 1.3755, 2.3695], device='cuda:0'), covar=tensor([0.0909, 0.1675, 0.0672, 0.1834, 0.2649, 0.0903, 0.3648, 0.0999], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0099, 0.0091, 0.0098, 0.0115, 0.0088, 0.0123, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 05:15:09,618 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9564, 2.1292, 2.7932, 2.4197, 2.9171, 2.7101, 2.7468, 2.5435], device='cuda:0'), covar=tensor([0.0875, 0.2655, 0.1124, 0.1573, 0.0612, 0.0853, 0.1124, 0.1588], device='cuda:0'), in_proj_covar=tensor([0.0358, 0.0317, 0.0401, 0.0307, 0.0380, 0.0325, 0.0365, 0.0307], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:15:15,692 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:15:20,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.180e+02 2.741e+02 3.315e+02 6.557e+02, threshold=5.482e+02, percent-clipped=1.0 2022-12-08 05:15:23,149 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:15:25,692 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8771, 4.7016, 4.5409, 4.9737, 4.4766, 4.2125, 4.9217, 4.8359], device='cuda:0'), covar=tensor([0.0541, 0.0636, 0.0762, 0.0467, 0.0779, 0.0553, 0.0537, 0.0538], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0135, 0.0141, 0.0152, 0.0141, 0.0118, 0.0161, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:15:41,211 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-100000.pt 2022-12-08 05:15:50,011 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7378, 2.1674, 2.5316, 1.4811, 2.2760, 2.5853, 2.7300, 2.1453], device='cuda:0'), covar=tensor([0.0790, 0.1046, 0.1027, 0.2011, 0.1132, 0.0822, 0.0614, 0.1579], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0174, 0.0136, 0.0125, 0.0138, 0.0149, 0.0127, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0006], device='cuda:0') 2022-12-08 05:15:50,677 INFO [train.py:873] (0/4) Epoch 14, batch 1700, loss[loss=0.1018, simple_loss=0.1431, pruned_loss=0.03029, over 14131.00 frames. ], tot_loss[loss=0.1176, simple_loss=0.1503, pruned_loss=0.04242, over 1951685.26 frames. ], batch size: 99, lr: 5.70e-03, grad_scale: 8.0 2022-12-08 05:15:50,840 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100007.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:16:09,972 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100028.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:11,523 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2022-12-08 05:16:16,925 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:30,341 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6161, 5.3225, 4.8932, 5.2292, 5.1141, 5.5063, 5.5167, 5.5653], device='cuda:0'), covar=tensor([0.0621, 0.0296, 0.1650, 0.2086, 0.0571, 0.0625, 0.0723, 0.0717], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0266, 0.0439, 0.0558, 0.0339, 0.0440, 0.0385, 0.0379], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:16:45,233 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100068.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 05:16:51,806 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100076.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:16:52,530 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.197e+02 2.632e+02 3.389e+02 8.054e+02, threshold=5.264e+02, percent-clipped=5.0 2022-12-08 05:17:19,680 INFO [train.py:873] (0/4) Epoch 14, batch 1800, loss[loss=0.121, simple_loss=0.1545, pruned_loss=0.04376, over 14380.00 frames. ], tot_loss[loss=0.1187, simple_loss=0.1512, pruned_loss=0.0431, over 1945662.75 frames. ], batch size: 73, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:17:34,622 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:17:34,743 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100124.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:18:17,537 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100172.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:18:21,977 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.047e+02 2.482e+02 3.140e+02 5.159e+02, threshold=4.965e+02, percent-clipped=0.0 2022-12-08 05:18:24,808 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4655, 2.3020, 3.1914, 2.5270, 3.2667, 3.1498, 3.0189, 2.6603], device='cuda:0'), covar=tensor([0.0816, 0.2763, 0.1005, 0.1943, 0.0838, 0.1010, 0.1061, 0.1827], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0316, 0.0399, 0.0307, 0.0379, 0.0325, 0.0363, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:18:48,335 INFO [train.py:873] (0/4) Epoch 14, batch 1900, loss[loss=0.1199, simple_loss=0.1534, pruned_loss=0.04315, over 14138.00 frames. ], tot_loss[loss=0.1187, simple_loss=0.1508, pruned_loss=0.04329, over 1882048.69 frames. ], batch size: 84, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:19:03,385 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100224.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:15,304 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:17,359 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 05:19:21,277 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100245.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:45,363 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:19:49,479 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.314e+02 2.818e+02 3.736e+02 1.298e+03, threshold=5.635e+02, percent-clipped=11.0 2022-12-08 05:19:54,015 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3148, 2.5124, 4.3830, 4.4437, 4.3864, 2.5791, 4.4777, 3.5374], device='cuda:0'), covar=tensor([0.0394, 0.1043, 0.0828, 0.0418, 0.0371, 0.1721, 0.0334, 0.0856], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0256, 0.0374, 0.0329, 0.0266, 0.0302, 0.0308, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:20:03,307 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100293.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:20:08,790 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:20:15,746 INFO [train.py:873] (0/4) Epoch 14, batch 2000, loss[loss=0.09688, simple_loss=0.1424, pruned_loss=0.02565, over 13998.00 frames. ], tot_loss[loss=0.1187, simple_loss=0.1509, pruned_loss=0.04324, over 1909305.26 frames. ], batch size: 26, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:20:18,033 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1559, 1.9785, 5.0380, 4.5048, 4.3589, 5.1121, 4.8647, 5.1007], device='cuda:0'), covar=tensor([0.1418, 0.1450, 0.0088, 0.0181, 0.0194, 0.0113, 0.0101, 0.0107], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0160, 0.0128, 0.0168, 0.0146, 0.0139, 0.0123, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 05:20:41,631 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:21:05,105 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100363.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:21:12,877 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 05:21:17,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 1.945e+02 2.468e+02 3.146e+02 6.937e+02, threshold=4.935e+02, percent-clipped=2.0 2022-12-08 05:21:19,867 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2022-12-08 05:21:23,771 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:21:39,868 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100402.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:21:44,076 INFO [train.py:873] (0/4) Epoch 14, batch 2100, loss[loss=0.09313, simple_loss=0.1344, pruned_loss=0.02593, over 14541.00 frames. ], tot_loss[loss=0.1163, simple_loss=0.1494, pruned_loss=0.04164, over 2003964.25 frames. ], batch size: 34, lr: 5.69e-03, grad_scale: 8.0 2022-12-08 05:22:34,504 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100463.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:22:46,710 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.182e+02 2.587e+02 3.120e+02 6.565e+02, threshold=5.173e+02, percent-clipped=5.0 2022-12-08 05:23:13,307 INFO [train.py:873] (0/4) Epoch 14, batch 2200, loss[loss=0.1105, simple_loss=0.1387, pruned_loss=0.04116, over 5997.00 frames. ], tot_loss[loss=0.1175, simple_loss=0.1496, pruned_loss=0.04267, over 1921389.76 frames. ], batch size: 100, lr: 5.68e-03, grad_scale: 8.0 2022-12-08 05:23:23,043 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6025, 1.9364, 3.6766, 2.6506, 3.5429, 1.9172, 2.7250, 3.5405], device='cuda:0'), covar=tensor([0.0728, 0.4324, 0.0620, 0.4799, 0.0837, 0.3368, 0.1523, 0.0597], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0206, 0.0211, 0.0278, 0.0229, 0.0208, 0.0207, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:23:29,419 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-08 05:23:34,151 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8246, 0.8563, 0.7964, 0.8707, 0.8710, 0.3332, 0.7903, 0.8965], device='cuda:0'), covar=tensor([0.0236, 0.0407, 0.0391, 0.0350, 0.0206, 0.0216, 0.0604, 0.0513], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0029, 0.0030, 0.0044, 0.0031, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 05:23:35,930 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4548, 2.8322, 2.6983, 2.8404, 2.3164, 2.8220, 2.6746, 1.5326], device='cuda:0'), covar=tensor([0.1466, 0.0732, 0.0952, 0.0588, 0.1060, 0.0758, 0.1060, 0.2285], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0084, 0.0067, 0.0071, 0.0098, 0.0083, 0.0099, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 05:23:37,594 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9192, 1.1757, 1.9850, 1.3110, 1.9655, 2.0662, 1.6575, 2.1464], device='cuda:0'), covar=tensor([0.0355, 0.2069, 0.0469, 0.1677, 0.0563, 0.0517, 0.1190, 0.0410], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0156, 0.0160, 0.0169, 0.0169, 0.0178, 0.0133, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:24:07,966 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 05:24:15,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 2.000e+02 2.520e+02 3.271e+02 5.030e+02, threshold=5.040e+02, percent-clipped=0.0 2022-12-08 05:24:24,756 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 05:24:30,057 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:24:41,407 INFO [train.py:873] (0/4) Epoch 14, batch 2300, loss[loss=0.1125, simple_loss=0.1269, pruned_loss=0.04903, over 2573.00 frames. ], tot_loss[loss=0.1175, simple_loss=0.1497, pruned_loss=0.04259, over 1928099.63 frames. ], batch size: 100, lr: 5.68e-03, grad_scale: 4.0 2022-12-08 05:25:05,948 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 05:25:30,732 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100663.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:25:31,750 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8492, 4.8753, 5.2271, 4.4404, 5.0332, 5.3089, 2.0671, 4.7594], device='cuda:0'), covar=tensor([0.0252, 0.0303, 0.0325, 0.0555, 0.0308, 0.0124, 0.2974, 0.0245], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0175, 0.0144, 0.0142, 0.0204, 0.0139, 0.0160, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 05:25:44,076 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 1.999e+02 2.526e+02 3.011e+02 4.450e+02, threshold=5.052e+02, percent-clipped=0.0 2022-12-08 05:25:56,116 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 05:26:09,986 INFO [train.py:873] (0/4) Epoch 14, batch 2400, loss[loss=0.09709, simple_loss=0.1423, pruned_loss=0.02595, over 13971.00 frames. ], tot_loss[loss=0.1168, simple_loss=0.1498, pruned_loss=0.04194, over 1939715.63 frames. ], batch size: 19, lr: 5.68e-03, grad_scale: 8.0 2022-12-08 05:26:13,587 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100711.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:26:30,438 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100730.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:26:55,103 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100758.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:27:08,073 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2022-12-08 05:27:13,002 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.043e+02 2.633e+02 3.365e+02 7.380e+02, threshold=5.267e+02, percent-clipped=6.0 2022-12-08 05:27:24,808 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100791.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 05:27:38,941 INFO [train.py:873] (0/4) Epoch 14, batch 2500, loss[loss=0.1389, simple_loss=0.1352, pruned_loss=0.07128, over 2657.00 frames. ], tot_loss[loss=0.1179, simple_loss=0.1504, pruned_loss=0.04271, over 1887115.94 frames. ], batch size: 100, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:28:01,777 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0250, 1.7469, 4.6302, 2.2123, 4.4255, 4.9533, 4.5515, 5.3541], device='cuda:0'), covar=tensor([0.0263, 0.3025, 0.0493, 0.2045, 0.0327, 0.0582, 0.0338, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0157, 0.0160, 0.0169, 0.0169, 0.0178, 0.0133, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:28:41,717 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.248e+02 2.659e+02 3.410e+02 6.665e+02, threshold=5.318e+02, percent-clipped=2.0 2022-12-08 05:28:55,525 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100894.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:29:06,706 INFO [train.py:873] (0/4) Epoch 14, batch 2600, loss[loss=0.107, simple_loss=0.1187, pruned_loss=0.04768, over 1256.00 frames. ], tot_loss[loss=0.1182, simple_loss=0.1506, pruned_loss=0.04294, over 1876858.32 frames. ], batch size: 100, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:29:25,172 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.88 vs. limit=5.0 2022-12-08 05:29:27,460 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9827, 2.0724, 1.9589, 2.1138, 1.7680, 1.9740, 2.0841, 2.0228], device='cuda:0'), covar=tensor([0.0998, 0.1171, 0.1134, 0.0841, 0.1647, 0.0913, 0.1020, 0.0938], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0134, 0.0140, 0.0154, 0.0142, 0.0118, 0.0160, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:29:30,452 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1020, 2.0294, 2.4734, 1.4613, 1.7422, 2.2867, 1.3314, 2.1887], device='cuda:0'), covar=tensor([0.0994, 0.1741, 0.0716, 0.2225, 0.2363, 0.0922, 0.3706, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0091, 0.0099, 0.0116, 0.0088, 0.0122, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 05:29:37,200 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100942.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:30:08,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.168e+02 2.110e+02 2.658e+02 3.162e+02 7.088e+02, threshold=5.316e+02, percent-clipped=3.0 2022-12-08 05:30:34,934 INFO [train.py:873] (0/4) Epoch 14, batch 2700, loss[loss=0.1057, simple_loss=0.1392, pruned_loss=0.03608, over 14232.00 frames. ], tot_loss[loss=0.1175, simple_loss=0.1502, pruned_loss=0.04245, over 1889721.52 frames. ], batch size: 37, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:30:44,180 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3808, 1.6254, 4.1572, 1.8782, 4.1369, 4.3479, 3.7146, 4.7723], device='cuda:0'), covar=tensor([0.0206, 0.2932, 0.0366, 0.2242, 0.0346, 0.0340, 0.0421, 0.0132], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0155, 0.0159, 0.0168, 0.0168, 0.0177, 0.0132, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:31:20,796 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:31:38,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.153e+02 2.205e+02 2.633e+02 3.607e+02 6.784e+02, threshold=5.265e+02, percent-clipped=5.0 2022-12-08 05:31:45,723 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101086.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:32:03,727 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:32:04,578 INFO [train.py:873] (0/4) Epoch 14, batch 2800, loss[loss=0.1086, simple_loss=0.1353, pruned_loss=0.04099, over 4951.00 frames. ], tot_loss[loss=0.1186, simple_loss=0.1513, pruned_loss=0.04301, over 1923522.43 frames. ], batch size: 100, lr: 5.67e-03, grad_scale: 8.0 2022-12-08 05:32:04,773 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101107.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:32:42,198 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-08 05:32:42,719 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3682, 1.4109, 1.5252, 1.5057, 1.6218, 0.9394, 1.2776, 1.3815], device='cuda:0'), covar=tensor([0.0697, 0.0641, 0.0558, 0.0824, 0.0566, 0.0923, 0.0893, 0.0568], device='cuda:0'), in_proj_covar=tensor([0.0029, 0.0029, 0.0032, 0.0028, 0.0029, 0.0043, 0.0030, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 05:32:58,697 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101168.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:33:03,779 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1136, 4.6984, 4.6771, 5.0778, 4.7670, 4.4310, 5.0519, 4.1414], device='cuda:0'), covar=tensor([0.0303, 0.0844, 0.0331, 0.0432, 0.0669, 0.0496, 0.0481, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0269, 0.0190, 0.0189, 0.0180, 0.0152, 0.0279, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 05:33:07,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.080e+02 2.537e+02 3.233e+02 5.412e+02, threshold=5.073e+02, percent-clipped=2.0 2022-12-08 05:33:15,763 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3603, 1.4814, 2.5319, 1.5950, 2.4832, 2.4779, 1.9933, 2.6501], device='cuda:0'), covar=tensor([0.0255, 0.2311, 0.0375, 0.1696, 0.0496, 0.0542, 0.1037, 0.0292], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0156, 0.0159, 0.0169, 0.0169, 0.0178, 0.0132, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:33:33,023 INFO [train.py:873] (0/4) Epoch 14, batch 2900, loss[loss=0.1148, simple_loss=0.149, pruned_loss=0.0403, over 14649.00 frames. ], tot_loss[loss=0.1186, simple_loss=0.1513, pruned_loss=0.04299, over 1952804.16 frames. ], batch size: 33, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:33:35,337 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.75 vs. limit=5.0 2022-12-08 05:34:20,727 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4814, 2.2441, 3.4970, 3.5826, 3.4601, 2.2993, 3.3911, 2.6141], device='cuda:0'), covar=tensor([0.0410, 0.0983, 0.0603, 0.0398, 0.0441, 0.1417, 0.0399, 0.0923], device='cuda:0'), in_proj_covar=tensor([0.0287, 0.0254, 0.0369, 0.0326, 0.0265, 0.0300, 0.0302, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:34:36,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.145e+02 2.615e+02 3.247e+02 7.058e+02, threshold=5.230e+02, percent-clipped=3.0 2022-12-08 05:34:52,392 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 05:34:55,961 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.52 vs. limit=5.0 2022-12-08 05:35:01,745 INFO [train.py:873] (0/4) Epoch 14, batch 3000, loss[loss=0.1862, simple_loss=0.165, pruned_loss=0.1037, over 1175.00 frames. ], tot_loss[loss=0.1183, simple_loss=0.1509, pruned_loss=0.04282, over 1939330.65 frames. ], batch size: 100, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:35:01,746 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 05:35:10,215 INFO [train.py:905] (0/4) Epoch 14, validation: loss=0.134, simple_loss=0.1722, pruned_loss=0.04793, over 857387.00 frames. 2022-12-08 05:35:10,215 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 05:35:32,037 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2022-12-08 05:35:46,254 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1565, 2.1308, 3.0132, 2.3496, 3.0008, 2.8958, 2.8366, 2.5074], device='cuda:0'), covar=tensor([0.0817, 0.2885, 0.0963, 0.2032, 0.0900, 0.1062, 0.1343, 0.1664], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0314, 0.0395, 0.0303, 0.0375, 0.0322, 0.0359, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:36:12,836 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.305e+02 3.036e+02 3.734e+02 6.961e+02, threshold=6.071e+02, percent-clipped=4.0 2022-12-08 05:36:14,586 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7784, 4.5758, 4.4874, 4.8087, 4.4758, 4.2576, 4.8665, 4.6273], device='cuda:0'), covar=tensor([0.0701, 0.0630, 0.0737, 0.0488, 0.0680, 0.0597, 0.0525, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0137, 0.0143, 0.0157, 0.0144, 0.0120, 0.0164, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:36:18,827 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101386.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:36:37,208 INFO [train.py:873] (0/4) Epoch 14, batch 3100, loss[loss=0.1755, simple_loss=0.1634, pruned_loss=0.09379, over 1197.00 frames. ], tot_loss[loss=0.1192, simple_loss=0.1517, pruned_loss=0.0433, over 1982651.72 frames. ], batch size: 100, lr: 5.66e-03, grad_scale: 4.0 2022-12-08 05:37:00,766 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101434.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:37:04,088 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5711, 1.7758, 1.9107, 1.3411, 1.2952, 1.7588, 1.0968, 1.6447], device='cuda:0'), covar=tensor([0.1402, 0.1822, 0.0867, 0.2032, 0.2918, 0.1001, 0.2781, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0099, 0.0092, 0.0099, 0.0116, 0.0087, 0.0122, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 05:37:26,059 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101463.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:37:39,841 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.043e+01 2.198e+02 2.600e+02 3.209e+02 7.473e+02, threshold=5.200e+02, percent-clipped=3.0 2022-12-08 05:37:59,989 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8356, 2.4611, 2.5671, 1.7404, 2.3042, 2.5451, 2.7792, 2.3432], device='cuda:0'), covar=tensor([0.0638, 0.0870, 0.0958, 0.1534, 0.1126, 0.0816, 0.0630, 0.1245], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0177, 0.0137, 0.0127, 0.0139, 0.0151, 0.0128, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 05:38:04,115 INFO [train.py:873] (0/4) Epoch 14, batch 3200, loss[loss=0.1008, simple_loss=0.1427, pruned_loss=0.02945, over 14597.00 frames. ], tot_loss[loss=0.1181, simple_loss=0.151, pruned_loss=0.04258, over 1989650.42 frames. ], batch size: 30, lr: 5.65e-03, grad_scale: 8.0 2022-12-08 05:38:47,052 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 05:39:00,115 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8754, 2.0015, 2.9385, 3.0388, 2.9728, 2.1473, 2.8603, 2.1888], device='cuda:0'), covar=tensor([0.0435, 0.1035, 0.0663, 0.0438, 0.0454, 0.1359, 0.0392, 0.0857], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0254, 0.0372, 0.0327, 0.0266, 0.0299, 0.0303, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:39:08,338 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.131e+02 2.492e+02 3.275e+02 6.158e+02, threshold=4.985e+02, percent-clipped=2.0 2022-12-08 05:39:20,315 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7574, 2.3762, 2.5655, 1.6981, 2.2790, 2.5352, 2.7547, 2.2749], device='cuda:0'), covar=tensor([0.0662, 0.0706, 0.0889, 0.1500, 0.1070, 0.0724, 0.0542, 0.1355], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0177, 0.0137, 0.0127, 0.0140, 0.0152, 0.0128, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 05:39:32,250 INFO [train.py:873] (0/4) Epoch 14, batch 3300, loss[loss=0.09667, simple_loss=0.1395, pruned_loss=0.02692, over 14260.00 frames. ], tot_loss[loss=0.1171, simple_loss=0.15, pruned_loss=0.04214, over 1961341.53 frames. ], batch size: 44, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:39:47,732 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0468, 3.7973, 3.4823, 3.6740, 3.8927, 3.9659, 4.0105, 3.9997], device='cuda:0'), covar=tensor([0.0890, 0.0584, 0.1932, 0.2545, 0.0774, 0.0828, 0.0932, 0.0870], device='cuda:0'), in_proj_covar=tensor([0.0378, 0.0264, 0.0434, 0.0562, 0.0341, 0.0435, 0.0386, 0.0378], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:40:02,740 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 05:40:36,072 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.167e+02 2.269e+02 2.791e+02 3.629e+02 1.218e+03, threshold=5.581e+02, percent-clipped=7.0 2022-12-08 05:40:39,600 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101684.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:40:59,540 INFO [train.py:873] (0/4) Epoch 14, batch 3400, loss[loss=0.1085, simple_loss=0.1429, pruned_loss=0.03705, over 14557.00 frames. ], tot_loss[loss=0.1161, simple_loss=0.1491, pruned_loss=0.04155, over 1948590.48 frames. ], batch size: 34, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:41:03,473 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8489, 2.4534, 3.6692, 2.8274, 3.7236, 3.5837, 3.4793, 3.0853], device='cuda:0'), covar=tensor([0.0818, 0.3000, 0.1107, 0.1881, 0.0830, 0.1019, 0.1459, 0.1807], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0314, 0.0395, 0.0300, 0.0373, 0.0322, 0.0360, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:41:32,727 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101745.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:41:48,189 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:42:02,723 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.178e+02 2.274e+02 2.665e+02 3.242e+02 5.348e+02, threshold=5.329e+02, percent-clipped=0.0 2022-12-08 05:42:26,267 INFO [train.py:873] (0/4) Epoch 14, batch 3500, loss[loss=0.1251, simple_loss=0.1379, pruned_loss=0.05611, over 3876.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1495, pruned_loss=0.04165, over 1997151.82 frames. ], batch size: 100, lr: 5.65e-03, grad_scale: 4.0 2022-12-08 05:42:29,433 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101811.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:43:29,694 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.210e+02 2.811e+02 3.369e+02 8.008e+02, threshold=5.622e+02, percent-clipped=4.0 2022-12-08 05:43:53,151 INFO [train.py:873] (0/4) Epoch 14, batch 3600, loss[loss=0.1386, simple_loss=0.1337, pruned_loss=0.0717, over 1282.00 frames. ], tot_loss[loss=0.1167, simple_loss=0.1497, pruned_loss=0.04185, over 1937194.14 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:44:05,158 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101920.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:44:09,408 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9355, 2.8536, 2.1775, 2.9748, 2.7854, 2.8723, 2.5443, 2.2886], device='cuda:0'), covar=tensor([0.0965, 0.1264, 0.2904, 0.0854, 0.0984, 0.0827, 0.1406, 0.2755], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0294, 0.0262, 0.0272, 0.0318, 0.0296, 0.0259, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:44:19,372 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3998, 2.0307, 2.4636, 2.5076, 2.2627, 1.9952, 2.4600, 2.1891], device='cuda:0'), covar=tensor([0.0451, 0.0871, 0.0519, 0.0436, 0.0603, 0.1078, 0.0449, 0.0588], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0254, 0.0370, 0.0326, 0.0266, 0.0300, 0.0302, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:44:53,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2022-12-08 05:44:57,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.121e+02 2.665e+02 3.565e+02 6.931e+02, threshold=5.329e+02, percent-clipped=3.0 2022-12-08 05:44:58,643 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101981.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:45:03,611 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:45:19,053 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.83 vs. limit=5.0 2022-12-08 05:45:21,975 INFO [train.py:873] (0/4) Epoch 14, batch 3700, loss[loss=0.1186, simple_loss=0.1492, pruned_loss=0.04402, over 10373.00 frames. ], tot_loss[loss=0.1179, simple_loss=0.1502, pruned_loss=0.04276, over 1914818.38 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:45:46,869 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3087, 2.1094, 3.2712, 3.4249, 3.3107, 2.2375, 3.2728, 2.4780], device='cuda:0'), covar=tensor([0.0434, 0.1018, 0.0666, 0.0443, 0.0433, 0.1479, 0.0405, 0.0940], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0257, 0.0374, 0.0329, 0.0269, 0.0303, 0.0305, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:45:50,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2022-12-08 05:45:51,414 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:45:58,301 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102048.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:46:26,400 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.149e+02 2.707e+02 3.533e+02 7.252e+02, threshold=5.415e+02, percent-clipped=4.0 2022-12-08 05:46:27,410 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6853, 4.5398, 4.2788, 4.7794, 4.3716, 3.9660, 4.7716, 4.5973], device='cuda:0'), covar=tensor([0.0689, 0.0757, 0.0869, 0.0572, 0.0763, 0.0724, 0.0587, 0.0680], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0135, 0.0140, 0.0154, 0.0141, 0.0119, 0.0162, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:46:49,971 INFO [train.py:873] (0/4) Epoch 14, batch 3800, loss[loss=0.1183, simple_loss=0.1475, pruned_loss=0.04455, over 14347.00 frames. ], tot_loss[loss=0.117, simple_loss=0.1498, pruned_loss=0.04214, over 1999072.58 frames. ], batch size: 73, lr: 5.64e-03, grad_scale: 8.0 2022-12-08 05:47:40,370 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0362, 3.1510, 3.3146, 3.0891, 3.1435, 2.7701, 1.4016, 3.0051], device='cuda:0'), covar=tensor([0.0439, 0.0434, 0.0395, 0.0464, 0.0403, 0.1062, 0.3311, 0.0331], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0174, 0.0143, 0.0142, 0.0204, 0.0139, 0.0160, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 05:47:56,010 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.049e+02 2.208e+02 2.673e+02 3.606e+02 1.148e+03, threshold=5.346e+02, percent-clipped=3.0 2022-12-08 05:48:01,422 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0404, 2.1432, 1.9626, 2.1824, 1.7938, 1.9063, 2.0915, 2.0871], device='cuda:0'), covar=tensor([0.0980, 0.1170, 0.1173, 0.0884, 0.1296, 0.0988, 0.1153, 0.0969], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0133, 0.0139, 0.0153, 0.0140, 0.0118, 0.0161, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:48:19,966 INFO [train.py:873] (0/4) Epoch 14, batch 3900, loss[loss=0.1304, simple_loss=0.1528, pruned_loss=0.05398, over 7795.00 frames. ], tot_loss[loss=0.1169, simple_loss=0.1496, pruned_loss=0.04208, over 2005058.14 frames. ], batch size: 100, lr: 5.64e-03, grad_scale: 4.0 2022-12-08 05:48:46,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6570, 2.2430, 3.6387, 3.7746, 3.5767, 2.4737, 3.5935, 2.6860], device='cuda:0'), covar=tensor([0.0441, 0.1152, 0.0808, 0.0559, 0.0500, 0.1434, 0.0473, 0.1035], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0258, 0.0373, 0.0328, 0.0268, 0.0303, 0.0305, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:49:08,239 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2022-12-08 05:49:22,111 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102276.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:49:26,437 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 2.029e+02 2.446e+02 2.972e+02 7.280e+02, threshold=4.892e+02, percent-clipped=2.0 2022-12-08 05:49:41,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-08 05:49:49,241 INFO [train.py:873] (0/4) Epoch 14, batch 4000, loss[loss=0.116, simple_loss=0.1367, pruned_loss=0.04764, over 6968.00 frames. ], tot_loss[loss=0.1168, simple_loss=0.15, pruned_loss=0.0418, over 1987057.79 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:50:18,901 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102340.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:50:21,414 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102343.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:50:52,947 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4995, 2.0112, 2.6099, 2.6037, 2.5122, 2.0136, 2.6146, 2.1108], device='cuda:0'), covar=tensor([0.0403, 0.0948, 0.0544, 0.0468, 0.0521, 0.1319, 0.0374, 0.0745], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0258, 0.0374, 0.0328, 0.0268, 0.0303, 0.0307, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:50:55,302 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.024e+02 2.365e+02 2.996e+02 7.946e+02, threshold=4.730e+02, percent-clipped=3.0 2022-12-08 05:51:01,406 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102388.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:51:17,028 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4351, 2.3222, 3.4189, 3.6008, 3.3873, 2.2159, 3.4244, 2.6503], device='cuda:0'), covar=tensor([0.0434, 0.1094, 0.0829, 0.0478, 0.0494, 0.1619, 0.0440, 0.1062], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0258, 0.0374, 0.0328, 0.0268, 0.0304, 0.0307, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:51:18,479 INFO [train.py:873] (0/4) Epoch 14, batch 4100, loss[loss=0.1145, simple_loss=0.1332, pruned_loss=0.04788, over 3848.00 frames. ], tot_loss[loss=0.1173, simple_loss=0.1502, pruned_loss=0.04226, over 1937269.09 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:51:26,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5837, 1.7517, 1.8423, 1.3573, 1.3147, 1.7608, 1.1580, 1.6116], device='cuda:0'), covar=tensor([0.1609, 0.2316, 0.0988, 0.2212, 0.2953, 0.1056, 0.3251, 0.1339], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0091, 0.0099, 0.0117, 0.0088, 0.0122, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 05:51:35,970 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 05:51:44,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2022-12-08 05:52:20,451 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2022-12-08 05:52:23,527 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.525e+01 2.118e+02 2.762e+02 3.343e+02 8.705e+02, threshold=5.523e+02, percent-clipped=3.0 2022-12-08 05:52:46,040 INFO [train.py:873] (0/4) Epoch 14, batch 4200, loss[loss=0.111, simple_loss=0.1398, pruned_loss=0.04108, over 3891.00 frames. ], tot_loss[loss=0.1176, simple_loss=0.1503, pruned_loss=0.04243, over 1939258.39 frames. ], batch size: 100, lr: 5.63e-03, grad_scale: 8.0 2022-12-08 05:52:56,361 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1795, 4.0560, 3.6341, 3.7863, 4.1544, 4.2356, 4.3275, 4.2339], device='cuda:0'), covar=tensor([0.1413, 0.0829, 0.2863, 0.3801, 0.1052, 0.1178, 0.1230, 0.1325], device='cuda:0'), in_proj_covar=tensor([0.0383, 0.0268, 0.0443, 0.0567, 0.0343, 0.0439, 0.0392, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:53:10,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9381, 2.7263, 4.9640, 3.2930, 4.5913, 2.5519, 3.5220, 4.7427], device='cuda:0'), covar=tensor([0.0443, 0.3349, 0.0271, 0.6332, 0.0647, 0.2838, 0.1121, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0205, 0.0210, 0.0279, 0.0229, 0.0208, 0.0204, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:53:15,727 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2022-12-08 05:53:25,842 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9152, 2.3760, 3.9477, 4.0918, 3.9269, 2.3855, 3.9916, 2.9100], device='cuda:0'), covar=tensor([0.0428, 0.1096, 0.0737, 0.0446, 0.0423, 0.1735, 0.0405, 0.1007], device='cuda:0'), in_proj_covar=tensor([0.0289, 0.0257, 0.0372, 0.0326, 0.0267, 0.0302, 0.0305, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:53:45,741 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102576.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 05:53:46,657 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9492, 1.7494, 4.1282, 3.8384, 3.8340, 4.2333, 3.6408, 4.1761], device='cuda:0'), covar=tensor([0.1508, 0.1506, 0.0125, 0.0209, 0.0228, 0.0127, 0.0211, 0.0135], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0161, 0.0131, 0.0169, 0.0148, 0.0142, 0.0124, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 05:53:49,866 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.062e+02 2.234e+02 2.939e+02 3.624e+02 9.310e+02, threshold=5.879e+02, percent-clipped=3.0 2022-12-08 05:53:56,003 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3265, 1.3505, 1.3227, 1.4409, 1.4591, 1.0268, 1.3794, 1.3613], device='cuda:0'), covar=tensor([0.0954, 0.0689, 0.1057, 0.0598, 0.0558, 0.0991, 0.0608, 0.0762], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 05:54:09,183 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7441, 1.4330, 2.8518, 1.5087, 2.9274, 2.8324, 2.1768, 3.0398], device='cuda:0'), covar=tensor([0.0293, 0.2669, 0.0439, 0.2139, 0.0385, 0.0511, 0.1047, 0.0265], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0157, 0.0160, 0.0170, 0.0170, 0.0179, 0.0134, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:54:13,398 INFO [train.py:873] (0/4) Epoch 14, batch 4300, loss[loss=0.1035, simple_loss=0.1447, pruned_loss=0.03113, over 14442.00 frames. ], tot_loss[loss=0.1168, simple_loss=0.1499, pruned_loss=0.04183, over 1994353.26 frames. ], batch size: 51, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:54:27,875 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102624.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 05:54:44,482 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:55:06,209 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 05:55:16,065 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.389e+01 1.915e+02 2.366e+02 2.869e+02 5.820e+02, threshold=4.732e+02, percent-clipped=0.0 2022-12-08 05:55:25,246 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:55:38,795 INFO [train.py:873] (0/4) Epoch 14, batch 4400, loss[loss=0.1253, simple_loss=0.1454, pruned_loss=0.05266, over 4952.00 frames. ], tot_loss[loss=0.1167, simple_loss=0.1497, pruned_loss=0.04191, over 1953756.36 frames. ], batch size: 100, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:55:38,925 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8704, 3.9500, 4.2467, 3.7049, 4.0621, 4.0811, 1.5703, 3.8296], device='cuda:0'), covar=tensor([0.0319, 0.0341, 0.0342, 0.0481, 0.0314, 0.0386, 0.3300, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0173, 0.0142, 0.0141, 0.0203, 0.0138, 0.0158, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 05:55:47,085 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 05:56:01,559 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.53 vs. limit=2.0 2022-12-08 05:56:32,140 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3818, 1.1123, 2.0019, 1.8023, 1.8501, 2.0131, 1.3440, 1.9506], device='cuda:0'), covar=tensor([0.1162, 0.1670, 0.0341, 0.0611, 0.0643, 0.0395, 0.0820, 0.0429], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0130, 0.0168, 0.0146, 0.0141, 0.0123, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 05:56:44,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.029e+02 2.612e+02 3.153e+02 6.081e+02, threshold=5.224e+02, percent-clipped=1.0 2022-12-08 05:57:07,853 INFO [train.py:873] (0/4) Epoch 14, batch 4500, loss[loss=0.1271, simple_loss=0.1314, pruned_loss=0.06145, over 1315.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1495, pruned_loss=0.04159, over 1949430.50 frames. ], batch size: 100, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:57:19,364 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7712, 3.5261, 3.4376, 3.7743, 3.6085, 3.7503, 3.8428, 3.1653], device='cuda:0'), covar=tensor([0.0430, 0.1044, 0.0522, 0.0487, 0.0745, 0.0352, 0.0564, 0.0575], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0272, 0.0191, 0.0190, 0.0181, 0.0154, 0.0280, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 05:57:24,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2022-12-08 05:57:32,271 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0335, 4.7440, 4.5019, 5.0663, 4.5382, 4.2133, 5.0459, 4.8671], device='cuda:0'), covar=tensor([0.0510, 0.0772, 0.0759, 0.0487, 0.0847, 0.0586, 0.0509, 0.0621], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0135, 0.0140, 0.0154, 0.0142, 0.0119, 0.0161, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:57:38,746 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5768, 2.0493, 2.4960, 2.6133, 2.4800, 1.9916, 2.5595, 2.2724], device='cuda:0'), covar=tensor([0.0379, 0.0820, 0.0467, 0.0357, 0.0509, 0.1051, 0.0408, 0.0617], device='cuda:0'), in_proj_covar=tensor([0.0289, 0.0256, 0.0372, 0.0326, 0.0268, 0.0302, 0.0306, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 05:57:46,983 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102851.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:58:12,667 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.344e+02 2.809e+02 3.466e+02 7.090e+02, threshold=5.617e+02, percent-clipped=4.0 2022-12-08 05:58:19,043 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102887.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:58:26,200 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6733, 2.7109, 2.8168, 2.7663, 2.7483, 2.5223, 1.5703, 2.5223], device='cuda:0'), covar=tensor([0.0498, 0.0435, 0.0410, 0.0367, 0.0384, 0.0905, 0.2374, 0.0341], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0173, 0.0144, 0.0141, 0.0204, 0.0139, 0.0159, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 05:58:36,490 INFO [train.py:873] (0/4) Epoch 14, batch 4600, loss[loss=0.148, simple_loss=0.1663, pruned_loss=0.06487, over 11182.00 frames. ], tot_loss[loss=0.1171, simple_loss=0.1502, pruned_loss=0.04207, over 1931270.26 frames. ], batch size: 100, lr: 5.62e-03, grad_scale: 8.0 2022-12-08 05:58:40,764 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102912.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:59:09,206 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6615, 1.8325, 1.8313, 1.7951, 1.8484, 1.1041, 1.5284, 1.7770], device='cuda:0'), covar=tensor([0.0804, 0.0630, 0.0663, 0.0724, 0.0715, 0.0955, 0.0729, 0.0703], device='cuda:0'), in_proj_covar=tensor([0.0030, 0.0030, 0.0033, 0.0028, 0.0030, 0.0043, 0.0031, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 05:59:12,693 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102948.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 05:59:17,739 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8482, 3.5705, 3.3316, 3.5051, 3.7287, 3.7411, 3.8257, 3.8088], device='cuda:0'), covar=tensor([0.0843, 0.0589, 0.1953, 0.2471, 0.0828, 0.0871, 0.0898, 0.0807], device='cuda:0'), in_proj_covar=tensor([0.0385, 0.0268, 0.0440, 0.0563, 0.0344, 0.0438, 0.0390, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:59:26,694 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0943, 1.0429, 1.1556, 1.0733, 0.9811, 0.8020, 0.9482, 0.9169], device='cuda:0'), covar=tensor([0.0198, 0.0210, 0.0180, 0.0194, 0.0248, 0.0383, 0.0246, 0.0388], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0019, 0.0017, 0.0018, 0.0018, 0.0029, 0.0024, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 05:59:29,539 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9933, 4.7223, 4.3443, 4.5598, 4.5400, 4.8223, 4.9460, 4.8947], device='cuda:0'), covar=tensor([0.0618, 0.0439, 0.2027, 0.2478, 0.0810, 0.0711, 0.0828, 0.0825], device='cuda:0'), in_proj_covar=tensor([0.0386, 0.0268, 0.0441, 0.0564, 0.0345, 0.0439, 0.0391, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 05:59:41,555 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 1.979e+02 2.486e+02 3.145e+02 5.919e+02, threshold=4.973e+02, percent-clipped=1.0 2022-12-08 05:59:46,539 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 06:00:04,660 INFO [train.py:873] (0/4) Epoch 14, batch 4700, loss[loss=0.1308, simple_loss=0.1665, pruned_loss=0.04756, over 14154.00 frames. ], tot_loss[loss=0.1175, simple_loss=0.1502, pruned_loss=0.04235, over 1909075.53 frames. ], batch size: 84, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:00:31,853 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2022-12-08 06:01:10,015 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.100e+02 2.622e+02 3.309e+02 6.848e+02, threshold=5.244e+02, percent-clipped=4.0 2022-12-08 06:01:25,295 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2022-12-08 06:01:25,840 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103098.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:01:33,435 INFO [train.py:873] (0/4) Epoch 14, batch 4800, loss[loss=0.1908, simple_loss=0.1619, pruned_loss=0.1099, over 1322.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1493, pruned_loss=0.04174, over 1920094.11 frames. ], batch size: 100, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:01:36,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2022-12-08 06:01:55,812 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:01:58,353 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 06:02:19,419 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103159.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:02:39,048 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.240e+02 2.787e+02 3.398e+02 6.860e+02, threshold=5.574e+02, percent-clipped=4.0 2022-12-08 06:02:49,612 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:02,607 INFO [train.py:873] (0/4) Epoch 14, batch 4900, loss[loss=0.09834, simple_loss=0.142, pruned_loss=0.02736, over 14615.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1492, pruned_loss=0.04182, over 1880305.57 frames. ], batch size: 22, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:03:02,702 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103207.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:02,755 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9121, 1.5236, 3.6350, 3.4004, 3.4459, 3.6582, 3.0687, 3.6863], device='cuda:0'), covar=tensor([0.1609, 0.1686, 0.0139, 0.0286, 0.0290, 0.0157, 0.0294, 0.0144], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0159, 0.0129, 0.0168, 0.0146, 0.0141, 0.0123, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:03:11,261 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1547, 2.0263, 2.0921, 2.1734, 2.0777, 2.0787, 2.2443, 1.8974], device='cuda:0'), covar=tensor([0.1121, 0.1525, 0.0797, 0.0808, 0.1196, 0.0751, 0.0920, 0.0763], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0269, 0.0190, 0.0189, 0.0180, 0.0154, 0.0278, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:03:33,649 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103243.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:03:52,379 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5697, 1.7714, 4.3778, 2.4210, 4.1947, 4.5366, 3.9521, 4.9510], device='cuda:0'), covar=tensor([0.0191, 0.2742, 0.0324, 0.1763, 0.0330, 0.0357, 0.0394, 0.0152], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0155, 0.0159, 0.0169, 0.0169, 0.0179, 0.0132, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:04:06,508 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.061e+02 2.525e+02 3.223e+02 5.765e+02, threshold=5.049e+02, percent-clipped=2.0 2022-12-08 06:04:29,068 INFO [train.py:873] (0/4) Epoch 14, batch 5000, loss[loss=0.1621, simple_loss=0.1483, pruned_loss=0.0879, over 1265.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1495, pruned_loss=0.04191, over 1888152.67 frames. ], batch size: 100, lr: 5.61e-03, grad_scale: 8.0 2022-12-08 06:05:03,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 06:05:20,802 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103366.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:05:33,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.256e+02 2.699e+02 3.304e+02 5.196e+02, threshold=5.398e+02, percent-clipped=1.0 2022-12-08 06:05:56,321 INFO [train.py:873] (0/4) Epoch 14, batch 5100, loss[loss=0.08483, simple_loss=0.1327, pruned_loss=0.01847, over 14138.00 frames. ], tot_loss[loss=0.1169, simple_loss=0.1493, pruned_loss=0.0423, over 1827802.23 frames. ], batch size: 29, lr: 5.60e-03, grad_scale: 4.0 2022-12-08 06:06:13,814 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103427.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 06:06:18,144 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8139, 3.2225, 3.8851, 2.8418, 2.5210, 3.5683, 1.8814, 3.6747], device='cuda:0'), covar=tensor([0.0564, 0.0931, 0.0547, 0.1738, 0.2040, 0.0715, 0.3255, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0099, 0.0090, 0.0098, 0.0115, 0.0087, 0.0119, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:06:37,605 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:07:01,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.364e+02 2.994e+02 3.604e+02 8.056e+02, threshold=5.989e+02, percent-clipped=5.0 2022-12-08 06:07:06,901 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103488.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:07:23,312 INFO [train.py:873] (0/4) Epoch 14, batch 5200, loss[loss=0.1337, simple_loss=0.1298, pruned_loss=0.06874, over 1300.00 frames. ], tot_loss[loss=0.1191, simple_loss=0.1507, pruned_loss=0.04371, over 1812879.86 frames. ], batch size: 100, lr: 5.60e-03, grad_scale: 8.0 2022-12-08 06:07:23,471 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103507.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:07:25,982 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8501, 1.1254, 1.2987, 1.2238, 1.0313, 1.3402, 1.1047, 0.9030], device='cuda:0'), covar=tensor([0.2136, 0.1248, 0.0450, 0.0421, 0.1757, 0.0850, 0.2168, 0.1386], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0086, 0.0068, 0.0072, 0.0098, 0.0084, 0.0099, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:07:31,885 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8376, 3.5474, 3.5045, 3.8293, 3.5864, 3.7833, 3.8554, 3.2599], device='cuda:0'), covar=tensor([0.0430, 0.0992, 0.0454, 0.0446, 0.0806, 0.0368, 0.0531, 0.0519], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0268, 0.0191, 0.0191, 0.0182, 0.0154, 0.0281, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:07:54,996 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103543.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:04,929 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103555.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:28,605 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 2.069e+02 2.548e+02 2.976e+02 7.652e+02, threshold=5.095e+02, percent-clipped=1.0 2022-12-08 06:08:36,186 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:08:50,464 INFO [train.py:873] (0/4) Epoch 14, batch 5300, loss[loss=0.1537, simple_loss=0.1528, pruned_loss=0.07735, over 1207.00 frames. ], tot_loss[loss=0.119, simple_loss=0.1508, pruned_loss=0.04356, over 1859074.53 frames. ], batch size: 100, lr: 5.60e-03, grad_scale: 8.0 2022-12-08 06:09:21,877 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7368, 3.3368, 2.6788, 3.8390, 3.7863, 3.7747, 3.2283, 2.6848], device='cuda:0'), covar=tensor([0.0760, 0.1317, 0.3170, 0.0600, 0.0746, 0.1070, 0.1292, 0.3149], device='cuda:0'), in_proj_covar=tensor([0.0273, 0.0292, 0.0262, 0.0274, 0.0318, 0.0298, 0.0255, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:09:28,425 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 06:09:47,706 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2022-12-08 06:09:55,897 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.008e+02 2.559e+02 3.052e+02 6.405e+02, threshold=5.118e+02, percent-clipped=1.0 2022-12-08 06:10:17,825 INFO [train.py:873] (0/4) Epoch 14, batch 5400, loss[loss=0.125, simple_loss=0.1554, pruned_loss=0.04737, over 14282.00 frames. ], tot_loss[loss=0.1167, simple_loss=0.1499, pruned_loss=0.0417, over 1952497.62 frames. ], batch size: 63, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:10:30,718 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103722.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:10:37,229 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4171, 2.8988, 2.6832, 2.9963, 2.1972, 2.9113, 2.7209, 1.5634], device='cuda:0'), covar=tensor([0.1528, 0.0777, 0.0949, 0.0463, 0.1143, 0.0577, 0.0969, 0.2267], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0086, 0.0068, 0.0072, 0.0098, 0.0084, 0.0099, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:10:47,303 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:10:54,125 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8334, 1.0686, 1.2804, 1.1992, 0.9199, 1.3509, 1.0961, 0.8698], device='cuda:0'), covar=tensor([0.2089, 0.1236, 0.0411, 0.0604, 0.2382, 0.0811, 0.2132, 0.1813], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0087, 0.0068, 0.0072, 0.0098, 0.0085, 0.0100, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:10:59,314 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103754.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:09,121 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:25,001 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.268e+02 2.622e+02 3.465e+02 5.931e+02, threshold=5.245e+02, percent-clipped=4.0 2022-12-08 06:11:29,583 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:41,180 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103801.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:41,938 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103802.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:11:46,663 INFO [train.py:873] (0/4) Epoch 14, batch 5500, loss[loss=0.1081, simple_loss=0.1469, pruned_loss=0.03472, over 14393.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.1494, pruned_loss=0.04093, over 2013589.85 frames. ], batch size: 44, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:11:48,531 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1707, 2.0579, 1.8245, 1.9033, 2.0774, 2.1279, 2.0866, 2.0791], device='cuda:0'), covar=tensor([0.1164, 0.1007, 0.3101, 0.2851, 0.1395, 0.1172, 0.1984, 0.1293], device='cuda:0'), in_proj_covar=tensor([0.0385, 0.0268, 0.0450, 0.0568, 0.0346, 0.0441, 0.0390, 0.0385], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:12:03,746 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103826.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:12,188 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103836.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:47,364 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103876.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:12:53,008 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.148e+02 2.635e+02 3.186e+02 7.478e+02, threshold=5.269e+02, percent-clipped=1.0 2022-12-08 06:12:53,220 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6086, 1.5875, 1.7492, 1.5059, 1.3336, 1.5266, 1.3410, 1.1271], device='cuda:0'), covar=tensor([0.0156, 0.0177, 0.0117, 0.0143, 0.0193, 0.0244, 0.0195, 0.0323], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0019, 0.0017, 0.0018, 0.0018, 0.0030, 0.0024, 0.0029], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:13:05,748 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2022-12-08 06:13:14,657 INFO [train.py:873] (0/4) Epoch 14, batch 5600, loss[loss=0.09541, simple_loss=0.1434, pruned_loss=0.0237, over 14214.00 frames. ], tot_loss[loss=0.1161, simple_loss=0.1494, pruned_loss=0.04136, over 2007823.11 frames. ], batch size: 32, lr: 5.59e-03, grad_scale: 8.0 2022-12-08 06:13:41,241 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103937.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 06:13:46,454 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1055, 3.7320, 2.9091, 4.4060, 4.0925, 4.2449, 3.6876, 2.9329], device='cuda:0'), covar=tensor([0.1060, 0.1384, 0.3758, 0.0560, 0.1482, 0.1273, 0.1268, 0.3786], device='cuda:0'), in_proj_covar=tensor([0.0274, 0.0292, 0.0263, 0.0274, 0.0318, 0.0296, 0.0255, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:14:21,890 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.081e+02 2.071e+02 2.747e+02 3.245e+02 6.809e+02, threshold=5.494e+02, percent-clipped=2.0 2022-12-08 06:14:27,046 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103990.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:14:32,978 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103997.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:14:42,174 INFO [train.py:873] (0/4) Epoch 14, batch 5700, loss[loss=0.1302, simple_loss=0.1639, pruned_loss=0.04829, over 14263.00 frames. ], tot_loss[loss=0.1174, simple_loss=0.1504, pruned_loss=0.04216, over 1988872.66 frames. ], batch size: 63, lr: 5.59e-03, grad_scale: 4.0 2022-12-08 06:14:49,102 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 06:14:55,477 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104022.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:15:18,608 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=104049.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:20,291 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104051.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:26,931 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:15:28,405 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8335, 2.1871, 2.1400, 2.0718, 2.2048, 1.9967, 1.7075, 1.7375], device='cuda:0'), covar=tensor([0.0434, 0.0976, 0.0542, 0.0364, 0.0330, 0.0494, 0.0469, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0017, 0.0018, 0.0019, 0.0030, 0.0024, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:15:36,680 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104070.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 06:15:49,004 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.112e+02 2.664e+02 3.363e+02 7.605e+02, threshold=5.327e+02, percent-clipped=4.0 2022-12-08 06:15:59,741 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104096.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:08,845 INFO [train.py:873] (0/4) Epoch 14, batch 5800, loss[loss=0.1201, simple_loss=0.1257, pruned_loss=0.05729, over 1259.00 frames. ], tot_loss[loss=0.1166, simple_loss=0.1499, pruned_loss=0.04167, over 1996994.24 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 4.0 2022-12-08 06:16:11,525 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:20,648 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104121.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:16:24,780 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3724, 2.3702, 1.9663, 2.3726, 2.2839, 2.3044, 2.1743, 2.0472], device='cuda:0'), covar=tensor([0.0864, 0.0878, 0.1978, 0.0889, 0.1263, 0.0943, 0.1338, 0.1713], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0292, 0.0264, 0.0275, 0.0321, 0.0299, 0.0257, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:17:14,047 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1040, 3.0567, 2.8409, 3.2492, 2.7644, 2.8556, 3.1386, 3.1148], device='cuda:0'), covar=tensor([0.0818, 0.0948, 0.0959, 0.0681, 0.1035, 0.0793, 0.0761, 0.0861], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0139, 0.0144, 0.0159, 0.0145, 0.0122, 0.0167, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:17:16,434 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.209e+02 2.706e+02 3.376e+02 1.574e+03, threshold=5.411e+02, percent-clipped=5.0 2022-12-08 06:17:36,928 INFO [train.py:873] (0/4) Epoch 14, batch 5900, loss[loss=0.1239, simple_loss=0.1589, pruned_loss=0.04448, over 14265.00 frames. ], tot_loss[loss=0.116, simple_loss=0.1493, pruned_loss=0.04136, over 1993831.24 frames. ], batch size: 76, lr: 5.58e-03, grad_scale: 4.0 2022-12-08 06:17:58,450 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104232.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:18:13,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6066, 3.3004, 3.1089, 2.3437, 2.9722, 3.2939, 3.5327, 2.8336], device='cuda:0'), covar=tensor([0.0558, 0.0988, 0.0862, 0.1255, 0.0935, 0.0607, 0.0794, 0.1133], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0176, 0.0138, 0.0127, 0.0140, 0.0152, 0.0129, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:18:19,381 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7797, 2.4406, 3.3240, 2.2179, 2.0029, 2.7983, 1.6811, 2.7411], device='cuda:0'), covar=tensor([0.1248, 0.1347, 0.0548, 0.2057, 0.2432, 0.0962, 0.3429, 0.1094], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0089, 0.0099, 0.0115, 0.0087, 0.0120, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:18:43,106 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.217e+02 2.661e+02 3.155e+02 4.728e+02, threshold=5.321e+02, percent-clipped=0.0 2022-12-08 06:19:04,079 INFO [train.py:873] (0/4) Epoch 14, batch 6000, loss[loss=0.1029, simple_loss=0.1182, pruned_loss=0.0438, over 2650.00 frames. ], tot_loss[loss=0.115, simple_loss=0.1487, pruned_loss=0.04068, over 2020986.98 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 8.0 2022-12-08 06:19:04,080 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 06:19:12,428 INFO [train.py:905] (0/4) Epoch 14, validation: loss=0.1346, simple_loss=0.1729, pruned_loss=0.04809, over 857387.00 frames. 2022-12-08 06:19:12,428 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 06:19:26,907 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=9.91 vs. limit=5.0 2022-12-08 06:19:38,120 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 06:19:45,990 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104346.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:19:52,281 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:18,638 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.150e+02 2.718e+02 3.385e+02 8.730e+02, threshold=5.435e+02, percent-clipped=4.0 2022-12-08 06:20:20,985 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9322, 1.4798, 2.9627, 2.6861, 2.8461, 2.9783, 2.1627, 2.9949], device='cuda:0'), covar=tensor([0.1221, 0.1365, 0.0168, 0.0408, 0.0359, 0.0187, 0.0566, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0130, 0.0170, 0.0146, 0.0142, 0.0123, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:20:21,861 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2731, 3.9870, 3.6780, 3.8881, 4.1059, 4.1211, 4.1912, 4.2009], device='cuda:0'), covar=tensor([0.0853, 0.0567, 0.2276, 0.2779, 0.0758, 0.0930, 0.0971, 0.0810], device='cuda:0'), in_proj_covar=tensor([0.0384, 0.0268, 0.0454, 0.0571, 0.0346, 0.0444, 0.0393, 0.0386], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:20:29,325 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:37,592 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104405.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:20:39,255 INFO [train.py:873] (0/4) Epoch 14, batch 6100, loss[loss=0.1273, simple_loss=0.1283, pruned_loss=0.06311, over 2618.00 frames. ], tot_loss[loss=0.1155, simple_loss=0.1488, pruned_loss=0.04105, over 1968728.44 frames. ], batch size: 100, lr: 5.58e-03, grad_scale: 8.0 2022-12-08 06:20:39,952 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2022-12-08 06:20:51,244 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104421.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:11,705 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104444.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:14,311 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9993, 1.9836, 2.2668, 1.4579, 1.5810, 2.0310, 1.3734, 1.9510], device='cuda:0'), covar=tensor([0.1058, 0.1483, 0.0618, 0.2131, 0.2485, 0.0915, 0.3297, 0.1071], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0099, 0.0090, 0.0099, 0.0116, 0.0087, 0.0120, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:21:32,844 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104469.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:21:46,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.353e+02 3.023e+02 3.584e+02 7.529e+02, threshold=6.047e+02, percent-clipped=6.0 2022-12-08 06:21:49,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9233, 4.0445, 4.2919, 3.6545, 4.1004, 4.2245, 1.7354, 3.8560], device='cuda:0'), covar=tensor([0.0346, 0.0310, 0.0332, 0.0617, 0.0288, 0.0254, 0.2906, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0172, 0.0143, 0.0144, 0.0203, 0.0139, 0.0158, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 06:21:54,202 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9250, 2.4465, 3.8861, 4.0499, 3.8769, 2.3260, 4.0228, 3.0771], device='cuda:0'), covar=tensor([0.0438, 0.1100, 0.0835, 0.0514, 0.0474, 0.1658, 0.0486, 0.0925], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0256, 0.0370, 0.0327, 0.0267, 0.0302, 0.0304, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:21:56,634 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1091, 2.0460, 1.7941, 1.8317, 2.0651, 2.0884, 2.0773, 2.0560], device='cuda:0'), covar=tensor([0.1296, 0.0874, 0.2585, 0.2754, 0.1339, 0.1199, 0.1747, 0.1368], device='cuda:0'), in_proj_covar=tensor([0.0383, 0.0266, 0.0448, 0.0568, 0.0346, 0.0442, 0.0390, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:21:59,997 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9179, 1.2896, 1.9886, 1.2263, 1.9728, 2.0652, 1.7041, 2.1334], device='cuda:0'), covar=tensor([0.0354, 0.2255, 0.0542, 0.2046, 0.0664, 0.0583, 0.1111, 0.0370], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0160, 0.0169, 0.0168, 0.0178, 0.0134, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:22:05,691 INFO [train.py:873] (0/4) Epoch 14, batch 6200, loss[loss=0.1208, simple_loss=0.1606, pruned_loss=0.0405, over 14023.00 frames. ], tot_loss[loss=0.1162, simple_loss=0.149, pruned_loss=0.04164, over 1960376.49 frames. ], batch size: 22, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:22:19,140 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 06:22:27,814 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104532.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 06:23:00,190 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8429, 1.6339, 1.6710, 1.7616, 1.8911, 1.1139, 1.6923, 1.7402], device='cuda:0'), covar=tensor([0.0746, 0.0707, 0.0679, 0.1075, 0.0635, 0.0961, 0.0643, 0.0611], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0031, 0.0034, 0.0029, 0.0031, 0.0044, 0.0032, 0.0035], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:23:09,239 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104580.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:23:12,641 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.150e+02 2.145e+02 2.527e+02 3.084e+02 9.249e+02, threshold=5.054e+02, percent-clipped=2.0 2022-12-08 06:23:26,388 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8412, 0.7616, 0.7707, 0.7916, 0.7341, 0.5247, 0.5936, 0.6529], device='cuda:0'), covar=tensor([0.0162, 0.0185, 0.0143, 0.0169, 0.0167, 0.0339, 0.0216, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0017, 0.0018, 0.0019, 0.0030, 0.0024, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:23:27,263 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4625, 1.4395, 3.5063, 1.5757, 3.3357, 3.5767, 2.3680, 3.7523], device='cuda:0'), covar=tensor([0.0256, 0.3106, 0.0365, 0.2305, 0.0711, 0.0321, 0.0973, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0157, 0.0160, 0.0169, 0.0168, 0.0178, 0.0134, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:23:29,800 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 06:23:33,265 INFO [train.py:873] (0/4) Epoch 14, batch 6300, loss[loss=0.1134, simple_loss=0.1338, pruned_loss=0.04646, over 3866.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.1488, pruned_loss=0.04121, over 2006192.75 frames. ], batch size: 100, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:24:07,384 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:13,525 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:20,382 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4567, 2.2919, 2.6742, 1.5782, 1.7630, 2.4327, 1.3745, 2.3383], device='cuda:0'), covar=tensor([0.1078, 0.1281, 0.0705, 0.2593, 0.2404, 0.0869, 0.3545, 0.1100], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0099, 0.0089, 0.0099, 0.0115, 0.0086, 0.0120, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:24:20,407 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4872, 2.1975, 2.4546, 1.6090, 2.1135, 2.4445, 2.5098, 2.1338], device='cuda:0'), covar=tensor([0.0817, 0.0727, 0.0890, 0.1479, 0.1047, 0.0842, 0.0570, 0.1332], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0175, 0.0137, 0.0126, 0.0140, 0.0151, 0.0128, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:24:40,912 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.535e+01 2.052e+02 2.556e+02 3.229e+02 6.200e+02, threshold=5.113e+02, percent-clipped=1.0 2022-12-08 06:24:49,226 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104694.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:55,238 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104701.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:24:58,722 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104705.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:25:00,314 INFO [train.py:873] (0/4) Epoch 14, batch 6400, loss[loss=0.07939, simple_loss=0.1247, pruned_loss=0.01706, over 14324.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1481, pruned_loss=0.04015, over 2009689.68 frames. ], batch size: 28, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:25:40,657 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104753.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:25:44,364 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8963, 1.8627, 1.6437, 1.9683, 1.8142, 1.8165, 1.7938, 1.6887], device='cuda:0'), covar=tensor([0.1159, 0.1028, 0.2325, 0.0781, 0.1127, 0.0669, 0.1579, 0.0900], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0293, 0.0263, 0.0275, 0.0320, 0.0300, 0.0256, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:26:07,308 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.032e+02 2.617e+02 3.308e+02 7.760e+02, threshold=5.234e+02, percent-clipped=3.0 2022-12-08 06:26:27,855 INFO [train.py:873] (0/4) Epoch 14, batch 6500, loss[loss=0.1278, simple_loss=0.1625, pruned_loss=0.04655, over 14243.00 frames. ], tot_loss[loss=0.1155, simple_loss=0.1494, pruned_loss=0.04085, over 2040468.37 frames. ], batch size: 94, lr: 5.57e-03, grad_scale: 8.0 2022-12-08 06:27:35,547 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.183e+02 2.623e+02 3.627e+02 5.274e+02, threshold=5.247e+02, percent-clipped=1.0 2022-12-08 06:27:54,697 INFO [train.py:873] (0/4) Epoch 14, batch 6600, loss[loss=0.1418, simple_loss=0.1372, pruned_loss=0.07317, over 1254.00 frames. ], tot_loss[loss=0.1158, simple_loss=0.1493, pruned_loss=0.04112, over 1983152.79 frames. ], batch size: 100, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:28:02,698 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 06:28:07,092 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 06:29:02,029 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.257e+02 2.787e+02 3.469e+02 5.148e+02, threshold=5.574e+02, percent-clipped=0.0 2022-12-08 06:29:16,829 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-105000.pt 2022-12-08 06:29:27,170 INFO [train.py:873] (0/4) Epoch 14, batch 6700, loss[loss=0.1144, simple_loss=0.1436, pruned_loss=0.04256, over 6913.00 frames. ], tot_loss[loss=0.1163, simple_loss=0.1495, pruned_loss=0.0416, over 1923810.92 frames. ], batch size: 100, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:30:02,635 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8492, 1.5613, 2.0817, 1.6247, 1.9983, 1.3764, 1.6640, 1.9307], device='cuda:0'), covar=tensor([0.2360, 0.2099, 0.0525, 0.1180, 0.0978, 0.1451, 0.1073, 0.0788], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0204, 0.0212, 0.0275, 0.0228, 0.0209, 0.0204, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:30:11,190 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9455, 1.5278, 2.9891, 2.7275, 2.8541, 2.9896, 2.1877, 2.9961], device='cuda:0'), covar=tensor([0.1167, 0.1320, 0.0182, 0.0366, 0.0376, 0.0186, 0.0519, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0130, 0.0169, 0.0146, 0.0141, 0.0123, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:30:12,650 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6697, 4.4474, 4.2813, 4.7411, 4.3387, 3.9203, 4.7081, 4.5203], device='cuda:0'), covar=tensor([0.0650, 0.0881, 0.0874, 0.0616, 0.0790, 0.0859, 0.0696, 0.0808], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0139, 0.0143, 0.0157, 0.0144, 0.0123, 0.0166, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:30:33,378 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6488, 3.7596, 4.0302, 3.4123, 3.8185, 3.8874, 1.5404, 3.6010], device='cuda:0'), covar=tensor([0.0338, 0.0355, 0.0316, 0.0557, 0.0349, 0.0429, 0.3179, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0174, 0.0145, 0.0145, 0.0205, 0.0140, 0.0159, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 06:30:35,610 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.155e+02 2.089e+02 2.481e+02 3.177e+02 6.321e+02, threshold=4.962e+02, percent-clipped=1.0 2022-12-08 06:30:54,206 INFO [train.py:873] (0/4) Epoch 14, batch 6800, loss[loss=0.137, simple_loss=0.1593, pruned_loss=0.05736, over 4942.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1491, pruned_loss=0.04186, over 1894592.73 frames. ], batch size: 100, lr: 5.56e-03, grad_scale: 8.0 2022-12-08 06:31:06,562 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2022-12-08 06:31:20,214 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0793, 1.2839, 1.3364, 0.9256, 0.9010, 1.1299, 0.8874, 1.1225], device='cuda:0'), covar=tensor([0.1964, 0.3197, 0.1452, 0.2668, 0.3380, 0.1400, 0.2545, 0.1364], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0091, 0.0101, 0.0117, 0.0088, 0.0122, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:31:53,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.96 vs. limit=5.0 2022-12-08 06:32:01,692 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.161e+02 2.579e+02 3.505e+02 6.694e+02, threshold=5.159e+02, percent-clipped=6.0 2022-12-08 06:32:21,681 INFO [train.py:873] (0/4) Epoch 14, batch 6900, loss[loss=0.1195, simple_loss=0.1609, pruned_loss=0.03901, over 14466.00 frames. ], tot_loss[loss=0.1171, simple_loss=0.1496, pruned_loss=0.04227, over 1921053.77 frames. ], batch size: 51, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:32:31,832 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6578, 3.4254, 3.3183, 3.7433, 3.3714, 3.3739, 3.6857, 3.5809], device='cuda:0'), covar=tensor([0.0664, 0.0980, 0.0897, 0.0643, 0.0865, 0.0670, 0.0759, 0.0820], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0138, 0.0142, 0.0157, 0.0143, 0.0122, 0.0165, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:32:56,679 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2022-12-08 06:33:29,082 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.377e+02 2.807e+02 3.459e+02 9.593e+02, threshold=5.615e+02, percent-clipped=4.0 2022-12-08 06:33:44,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9709, 1.8606, 4.3450, 4.0251, 3.9985, 4.4055, 3.9260, 4.4311], device='cuda:0'), covar=tensor([0.1489, 0.1447, 0.0101, 0.0205, 0.0220, 0.0114, 0.0170, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0130, 0.0171, 0.0147, 0.0141, 0.0124, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:33:48,170 INFO [train.py:873] (0/4) Epoch 14, batch 7000, loss[loss=0.12, simple_loss=0.1525, pruned_loss=0.04379, over 12753.00 frames. ], tot_loss[loss=0.1157, simple_loss=0.149, pruned_loss=0.04124, over 1973863.27 frames. ], batch size: 100, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:34:07,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1167, 2.1607, 2.2117, 2.4589, 1.9700, 2.1942, 1.9957, 2.2172], device='cuda:0'), covar=tensor([0.0632, 0.0754, 0.0363, 0.0316, 0.0432, 0.0367, 0.0465, 0.0530], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:34:40,423 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-08 06:34:56,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.288e+01 2.265e+02 2.868e+02 3.572e+02 1.076e+03, threshold=5.736e+02, percent-clipped=4.0 2022-12-08 06:34:59,503 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105388.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:35:06,844 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4711, 1.4022, 3.4785, 1.6408, 3.3502, 3.5432, 2.4907, 3.8057], device='cuda:0'), covar=tensor([0.0256, 0.3042, 0.0440, 0.2141, 0.0730, 0.0410, 0.0947, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0155, 0.0158, 0.0167, 0.0166, 0.0175, 0.0133, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:35:08,521 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2147, 3.0142, 3.0301, 3.2644, 3.0906, 3.1136, 3.3081, 2.7555], device='cuda:0'), covar=tensor([0.0624, 0.1117, 0.0584, 0.0576, 0.0865, 0.0616, 0.0663, 0.0692], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0270, 0.0193, 0.0191, 0.0183, 0.0155, 0.0283, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:35:16,755 INFO [train.py:873] (0/4) Epoch 14, batch 7100, loss[loss=0.1136, simple_loss=0.1552, pruned_loss=0.03598, over 14299.00 frames. ], tot_loss[loss=0.1156, simple_loss=0.1488, pruned_loss=0.04114, over 1907453.64 frames. ], batch size: 31, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:35:19,654 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0668, 1.9938, 2.0680, 2.1601, 2.0670, 1.9434, 2.1947, 1.8222], device='cuda:0'), covar=tensor([0.1164, 0.1277, 0.0717, 0.0802, 0.1017, 0.0970, 0.0871, 0.0804], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0269, 0.0193, 0.0190, 0.0182, 0.0154, 0.0282, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:35:53,778 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105449.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:36:10,135 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.39 vs. limit=5.0 2022-12-08 06:36:25,709 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.061e+02 2.137e+02 2.673e+02 3.373e+02 4.675e+02, threshold=5.347e+02, percent-clipped=1.0 2022-12-08 06:36:25,865 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105486.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:36:39,552 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105502.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:36:43,629 INFO [train.py:873] (0/4) Epoch 14, batch 7200, loss[loss=0.1195, simple_loss=0.1428, pruned_loss=0.0481, over 5940.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1494, pruned_loss=0.04173, over 1912388.88 frames. ], batch size: 100, lr: 5.55e-03, grad_scale: 8.0 2022-12-08 06:36:56,364 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3372, 4.2415, 4.1142, 4.4404, 4.0802, 3.8699, 4.4339, 4.2134], device='cuda:0'), covar=tensor([0.0665, 0.0842, 0.0834, 0.0588, 0.0797, 0.0719, 0.0662, 0.0821], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0140, 0.0144, 0.0158, 0.0145, 0.0122, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:36:57,909 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.27 vs. limit=5.0 2022-12-08 06:37:13,614 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0487, 1.2786, 1.4057, 1.0421, 0.9196, 1.2144, 0.9551, 1.3061], device='cuda:0'), covar=tensor([0.2225, 0.3181, 0.0996, 0.2487, 0.2881, 0.1099, 0.1754, 0.1052], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0090, 0.0099, 0.0117, 0.0088, 0.0121, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:37:17,882 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105547.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:37:32,154 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105563.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:37:51,600 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.276e+02 2.781e+02 3.572e+02 7.841e+02, threshold=5.562e+02, percent-clipped=2.0 2022-12-08 06:38:10,274 INFO [train.py:873] (0/4) Epoch 14, batch 7300, loss[loss=0.108, simple_loss=0.1449, pruned_loss=0.03552, over 14400.00 frames. ], tot_loss[loss=0.1154, simple_loss=0.1487, pruned_loss=0.04109, over 1974701.01 frames. ], batch size: 41, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:38:54,901 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9981, 1.1560, 0.9856, 1.0054, 1.1237, 0.7167, 0.9375, 1.0895], device='cuda:0'), covar=tensor([0.0548, 0.0494, 0.0501, 0.0427, 0.0449, 0.0417, 0.1043, 0.0751], device='cuda:0'), in_proj_covar=tensor([0.0032, 0.0032, 0.0034, 0.0029, 0.0031, 0.0044, 0.0032, 0.0035], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 06:38:56,664 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4144, 2.6345, 2.6731, 2.7346, 2.2644, 2.7290, 2.6307, 1.5647], device='cuda:0'), covar=tensor([0.1156, 0.1011, 0.0686, 0.0560, 0.0908, 0.0482, 0.0865, 0.1874], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0085, 0.0067, 0.0070, 0.0095, 0.0082, 0.0097, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:39:20,010 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.190e+02 2.577e+02 3.146e+02 6.129e+02, threshold=5.153e+02, percent-clipped=1.0 2022-12-08 06:39:38,185 INFO [train.py:873] (0/4) Epoch 14, batch 7400, loss[loss=0.1121, simple_loss=0.1498, pruned_loss=0.03715, over 14264.00 frames. ], tot_loss[loss=0.1164, simple_loss=0.1495, pruned_loss=0.04162, over 1972984.66 frames. ], batch size: 76, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:39:39,081 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5391, 1.8034, 1.8961, 1.3521, 1.2804, 1.7045, 1.3045, 1.6594], device='cuda:0'), covar=tensor([0.1545, 0.2066, 0.0880, 0.2066, 0.2765, 0.1163, 0.2885, 0.1157], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0098, 0.0089, 0.0098, 0.0116, 0.0086, 0.0120, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 06:39:53,908 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-08 06:40:11,113 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105744.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 06:40:48,154 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.057e+02 2.114e+02 2.694e+02 3.325e+02 8.535e+02, threshold=5.387e+02, percent-clipped=3.0 2022-12-08 06:40:53,295 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105792.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:40:59,940 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105799.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:07,345 INFO [train.py:873] (0/4) Epoch 14, batch 7500, loss[loss=0.1112, simple_loss=0.1516, pruned_loss=0.03541, over 14265.00 frames. ], tot_loss[loss=0.1168, simple_loss=0.1496, pruned_loss=0.04204, over 1987655.00 frames. ], batch size: 80, lr: 5.54e-03, grad_scale: 8.0 2022-12-08 06:41:37,011 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105842.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:44,730 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105853.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:46,902 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 06:41:47,026 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105858.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:48,039 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105860.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:41:53,644 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-14.pt 2022-12-08 06:42:35,201 INFO [train.py:873] (0/4) Epoch 15, batch 0, loss[loss=0.1447, simple_loss=0.169, pruned_loss=0.06023, over 7761.00 frames. ], tot_loss[loss=0.1447, simple_loss=0.169, pruned_loss=0.06023, over 7761.00 frames. ], batch size: 100, lr: 5.35e-03, grad_scale: 8.0 2022-12-08 06:42:35,202 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 06:42:39,187 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1125, 3.8428, 3.7021, 3.0816, 3.5812, 3.7482, 4.3229, 3.5265], device='cuda:0'), covar=tensor([0.0493, 0.1105, 0.0659, 0.1098, 0.0800, 0.0539, 0.0521, 0.0899], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0175, 0.0137, 0.0125, 0.0139, 0.0151, 0.0128, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:42:42,694 INFO [train.py:905] (0/4) Epoch 15, validation: loss=0.1381, simple_loss=0.1782, pruned_loss=0.049, over 857387.00 frames. 2022-12-08 06:42:42,694 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 06:42:58,146 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.674e+01 2.091e+02 3.002e+02 4.364e+02 1.099e+03, threshold=6.004e+02, percent-clipped=13.0 2022-12-08 06:42:59,262 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4164, 1.9253, 2.3034, 1.4829, 2.0116, 2.2134, 2.3782, 2.0352], device='cuda:0'), covar=tensor([0.0880, 0.0838, 0.0967, 0.1585, 0.1294, 0.0775, 0.0719, 0.1560], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0175, 0.0138, 0.0126, 0.0140, 0.0152, 0.0129, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:43:45,358 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0353, 2.1001, 1.9680, 2.1454, 1.7472, 1.9449, 2.1309, 2.0867], device='cuda:0'), covar=tensor([0.0891, 0.1207, 0.1199, 0.0943, 0.1615, 0.0974, 0.0922, 0.0923], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0140, 0.0145, 0.0158, 0.0146, 0.0123, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:44:11,360 INFO [train.py:873] (0/4) Epoch 15, batch 100, loss[loss=0.1068, simple_loss=0.1462, pruned_loss=0.03369, over 14314.00 frames. ], tot_loss[loss=0.1154, simple_loss=0.1505, pruned_loss=0.04019, over 867951.81 frames. ], batch size: 28, lr: 5.35e-03, grad_scale: 8.0 2022-12-08 06:44:26,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.570e+02 2.979e+02 3.765e+02 9.423e+02, threshold=5.958e+02, percent-clipped=4.0 2022-12-08 06:45:18,031 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106044.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:45:40,385 INFO [train.py:873] (0/4) Epoch 15, batch 200, loss[loss=0.1073, simple_loss=0.1506, pruned_loss=0.03198, over 13890.00 frames. ], tot_loss[loss=0.1155, simple_loss=0.1499, pruned_loss=0.04059, over 1332811.61 frames. ], batch size: 23, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:45:47,347 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106077.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:45:55,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.218e+02 2.817e+02 3.547e+02 5.569e+02, threshold=5.634e+02, percent-clipped=0.0 2022-12-08 06:46:00,773 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:41,869 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106138.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:45,013 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:48,436 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8207, 1.4155, 2.9992, 2.6898, 2.8740, 2.9986, 2.2446, 3.0033], device='cuda:0'), covar=tensor([0.1229, 0.1475, 0.0162, 0.0394, 0.0351, 0.0181, 0.0499, 0.0203], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0160, 0.0131, 0.0172, 0.0148, 0.0143, 0.0124, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:46:50,467 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106148.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:56,475 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106155.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:46:59,134 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:47:08,665 INFO [train.py:873] (0/4) Epoch 15, batch 300, loss[loss=0.1083, simple_loss=0.1405, pruned_loss=0.03809, over 6908.00 frames. ], tot_loss[loss=0.1154, simple_loss=0.1491, pruned_loss=0.04088, over 1594422.41 frames. ], batch size: 100, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:47:08,776 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5131, 4.1452, 3.9981, 4.5446, 4.2496, 4.0729, 4.5593, 3.7517], device='cuda:0'), covar=tensor([0.0431, 0.1046, 0.0440, 0.0428, 0.0765, 0.0831, 0.0527, 0.0535], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0268, 0.0192, 0.0189, 0.0181, 0.0152, 0.0278, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:47:20,434 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4023, 0.9909, 1.1770, 0.8115, 1.1085, 1.3743, 1.0844, 1.1087], device='cuda:0'), covar=tensor([0.0449, 0.1004, 0.0730, 0.0551, 0.1134, 0.0767, 0.0615, 0.1206], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0176, 0.0138, 0.0126, 0.0141, 0.0153, 0.0129, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:47:23,718 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.130e+02 2.529e+02 3.253e+02 6.005e+02, threshold=5.059e+02, percent-clipped=1.0 2022-12-08 06:47:27,432 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106190.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:47:33,811 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7561, 1.9751, 3.7969, 2.6037, 3.6554, 1.9210, 2.7772, 3.7140], device='cuda:0'), covar=tensor([0.0582, 0.4105, 0.0498, 0.5488, 0.0702, 0.3478, 0.1282, 0.0419], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0207, 0.0212, 0.0277, 0.0230, 0.0211, 0.0205, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:47:42,121 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:48:14,606 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9504, 5.3705, 5.3707, 5.8776, 5.5227, 4.8294, 5.8360, 4.9220], device='cuda:0'), covar=tensor([0.0318, 0.0841, 0.0365, 0.0380, 0.0743, 0.0380, 0.0394, 0.0440], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0267, 0.0191, 0.0189, 0.0180, 0.0151, 0.0278, 0.0164], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:48:37,051 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:48:37,735 INFO [train.py:873] (0/4) Epoch 15, batch 400, loss[loss=0.1198, simple_loss=0.1475, pruned_loss=0.04611, over 14634.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.1491, pruned_loss=0.04061, over 1800864.19 frames. ], batch size: 33, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:48:40,494 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 06:48:52,653 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.076e+02 2.571e+02 3.282e+02 8.496e+02, threshold=5.142e+02, percent-clipped=2.0 2022-12-08 06:49:28,446 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1693, 2.0500, 2.1308, 2.2086, 2.1417, 2.1321, 2.2827, 1.8094], device='cuda:0'), covar=tensor([0.1015, 0.1346, 0.0714, 0.0877, 0.1082, 0.0712, 0.0835, 0.0943], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0269, 0.0193, 0.0190, 0.0181, 0.0151, 0.0280, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 06:49:30,168 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:49:36,168 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106336.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:49:37,090 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5358, 2.2607, 4.5166, 3.0844, 4.2508, 2.0998, 3.2082, 4.3508], device='cuda:0'), covar=tensor([0.0594, 0.4097, 0.0426, 0.6086, 0.0655, 0.3473, 0.1223, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0207, 0.0212, 0.0277, 0.0227, 0.0210, 0.0204, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 06:50:06,075 INFO [train.py:873] (0/4) Epoch 15, batch 500, loss[loss=0.101, simple_loss=0.1432, pruned_loss=0.02936, over 14023.00 frames. ], tot_loss[loss=0.1153, simple_loss=0.1492, pruned_loss=0.04075, over 1893996.61 frames. ], batch size: 22, lr: 5.34e-03, grad_scale: 8.0 2022-12-08 06:50:21,314 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.064e+02 2.651e+02 3.398e+02 5.648e+02, threshold=5.301e+02, percent-clipped=3.0 2022-12-08 06:50:31,300 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106397.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:03,961 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106433.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:17,617 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106448.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:22,869 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106454.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:23,728 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106455.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:51:36,097 INFO [train.py:873] (0/4) Epoch 15, batch 600, loss[loss=0.1029, simple_loss=0.1397, pruned_loss=0.03302, over 11978.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.149, pruned_loss=0.04071, over 1965288.69 frames. ], batch size: 100, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:51:51,214 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.156e+02 2.802e+02 3.401e+02 5.738e+02, threshold=5.604e+02, percent-clipped=2.0 2022-12-08 06:51:57,600 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0588, 1.5951, 2.0652, 1.3742, 1.7318, 2.0559, 1.9065, 1.8488], device='cuda:0'), covar=tensor([0.1007, 0.0903, 0.0965, 0.1483, 0.1621, 0.1018, 0.0939, 0.1724], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0178, 0.0139, 0.0128, 0.0143, 0.0155, 0.0132, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:52:00,111 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106496.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:52:06,422 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106503.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:52:17,405 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106515.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:53:05,159 INFO [train.py:873] (0/4) Epoch 15, batch 700, loss[loss=0.1548, simple_loss=0.1446, pruned_loss=0.08246, over 2565.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.1487, pruned_loss=0.04081, over 1911212.40 frames. ], batch size: 100, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:53:19,980 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.992e+01 2.009e+02 2.731e+02 3.273e+02 5.771e+02, threshold=5.463e+02, percent-clipped=3.0 2022-12-08 06:53:53,090 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106624.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:54:25,182 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1532, 1.3759, 1.6958, 1.6081, 1.5537, 1.5865, 1.3565, 1.3100], device='cuda:0'), covar=tensor([0.1693, 0.1339, 0.0369, 0.0713, 0.1228, 0.1072, 0.1955, 0.1882], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0085, 0.0067, 0.0071, 0.0096, 0.0083, 0.0098, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0006, 0.0006], device='cuda:0') 2022-12-08 06:54:32,758 INFO [train.py:873] (0/4) Epoch 15, batch 800, loss[loss=0.1269, simple_loss=0.1587, pruned_loss=0.04756, over 14263.00 frames. ], tot_loss[loss=0.1151, simple_loss=0.1483, pruned_loss=0.04096, over 1902389.07 frames. ], batch size: 63, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:54:39,275 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5551, 1.0622, 2.0017, 1.7875, 1.9186, 2.0330, 1.3859, 2.0591], device='cuda:0'), covar=tensor([0.0749, 0.1445, 0.0259, 0.0551, 0.0574, 0.0306, 0.0741, 0.0269], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0130, 0.0170, 0.0148, 0.0142, 0.0124, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 06:54:47,971 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.230e+02 2.748e+02 3.448e+02 5.498e+02, threshold=5.496e+02, percent-clipped=1.0 2022-12-08 06:54:49,559 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.48 vs. limit=5.0 2022-12-08 06:54:53,532 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:55:05,690 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1009, 3.8865, 3.8272, 4.1773, 3.7057, 3.4156, 4.2229, 4.0366], device='cuda:0'), covar=tensor([0.0772, 0.0911, 0.0852, 0.0645, 0.0810, 0.0770, 0.0604, 0.0837], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0138, 0.0142, 0.0155, 0.0144, 0.0121, 0.0164, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:55:30,002 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106733.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:55:32,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3107, 2.1126, 3.2566, 3.4009, 3.2708, 2.2798, 3.2662, 2.4564], device='cuda:0'), covar=tensor([0.0476, 0.1130, 0.0774, 0.0512, 0.0511, 0.1525, 0.0540, 0.1066], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0256, 0.0371, 0.0327, 0.0268, 0.0301, 0.0305, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 06:56:02,202 INFO [train.py:873] (0/4) Epoch 15, batch 900, loss[loss=0.09909, simple_loss=0.1203, pruned_loss=0.03894, over 3879.00 frames. ], tot_loss[loss=0.1151, simple_loss=0.1478, pruned_loss=0.04119, over 1832074.75 frames. ], batch size: 100, lr: 5.33e-03, grad_scale: 8.0 2022-12-08 06:56:12,629 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:56:16,790 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 2.159e+02 2.563e+02 3.204e+02 7.042e+02, threshold=5.126e+02, percent-clipped=4.0 2022-12-08 06:56:38,457 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106810.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:56:53,839 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 06:57:10,274 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2022-12-08 06:57:30,125 INFO [train.py:873] (0/4) Epoch 15, batch 1000, loss[loss=0.1098, simple_loss=0.1488, pruned_loss=0.03541, over 14390.00 frames. ], tot_loss[loss=0.1145, simple_loss=0.1476, pruned_loss=0.04074, over 1869566.39 frames. ], batch size: 53, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 06:57:44,838 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.224e+02 2.651e+02 3.456e+02 5.513e+02, threshold=5.301e+02, percent-clipped=2.0 2022-12-08 06:58:18,766 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106924.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:58:58,495 INFO [train.py:873] (0/4) Epoch 15, batch 1100, loss[loss=0.1138, simple_loss=0.1475, pruned_loss=0.04006, over 6945.00 frames. ], tot_loss[loss=0.1139, simple_loss=0.1475, pruned_loss=0.04012, over 1939025.53 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 06:59:01,481 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:59:13,542 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 2.263e+02 2.855e+02 3.604e+02 7.949e+02, threshold=5.710e+02, percent-clipped=5.0 2022-12-08 06:59:19,133 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106992.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 06:59:38,410 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6002, 1.7910, 1.8688, 1.3466, 1.3174, 1.6408, 1.1886, 1.6526], device='cuda:0'), covar=tensor([0.1418, 0.1942, 0.0764, 0.2184, 0.2490, 0.1071, 0.2707, 0.1042], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0091, 0.0098, 0.0116, 0.0088, 0.0120, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 07:00:02,193 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:00:28,132 INFO [train.py:873] (0/4) Epoch 15, batch 1200, loss[loss=0.1892, simple_loss=0.1625, pruned_loss=0.108, over 1248.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1473, pruned_loss=0.03988, over 2029430.42 frames. ], batch size: 100, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 07:00:42,788 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 2.082e+02 2.691e+02 3.251e+02 9.387e+02, threshold=5.382e+02, percent-clipped=1.0 2022-12-08 07:00:55,225 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1286, 1.9329, 1.9862, 2.2753, 1.8454, 2.1075, 1.9110, 2.0183], device='cuda:0'), covar=tensor([0.0338, 0.1633, 0.0512, 0.0315, 0.0297, 0.0503, 0.0557, 0.0520], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:01:04,311 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:30,762 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107140.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:01:41,332 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107152.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:46,369 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:01:56,715 INFO [train.py:873] (0/4) Epoch 15, batch 1300, loss[loss=0.104, simple_loss=0.1436, pruned_loss=0.03222, over 14176.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1479, pruned_loss=0.04023, over 2005100.03 frames. ], batch size: 35, lr: 5.32e-03, grad_scale: 8.0 2022-12-08 07:02:12,112 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.125e+02 2.037e+02 2.513e+02 3.265e+02 6.018e+02, threshold=5.026e+02, percent-clipped=1.0 2022-12-08 07:02:25,632 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107201.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:02:36,953 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107213.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:03:16,252 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1637, 2.8168, 2.8344, 1.8185, 2.6370, 2.9068, 3.1811, 2.4608], device='cuda:0'), covar=tensor([0.0591, 0.0930, 0.0950, 0.1531, 0.1058, 0.0715, 0.0621, 0.1323], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0176, 0.0139, 0.0127, 0.0141, 0.0152, 0.0130, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:03:19,224 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 07:03:26,075 INFO [train.py:873] (0/4) Epoch 15, batch 1400, loss[loss=0.1112, simple_loss=0.151, pruned_loss=0.03567, over 14273.00 frames. ], tot_loss[loss=0.1152, simple_loss=0.1486, pruned_loss=0.04093, over 1946382.73 frames. ], batch size: 63, lr: 5.31e-03, grad_scale: 4.0 2022-12-08 07:03:41,696 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 2.191e+02 2.804e+02 3.719e+02 1.162e+03, threshold=5.608e+02, percent-clipped=5.0 2022-12-08 07:04:29,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 07:04:54,154 INFO [train.py:873] (0/4) Epoch 15, batch 1500, loss[loss=0.1586, simple_loss=0.1487, pruned_loss=0.0842, over 1208.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1479, pruned_loss=0.04029, over 1954954.00 frames. ], batch size: 100, lr: 5.31e-03, grad_scale: 4.0 2022-12-08 07:05:04,412 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9946, 2.5248, 3.7698, 2.8428, 3.8247, 3.7200, 3.6004, 3.1873], device='cuda:0'), covar=tensor([0.0759, 0.2839, 0.1153, 0.1905, 0.0855, 0.0894, 0.1301, 0.1730], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0316, 0.0394, 0.0302, 0.0376, 0.0324, 0.0362, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:05:10,370 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.250e+02 2.599e+02 3.364e+02 1.423e+03, threshold=5.197e+02, percent-clipped=1.0 2022-12-08 07:05:36,271 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3301, 5.1046, 4.6410, 4.9404, 4.9036, 5.2007, 5.3289, 5.2665], device='cuda:0'), covar=tensor([0.0610, 0.0362, 0.2044, 0.2360, 0.0636, 0.0669, 0.0674, 0.0732], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0259, 0.0444, 0.0562, 0.0339, 0.0437, 0.0390, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:06:23,610 INFO [train.py:873] (0/4) Epoch 15, batch 1600, loss[loss=0.1048, simple_loss=0.1447, pruned_loss=0.03248, over 14284.00 frames. ], tot_loss[loss=0.1141, simple_loss=0.1481, pruned_loss=0.04009, over 2022367.44 frames. ], batch size: 39, lr: 5.31e-03, grad_scale: 8.0 2022-12-08 07:06:28,007 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9460, 3.7349, 3.6514, 4.0347, 3.7603, 3.5937, 4.0586, 3.3269], device='cuda:0'), covar=tensor([0.0606, 0.0861, 0.0451, 0.0401, 0.0812, 0.1483, 0.0483, 0.0593], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0274, 0.0197, 0.0195, 0.0184, 0.0155, 0.0286, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 07:06:28,077 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0152, 2.1980, 2.2684, 2.3270, 2.0807, 2.2486, 2.1707, 1.4181], device='cuda:0'), covar=tensor([0.0901, 0.0911, 0.0726, 0.0500, 0.0992, 0.0637, 0.1128, 0.2127], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0087, 0.0068, 0.0071, 0.0098, 0.0085, 0.0100, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:06:33,940 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5048, 5.2729, 4.9806, 5.4798, 5.0120, 4.7489, 5.5019, 5.3224], device='cuda:0'), covar=tensor([0.0518, 0.0718, 0.0786, 0.0566, 0.0674, 0.0487, 0.0554, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0139, 0.0144, 0.0158, 0.0146, 0.0122, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:06:39,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.559e+01 2.229e+02 2.676e+02 3.216e+02 6.513e+02, threshold=5.353e+02, percent-clipped=3.0 2022-12-08 07:06:47,572 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107496.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:06:58,086 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107508.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:06:59,863 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:07:13,125 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-08 07:07:51,201 INFO [train.py:873] (0/4) Epoch 15, batch 1700, loss[loss=0.1206, simple_loss=0.1551, pruned_loss=0.04307, over 14372.00 frames. ], tot_loss[loss=0.1145, simple_loss=0.1479, pruned_loss=0.0405, over 1915448.04 frames. ], batch size: 73, lr: 5.31e-03, grad_scale: 8.0 2022-12-08 07:07:53,455 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107571.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:08:07,305 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 2.065e+02 2.454e+02 2.883e+02 5.170e+02, threshold=4.908e+02, percent-clipped=0.0 2022-12-08 07:09:11,893 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:09:14,819 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.69 vs. limit=5.0 2022-12-08 07:09:20,339 INFO [train.py:873] (0/4) Epoch 15, batch 1800, loss[loss=0.1411, simple_loss=0.1615, pruned_loss=0.06039, over 14221.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1478, pruned_loss=0.04013, over 1957986.62 frames. ], batch size: 35, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:09:35,244 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.000e+02 2.283e+02 3.065e+02 3.850e+02 1.119e+03, threshold=6.129e+02, percent-clipped=8.0 2022-12-08 07:10:04,683 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:10:08,387 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 07:10:20,195 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107737.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:10:48,099 INFO [train.py:873] (0/4) Epoch 15, batch 1900, loss[loss=0.1085, simple_loss=0.13, pruned_loss=0.04351, over 4943.00 frames. ], tot_loss[loss=0.115, simple_loss=0.1481, pruned_loss=0.04097, over 1902259.74 frames. ], batch size: 100, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:11:04,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.201e+02 2.753e+02 3.141e+02 7.986e+02, threshold=5.507e+02, percent-clipped=3.0 2022-12-08 07:11:12,095 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107796.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:11:14,014 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107798.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:11:22,822 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107808.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:11:54,580 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107844.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:12:05,284 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107856.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:12:13,842 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107866.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:12:16,483 INFO [train.py:873] (0/4) Epoch 15, batch 2000, loss[loss=0.12, simple_loss=0.15, pruned_loss=0.04498, over 12760.00 frames. ], tot_loss[loss=0.1168, simple_loss=0.1491, pruned_loss=0.04227, over 1861310.78 frames. ], batch size: 100, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:12:31,885 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 1.995e+02 2.775e+02 3.641e+02 9.267e+02, threshold=5.550e+02, percent-clipped=5.0 2022-12-08 07:13:43,634 INFO [train.py:873] (0/4) Epoch 15, batch 2100, loss[loss=0.09258, simple_loss=0.1367, pruned_loss=0.02425, over 14391.00 frames. ], tot_loss[loss=0.1155, simple_loss=0.1486, pruned_loss=0.04125, over 1942885.75 frames. ], batch size: 41, lr: 5.30e-03, grad_scale: 8.0 2022-12-08 07:13:59,441 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 1.955e+02 2.546e+02 3.190e+02 6.620e+02, threshold=5.091e+02, percent-clipped=1.0 2022-12-08 07:14:24,274 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:15:11,288 INFO [train.py:873] (0/4) Epoch 15, batch 2200, loss[loss=0.1092, simple_loss=0.1429, pruned_loss=0.03778, over 11133.00 frames. ], tot_loss[loss=0.1146, simple_loss=0.1481, pruned_loss=0.04057, over 2015981.33 frames. ], batch size: 100, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:15:24,002 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4079, 3.2797, 3.1824, 3.4908, 3.0647, 2.9643, 3.4921, 3.3839], device='cuda:0'), covar=tensor([0.0670, 0.0942, 0.0947, 0.0609, 0.1015, 0.0772, 0.0626, 0.0739], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0137, 0.0141, 0.0154, 0.0142, 0.0120, 0.0162, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:15:26,510 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.145e+02 2.568e+02 3.392e+02 6.463e+02, threshold=5.135e+02, percent-clipped=2.0 2022-12-08 07:15:32,323 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108093.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:15:40,515 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6310, 1.7726, 1.9045, 1.6006, 1.5724, 1.6430, 1.4198, 1.1683], device='cuda:0'), covar=tensor([0.0216, 0.0226, 0.0180, 0.0252, 0.0259, 0.0327, 0.0276, 0.0381], device='cuda:0'), in_proj_covar=tensor([0.0019, 0.0020, 0.0018, 0.0019, 0.0019, 0.0030, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:16:16,338 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1449, 3.0511, 2.5926, 2.7802, 3.1319, 3.1343, 3.1986, 3.1338], device='cuda:0'), covar=tensor([0.1564, 0.0854, 0.3189, 0.3639, 0.1139, 0.1505, 0.1569, 0.1547], device='cuda:0'), in_proj_covar=tensor([0.0385, 0.0258, 0.0443, 0.0561, 0.0341, 0.0436, 0.0389, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:16:36,146 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108166.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:16:39,019 INFO [train.py:873] (0/4) Epoch 15, batch 2300, loss[loss=0.1094, simple_loss=0.1259, pruned_loss=0.04645, over 2622.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1477, pruned_loss=0.0401, over 2035087.10 frames. ], batch size: 100, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:16:54,803 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.030e+02 2.596e+02 3.385e+02 5.799e+02, threshold=5.191e+02, percent-clipped=3.0 2022-12-08 07:17:18,544 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108214.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:18:06,534 INFO [train.py:873] (0/4) Epoch 15, batch 2400, loss[loss=0.1252, simple_loss=0.1596, pruned_loss=0.04543, over 14287.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1482, pruned_loss=0.04013, over 2012580.94 frames. ], batch size: 39, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:18:22,283 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.197e+02 2.869e+02 3.470e+02 6.788e+02, threshold=5.738e+02, percent-clipped=3.0 2022-12-08 07:18:26,536 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.96 vs. limit=5.0 2022-12-08 07:18:47,584 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:18:50,839 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6222, 2.5069, 2.2737, 2.3709, 2.5519, 2.5782, 2.5869, 2.5813], device='cuda:0'), covar=tensor([0.1177, 0.0877, 0.2904, 0.3060, 0.1229, 0.1340, 0.1665, 0.1228], device='cuda:0'), in_proj_covar=tensor([0.0386, 0.0261, 0.0445, 0.0564, 0.0342, 0.0439, 0.0389, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:18:55,782 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108325.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:29,550 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:34,871 INFO [train.py:873] (0/4) Epoch 15, batch 2500, loss[loss=0.0991, simple_loss=0.1131, pruned_loss=0.04253, over 1180.00 frames. ], tot_loss[loss=0.1139, simple_loss=0.1479, pruned_loss=0.03996, over 1963166.39 frames. ], batch size: 100, lr: 5.29e-03, grad_scale: 8.0 2022-12-08 07:19:50,354 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108386.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:19:50,976 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.088e+02 2.601e+02 3.226e+02 6.971e+02, threshold=5.203e+02, percent-clipped=3.0 2022-12-08 07:19:56,357 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108393.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:20:31,678 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:20:38,929 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:21:03,651 INFO [train.py:873] (0/4) Epoch 15, batch 2600, loss[loss=0.1211, simple_loss=0.1467, pruned_loss=0.04773, over 13954.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1482, pruned_loss=0.04014, over 2000030.82 frames. ], batch size: 19, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:21:19,264 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.775e+01 2.028e+02 2.663e+02 3.413e+02 6.606e+02, threshold=5.327e+02, percent-clipped=6.0 2022-12-08 07:21:24,979 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108493.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:22:11,654 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7713, 3.5402, 3.2671, 2.6473, 3.2996, 3.5572, 4.0224, 3.0075], device='cuda:0'), covar=tensor([0.0540, 0.1130, 0.0903, 0.1222, 0.0746, 0.0584, 0.0582, 0.1072], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0174, 0.0139, 0.0126, 0.0140, 0.0151, 0.0130, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:22:32,111 INFO [train.py:873] (0/4) Epoch 15, batch 2700, loss[loss=0.1176, simple_loss=0.1357, pruned_loss=0.0498, over 3849.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1477, pruned_loss=0.04015, over 1973297.75 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 4.0 2022-12-08 07:22:39,979 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:22:49,625 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.154e+02 2.629e+02 3.316e+02 6.561e+02, threshold=5.258e+02, percent-clipped=1.0 2022-12-08 07:23:34,001 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:23:42,405 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6823, 3.5941, 3.4110, 3.7701, 3.3143, 3.1800, 3.7481, 3.6509], device='cuda:0'), covar=tensor([0.0651, 0.0902, 0.0887, 0.0611, 0.0923, 0.0752, 0.0609, 0.0762], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0139, 0.0143, 0.0156, 0.0143, 0.0122, 0.0164, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:24:01,225 INFO [train.py:873] (0/4) Epoch 15, batch 2800, loss[loss=0.1185, simple_loss=0.1405, pruned_loss=0.04823, over 4951.00 frames. ], tot_loss[loss=0.1137, simple_loss=0.1478, pruned_loss=0.03985, over 2002504.96 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:24:12,118 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108681.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:24:17,714 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.202e+02 2.137e+02 2.731e+02 3.357e+02 8.128e+02, threshold=5.462e+02, percent-clipped=3.0 2022-12-08 07:24:22,294 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 07:24:28,145 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7529, 2.3933, 3.6351, 2.7296, 3.6690, 3.5558, 3.3801, 3.0425], device='cuda:0'), covar=tensor([0.0866, 0.3182, 0.1122, 0.1924, 0.0829, 0.0976, 0.1329, 0.2008], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0317, 0.0395, 0.0299, 0.0375, 0.0322, 0.0362, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:25:04,984 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9404, 4.7140, 4.5212, 4.9033, 4.4326, 4.1172, 4.9459, 4.7645], device='cuda:0'), covar=tensor([0.0585, 0.0746, 0.0806, 0.0595, 0.0786, 0.0701, 0.0560, 0.0706], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0138, 0.0142, 0.0155, 0.0142, 0.0121, 0.0163, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:25:29,353 INFO [train.py:873] (0/4) Epoch 15, batch 2900, loss[loss=0.1708, simple_loss=0.1524, pruned_loss=0.09462, over 1269.00 frames. ], tot_loss[loss=0.1133, simple_loss=0.1475, pruned_loss=0.03961, over 1984789.79 frames. ], batch size: 100, lr: 5.28e-03, grad_scale: 8.0 2022-12-08 07:25:30,408 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5150, 1.3670, 3.5441, 1.7519, 3.3986, 3.5949, 2.6167, 3.8437], device='cuda:0'), covar=tensor([0.0264, 0.3254, 0.0421, 0.2230, 0.0759, 0.0408, 0.0969, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0160, 0.0169, 0.0167, 0.0178, 0.0133, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:25:34,995 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 07:25:40,302 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 07:25:45,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.119e+02 2.529e+02 3.118e+02 5.786e+02, threshold=5.058e+02, percent-clipped=1.0 2022-12-08 07:25:45,646 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108788.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:25:48,287 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108791.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:26:41,812 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 07:26:42,268 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:26:56,661 INFO [train.py:873] (0/4) Epoch 15, batch 3000, loss[loss=0.1282, simple_loss=0.1585, pruned_loss=0.04897, over 14213.00 frames. ], tot_loss[loss=0.1133, simple_loss=0.1473, pruned_loss=0.03962, over 2017154.92 frames. ], batch size: 94, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:26:56,662 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 07:27:02,633 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8336, 3.6238, 3.3633, 2.6636, 3.3323, 3.6705, 3.9888, 2.9760], device='cuda:0'), covar=tensor([0.0546, 0.0806, 0.0946, 0.1290, 0.0874, 0.0630, 0.0617, 0.1244], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0173, 0.0138, 0.0125, 0.0140, 0.0151, 0.0129, 0.0138], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:27:05,082 INFO [train.py:905] (0/4) Epoch 15, validation: loss=0.1368, simple_loss=0.1737, pruned_loss=0.04998, over 857387.00 frames. 2022-12-08 07:27:05,082 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 07:27:21,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 2.224e+02 2.739e+02 3.295e+02 6.855e+02, threshold=5.478e+02, percent-clipped=4.0 2022-12-08 07:27:59,308 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9456, 2.4738, 2.2309, 2.2316, 2.0969, 1.9722, 1.7834, 1.9922], device='cuda:0'), covar=tensor([0.0425, 0.0276, 0.0444, 0.0260, 0.0393, 0.0471, 0.0397, 0.0450], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:28:00,916 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:28:32,782 INFO [train.py:873] (0/4) Epoch 15, batch 3100, loss[loss=0.1116, simple_loss=0.1348, pruned_loss=0.04414, over 3897.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1466, pruned_loss=0.03894, over 2034205.18 frames. ], batch size: 100, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:28:40,916 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.85 vs. limit=5.0 2022-12-08 07:28:42,861 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108981.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:28:48,816 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.020e+02 2.542e+02 3.088e+02 8.012e+02, threshold=5.083e+02, percent-clipped=4.0 2022-12-08 07:29:15,409 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2022-12-08 07:29:25,225 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109029.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:29:32,488 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109037.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:30:00,244 INFO [train.py:873] (0/4) Epoch 15, batch 3200, loss[loss=0.1109, simple_loss=0.1394, pruned_loss=0.04124, over 6969.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1468, pruned_loss=0.03897, over 2058860.07 frames. ], batch size: 100, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:30:16,994 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 2.174e+02 2.531e+02 2.915e+02 6.335e+02, threshold=5.062e+02, percent-clipped=1.0 2022-12-08 07:30:17,206 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:30:25,164 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2022-12-08 07:30:25,648 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109098.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:30:58,805 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:31:08,437 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:31:27,857 INFO [train.py:873] (0/4) Epoch 15, batch 3300, loss[loss=0.09886, simple_loss=0.1404, pruned_loss=0.02868, over 14173.00 frames. ], tot_loss[loss=0.1132, simple_loss=0.1471, pruned_loss=0.03965, over 2015908.92 frames. ], batch size: 89, lr: 5.27e-03, grad_scale: 8.0 2022-12-08 07:31:41,844 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2717, 1.3477, 1.3655, 1.4933, 1.5175, 0.9715, 1.1566, 1.3624], device='cuda:0'), covar=tensor([0.0783, 0.0932, 0.0735, 0.0812, 0.0478, 0.0928, 0.0990, 0.0800], device='cuda:0'), in_proj_covar=tensor([0.0032, 0.0032, 0.0035, 0.0030, 0.0032, 0.0045, 0.0033, 0.0036], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:31:44,097 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.026e+02 2.178e+02 2.677e+02 3.330e+02 7.070e+02, threshold=5.354e+02, percent-clipped=2.0 2022-12-08 07:31:49,453 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109194.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:23,648 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:43,215 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109255.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:32:55,126 INFO [train.py:873] (0/4) Epoch 15, batch 3400, loss[loss=0.08457, simple_loss=0.122, pruned_loss=0.02357, over 6937.00 frames. ], tot_loss[loss=0.1124, simple_loss=0.1463, pruned_loss=0.03922, over 1905624.03 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 4.0 2022-12-08 07:33:05,753 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:33:12,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4935, 1.0552, 2.0442, 1.8159, 1.8544, 2.0623, 1.5006, 2.0518], device='cuda:0'), covar=tensor([0.0863, 0.1536, 0.0275, 0.0513, 0.0642, 0.0283, 0.0695, 0.0305], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0161, 0.0131, 0.0172, 0.0149, 0.0142, 0.0125, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 07:33:12,975 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.873e+01 2.037e+02 2.446e+02 3.255e+02 7.608e+02, threshold=4.893e+02, percent-clipped=4.0 2022-12-08 07:33:17,246 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109294.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:33:38,113 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:34:10,937 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109355.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:34:23,255 INFO [train.py:873] (0/4) Epoch 15, batch 3500, loss[loss=0.1192, simple_loss=0.1494, pruned_loss=0.04451, over 12789.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1465, pruned_loss=0.039, over 1973318.69 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 4.0 2022-12-08 07:34:32,302 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109379.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:34:40,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.264e+02 2.745e+02 3.644e+02 1.157e+03, threshold=5.490e+02, percent-clipped=6.0 2022-12-08 07:34:44,039 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109393.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 07:34:59,779 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9244, 1.8165, 2.1400, 1.8955, 1.8036, 1.6316, 1.8126, 1.2874], device='cuda:0'), covar=tensor([0.0182, 0.0313, 0.0179, 0.0283, 0.0216, 0.0287, 0.0244, 0.0384], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0020, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:35:31,891 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109447.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:35:39,041 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4973, 1.3759, 1.5002, 1.3509, 1.3121, 1.2873, 1.1359, 1.0818], device='cuda:0'), covar=tensor([0.0197, 0.0311, 0.0261, 0.0241, 0.0236, 0.0404, 0.0322, 0.0470], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0020, 0.0018, 0.0020, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:35:40,120 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2022-12-08 07:35:50,855 INFO [train.py:873] (0/4) Epoch 15, batch 3600, loss[loss=0.1425, simple_loss=0.1333, pruned_loss=0.07584, over 1282.00 frames. ], tot_loss[loss=0.1135, simple_loss=0.1473, pruned_loss=0.03989, over 1941915.46 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:36:04,914 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3175, 1.4605, 3.4937, 1.5877, 3.1891, 3.4149, 2.5484, 3.6646], device='cuda:0'), covar=tensor([0.0260, 0.2957, 0.0343, 0.2182, 0.0944, 0.0404, 0.0877, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0155, 0.0158, 0.0167, 0.0167, 0.0178, 0.0131, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:36:08,874 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.104e+02 2.724e+02 3.497e+02 9.894e+02, threshold=5.448e+02, percent-clipped=5.0 2022-12-08 07:36:14,300 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:36:46,169 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0912, 2.6980, 5.1319, 3.4256, 4.8209, 2.3516, 3.9005, 4.7853], device='cuda:0'), covar=tensor([0.0370, 0.3514, 0.0324, 0.5744, 0.0540, 0.3195, 0.1029, 0.0345], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0206, 0.0213, 0.0278, 0.0231, 0.0208, 0.0206, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:37:01,360 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5344, 3.2332, 2.4859, 3.6461, 3.4868, 3.5585, 3.1165, 2.4638], device='cuda:0'), covar=tensor([0.0808, 0.1290, 0.3172, 0.0547, 0.0894, 0.0914, 0.1214, 0.3217], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0286, 0.0259, 0.0275, 0.0319, 0.0299, 0.0255, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:37:02,969 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109550.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:37:03,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7094, 1.4704, 3.2991, 3.0961, 3.2130, 3.4096, 2.8035, 3.2274], device='cuda:0'), covar=tensor([0.2002, 0.2080, 0.0264, 0.0419, 0.0406, 0.0268, 0.0361, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0160, 0.0130, 0.0172, 0.0148, 0.0141, 0.0124, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 07:37:04,533 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1235, 2.3680, 2.3523, 2.3786, 2.1100, 2.4003, 2.2677, 1.3903], device='cuda:0'), covar=tensor([0.1412, 0.1042, 0.0747, 0.0857, 0.0959, 0.0672, 0.1016, 0.2115], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0089, 0.0070, 0.0072, 0.0099, 0.0087, 0.0101, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:37:19,585 INFO [train.py:873] (0/4) Epoch 15, batch 3700, loss[loss=0.1313, simple_loss=0.1467, pruned_loss=0.05795, over 3903.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.1475, pruned_loss=0.04001, over 1947435.11 frames. ], batch size: 100, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:37:28,824 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109579.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:37:37,109 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.068e+02 2.547e+02 3.153e+02 5.371e+02, threshold=5.093e+02, percent-clipped=0.0 2022-12-08 07:37:39,286 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7222, 4.2659, 3.3898, 5.1228, 4.4822, 4.8483, 4.2895, 3.6382], device='cuda:0'), covar=tensor([0.0702, 0.0981, 0.3275, 0.0368, 0.0828, 0.1131, 0.1061, 0.2614], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0287, 0.0260, 0.0275, 0.0320, 0.0300, 0.0256, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:37:39,615 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2022-12-08 07:37:57,427 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3168, 1.4852, 2.5251, 1.4832, 2.4590, 2.4693, 1.9780, 2.6328], device='cuda:0'), covar=tensor([0.0331, 0.2462, 0.0421, 0.1845, 0.0512, 0.0634, 0.1170, 0.0327], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0158, 0.0168, 0.0167, 0.0178, 0.0133, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:38:23,069 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109640.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:32,112 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:39,806 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:38:40,415 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2022-12-08 07:38:42,966 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2022-12-08 07:38:48,151 INFO [train.py:873] (0/4) Epoch 15, batch 3800, loss[loss=0.109, simple_loss=0.1466, pruned_loss=0.03568, over 14222.00 frames. ], tot_loss[loss=0.1141, simple_loss=0.1475, pruned_loss=0.04033, over 1890259.99 frames. ], batch size: 94, lr: 5.26e-03, grad_scale: 8.0 2022-12-08 07:38:53,040 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109674.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:39:06,079 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.193e+02 2.777e+02 3.474e+02 7.448e+02, threshold=5.555e+02, percent-clipped=3.0 2022-12-08 07:39:07,292 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7929, 2.3008, 3.6874, 3.9882, 3.7917, 2.2345, 3.9171, 2.7692], device='cuda:0'), covar=tensor([0.0471, 0.1282, 0.0956, 0.0498, 0.0476, 0.2028, 0.0406, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0256, 0.0371, 0.0326, 0.0268, 0.0301, 0.0308, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:39:08,456 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2022-12-08 07:39:10,088 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8062, 3.5241, 3.3015, 3.4605, 3.6551, 3.7076, 3.7374, 3.7442], device='cuda:0'), covar=tensor([0.0805, 0.0604, 0.2147, 0.2551, 0.0834, 0.0960, 0.1140, 0.0896], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0262, 0.0445, 0.0558, 0.0338, 0.0439, 0.0387, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:39:10,133 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109693.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:39:33,904 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:39:52,257 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109741.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 07:40:16,359 INFO [train.py:873] (0/4) Epoch 15, batch 3900, loss[loss=0.1195, simple_loss=0.1514, pruned_loss=0.04385, over 14249.00 frames. ], tot_loss[loss=0.1142, simple_loss=0.1477, pruned_loss=0.04037, over 1927438.74 frames. ], batch size: 39, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:40:34,143 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.235e+02 2.763e+02 3.267e+02 1.043e+03, threshold=5.525e+02, percent-clipped=3.0 2022-12-08 07:40:51,630 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8489, 1.6363, 1.9126, 1.6622, 1.9473, 1.7835, 1.6821, 1.8436], device='cuda:0'), covar=tensor([0.0557, 0.1374, 0.0404, 0.0434, 0.0477, 0.0781, 0.0303, 0.0407], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0317, 0.0395, 0.0301, 0.0375, 0.0324, 0.0361, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:41:28,444 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:41:30,227 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1308, 3.0184, 4.9257, 3.7496, 4.9840, 4.7957, 4.5506, 4.4376], device='cuda:0'), covar=tensor([0.0540, 0.3213, 0.0650, 0.1357, 0.0574, 0.0867, 0.1364, 0.1370], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0315, 0.0391, 0.0299, 0.0372, 0.0321, 0.0358, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:41:45,243 INFO [train.py:873] (0/4) Epoch 15, batch 4000, loss[loss=0.1054, simple_loss=0.1361, pruned_loss=0.03734, over 6006.00 frames. ], tot_loss[loss=0.1144, simple_loss=0.1481, pruned_loss=0.04032, over 2020185.71 frames. ], batch size: 100, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:41:56,055 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3767, 3.6880, 2.9122, 4.6182, 4.2359, 4.4296, 3.8789, 3.0323], device='cuda:0'), covar=tensor([0.0710, 0.1414, 0.3706, 0.0629, 0.0998, 0.1310, 0.1216, 0.3588], device='cuda:0'), in_proj_covar=tensor([0.0278, 0.0291, 0.0262, 0.0280, 0.0322, 0.0303, 0.0257, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:41:59,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0274, 1.6363, 3.8381, 3.5257, 3.6050, 3.8946, 3.3358, 3.8588], device='cuda:0'), covar=tensor([0.1476, 0.1605, 0.0117, 0.0274, 0.0264, 0.0147, 0.0243, 0.0144], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0158, 0.0129, 0.0170, 0.0147, 0.0141, 0.0124, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 07:42:02,600 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.030e+02 2.144e+02 2.556e+02 3.390e+02 6.366e+02, threshold=5.112e+02, percent-clipped=4.0 2022-12-08 07:42:10,895 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109898.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:42:43,643 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109935.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:42:44,479 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9451, 2.7256, 2.7150, 2.9686, 2.8456, 2.8658, 3.0134, 2.4235], device='cuda:0'), covar=tensor([0.0620, 0.1094, 0.0575, 0.0551, 0.0820, 0.0489, 0.0639, 0.0669], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0270, 0.0197, 0.0194, 0.0185, 0.0154, 0.0285, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 07:42:57,039 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:13,346 INFO [train.py:873] (0/4) Epoch 15, batch 4100, loss[loss=0.1014, simple_loss=0.1414, pruned_loss=0.03067, over 14377.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1466, pruned_loss=0.03915, over 2031599.19 frames. ], batch size: 31, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:43:18,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:21,038 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3463, 2.3438, 1.9502, 2.3467, 2.2295, 2.2833, 2.1483, 1.9982], device='cuda:0'), covar=tensor([0.0976, 0.0967, 0.2048, 0.0917, 0.1155, 0.0825, 0.1285, 0.1482], device='cuda:0'), in_proj_covar=tensor([0.0277, 0.0290, 0.0261, 0.0278, 0.0322, 0.0302, 0.0256, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:43:31,525 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.583e+01 2.156e+02 2.609e+02 3.160e+02 7.009e+02, threshold=5.218e+02, percent-clipped=5.0 2022-12-08 07:43:39,878 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:43:41,876 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-110000.pt 2022-12-08 07:43:46,206 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5099, 1.9218, 2.0099, 2.1064, 1.8781, 2.0557, 1.7554, 1.2333], device='cuda:0'), covar=tensor([0.1370, 0.1251, 0.0919, 0.0828, 0.1342, 0.0935, 0.1771, 0.2696], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0087, 0.0068, 0.0072, 0.0098, 0.0086, 0.0100, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:43:58,990 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:00,700 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110017.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:05,351 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110022.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:12,984 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8977, 4.0543, 4.2480, 3.6420, 4.0638, 4.2076, 1.6324, 3.8562], device='cuda:0'), covar=tensor([0.0322, 0.0350, 0.0318, 0.0489, 0.0321, 0.0256, 0.3114, 0.0292], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0173, 0.0144, 0.0144, 0.0202, 0.0140, 0.0158, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 07:44:13,076 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8892, 1.8497, 1.9914, 1.9549, 1.8658, 1.5858, 1.5024, 1.3281], device='cuda:0'), covar=tensor([0.0240, 0.0374, 0.0224, 0.0220, 0.0228, 0.0275, 0.0289, 0.0459], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:44:24,386 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110044.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:44:46,260 INFO [train.py:873] (0/4) Epoch 15, batch 4200, loss[loss=0.1498, simple_loss=0.1327, pruned_loss=0.08347, over 1248.00 frames. ], tot_loss[loss=0.1129, simple_loss=0.1467, pruned_loss=0.03953, over 1941416.21 frames. ], batch size: 100, lr: 5.25e-03, grad_scale: 8.0 2022-12-08 07:44:54,479 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110078.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:02,765 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.94 vs. limit=5.0 2022-12-08 07:45:03,731 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.176e+02 2.682e+02 3.403e+02 7.328e+02, threshold=5.365e+02, percent-clipped=6.0 2022-12-08 07:45:18,580 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110105.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:22,697 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:45:25,489 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1551, 2.3162, 2.2716, 1.7815, 2.1928, 1.9512, 1.7566, 1.8271], device='cuda:0'), covar=tensor([0.0266, 0.0271, 0.0433, 0.0800, 0.0411, 0.0524, 0.0575, 0.0644], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 07:45:25,554 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1102, 2.4806, 4.3057, 4.3911, 4.2084, 2.2646, 4.3911, 3.2444], device='cuda:0'), covar=tensor([0.0436, 0.1142, 0.0735, 0.0393, 0.0456, 0.1906, 0.0331, 0.1026], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0256, 0.0370, 0.0327, 0.0268, 0.0301, 0.0307, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:45:27,164 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:46:13,633 INFO [train.py:873] (0/4) Epoch 15, batch 4300, loss[loss=0.1251, simple_loss=0.1563, pruned_loss=0.04697, over 11178.00 frames. ], tot_loss[loss=0.1131, simple_loss=0.1474, pruned_loss=0.03946, over 1944933.59 frames. ], batch size: 100, lr: 5.24e-03, grad_scale: 4.0 2022-12-08 07:46:15,358 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:46:20,255 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:46:32,112 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 2.281e+02 2.830e+02 3.301e+02 1.295e+03, threshold=5.661e+02, percent-clipped=5.0 2022-12-08 07:46:36,664 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1105, 1.9032, 3.2582, 2.3895, 3.1540, 1.8273, 2.5730, 3.0933], device='cuda:0'), covar=tensor([0.0849, 0.3905, 0.0604, 0.4311, 0.0852, 0.3340, 0.1281, 0.0633], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0207, 0.0214, 0.0277, 0.0232, 0.0207, 0.0207, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:47:12,449 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110235.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:47:41,817 INFO [train.py:873] (0/4) Epoch 15, batch 4400, loss[loss=0.1023, simple_loss=0.1436, pruned_loss=0.03051, over 14432.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.147, pruned_loss=0.03988, over 1889696.39 frames. ], batch size: 41, lr: 5.24e-03, grad_scale: 8.0 2022-12-08 07:47:54,410 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:47:54,519 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:00,237 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.123e+01 2.204e+02 2.570e+02 3.174e+02 5.425e+02, threshold=5.140e+02, percent-clipped=0.0 2022-12-08 07:48:12,336 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110303.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:22,577 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:25,414 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:48:41,228 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1464, 4.9409, 4.7100, 5.2720, 4.7375, 4.6035, 5.2642, 4.9875], device='cuda:0'), covar=tensor([0.0625, 0.0728, 0.0740, 0.0428, 0.0725, 0.0514, 0.0498, 0.0620], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0141, 0.0145, 0.0157, 0.0146, 0.0123, 0.0168, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:48:48,570 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110344.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:05,186 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:06,159 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:08,770 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7574, 2.6502, 2.0259, 2.7998, 2.5770, 2.6719, 2.3753, 2.1104], device='cuda:0'), covar=tensor([0.0987, 0.1289, 0.3065, 0.0863, 0.1179, 0.0890, 0.1583, 0.2922], device='cuda:0'), in_proj_covar=tensor([0.0277, 0.0289, 0.0262, 0.0279, 0.0322, 0.0302, 0.0256, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:49:10,332 INFO [train.py:873] (0/4) Epoch 15, batch 4500, loss[loss=0.1232, simple_loss=0.1483, pruned_loss=0.049, over 5970.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1466, pruned_loss=0.03917, over 1910840.09 frames. ], batch size: 100, lr: 5.24e-03, grad_scale: 8.0 2022-12-08 07:49:14,175 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110373.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:20,087 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110379.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:49:28,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5810, 1.6637, 4.1962, 2.4562, 4.2773, 4.5032, 4.1033, 5.0521], device='cuda:0'), covar=tensor([0.0203, 0.3107, 0.0472, 0.1929, 0.0337, 0.0354, 0.0378, 0.0148], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0161, 0.0171, 0.0170, 0.0180, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:49:29,134 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.023e+02 2.685e+02 3.179e+02 5.561e+02, threshold=5.371e+02, percent-clipped=3.0 2022-12-08 07:49:38,318 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110400.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:50:15,557 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 07:50:36,568 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:50:39,117 INFO [train.py:873] (0/4) Epoch 15, batch 4600, loss[loss=0.0957, simple_loss=0.14, pruned_loss=0.02569, over 14441.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1469, pruned_loss=0.03934, over 1983818.24 frames. ], batch size: 53, lr: 5.24e-03, grad_scale: 4.0 2022-12-08 07:50:40,954 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:50:58,404 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.355e+02 2.918e+02 3.425e+02 9.705e+02, threshold=5.836e+02, percent-clipped=5.0 2022-12-08 07:51:23,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8637, 2.7148, 2.7359, 2.9250, 2.8126, 2.8402, 3.0018, 2.5230], device='cuda:0'), covar=tensor([0.0807, 0.1118, 0.0633, 0.0599, 0.0896, 0.0608, 0.0601, 0.0664], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0274, 0.0199, 0.0196, 0.0188, 0.0157, 0.0289, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 07:51:49,557 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4884, 3.9117, 3.6096, 3.4435, 2.7403, 3.8224, 3.5048, 2.1109], device='cuda:0'), covar=tensor([0.1578, 0.0551, 0.0739, 0.0803, 0.0923, 0.0470, 0.1095, 0.2017], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0088, 0.0068, 0.0071, 0.0097, 0.0085, 0.0099, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:52:06,647 INFO [train.py:873] (0/4) Epoch 15, batch 4700, loss[loss=0.1203, simple_loss=0.1505, pruned_loss=0.04503, over 6964.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1466, pruned_loss=0.03884, over 1998794.57 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 4.0 2022-12-08 07:52:26,207 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.034e+02 2.470e+02 3.158e+02 1.519e+03, threshold=4.939e+02, percent-clipped=4.0 2022-12-08 07:52:44,268 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110611.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:01,243 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0388, 2.0605, 4.0045, 2.8263, 3.9005, 1.9402, 2.9974, 3.8816], device='cuda:0'), covar=tensor([0.0692, 0.4325, 0.0554, 0.5477, 0.0615, 0.3652, 0.1482, 0.0512], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0204, 0.0213, 0.0275, 0.0230, 0.0207, 0.0206, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:53:09,039 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110639.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:26,642 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:34,854 INFO [train.py:873] (0/4) Epoch 15, batch 4800, loss[loss=0.1318, simple_loss=0.1326, pruned_loss=0.06547, over 2559.00 frames. ], tot_loss[loss=0.1127, simple_loss=0.1467, pruned_loss=0.03932, over 1952810.84 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 8.0 2022-12-08 07:53:37,787 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110672.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:38,548 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110673.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:39,412 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110674.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:53:53,745 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.088e+02 2.309e+02 2.792e+02 3.592e+02 7.510e+02, threshold=5.583e+02, percent-clipped=5.0 2022-12-08 07:53:56,327 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2022-12-08 07:54:02,004 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110700.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:15,555 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4324, 3.1708, 3.8941, 2.7801, 2.3886, 3.2817, 1.8634, 3.5117], device='cuda:0'), covar=tensor([0.0954, 0.0954, 0.0693, 0.2268, 0.2311, 0.0991, 0.3239, 0.0897], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0100, 0.0092, 0.0098, 0.0114, 0.0088, 0.0119, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 07:54:18,067 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9524, 3.7317, 3.6775, 4.0053, 3.5756, 3.3800, 4.0486, 3.8512], device='cuda:0'), covar=tensor([0.0701, 0.1111, 0.0943, 0.0668, 0.1093, 0.0800, 0.0667, 0.0836], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0140, 0.0144, 0.0156, 0.0144, 0.0122, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:54:20,446 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110721.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:25,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0593, 2.5162, 5.0824, 3.3210, 4.7713, 2.2557, 3.4615, 4.7567], device='cuda:0'), covar=tensor([0.0452, 0.3930, 0.0324, 0.6165, 0.0431, 0.3320, 0.1411, 0.0320], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0206, 0.0214, 0.0277, 0.0232, 0.0208, 0.0206, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:54:33,618 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:54:44,436 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110748.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:00,618 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:03,180 INFO [train.py:873] (0/4) Epoch 15, batch 4900, loss[loss=0.09297, simple_loss=0.1379, pruned_loss=0.02404, over 14615.00 frames. ], tot_loss[loss=0.1119, simple_loss=0.1466, pruned_loss=0.03857, over 1981599.84 frames. ], batch size: 22, lr: 5.23e-03, grad_scale: 8.0 2022-12-08 07:55:04,784 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110771.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:22,714 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 1.962e+02 2.570e+02 3.049e+02 5.982e+02, threshold=5.139e+02, percent-clipped=2.0 2022-12-08 07:55:28,509 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110797.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 07:55:29,323 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5349, 4.2718, 4.0064, 4.1746, 4.3102, 4.4485, 4.5057, 4.5187], device='cuda:0'), covar=tensor([0.0713, 0.0469, 0.1974, 0.2473, 0.0703, 0.0841, 0.0713, 0.0689], device='cuda:0'), in_proj_covar=tensor([0.0380, 0.0261, 0.0443, 0.0553, 0.0338, 0.0440, 0.0386, 0.0381], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:55:34,894 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4990, 2.5843, 4.2998, 4.4904, 4.4163, 2.6140, 4.5364, 3.5611], device='cuda:0'), covar=tensor([0.0332, 0.1112, 0.0750, 0.0369, 0.0375, 0.1613, 0.0320, 0.0835], device='cuda:0'), in_proj_covar=tensor([0.0287, 0.0254, 0.0367, 0.0324, 0.0267, 0.0300, 0.0307, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 07:55:43,505 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:55:48,021 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110819.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:56:19,650 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9198, 2.4427, 3.6497, 2.7307, 3.6939, 3.5645, 3.4869, 3.1133], device='cuda:0'), covar=tensor([0.0874, 0.3001, 0.1087, 0.2068, 0.1060, 0.1099, 0.1551, 0.1725], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0318, 0.0397, 0.0302, 0.0378, 0.0326, 0.0363, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:56:31,171 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110868.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:56:31,858 INFO [train.py:873] (0/4) Epoch 15, batch 5000, loss[loss=0.1083, simple_loss=0.1267, pruned_loss=0.04492, over 3877.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1469, pruned_loss=0.0389, over 1929710.72 frames. ], batch size: 100, lr: 5.23e-03, grad_scale: 4.0 2022-12-08 07:56:38,100 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0310, 3.3723, 3.3074, 3.3236, 2.4937, 3.2881, 3.1121, 1.7236], device='cuda:0'), covar=tensor([0.1534, 0.0696, 0.1007, 0.0631, 0.0997, 0.0579, 0.1230, 0.2247], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0088, 0.0069, 0.0072, 0.0098, 0.0086, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 07:56:52,356 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.709e+01 2.105e+02 2.624e+02 3.225e+02 7.187e+02, threshold=5.248e+02, percent-clipped=2.0 2022-12-08 07:57:02,681 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3988, 3.9582, 3.1176, 4.7527, 4.3890, 4.5120, 4.0217, 3.4071], device='cuda:0'), covar=tensor([0.0773, 0.1205, 0.3679, 0.0533, 0.0952, 0.1246, 0.1243, 0.3057], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0293, 0.0263, 0.0280, 0.0323, 0.0303, 0.0257, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:57:25,014 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:33,742 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110939.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:50,964 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110958.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:51,703 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110959.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:57:58,574 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110967.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:00,303 INFO [train.py:873] (0/4) Epoch 15, batch 5100, loss[loss=0.1263, simple_loss=0.1227, pruned_loss=0.06495, over 1256.00 frames. ], tot_loss[loss=0.1127, simple_loss=0.1471, pruned_loss=0.0392, over 1993078.52 frames. ], batch size: 100, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 07:58:04,899 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:16,276 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:20,532 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.198e+02 2.177e+02 2.703e+02 3.385e+02 6.811e+02, threshold=5.407e+02, percent-clipped=6.0 2022-12-08 07:58:23,738 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8875, 4.1064, 4.2503, 3.7722, 4.1044, 4.1995, 1.7721, 3.8955], device='cuda:0'), covar=tensor([0.0334, 0.0307, 0.0305, 0.0393, 0.0283, 0.0297, 0.2963, 0.0256], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0172, 0.0142, 0.0144, 0.0202, 0.0139, 0.0157, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 07:58:33,407 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:44,165 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111019.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:58:45,296 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.17 vs. limit=2.0 2022-12-08 07:58:46,623 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111022.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 07:59:05,549 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2022-12-08 07:59:28,410 INFO [train.py:873] (0/4) Epoch 15, batch 5200, loss[loss=0.08675, simple_loss=0.1255, pruned_loss=0.02399, over 12054.00 frames. ], tot_loss[loss=0.1118, simple_loss=0.1465, pruned_loss=0.03855, over 1983913.69 frames. ], batch size: 15, lr: 5.22e-03, grad_scale: 8.0 2022-12-08 07:59:35,770 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2022-12-08 07:59:37,094 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6340, 2.4907, 2.2129, 2.3032, 2.5771, 2.6027, 2.5816, 2.5606], device='cuda:0'), covar=tensor([0.1032, 0.0733, 0.2371, 0.2605, 0.1017, 0.1181, 0.1346, 0.1033], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0263, 0.0442, 0.0559, 0.0341, 0.0442, 0.0389, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 07:59:46,669 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7856, 1.5272, 3.5377, 3.2614, 3.4055, 3.6171, 2.7792, 3.5814], device='cuda:0'), covar=tensor([0.2370, 0.2515, 0.0239, 0.0480, 0.0430, 0.0298, 0.0541, 0.0242], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0158, 0.0129, 0.0169, 0.0146, 0.0140, 0.0121, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 07:59:48,545 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.014e+01 2.007e+02 2.591e+02 3.079e+02 5.022e+02, threshold=5.182e+02, percent-clipped=0.0 2022-12-08 07:59:48,694 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111092.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 07:59:50,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 08:00:30,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 08:00:55,994 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 08:00:57,147 INFO [train.py:873] (0/4) Epoch 15, batch 5300, loss[loss=0.1195, simple_loss=0.1512, pruned_loss=0.04391, over 10369.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.147, pruned_loss=0.03887, over 1959182.91 frames. ], batch size: 100, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 08:01:10,459 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4231, 4.1035, 3.9761, 4.4535, 4.1087, 3.9323, 4.4610, 3.6845], device='cuda:0'), covar=tensor([0.0420, 0.0966, 0.0450, 0.0413, 0.0795, 0.0923, 0.0492, 0.0582], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0270, 0.0197, 0.0194, 0.0185, 0.0155, 0.0285, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 08:01:18,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.026e+02 2.377e+02 2.924e+02 1.186e+03, threshold=4.754e+02, percent-clipped=4.0 2022-12-08 08:01:39,030 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6732, 2.0822, 2.6183, 2.7403, 2.6008, 1.9981, 2.7402, 2.2711], device='cuda:0'), covar=tensor([0.0458, 0.0993, 0.0632, 0.0471, 0.0605, 0.1370, 0.0459, 0.0825], device='cuda:0'), in_proj_covar=tensor([0.0289, 0.0255, 0.0369, 0.0325, 0.0269, 0.0302, 0.0307, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:01:46,181 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111224.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:01:46,912 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9899, 5.3685, 5.4078, 5.9464, 5.5086, 4.8158, 5.8204, 4.8924], device='cuda:0'), covar=tensor([0.0271, 0.0882, 0.0303, 0.0332, 0.0738, 0.0336, 0.0427, 0.0451], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0271, 0.0197, 0.0194, 0.0185, 0.0155, 0.0286, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 08:01:47,025 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:23,960 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111267.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:25,564 INFO [train.py:873] (0/4) Epoch 15, batch 5400, loss[loss=0.1196, simple_loss=0.1544, pruned_loss=0.04244, over 14269.00 frames. ], tot_loss[loss=0.1131, simple_loss=0.1474, pruned_loss=0.03942, over 1934271.60 frames. ], batch size: 69, lr: 5.22e-03, grad_scale: 4.0 2022-12-08 08:02:40,895 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:02:46,809 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.071e+02 2.546e+02 3.124e+02 5.888e+02, threshold=5.092e+02, percent-clipped=1.0 2022-12-08 08:03:05,559 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111314.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:03:06,427 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:03:36,998 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2027, 1.7728, 2.2125, 1.9315, 2.3058, 2.0512, 2.0232, 2.0961], device='cuda:0'), covar=tensor([0.0649, 0.2116, 0.0607, 0.0831, 0.0475, 0.1070, 0.0455, 0.0592], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0311, 0.0388, 0.0296, 0.0371, 0.0321, 0.0358, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:03:53,189 INFO [train.py:873] (0/4) Epoch 15, batch 5500, loss[loss=0.1053, simple_loss=0.1279, pruned_loss=0.04137, over 4932.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1477, pruned_loss=0.03958, over 1967876.77 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:04:13,960 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111392.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 08:04:14,624 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.065e+02 2.464e+02 3.206e+02 5.854e+02, threshold=4.928e+02, percent-clipped=3.0 2022-12-08 08:04:27,077 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4519, 3.7149, 3.0037, 4.8306, 4.2656, 4.5395, 3.9759, 3.2992], device='cuda:0'), covar=tensor([0.0666, 0.1468, 0.3345, 0.0638, 0.0955, 0.1359, 0.1125, 0.3025], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0295, 0.0264, 0.0282, 0.0326, 0.0305, 0.0258, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:04:36,657 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7190, 2.4707, 4.7718, 3.1756, 4.5188, 2.3702, 3.5653, 4.4984], device='cuda:0'), covar=tensor([0.0468, 0.4057, 0.0298, 0.6570, 0.0486, 0.3280, 0.1378, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0206, 0.0215, 0.0277, 0.0233, 0.0206, 0.0207, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:04:56,354 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111440.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:05:20,853 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:05:21,569 INFO [train.py:873] (0/4) Epoch 15, batch 5600, loss[loss=0.1094, simple_loss=0.117, pruned_loss=0.05095, over 2658.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.1478, pruned_loss=0.03995, over 1947589.27 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 8.0 2022-12-08 08:05:42,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.013e+02 2.578e+02 3.220e+02 8.105e+02, threshold=5.156e+02, percent-clipped=4.0 2022-12-08 08:05:47,320 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2022-12-08 08:05:58,403 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8021, 3.3556, 2.7178, 3.9930, 3.8023, 3.8118, 3.2990, 2.8120], device='cuda:0'), covar=tensor([0.0786, 0.1421, 0.3010, 0.0615, 0.0894, 0.1298, 0.1325, 0.2759], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0295, 0.0264, 0.0283, 0.0327, 0.0304, 0.0259, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:06:10,243 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:14,483 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:48,902 INFO [train.py:873] (0/4) Epoch 15, batch 5700, loss[loss=0.1128, simple_loss=0.1383, pruned_loss=0.04363, over 6020.00 frames. ], tot_loss[loss=0.114, simple_loss=0.1477, pruned_loss=0.04015, over 1924385.10 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:06:51,474 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111572.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:06:59,414 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111581.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:07:10,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.509e+01 2.357e+02 2.792e+02 3.438e+02 6.748e+02, threshold=5.583e+02, percent-clipped=1.0 2022-12-08 08:07:28,409 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111614.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:08:10,714 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111662.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:08:17,295 INFO [train.py:873] (0/4) Epoch 15, batch 5800, loss[loss=0.09645, simple_loss=0.1201, pruned_loss=0.03637, over 3838.00 frames. ], tot_loss[loss=0.1147, simple_loss=0.1484, pruned_loss=0.04054, over 1950845.57 frames. ], batch size: 100, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:08:32,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 08:08:37,099 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 08:08:38,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.211e+02 2.861e+02 3.266e+02 8.310e+02, threshold=5.722e+02, percent-clipped=4.0 2022-12-08 08:08:47,654 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2022-12-08 08:08:51,944 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5724, 2.3131, 4.5695, 3.2397, 4.3453, 2.2836, 3.3905, 4.3124], device='cuda:0'), covar=tensor([0.0646, 0.4130, 0.0386, 0.5710, 0.0573, 0.3213, 0.1370, 0.0546], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0205, 0.0215, 0.0275, 0.0231, 0.0205, 0.0206, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:09:00,026 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3235, 2.2946, 4.3185, 3.0259, 4.1545, 2.3376, 3.2764, 4.1056], device='cuda:0'), covar=tensor([0.0603, 0.3838, 0.0471, 0.5598, 0.0520, 0.3121, 0.1354, 0.0454], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0205, 0.0215, 0.0275, 0.0232, 0.0206, 0.0206, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:09:16,754 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4323, 2.7214, 4.2766, 3.1933, 4.2410, 4.0651, 3.9925, 3.6090], device='cuda:0'), covar=tensor([0.0750, 0.2866, 0.0842, 0.1636, 0.0692, 0.0901, 0.1505, 0.1547], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0313, 0.0392, 0.0299, 0.0372, 0.0324, 0.0361, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:09:39,410 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8835, 2.3747, 3.6407, 2.8184, 3.7221, 3.5589, 3.4650, 3.0056], device='cuda:0'), covar=tensor([0.1088, 0.3479, 0.1021, 0.1917, 0.0949, 0.1203, 0.1569, 0.2094], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0315, 0.0394, 0.0301, 0.0375, 0.0327, 0.0363, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:09:44,962 INFO [train.py:873] (0/4) Epoch 15, batch 5900, loss[loss=0.1151, simple_loss=0.1459, pruned_loss=0.04212, over 14231.00 frames. ], tot_loss[loss=0.1127, simple_loss=0.1474, pruned_loss=0.03895, over 2060036.89 frames. ], batch size: 89, lr: 5.21e-03, grad_scale: 4.0 2022-12-08 08:10:04,849 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111791.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 08:10:07,183 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.002e+02 2.082e+02 2.433e+02 3.039e+02 5.689e+02, threshold=4.867e+02, percent-clipped=0.0 2022-12-08 08:10:33,515 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:10:57,538 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111852.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:11:12,396 INFO [train.py:873] (0/4) Epoch 15, batch 6000, loss[loss=0.1155, simple_loss=0.1459, pruned_loss=0.0425, over 6933.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.147, pruned_loss=0.03879, over 2041892.62 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 8.0 2022-12-08 08:11:12,397 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 08:11:20,777 INFO [train.py:905] (0/4) Epoch 15, validation: loss=0.1363, simple_loss=0.1737, pruned_loss=0.04946, over 857387.00 frames. 2022-12-08 08:11:20,777 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 08:11:31,660 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:11:42,936 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.150e+02 2.606e+02 3.231e+02 6.699e+02, threshold=5.212e+02, percent-clipped=6.0 2022-12-08 08:12:13,675 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:12:48,569 INFO [train.py:873] (0/4) Epoch 15, batch 6100, loss[loss=0.137, simple_loss=0.1344, pruned_loss=0.06986, over 1284.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1471, pruned_loss=0.03892, over 2079034.52 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:13:10,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.212e+02 2.710e+02 3.481e+02 4.963e+02, threshold=5.420e+02, percent-clipped=0.0 2022-12-08 08:13:29,266 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8841, 1.8587, 1.7554, 2.0262, 1.8093, 1.7335, 1.2662, 1.6426], device='cuda:0'), covar=tensor([0.1123, 0.0947, 0.1114, 0.0627, 0.1087, 0.1786, 0.3298, 0.1273], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0173, 0.0143, 0.0144, 0.0204, 0.0139, 0.0158, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 08:14:13,887 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:14:16,349 INFO [train.py:873] (0/4) Epoch 15, batch 6200, loss[loss=0.1471, simple_loss=0.1581, pruned_loss=0.06807, over 5959.00 frames. ], tot_loss[loss=0.1132, simple_loss=0.1474, pruned_loss=0.0395, over 2044004.55 frames. ], batch size: 100, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:14:39,268 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 2.103e+02 2.700e+02 3.313e+02 6.071e+02, threshold=5.399e+02, percent-clipped=3.0 2022-12-08 08:15:05,071 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:15:05,920 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8808, 1.5186, 3.8385, 1.6816, 3.7349, 3.9738, 2.9378, 4.2048], device='cuda:0'), covar=tensor([0.0237, 0.3155, 0.0450, 0.2334, 0.0555, 0.0381, 0.0726, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0160, 0.0168, 0.0167, 0.0179, 0.0132, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:15:07,646 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:15:25,310 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112147.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:15:44,577 INFO [train.py:873] (0/4) Epoch 15, batch 6300, loss[loss=0.1031, simple_loss=0.144, pruned_loss=0.03112, over 14071.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1473, pruned_loss=0.03884, over 2093410.99 frames. ], batch size: 29, lr: 5.20e-03, grad_scale: 4.0 2022-12-08 08:15:47,438 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:16:07,628 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.242e+02 2.167e+02 2.529e+02 3.043e+02 7.690e+02, threshold=5.057e+02, percent-clipped=2.0 2022-12-08 08:17:13,384 INFO [train.py:873] (0/4) Epoch 15, batch 6400, loss[loss=0.154, simple_loss=0.1695, pruned_loss=0.06925, over 8614.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1474, pruned_loss=0.03906, over 2050798.09 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:17:35,982 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 2.167e+02 2.735e+02 3.656e+02 1.132e+03, threshold=5.470e+02, percent-clipped=11.0 2022-12-08 08:17:48,955 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1927, 2.3706, 2.0579, 2.0465, 1.7556, 1.9541, 1.9132, 1.8504], device='cuda:0'), covar=tensor([0.0279, 0.1062, 0.0449, 0.0493, 0.0531, 0.0579, 0.0478, 0.0697], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:18:15,170 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3763, 4.4305, 4.7966, 4.0404, 4.5671, 4.8421, 1.6874, 4.3231], device='cuda:0'), covar=tensor([0.0276, 0.0292, 0.0277, 0.0436, 0.0289, 0.0177, 0.3265, 0.0266], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0174, 0.0144, 0.0146, 0.0206, 0.0140, 0.0159, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 08:18:41,068 INFO [train.py:873] (0/4) Epoch 15, batch 6500, loss[loss=0.1058, simple_loss=0.118, pruned_loss=0.04679, over 2670.00 frames. ], tot_loss[loss=0.1126, simple_loss=0.1475, pruned_loss=0.03885, over 2072719.34 frames. ], batch size: 100, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:18:42,849 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.80 vs. limit=5.0 2022-12-08 08:18:59,408 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112389.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:19:04,118 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 2.163e+02 2.578e+02 3.269e+02 5.158e+02, threshold=5.155e+02, percent-clipped=0.0 2022-12-08 08:19:28,050 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112422.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:19:35,625 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0171, 3.1029, 3.2127, 3.0678, 3.1529, 3.0281, 1.4996, 2.9444], device='cuda:0'), covar=tensor([0.0422, 0.0433, 0.0382, 0.0426, 0.0347, 0.0841, 0.2944, 0.0376], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0174, 0.0143, 0.0146, 0.0206, 0.0140, 0.0159, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 08:19:50,459 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112447.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:19:53,073 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:20:04,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2022-12-08 08:20:07,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4500, 5.0427, 4.8444, 5.4105, 4.9292, 4.7204, 5.3874, 4.4979], device='cuda:0'), covar=tensor([0.0331, 0.0950, 0.0378, 0.0355, 0.0859, 0.0418, 0.0465, 0.0480], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0268, 0.0195, 0.0191, 0.0185, 0.0155, 0.0281, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 08:20:09,726 INFO [train.py:873] (0/4) Epoch 15, batch 6600, loss[loss=0.1154, simple_loss=0.1529, pruned_loss=0.03892, over 14272.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.147, pruned_loss=0.03866, over 2052591.14 frames. ], batch size: 57, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:20:21,531 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7696, 3.5723, 3.2818, 2.4925, 3.2490, 3.5047, 3.8854, 3.0303], device='cuda:0'), covar=tensor([0.0546, 0.0959, 0.0858, 0.1295, 0.0787, 0.0607, 0.0689, 0.1129], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0172, 0.0138, 0.0125, 0.0139, 0.0151, 0.0131, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 08:20:32,876 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.043e+02 2.558e+02 3.259e+02 6.600e+02, threshold=5.115e+02, percent-clipped=3.0 2022-12-08 08:20:32,968 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112495.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:21:25,368 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2022-12-08 08:21:31,010 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 08:21:32,150 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112562.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:21:37,740 INFO [train.py:873] (0/4) Epoch 15, batch 6700, loss[loss=0.1111, simple_loss=0.1437, pruned_loss=0.03926, over 14313.00 frames. ], tot_loss[loss=0.1121, simple_loss=0.1469, pruned_loss=0.03863, over 2109372.17 frames. ], batch size: 46, lr: 5.19e-03, grad_scale: 8.0 2022-12-08 08:21:44,911 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:21:56,142 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3563, 3.0254, 3.7887, 2.8107, 2.4307, 3.2998, 2.1074, 3.4567], device='cuda:0'), covar=tensor([0.0880, 0.1047, 0.0521, 0.1509, 0.1916, 0.0846, 0.2618, 0.0812], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0093, 0.0099, 0.0116, 0.0089, 0.0120, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 08:21:59,524 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9073, 0.7981, 0.8262, 0.8983, 0.8244, 0.5137, 0.5341, 0.6923], device='cuda:0'), covar=tensor([0.0118, 0.0124, 0.0112, 0.0128, 0.0119, 0.0221, 0.0144, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0019, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:22:01,002 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.231e+02 2.690e+02 3.411e+02 7.258e+02, threshold=5.380e+02, percent-clipped=8.0 2022-12-08 08:22:20,659 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:22:26,031 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112623.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:22:39,559 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112638.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:22:45,878 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2022-12-08 08:22:48,829 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8740, 1.6163, 1.8442, 1.6501, 1.9041, 1.7866, 1.5716, 1.8404], device='cuda:0'), covar=tensor([0.0548, 0.1141, 0.0486, 0.0436, 0.0607, 0.0698, 0.0284, 0.0380], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0308, 0.0391, 0.0296, 0.0368, 0.0323, 0.0359, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:23:06,603 INFO [train.py:873] (0/4) Epoch 15, batch 6800, loss[loss=0.1328, simple_loss=0.1534, pruned_loss=0.05611, over 4942.00 frames. ], tot_loss[loss=0.1125, simple_loss=0.1471, pruned_loss=0.03899, over 2042998.91 frames. ], batch size: 100, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:23:14,248 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112678.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 08:23:26,478 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6695, 1.1005, 1.2626, 1.1295, 0.9078, 1.2440, 0.9756, 0.7756], device='cuda:0'), covar=tensor([0.2097, 0.1150, 0.0431, 0.0488, 0.1881, 0.1139, 0.1532, 0.1563], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0089, 0.0069, 0.0072, 0.0098, 0.0087, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 08:23:28,859 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.249e+02 2.719e+02 3.532e+02 9.106e+02, threshold=5.438e+02, percent-clipped=9.0 2022-12-08 08:23:36,472 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-08 08:23:42,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3696, 1.2958, 1.2835, 1.4075, 1.4289, 0.9900, 1.2179, 1.2581], device='cuda:0'), covar=tensor([0.0572, 0.0913, 0.0575, 0.0717, 0.0749, 0.1037, 0.0953, 0.0859], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:23:52,908 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112722.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:24:13,195 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112745.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:24:34,753 INFO [train.py:873] (0/4) Epoch 15, batch 6900, loss[loss=0.1153, simple_loss=0.1372, pruned_loss=0.04666, over 7768.00 frames. ], tot_loss[loss=0.1132, simple_loss=0.1473, pruned_loss=0.03955, over 2013782.55 frames. ], batch size: 100, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:24:35,710 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:24:57,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.169e+02 2.514e+02 3.179e+02 8.231e+02, threshold=5.028e+02, percent-clipped=2.0 2022-12-08 08:25:17,704 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2022-12-08 08:25:50,131 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2022-12-08 08:26:02,334 INFO [train.py:873] (0/4) Epoch 15, batch 7000, loss[loss=0.1535, simple_loss=0.1447, pruned_loss=0.08119, over 1188.00 frames. ], tot_loss[loss=0.1131, simple_loss=0.1471, pruned_loss=0.0396, over 1955873.56 frames. ], batch size: 100, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:26:25,455 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.195e+02 2.715e+02 3.350e+02 5.262e+02, threshold=5.431e+02, percent-clipped=3.0 2022-12-08 08:26:36,764 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2561, 5.0232, 4.7964, 5.2628, 4.8345, 4.6215, 5.3314, 5.0309], device='cuda:0'), covar=tensor([0.0601, 0.0704, 0.0689, 0.0456, 0.0630, 0.0450, 0.0506, 0.0618], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0141, 0.0146, 0.0160, 0.0147, 0.0123, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:26:39,407 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 08:26:39,599 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-08 08:26:46,153 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:26:48,910 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112921.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:26:59,620 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112933.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 08:27:06,742 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9808, 3.7855, 3.7229, 4.0702, 3.6224, 3.3633, 4.1113, 3.9002], device='cuda:0'), covar=tensor([0.0774, 0.1114, 0.0879, 0.0645, 0.0910, 0.0848, 0.0665, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0142, 0.0146, 0.0160, 0.0147, 0.0123, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:27:31,512 INFO [train.py:873] (0/4) Epoch 15, batch 7100, loss[loss=0.1247, simple_loss=0.155, pruned_loss=0.04714, over 14277.00 frames. ], tot_loss[loss=0.1133, simple_loss=0.1474, pruned_loss=0.03961, over 1981058.13 frames. ], batch size: 66, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:27:35,245 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112973.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 08:27:43,026 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112982.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:27:54,134 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.181e+02 2.817e+02 3.355e+02 7.638e+02, threshold=5.634e+02, percent-clipped=2.0 2022-12-08 08:28:17,448 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4822, 2.2582, 2.7905, 1.8728, 1.8719, 2.5398, 1.5493, 2.4234], device='cuda:0'), covar=tensor([0.1088, 0.1627, 0.0654, 0.1715, 0.2239, 0.0943, 0.3303, 0.1186], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0093, 0.0099, 0.0116, 0.0088, 0.0120, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 08:28:29,484 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.52 vs. limit=5.0 2022-12-08 08:28:38,844 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113045.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:28:43,209 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7073, 3.3539, 2.6463, 3.9637, 3.7823, 3.8154, 3.3033, 2.5208], device='cuda:0'), covar=tensor([0.1006, 0.1367, 0.3370, 0.0506, 0.0876, 0.0973, 0.1323, 0.3830], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0290, 0.0258, 0.0278, 0.0322, 0.0299, 0.0251, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:28:51,901 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 08:28:59,911 INFO [train.py:873] (0/4) Epoch 15, batch 7200, loss[loss=0.1294, simple_loss=0.1568, pruned_loss=0.05099, over 8603.00 frames. ], tot_loss[loss=0.1138, simple_loss=0.1475, pruned_loss=0.04001, over 1918051.94 frames. ], batch size: 100, lr: 5.18e-03, grad_scale: 8.0 2022-12-08 08:29:21,087 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113093.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:29:23,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.228e+02 2.733e+02 3.389e+02 6.488e+02, threshold=5.466e+02, percent-clipped=1.0 2022-12-08 08:30:28,628 INFO [train.py:873] (0/4) Epoch 15, batch 7300, loss[loss=0.1068, simple_loss=0.1429, pruned_loss=0.03533, over 14292.00 frames. ], tot_loss[loss=0.113, simple_loss=0.147, pruned_loss=0.03952, over 1958048.98 frames. ], batch size: 76, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:30:52,685 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.057e+02 2.452e+02 3.236e+02 1.011e+03, threshold=4.903e+02, percent-clipped=3.0 2022-12-08 08:31:04,815 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0486, 1.9736, 1.9658, 2.0576, 1.8898, 1.2906, 2.0226, 2.1793], device='cuda:0'), covar=tensor([0.1177, 0.2230, 0.0716, 0.2609, 0.1192, 0.0935, 0.0728, 0.1470], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:31:12,628 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:22,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8173, 0.7300, 0.7801, 0.8110, 0.8353, 0.6066, 0.5595, 0.7722], device='cuda:0'), covar=tensor([0.0136, 0.0145, 0.0130, 0.0147, 0.0157, 0.0245, 0.0186, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:31:25,823 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:26,794 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113234.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:55,288 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113266.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:31:57,792 INFO [train.py:873] (0/4) Epoch 15, batch 7400, loss[loss=0.1627, simple_loss=0.1443, pruned_loss=0.09055, over 1228.00 frames. ], tot_loss[loss=0.113, simple_loss=0.1467, pruned_loss=0.03961, over 1940870.14 frames. ], batch size: 100, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:32:01,577 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113273.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 08:32:05,498 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113277.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:08,864 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113281.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:17,550 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 08:32:21,664 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113295.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:32:22,353 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.108e+02 1.996e+02 2.516e+02 3.094e+02 5.431e+02, threshold=5.033e+02, percent-clipped=1.0 2022-12-08 08:32:27,895 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5492, 3.2479, 2.5964, 3.6960, 3.5973, 3.6524, 3.1524, 2.6175], device='cuda:0'), covar=tensor([0.0837, 0.1334, 0.3247, 0.0674, 0.0872, 0.0925, 0.1266, 0.3374], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0294, 0.0262, 0.0284, 0.0326, 0.0304, 0.0256, 0.0249], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:32:44,276 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:33:25,701 INFO [train.py:873] (0/4) Epoch 15, batch 7500, loss[loss=0.09952, simple_loss=0.1415, pruned_loss=0.02877, over 14391.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1464, pruned_loss=0.03956, over 1905243.03 frames. ], batch size: 73, lr: 5.17e-03, grad_scale: 4.0 2022-12-08 08:33:48,620 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 2.151e+02 2.525e+02 3.725e+02 6.084e+02, threshold=5.049e+02, percent-clipped=3.0 2022-12-08 08:34:12,700 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-15.pt 2022-12-08 08:34:54,699 INFO [train.py:873] (0/4) Epoch 16, batch 0, loss[loss=0.1284, simple_loss=0.167, pruned_loss=0.04495, over 13569.00 frames. ], tot_loss[loss=0.1284, simple_loss=0.167, pruned_loss=0.04495, over 13569.00 frames. ], batch size: 100, lr: 5.00e-03, grad_scale: 8.0 2022-12-08 08:34:54,699 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 08:35:01,997 INFO [train.py:905] (0/4) Epoch 16, validation: loss=0.1445, simple_loss=0.1858, pruned_loss=0.05158, over 857387.00 frames. 2022-12-08 08:35:01,997 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 08:35:02,510 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2022-12-08 08:35:37,956 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1058, 1.5311, 4.0764, 1.8456, 4.0084, 4.1950, 3.2361, 4.5066], device='cuda:0'), covar=tensor([0.0215, 0.3049, 0.0398, 0.2074, 0.0400, 0.0440, 0.0650, 0.0149], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0155, 0.0159, 0.0168, 0.0167, 0.0179, 0.0133, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:36:01,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.044e+01 2.013e+02 2.778e+02 4.051e+02 1.103e+03, threshold=5.556e+02, percent-clipped=15.0 2022-12-08 08:36:31,855 INFO [train.py:873] (0/4) Epoch 16, batch 100, loss[loss=0.0997, simple_loss=0.1422, pruned_loss=0.02858, over 14314.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1474, pruned_loss=0.03914, over 862431.11 frames. ], batch size: 28, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:37:13,524 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:20,437 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9806, 1.5860, 3.2180, 2.9246, 3.1435, 3.2876, 2.6001, 3.2053], device='cuda:0'), covar=tensor([0.1276, 0.1432, 0.0167, 0.0348, 0.0302, 0.0189, 0.0370, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0160, 0.0132, 0.0171, 0.0148, 0.0143, 0.0125, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 08:37:25,053 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113590.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:31,129 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 2.137e+02 2.757e+02 3.201e+02 6.700e+02, threshold=5.514e+02, percent-clipped=1.0 2022-12-08 08:37:42,226 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5667, 1.6112, 1.6422, 1.5304, 1.4137, 1.4376, 1.2622, 1.1669], device='cuda:0'), covar=tensor([0.0176, 0.0197, 0.0179, 0.0170, 0.0174, 0.0281, 0.0219, 0.0336], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0019, 0.0031, 0.0025, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:37:45,093 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.84 vs. limit=5.0 2022-12-08 08:37:49,133 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113617.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:37:56,322 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:01,673 INFO [train.py:873] (0/4) Epoch 16, batch 200, loss[loss=0.1536, simple_loss=0.173, pruned_loss=0.06716, over 8626.00 frames. ], tot_loss[loss=0.1129, simple_loss=0.1472, pruned_loss=0.03932, over 1251947.84 frames. ], batch size: 100, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:38:14,863 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113646.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:20,657 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 08:38:43,101 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:38:49,901 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1430, 1.1476, 1.0966, 1.1048, 1.2274, 0.7595, 1.0416, 1.0983], device='cuda:0'), covar=tensor([0.0589, 0.0675, 0.0550, 0.0399, 0.0431, 0.0658, 0.0954, 0.0841], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:38:59,610 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.233e+02 2.172e+02 2.621e+02 3.364e+02 2.345e+03, threshold=5.243e+02, percent-clipped=6.0 2022-12-08 08:39:09,092 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:39:17,589 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2812, 3.9572, 3.9070, 4.2819, 4.0334, 3.8774, 4.3314, 3.6552], device='cuda:0'), covar=tensor([0.0459, 0.1005, 0.0432, 0.0465, 0.0791, 0.0915, 0.0521, 0.0524], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0270, 0.0197, 0.0191, 0.0185, 0.0156, 0.0282, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 08:39:30,009 INFO [train.py:873] (0/4) Epoch 16, batch 300, loss[loss=0.1011, simple_loss=0.1439, pruned_loss=0.02915, over 14227.00 frames. ], tot_loss[loss=0.1134, simple_loss=0.1471, pruned_loss=0.03987, over 1504353.77 frames. ], batch size: 60, lr: 5.00e-03, grad_scale: 4.0 2022-12-08 08:40:28,277 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 1.970e+02 2.457e+02 3.061e+02 5.224e+02, threshold=4.913e+02, percent-clipped=0.0 2022-12-08 08:40:57,902 INFO [train.py:873] (0/4) Epoch 16, batch 400, loss[loss=0.2133, simple_loss=0.1846, pruned_loss=0.121, over 1230.00 frames. ], tot_loss[loss=0.1128, simple_loss=0.1468, pruned_loss=0.03934, over 1672353.86 frames. ], batch size: 100, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:41:46,426 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5273, 2.2170, 2.4603, 1.6438, 2.1498, 2.4523, 2.5451, 2.1479], device='cuda:0'), covar=tensor([0.0809, 0.0659, 0.0914, 0.1416, 0.1069, 0.0778, 0.0617, 0.1222], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0175, 0.0140, 0.0128, 0.0143, 0.0155, 0.0132, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 08:41:50,228 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113890.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:41:56,841 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.079e+02 2.042e+02 2.566e+02 3.444e+02 7.891e+02, threshold=5.133e+02, percent-clipped=8.0 2022-12-08 08:42:25,885 INFO [train.py:873] (0/4) Epoch 16, batch 500, loss[loss=0.1198, simple_loss=0.1313, pruned_loss=0.0541, over 3895.00 frames. ], tot_loss[loss=0.1115, simple_loss=0.1464, pruned_loss=0.03832, over 1802828.82 frames. ], batch size: 100, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:42:32,144 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113938.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:42:32,275 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5827, 1.9758, 2.6157, 2.6286, 2.4991, 1.9402, 2.5793, 2.1731], device='cuda:0'), covar=tensor([0.0387, 0.1122, 0.0537, 0.0506, 0.0604, 0.1471, 0.0415, 0.0766], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0259, 0.0374, 0.0329, 0.0272, 0.0306, 0.0310, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:42:54,345 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0552, 2.4169, 3.9592, 4.1583, 3.9485, 2.3664, 4.2373, 3.1905], device='cuda:0'), covar=tensor([0.0503, 0.1229, 0.1244, 0.0500, 0.0585, 0.1877, 0.0406, 0.0979], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0259, 0.0374, 0.0329, 0.0272, 0.0305, 0.0310, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:43:00,767 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0930, 3.1211, 4.7429, 3.6659, 4.8480, 4.6049, 4.5123, 4.1989], device='cuda:0'), covar=tensor([0.0532, 0.2975, 0.0789, 0.1467, 0.0646, 0.0847, 0.1271, 0.1669], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0309, 0.0389, 0.0296, 0.0367, 0.0320, 0.0359, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:43:02,320 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:43:18,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2022-12-08 08:43:24,305 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.003e+02 2.124e+02 2.547e+02 2.966e+02 5.657e+02, threshold=5.095e+02, percent-clipped=2.0 2022-12-08 08:43:28,088 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:43:53,990 INFO [train.py:873] (0/4) Epoch 16, batch 600, loss[loss=0.1322, simple_loss=0.1633, pruned_loss=0.05058, over 14301.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1472, pruned_loss=0.03868, over 1939202.45 frames. ], batch size: 35, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:44:41,126 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2022-12-08 08:44:52,421 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.140e+02 2.657e+02 3.260e+02 5.874e+02, threshold=5.314e+02, percent-clipped=5.0 2022-12-08 08:45:18,442 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:45:19,739 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2022-12-08 08:45:21,945 INFO [train.py:873] (0/4) Epoch 16, batch 700, loss[loss=0.0936, simple_loss=0.1353, pruned_loss=0.02597, over 14233.00 frames. ], tot_loss[loss=0.1115, simple_loss=0.1461, pruned_loss=0.03838, over 1943768.10 frames. ], batch size: 69, lr: 4.99e-03, grad_scale: 4.0 2022-12-08 08:45:54,225 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6905, 2.3222, 2.5868, 1.6678, 2.2060, 2.4396, 2.6423, 2.2710], device='cuda:0'), covar=tensor([0.0720, 0.0697, 0.0876, 0.1474, 0.1197, 0.0836, 0.0779, 0.1243], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0175, 0.0139, 0.0127, 0.0143, 0.0155, 0.0133, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 08:46:08,151 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1948, 2.0806, 4.7577, 4.2162, 4.1254, 4.8597, 4.4793, 4.7963], device='cuda:0'), covar=tensor([0.1319, 0.1296, 0.0086, 0.0196, 0.0211, 0.0094, 0.0146, 0.0102], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0131, 0.0170, 0.0147, 0.0142, 0.0125, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 08:46:12,500 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114188.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:46:20,991 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.016e+02 2.560e+02 3.238e+02 1.602e+03, threshold=5.120e+02, percent-clipped=5.0 2022-12-08 08:46:50,899 INFO [train.py:873] (0/4) Epoch 16, batch 800, loss[loss=0.1055, simple_loss=0.1446, pruned_loss=0.03319, over 14263.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1463, pruned_loss=0.03906, over 1944713.57 frames. ], batch size: 63, lr: 4.98e-03, grad_scale: 8.0 2022-12-08 08:47:28,699 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:47:50,274 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.518e+01 2.109e+02 2.499e+02 3.135e+02 7.663e+02, threshold=4.999e+02, percent-clipped=3.0 2022-12-08 08:47:54,017 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114302.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:48:01,804 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7385, 2.6899, 2.1074, 2.7805, 2.6608, 2.7416, 2.4176, 2.2128], device='cuda:0'), covar=tensor([0.0970, 0.1266, 0.2953, 0.1041, 0.1199, 0.0977, 0.1430, 0.2529], device='cuda:0'), in_proj_covar=tensor([0.0280, 0.0293, 0.0259, 0.0280, 0.0321, 0.0300, 0.0253, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:48:09,876 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:48:18,623 INFO [train.py:873] (0/4) Epoch 16, batch 900, loss[loss=0.09397, simple_loss=0.1342, pruned_loss=0.02687, over 6949.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1462, pruned_loss=0.03909, over 1940394.61 frames. ], batch size: 100, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:48:28,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.23 vs. limit=5.0 2022-12-08 08:48:35,141 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:48:40,987 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4415, 2.1191, 2.1166, 2.1521, 2.0903, 1.2115, 2.4511, 2.4316], device='cuda:0'), covar=tensor([0.0921, 0.0618, 0.1157, 0.2176, 0.1662, 0.0879, 0.0709, 0.1237], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:48:46,092 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8606, 1.7182, 1.7508, 1.8005, 1.7206, 1.0842, 1.6201, 1.6444], device='cuda:0'), covar=tensor([0.0667, 0.0778, 0.0652, 0.0621, 0.0936, 0.0950, 0.0720, 0.1094], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:49:17,095 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.233e+02 2.852e+02 3.607e+02 7.620e+02, threshold=5.705e+02, percent-clipped=7.0 2022-12-08 08:49:45,661 INFO [train.py:873] (0/4) Epoch 16, batch 1000, loss[loss=0.106, simple_loss=0.1492, pruned_loss=0.03144, over 14201.00 frames. ], tot_loss[loss=0.112, simple_loss=0.146, pruned_loss=0.03896, over 1912506.72 frames. ], batch size: 46, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:50:30,761 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:50:43,864 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 08:50:45,034 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.045e+02 2.472e+02 3.169e+02 1.892e+03, threshold=4.944e+02, percent-clipped=6.0 2022-12-08 08:51:13,047 INFO [train.py:873] (0/4) Epoch 16, batch 1100, loss[loss=0.1493, simple_loss=0.1403, pruned_loss=0.07918, over 1265.00 frames. ], tot_loss[loss=0.1119, simple_loss=0.1459, pruned_loss=0.03893, over 1914502.25 frames. ], batch size: 100, lr: 4.98e-03, grad_scale: 4.0 2022-12-08 08:51:35,783 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0852, 2.0214, 2.2196, 2.0598, 1.9071, 1.7608, 1.6403, 1.5753], device='cuda:0'), covar=tensor([0.0240, 0.0477, 0.0323, 0.0539, 0.0366, 0.0515, 0.0478, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0020, 0.0021, 0.0018, 0.0020, 0.0020, 0.0031, 0.0026, 0.0030], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 08:51:57,961 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 08:52:12,292 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.285e+02 2.811e+02 4.159e+02 1.083e+03, threshold=5.623e+02, percent-clipped=11.0 2022-12-08 08:52:41,095 INFO [train.py:873] (0/4) Epoch 16, batch 1200, loss[loss=0.1101, simple_loss=0.1499, pruned_loss=0.03515, over 14097.00 frames. ], tot_loss[loss=0.1124, simple_loss=0.1465, pruned_loss=0.03921, over 1940985.44 frames. ], batch size: 29, lr: 4.98e-03, grad_scale: 8.0 2022-12-08 08:53:25,118 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2022-12-08 08:53:41,425 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 1.943e+02 2.456e+02 3.315e+02 7.474e+02, threshold=4.912e+02, percent-clipped=2.0 2022-12-08 08:53:56,730 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114716.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:54:09,742 INFO [train.py:873] (0/4) Epoch 16, batch 1300, loss[loss=0.1131, simple_loss=0.15, pruned_loss=0.03807, over 14284.00 frames. ], tot_loss[loss=0.112, simple_loss=0.1461, pruned_loss=0.03899, over 1925923.29 frames. ], batch size: 80, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:54:25,310 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8440, 1.5627, 1.7910, 1.9574, 1.3509, 1.7181, 1.6786, 1.8672], device='cuda:0'), covar=tensor([0.0246, 0.0344, 0.0200, 0.0195, 0.0429, 0.0455, 0.0278, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0259, 0.0375, 0.0328, 0.0272, 0.0306, 0.0310, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 08:54:42,648 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4493, 5.2285, 4.7563, 5.0336, 5.0628, 5.3638, 5.4299, 5.3976], device='cuda:0'), covar=tensor([0.0679, 0.0407, 0.2203, 0.2486, 0.0720, 0.0759, 0.0743, 0.0815], device='cuda:0'), in_proj_covar=tensor([0.0387, 0.0266, 0.0445, 0.0567, 0.0343, 0.0442, 0.0391, 0.0389], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:54:49,597 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 08:54:50,244 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:54:56,031 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114783.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:55:09,769 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.098e+02 2.728e+02 3.262e+02 8.407e+02, threshold=5.455e+02, percent-clipped=4.0 2022-12-08 08:55:38,357 INFO [train.py:873] (0/4) Epoch 16, batch 1400, loss[loss=0.1327, simple_loss=0.1586, pruned_loss=0.05337, over 9522.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.146, pruned_loss=0.03829, over 1936141.75 frames. ], batch size: 100, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:55:38,432 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114831.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:55:57,999 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4389, 4.2194, 3.8634, 4.0647, 4.2599, 4.3446, 4.4022, 4.4308], device='cuda:0'), covar=tensor([0.0781, 0.0521, 0.2176, 0.2646, 0.0716, 0.0804, 0.0861, 0.0740], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0266, 0.0444, 0.0565, 0.0344, 0.0442, 0.0390, 0.0389], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 08:56:14,163 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114871.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:56:24,040 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6094, 2.2355, 5.1293, 4.5821, 4.4490, 5.1146, 4.9031, 5.1684], device='cuda:0'), covar=tensor([0.1204, 0.1235, 0.0080, 0.0159, 0.0189, 0.0121, 0.0110, 0.0095], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0131, 0.0170, 0.0147, 0.0142, 0.0125, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 08:56:38,337 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 2.101e+02 2.672e+02 3.477e+02 6.209e+02, threshold=5.345e+02, percent-clipped=3.0 2022-12-08 08:57:06,855 INFO [train.py:873] (0/4) Epoch 16, batch 1500, loss[loss=0.1321, simple_loss=0.1402, pruned_loss=0.06197, over 2650.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1464, pruned_loss=0.03898, over 1920480.65 frames. ], batch size: 100, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:57:07,907 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114932.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:58:07,016 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.137e+02 2.682e+02 3.588e+02 7.078e+02, threshold=5.364e+02, percent-clipped=5.0 2022-12-08 08:58:08,312 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-115000.pt 2022-12-08 08:58:39,013 INFO [train.py:873] (0/4) Epoch 16, batch 1600, loss[loss=0.1075, simple_loss=0.122, pruned_loss=0.04654, over 3865.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1459, pruned_loss=0.03926, over 1916608.71 frames. ], batch size: 100, lr: 4.97e-03, grad_scale: 8.0 2022-12-08 08:59:15,314 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:59:27,657 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115086.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 08:59:39,794 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.275e+02 2.932e+02 3.523e+02 7.695e+02, threshold=5.865e+02, percent-clipped=6.0 2022-12-08 09:00:07,459 INFO [train.py:873] (0/4) Epoch 16, batch 1700, loss[loss=0.1548, simple_loss=0.1463, pruned_loss=0.08168, over 1198.00 frames. ], tot_loss[loss=0.1123, simple_loss=0.1463, pruned_loss=0.03914, over 1971145.06 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:00:21,818 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:00:23,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 09:01:04,127 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7519, 3.4522, 3.2397, 2.3550, 3.1432, 3.4637, 3.7874, 2.9900], device='cuda:0'), covar=tensor([0.0529, 0.0982, 0.0790, 0.1319, 0.0801, 0.0547, 0.0639, 0.1032], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0174, 0.0138, 0.0126, 0.0141, 0.0154, 0.0132, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:01:09,161 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.404e+01 2.088e+02 2.519e+02 3.191e+02 6.447e+02, threshold=5.039e+02, percent-clipped=1.0 2022-12-08 09:01:33,686 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115227.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:01:36,927 INFO [train.py:873] (0/4) Epoch 16, batch 1800, loss[loss=0.1104, simple_loss=0.1505, pruned_loss=0.0351, over 14301.00 frames. ], tot_loss[loss=0.1116, simple_loss=0.1459, pruned_loss=0.03864, over 1948915.02 frames. ], batch size: 31, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:02:05,263 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9078, 1.7353, 2.0396, 1.9992, 1.7273, 1.6888, 1.5875, 1.2078], device='cuda:0'), covar=tensor([0.0173, 0.0477, 0.0185, 0.0199, 0.0271, 0.0308, 0.0270, 0.0424], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:02:38,559 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.091e+02 1.953e+02 2.460e+02 2.998e+02 4.316e+02, threshold=4.920e+02, percent-clipped=0.0 2022-12-08 09:02:46,464 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8844, 1.8537, 1.6138, 1.8514, 1.7618, 1.8320, 1.7857, 1.7320], device='cuda:0'), covar=tensor([0.1009, 0.0863, 0.1890, 0.0900, 0.1175, 0.0692, 0.1387, 0.0941], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0294, 0.0261, 0.0281, 0.0323, 0.0302, 0.0254, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:03:06,676 INFO [train.py:873] (0/4) Epoch 16, batch 1900, loss[loss=0.1706, simple_loss=0.1493, pruned_loss=0.09595, over 1224.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1451, pruned_loss=0.03812, over 1937788.81 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:03:41,560 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115370.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:03:42,474 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:03:43,314 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115372.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:03,390 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9215, 2.1083, 1.9201, 1.6934, 2.1034, 1.2581, 2.1696, 2.0971], device='cuda:0'), covar=tensor([0.1260, 0.0738, 0.0860, 0.2531, 0.1388, 0.1037, 0.0932, 0.1536], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0033, 0.0036, 0.0031, 0.0032, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:04:08,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 2.103e+02 2.568e+02 3.222e+02 1.079e+03, threshold=5.135e+02, percent-clipped=6.0 2022-12-08 09:04:17,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3963, 3.6884, 2.9650, 4.5368, 4.1983, 4.4092, 3.7428, 3.0866], device='cuda:0'), covar=tensor([0.0501, 0.1134, 0.2883, 0.0460, 0.0924, 0.1021, 0.1105, 0.2746], device='cuda:0'), in_proj_covar=tensor([0.0278, 0.0290, 0.0257, 0.0278, 0.0318, 0.0297, 0.0251, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:04:26,546 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:36,093 INFO [train.py:873] (0/4) Epoch 16, batch 2000, loss[loss=0.08887, simple_loss=0.1342, pruned_loss=0.02179, over 14359.00 frames. ], tot_loss[loss=0.1116, simple_loss=0.1459, pruned_loss=0.03858, over 1992925.13 frames. ], batch size: 31, lr: 4.96e-03, grad_scale: 8.0 2022-12-08 09:04:36,291 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115431.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:37,111 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:04:41,690 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0620, 1.8602, 2.0962, 2.3320, 2.0397, 1.8800, 1.7049, 1.2326], device='cuda:0'), covar=tensor([0.0286, 0.0771, 0.0383, 0.0242, 0.0304, 0.0372, 0.0359, 0.0734], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:04:45,844 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115442.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:05:07,771 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9147, 1.8918, 3.0348, 2.3751, 2.8866, 1.9054, 2.4886, 2.9156], device='cuda:0'), covar=tensor([0.1227, 0.3726, 0.0650, 0.3442, 0.1262, 0.2573, 0.1063, 0.0937], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0203, 0.0214, 0.0274, 0.0232, 0.0203, 0.0204, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:05:36,028 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.145e+01 1.975e+02 2.632e+02 3.322e+02 6.213e+02, threshold=5.264e+02, percent-clipped=1.0 2022-12-08 09:05:57,655 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 09:05:59,696 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115527.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:06:03,272 INFO [train.py:873] (0/4) Epoch 16, batch 2100, loss[loss=0.1151, simple_loss=0.1495, pruned_loss=0.04029, over 10349.00 frames. ], tot_loss[loss=0.1115, simple_loss=0.1463, pruned_loss=0.03834, over 1982381.35 frames. ], batch size: 100, lr: 4.96e-03, grad_scale: 4.0 2022-12-08 09:06:42,169 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:06:56,940 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2022-12-08 09:07:05,101 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 1.971e+02 2.573e+02 2.986e+02 1.089e+03, threshold=5.147e+02, percent-clipped=6.0 2022-12-08 09:07:06,398 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2022-12-08 09:07:31,194 INFO [train.py:873] (0/4) Epoch 16, batch 2200, loss[loss=0.1056, simple_loss=0.1441, pruned_loss=0.03355, over 14152.00 frames. ], tot_loss[loss=0.1119, simple_loss=0.1464, pruned_loss=0.03867, over 1980522.01 frames. ], batch size: 37, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:08:08,341 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6922, 3.4493, 3.3553, 3.6918, 3.4609, 3.6697, 3.7409, 3.1570], device='cuda:0'), covar=tensor([0.0624, 0.1020, 0.0530, 0.0530, 0.0856, 0.0394, 0.0576, 0.0567], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0268, 0.0194, 0.0191, 0.0181, 0.0155, 0.0279, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 09:08:32,753 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.098e+02 2.631e+02 3.239e+02 6.147e+02, threshold=5.263e+02, percent-clipped=2.0 2022-12-08 09:08:54,866 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115726.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:08:55,678 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115727.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:08:58,973 INFO [train.py:873] (0/4) Epoch 16, batch 2300, loss[loss=0.1358, simple_loss=0.1541, pruned_loss=0.05876, over 5011.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1452, pruned_loss=0.03807, over 1957734.97 frames. ], batch size: 100, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:09:00,907 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8112, 2.2251, 3.7531, 3.8933, 3.7123, 2.3038, 3.7901, 2.7766], device='cuda:0'), covar=tensor([0.0496, 0.1307, 0.0918, 0.0507, 0.0566, 0.1930, 0.0578, 0.1109], device='cuda:0'), in_proj_covar=tensor([0.0290, 0.0258, 0.0376, 0.0327, 0.0271, 0.0306, 0.0309, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:09:08,323 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115742.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:09:50,743 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115790.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:10:00,477 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.978e+01 2.138e+02 2.481e+02 3.114e+02 7.710e+02, threshold=4.963e+02, percent-clipped=3.0 2022-12-08 09:10:26,760 INFO [train.py:873] (0/4) Epoch 16, batch 2400, loss[loss=0.08872, simple_loss=0.1316, pruned_loss=0.02292, over 14642.00 frames. ], tot_loss[loss=0.1109, simple_loss=0.1457, pruned_loss=0.03812, over 1965887.08 frames. ], batch size: 33, lr: 4.95e-03, grad_scale: 8.0 2022-12-08 09:11:19,552 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115892.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:11:27,448 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.676e+01 2.182e+02 2.667e+02 3.375e+02 7.141e+02, threshold=5.334e+02, percent-clipped=4.0 2022-12-08 09:11:54,002 INFO [train.py:873] (0/4) Epoch 16, batch 2500, loss[loss=0.1058, simple_loss=0.1419, pruned_loss=0.03481, over 11975.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1444, pruned_loss=0.03686, over 1975388.25 frames. ], batch size: 100, lr: 4.95e-03, grad_scale: 8.0 2022-12-08 09:12:13,600 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115953.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:12:37,005 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1058, 1.0764, 1.3068, 1.0524, 1.1079, 0.8785, 0.8349, 0.9204], device='cuda:0'), covar=tensor([0.0246, 0.0306, 0.0187, 0.0327, 0.0259, 0.0432, 0.0377, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0018, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:12:56,497 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.068e+02 2.436e+02 3.138e+02 6.345e+02, threshold=4.873e+02, percent-clipped=1.0 2022-12-08 09:13:06,115 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3442, 4.8765, 4.7395, 5.2537, 4.8976, 4.5754, 5.2596, 4.4038], device='cuda:0'), covar=tensor([0.0333, 0.0867, 0.0403, 0.0411, 0.0740, 0.0462, 0.0468, 0.0472], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0268, 0.0195, 0.0191, 0.0181, 0.0154, 0.0278, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 09:13:07,026 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116014.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:17,508 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:18,394 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:13:21,682 INFO [train.py:873] (0/4) Epoch 16, batch 2600, loss[loss=0.1297, simple_loss=0.1595, pruned_loss=0.04999, over 14315.00 frames. ], tot_loss[loss=0.1103, simple_loss=0.1451, pruned_loss=0.03776, over 1987059.86 frames. ], batch size: 46, lr: 4.95e-03, grad_scale: 4.0 2022-12-08 09:13:59,629 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116074.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:00,544 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:00,626 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116075.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 09:14:23,795 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.134e+02 2.737e+02 3.520e+02 9.731e+02, threshold=5.473e+02, percent-clipped=5.0 2022-12-08 09:14:26,825 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116105.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:36,379 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116116.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:14:49,176 INFO [train.py:873] (0/4) Epoch 16, batch 2700, loss[loss=0.1095, simple_loss=0.1423, pruned_loss=0.03833, over 14232.00 frames. ], tot_loss[loss=0.1109, simple_loss=0.1454, pruned_loss=0.03816, over 1916346.38 frames. ], batch size: 37, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:15:16,483 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8472, 3.6616, 3.5361, 3.9031, 3.5008, 3.2203, 3.9073, 3.7495], device='cuda:0'), covar=tensor([0.0635, 0.0895, 0.0845, 0.0561, 0.0832, 0.0779, 0.0613, 0.0721], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0142, 0.0144, 0.0158, 0.0147, 0.0124, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:15:20,057 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116166.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:15:27,862 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9617, 3.0320, 3.0013, 2.9260, 2.9722, 2.9237, 1.2956, 2.7298], device='cuda:0'), covar=tensor([0.0612, 0.0619, 0.0644, 0.0674, 0.0684, 0.1040, 0.3934, 0.0650], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0172, 0.0145, 0.0145, 0.0205, 0.0141, 0.0156, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 09:15:29,370 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116177.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:15:51,299 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.057e+02 2.028e+02 2.481e+02 3.066e+02 5.287e+02, threshold=4.961e+02, percent-clipped=0.0 2022-12-08 09:16:16,288 INFO [train.py:873] (0/4) Epoch 16, batch 2800, loss[loss=0.1167, simple_loss=0.1497, pruned_loss=0.04185, over 14420.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1453, pruned_loss=0.03802, over 1911167.30 frames. ], batch size: 73, lr: 4.94e-03, grad_scale: 8.0 2022-12-08 09:16:17,405 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6650, 1.9797, 2.6577, 2.7216, 2.6394, 2.0420, 2.7234, 2.1987], device='cuda:0'), covar=tensor([0.0438, 0.1069, 0.0626, 0.0503, 0.0592, 0.1446, 0.0462, 0.0915], device='cuda:0'), in_proj_covar=tensor([0.0294, 0.0262, 0.0380, 0.0331, 0.0275, 0.0308, 0.0311, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:16:31,668 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116248.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:17:19,503 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.003e+02 2.440e+02 3.142e+02 5.921e+02, threshold=4.880e+02, percent-clipped=3.0 2022-12-08 09:17:37,581 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116323.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:17:44,502 INFO [train.py:873] (0/4) Epoch 16, batch 2900, loss[loss=0.1067, simple_loss=0.1294, pruned_loss=0.04194, over 3869.00 frames. ], tot_loss[loss=0.1114, simple_loss=0.1459, pruned_loss=0.03846, over 1927191.57 frames. ], batch size: 100, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:18:02,542 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4313, 2.8414, 4.3258, 3.2956, 4.2041, 4.2416, 4.0338, 3.6887], device='cuda:0'), covar=tensor([0.1053, 0.2941, 0.0953, 0.1645, 0.0791, 0.0880, 0.1712, 0.1648], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0310, 0.0391, 0.0297, 0.0367, 0.0321, 0.0361, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:18:18,460 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116370.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 09:18:21,910 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9287, 2.3975, 3.2285, 2.2266, 2.0371, 2.9602, 1.4380, 2.8247], device='cuda:0'), covar=tensor([0.0980, 0.1348, 0.0577, 0.1664, 0.2089, 0.0791, 0.3371, 0.1032], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0100, 0.0093, 0.0098, 0.0114, 0.0088, 0.0118, 0.0091], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 09:18:30,424 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:18:47,320 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.101e+02 2.585e+02 3.248e+02 5.004e+02, threshold=5.170e+02, percent-clipped=1.0 2022-12-08 09:19:07,523 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6527, 4.5696, 4.2698, 4.7738, 4.2399, 4.1133, 4.7738, 4.4963], device='cuda:0'), covar=tensor([0.0611, 0.0756, 0.0829, 0.0481, 0.0772, 0.0640, 0.0546, 0.0645], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0143, 0.0145, 0.0159, 0.0147, 0.0124, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:19:11,754 INFO [train.py:873] (0/4) Epoch 16, batch 3000, loss[loss=0.1064, simple_loss=0.1468, pruned_loss=0.033, over 13994.00 frames. ], tot_loss[loss=0.1108, simple_loss=0.1456, pruned_loss=0.03793, over 1978312.57 frames. ], batch size: 22, lr: 4.94e-03, grad_scale: 4.0 2022-12-08 09:19:11,755 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 09:19:20,203 INFO [train.py:905] (0/4) Epoch 16, validation: loss=0.1369, simple_loss=0.1741, pruned_loss=0.04986, over 857387.00 frames. 2022-12-08 09:19:20,204 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 09:19:32,934 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116446.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:19:45,757 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116461.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:19:51,348 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 09:19:55,312 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:20:22,409 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.145e+02 2.640e+02 3.351e+02 6.789e+02, threshold=5.280e+02, percent-clipped=3.0 2022-12-08 09:20:24,278 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:20:26,035 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116507.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 09:20:46,689 INFO [train.py:873] (0/4) Epoch 16, batch 3100, loss[loss=0.1269, simple_loss=0.1545, pruned_loss=0.0496, over 14210.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1458, pruned_loss=0.03835, over 1991880.81 frames. ], batch size: 94, lr: 4.93e-03, grad_scale: 4.0 2022-12-08 09:21:01,885 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116548.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:21:17,539 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116566.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:21:43,604 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116596.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:21:49,521 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.181e+02 2.695e+02 3.453e+02 7.270e+02, threshold=5.390e+02, percent-clipped=4.0 2022-12-08 09:21:51,431 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7762, 2.5016, 2.6909, 1.7501, 2.2832, 2.6887, 2.7819, 2.3394], device='cuda:0'), covar=tensor([0.0798, 0.0830, 0.0868, 0.1374, 0.1114, 0.0674, 0.0752, 0.1184], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0174, 0.0139, 0.0126, 0.0142, 0.0153, 0.0133, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:22:02,729 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116618.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:22:13,975 INFO [train.py:873] (0/4) Epoch 16, batch 3200, loss[loss=0.1031, simple_loss=0.1449, pruned_loss=0.0306, over 14405.00 frames. ], tot_loss[loss=0.1114, simple_loss=0.1459, pruned_loss=0.03843, over 1954310.54 frames. ], batch size: 53, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:22:30,255 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2022-12-08 09:22:48,035 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116670.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:22:55,579 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:22:55,650 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:23:16,520 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.227e+02 2.604e+02 3.188e+02 6.019e+02, threshold=5.207e+02, percent-clipped=1.0 2022-12-08 09:23:17,670 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:23:29,706 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:23:37,626 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0673, 2.7116, 2.8559, 1.8759, 2.5310, 2.8451, 3.0953, 2.5110], device='cuda:0'), covar=tensor([0.0655, 0.1034, 0.0802, 0.1412, 0.1020, 0.0687, 0.0596, 0.1137], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0173, 0.0139, 0.0126, 0.0142, 0.0153, 0.0133, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:23:41,059 INFO [train.py:873] (0/4) Epoch 16, batch 3300, loss[loss=0.1015, simple_loss=0.1423, pruned_loss=0.03031, over 14588.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1453, pruned_loss=0.03793, over 1974995.35 frames. ], batch size: 21, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:24:07,215 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116761.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:10,649 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:16,803 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:43,071 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116802.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:24:43,146 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116802.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:43,678 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 1.987e+02 2.396e+02 2.955e+02 5.766e+02, threshold=4.792e+02, percent-clipped=2.0 2022-12-08 09:24:48,965 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116809.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:24:58,329 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:25:00,524 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5455, 2.3513, 2.7847, 1.8740, 1.8607, 2.4973, 1.4637, 2.4765], device='cuda:0'), covar=tensor([0.0898, 0.1414, 0.0697, 0.2229, 0.2444, 0.0942, 0.3383, 0.0866], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0094, 0.0099, 0.0116, 0.0089, 0.0119, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 09:25:02,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 09:25:08,247 INFO [train.py:873] (0/4) Epoch 16, batch 3400, loss[loss=0.1282, simple_loss=0.1602, pruned_loss=0.04806, over 14673.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.145, pruned_loss=0.03764, over 1984177.35 frames. ], batch size: 23, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:25:34,032 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116861.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:25:35,791 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116863.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:26:10,487 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.102e+02 2.045e+02 2.562e+02 3.256e+02 5.626e+02, threshold=5.125e+02, percent-clipped=2.0 2022-12-08 09:26:35,745 INFO [train.py:873] (0/4) Epoch 16, batch 3500, loss[loss=0.1083, simple_loss=0.1469, pruned_loss=0.03483, over 14649.00 frames. ], tot_loss[loss=0.1105, simple_loss=0.145, pruned_loss=0.038, over 1928238.74 frames. ], batch size: 23, lr: 4.93e-03, grad_scale: 8.0 2022-12-08 09:26:59,093 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9159, 5.8226, 5.5206, 6.1259, 5.2689, 5.3184, 6.1094, 5.6613], device='cuda:0'), covar=tensor([0.0767, 0.0680, 0.0771, 0.0511, 0.0954, 0.0476, 0.0527, 0.0832], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0143, 0.0145, 0.0159, 0.0146, 0.0124, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:27:12,586 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:27:16,942 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116979.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:27:38,633 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.091e+02 2.141e+02 2.605e+02 3.184e+02 8.184e+02, threshold=5.210e+02, percent-clipped=2.0 2022-12-08 09:27:58,644 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:28:02,010 INFO [train.py:873] (0/4) Epoch 16, batch 3600, loss[loss=0.1043, simple_loss=0.1388, pruned_loss=0.03489, over 10311.00 frames. ], tot_loss[loss=0.111, simple_loss=0.1453, pruned_loss=0.03833, over 1925188.41 frames. ], batch size: 100, lr: 4.92e-03, grad_scale: 8.0 2022-12-08 09:28:12,290 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 09:28:27,284 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117060.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:28:40,563 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3354, 1.3102, 1.2241, 1.4036, 1.4002, 1.0110, 1.1980, 1.3544], device='cuda:0'), covar=tensor([0.0767, 0.0920, 0.0845, 0.0815, 0.0608, 0.1136, 0.1101, 0.0771], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0033, 0.0036, 0.0031, 0.0033, 0.0046, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:28:43,118 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0580, 1.4655, 3.2089, 1.5780, 3.1197, 3.1895, 2.2247, 3.4117], device='cuda:0'), covar=tensor([0.0296, 0.3041, 0.0410, 0.2160, 0.0780, 0.0453, 0.1166, 0.0260], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0157, 0.0161, 0.0170, 0.0169, 0.0182, 0.0135, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:29:03,673 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117102.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:06,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.154e+02 2.650e+02 3.296e+02 7.753e+02, threshold=5.300e+02, percent-clipped=2.0 2022-12-08 09:29:28,804 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2022-12-08 09:29:28,980 INFO [train.py:873] (0/4) Epoch 16, batch 3700, loss[loss=0.1493, simple_loss=0.1702, pruned_loss=0.06424, over 9502.00 frames. ], tot_loss[loss=0.1113, simple_loss=0.1455, pruned_loss=0.03852, over 1907796.24 frames. ], batch size: 100, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:29:45,672 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117150.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:52,527 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117158.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:29:55,064 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117161.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:08,159 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9138, 2.3675, 3.9195, 4.0869, 3.9182, 2.4515, 3.9192, 3.1600], device='cuda:0'), covar=tensor([0.0467, 0.1207, 0.0935, 0.0476, 0.0463, 0.1733, 0.0482, 0.0909], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0259, 0.0376, 0.0328, 0.0272, 0.0306, 0.0310, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:30:11,764 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2022-12-08 09:30:34,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.079e+02 2.495e+02 3.268e+02 6.334e+02, threshold=4.990e+02, percent-clipped=4.0 2022-12-08 09:30:38,006 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117209.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:48,088 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:30:57,050 INFO [train.py:873] (0/4) Epoch 16, batch 3800, loss[loss=0.1088, simple_loss=0.146, pruned_loss=0.03579, over 14290.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.1455, pruned_loss=0.03829, over 1922002.13 frames. ], batch size: 76, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:31:29,636 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5321, 2.3867, 2.0352, 2.1531, 2.4552, 2.4559, 2.4820, 2.4753], device='cuda:0'), covar=tensor([0.1372, 0.1080, 0.3445, 0.3283, 0.1521, 0.1719, 0.1984, 0.1326], device='cuda:0'), in_proj_covar=tensor([0.0386, 0.0269, 0.0449, 0.0562, 0.0346, 0.0444, 0.0387, 0.0391], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:31:34,926 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117274.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:31:41,899 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117282.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:32:02,069 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 2.002e+02 2.660e+02 3.371e+02 8.415e+02, threshold=5.321e+02, percent-clipped=3.0 2022-12-08 09:32:17,220 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117322.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:32:23,629 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:32:25,200 INFO [train.py:873] (0/4) Epoch 16, batch 3900, loss[loss=0.0844, simple_loss=0.1316, pruned_loss=0.0186, over 13580.00 frames. ], tot_loss[loss=0.1105, simple_loss=0.1451, pruned_loss=0.03794, over 1928467.34 frames. ], batch size: 17, lr: 4.92e-03, grad_scale: 4.0 2022-12-08 09:32:37,325 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6040, 1.6528, 2.7792, 2.1564, 2.5939, 1.7083, 2.2814, 2.6044], device='cuda:0'), covar=tensor([0.1365, 0.3817, 0.0658, 0.3553, 0.1080, 0.2906, 0.1096, 0.0868], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0201, 0.0213, 0.0271, 0.0231, 0.0204, 0.0201, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 09:32:51,281 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117360.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:17,700 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:18,724 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 09:33:24,087 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5408, 1.9406, 3.5847, 2.5654, 3.4617, 1.8494, 2.7288, 3.5067], device='cuda:0'), covar=tensor([0.0768, 0.3808, 0.0600, 0.5174, 0.0753, 0.3189, 0.1332, 0.0569], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0203, 0.0215, 0.0273, 0.0232, 0.0206, 0.0203, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 09:33:30,799 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 1.882e+02 2.302e+02 2.791e+02 6.246e+02, threshold=4.605e+02, percent-clipped=1.0 2022-12-08 09:33:33,449 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117408.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:33:46,456 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7671, 4.6332, 4.0409, 4.3008, 4.5620, 4.7636, 4.8621, 4.8592], device='cuda:0'), covar=tensor([0.1301, 0.0583, 0.2519, 0.3707, 0.0944, 0.1158, 0.1184, 0.1040], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0269, 0.0449, 0.0562, 0.0346, 0.0444, 0.0386, 0.0390], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:33:53,716 INFO [train.py:873] (0/4) Epoch 16, batch 4000, loss[loss=0.1293, simple_loss=0.1545, pruned_loss=0.05204, over 11171.00 frames. ], tot_loss[loss=0.1103, simple_loss=0.1449, pruned_loss=0.03788, over 1938028.17 frames. ], batch size: 100, lr: 4.92e-03, grad_scale: 8.0 2022-12-08 09:34:17,494 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117458.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:34:31,428 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 09:34:58,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.165e+02 2.594e+02 3.323e+02 1.177e+03, threshold=5.188e+02, percent-clipped=6.0 2022-12-08 09:35:00,049 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:35:22,146 INFO [train.py:873] (0/4) Epoch 16, batch 4100, loss[loss=0.1356, simple_loss=0.1653, pruned_loss=0.05292, over 10317.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.1445, pruned_loss=0.03782, over 1901576.23 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:35:37,668 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2453, 2.1875, 3.1386, 3.3239, 3.2191, 2.1232, 3.2191, 2.4514], device='cuda:0'), covar=tensor([0.0512, 0.1252, 0.0901, 0.0498, 0.0640, 0.1850, 0.0472, 0.1157], device='cuda:0'), in_proj_covar=tensor([0.0292, 0.0259, 0.0375, 0.0328, 0.0272, 0.0305, 0.0308, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:35:37,961 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 09:36:02,192 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117577.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:36:12,044 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1034, 1.0558, 1.0966, 0.9460, 1.0044, 0.7951, 0.8690, 0.8642], device='cuda:0'), covar=tensor([0.0232, 0.0216, 0.0171, 0.0224, 0.0232, 0.0406, 0.0284, 0.0352], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:36:27,068 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.013e+02 2.651e+02 3.289e+02 6.397e+02, threshold=5.302e+02, percent-clipped=7.0 2022-12-08 09:36:49,546 INFO [train.py:873] (0/4) Epoch 16, batch 4200, loss[loss=0.1218, simple_loss=0.1493, pruned_loss=0.04709, over 14019.00 frames. ], tot_loss[loss=0.1104, simple_loss=0.1449, pruned_loss=0.03794, over 1906447.11 frames. ], batch size: 19, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:37:37,435 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:37:53,554 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7514, 1.6200, 1.8039, 1.6707, 1.5963, 1.5574, 1.5014, 1.2119], device='cuda:0'), covar=tensor([0.0215, 0.0250, 0.0212, 0.0205, 0.0203, 0.0334, 0.0266, 0.0435], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:37:54,404 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.263e+02 2.184e+02 2.504e+02 3.069e+02 5.464e+02, threshold=5.008e+02, percent-clipped=1.0 2022-12-08 09:38:17,332 INFO [train.py:873] (0/4) Epoch 16, batch 4300, loss[loss=0.1016, simple_loss=0.1381, pruned_loss=0.03254, over 13969.00 frames. ], tot_loss[loss=0.1115, simple_loss=0.1458, pruned_loss=0.03864, over 1938952.89 frames. ], batch size: 19, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:39:22,644 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0323, 3.8226, 3.7868, 4.0845, 3.7234, 3.4719, 4.0987, 3.8921], device='cuda:0'), covar=tensor([0.0658, 0.0897, 0.0841, 0.0685, 0.0916, 0.0803, 0.0623, 0.0816], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0143, 0.0146, 0.0161, 0.0146, 0.0124, 0.0169, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:39:23,470 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.083e+02 2.539e+02 3.211e+02 7.278e+02, threshold=5.077e+02, percent-clipped=7.0 2022-12-08 09:39:32,438 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3700, 3.6435, 3.6325, 3.4036, 2.6831, 3.7643, 3.4703, 2.0460], device='cuda:0'), covar=tensor([0.1476, 0.0919, 0.0677, 0.0858, 0.0857, 0.0422, 0.1128, 0.1753], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0088, 0.0069, 0.0072, 0.0098, 0.0087, 0.0100, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:39:46,099 INFO [train.py:873] (0/4) Epoch 16, batch 4400, loss[loss=0.1146, simple_loss=0.1504, pruned_loss=0.03941, over 14156.00 frames. ], tot_loss[loss=0.1111, simple_loss=0.1454, pruned_loss=0.03845, over 1935881.68 frames. ], batch size: 84, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:40:09,310 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1407, 3.9454, 3.5956, 3.7943, 4.0188, 4.0918, 4.1353, 4.1549], device='cuda:0'), covar=tensor([0.0929, 0.0554, 0.2057, 0.2679, 0.0747, 0.0873, 0.0848, 0.0774], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0268, 0.0448, 0.0563, 0.0347, 0.0443, 0.0384, 0.0390], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:40:26,870 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117877.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:40:32,149 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1362, 3.1143, 2.9756, 3.2392, 2.8279, 2.8365, 3.1946, 3.0955], device='cuda:0'), covar=tensor([0.0725, 0.1043, 0.0917, 0.0669, 0.1206, 0.0840, 0.0833, 0.0844], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0143, 0.0145, 0.0160, 0.0146, 0.0124, 0.0169, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:40:50,806 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.193e+02 2.093e+02 2.556e+02 3.184e+02 6.806e+02, threshold=5.111e+02, percent-clipped=1.0 2022-12-08 09:41:09,096 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:41:14,049 INFO [train.py:873] (0/4) Epoch 16, batch 4500, loss[loss=0.1218, simple_loss=0.147, pruned_loss=0.0483, over 5970.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1452, pruned_loss=0.03801, over 1982913.58 frames. ], batch size: 100, lr: 4.91e-03, grad_scale: 8.0 2022-12-08 09:42:01,224 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:42:18,930 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.027e+02 2.156e+02 2.695e+02 3.844e+02 9.084e+02, threshold=5.389e+02, percent-clipped=7.0 2022-12-08 09:42:30,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9087, 2.5125, 3.6962, 2.8095, 3.7376, 3.6289, 3.5469, 3.1776], device='cuda:0'), covar=tensor([0.0933, 0.2955, 0.1063, 0.1779, 0.0773, 0.0987, 0.1401, 0.1580], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0310, 0.0393, 0.0299, 0.0367, 0.0323, 0.0361, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:42:41,604 INFO [train.py:873] (0/4) Epoch 16, batch 4600, loss[loss=0.09139, simple_loss=0.1387, pruned_loss=0.02202, over 14459.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1454, pruned_loss=0.03788, over 1958593.92 frames. ], batch size: 24, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:42:43,371 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:42:50,031 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:43:07,147 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0370, 2.8968, 5.0131, 3.3824, 4.8177, 2.4991, 3.8717, 4.8068], device='cuda:0'), covar=tensor([0.0480, 0.3275, 0.0422, 0.5825, 0.0529, 0.3077, 0.1136, 0.0433], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0204, 0.0217, 0.0273, 0.0236, 0.0209, 0.0204, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 09:43:10,910 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.53 vs. limit=5.0 2022-12-08 09:43:42,429 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118101.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:43:45,464 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.071e+02 2.553e+02 3.122e+02 5.122e+02, threshold=5.107e+02, percent-clipped=0.0 2022-12-08 09:44:08,015 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2022-12-08 09:44:08,310 INFO [train.py:873] (0/4) Epoch 16, batch 4700, loss[loss=0.126, simple_loss=0.1325, pruned_loss=0.0597, over 3813.00 frames. ], tot_loss[loss=0.1102, simple_loss=0.145, pruned_loss=0.03765, over 1952130.74 frames. ], batch size: 100, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:44:09,179 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1001, 3.8842, 3.8405, 4.1121, 3.6316, 3.4479, 4.1461, 3.9294], device='cuda:0'), covar=tensor([0.0647, 0.0955, 0.0781, 0.0632, 0.0931, 0.0728, 0.0554, 0.0744], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0142, 0.0144, 0.0159, 0.0145, 0.0123, 0.0166, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:44:40,861 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118168.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:45:13,140 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 2.277e+02 2.708e+02 3.256e+02 5.977e+02, threshold=5.416e+02, percent-clipped=1.0 2022-12-08 09:45:33,927 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118229.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 09:45:35,651 INFO [train.py:873] (0/4) Epoch 16, batch 4800, loss[loss=0.1232, simple_loss=0.1585, pruned_loss=0.04394, over 14256.00 frames. ], tot_loss[loss=0.1108, simple_loss=0.1455, pruned_loss=0.03807, over 1970412.79 frames. ], batch size: 76, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:46:08,490 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:46:18,493 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2022-12-08 09:46:26,225 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:46:39,569 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 2.128e+02 2.595e+02 3.127e+02 6.386e+02, threshold=5.189e+02, percent-clipped=3.0 2022-12-08 09:47:00,719 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:01,549 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118330.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:02,255 INFO [train.py:873] (0/4) Epoch 16, batch 4900, loss[loss=0.1187, simple_loss=0.142, pruned_loss=0.04768, over 4974.00 frames. ], tot_loss[loss=0.1104, simple_loss=0.1452, pruned_loss=0.03781, over 1961831.38 frames. ], batch size: 100, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:47:18,631 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:53,105 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:47:58,323 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:48:06,579 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.054e+02 2.113e+02 2.635e+02 3.192e+02 6.518e+02, threshold=5.270e+02, percent-clipped=3.0 2022-12-08 09:48:29,343 INFO [train.py:873] (0/4) Epoch 16, batch 5000, loss[loss=0.1101, simple_loss=0.1509, pruned_loss=0.03464, over 14209.00 frames. ], tot_loss[loss=0.11, simple_loss=0.145, pruned_loss=0.03754, over 1905416.17 frames. ], batch size: 94, lr: 4.90e-03, grad_scale: 8.0 2022-12-08 09:48:56,330 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9254, 1.7834, 2.1636, 1.9355, 1.6449, 1.6375, 1.3406, 1.1169], device='cuda:0'), covar=tensor([0.0282, 0.0596, 0.0207, 0.0291, 0.0487, 0.0357, 0.0354, 0.0715], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0026, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:48:59,615 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0452, 3.8199, 3.7398, 4.1237, 3.8545, 3.5884, 4.1342, 3.4960], device='cuda:0'), covar=tensor([0.0528, 0.0794, 0.0420, 0.0424, 0.0703, 0.1591, 0.0508, 0.0521], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0271, 0.0197, 0.0190, 0.0182, 0.0155, 0.0281, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 09:49:34,275 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.516e+01 2.019e+02 2.454e+02 2.999e+02 6.382e+02, threshold=4.907e+02, percent-clipped=1.0 2022-12-08 09:49:45,152 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6910, 4.0398, 3.0987, 4.9208, 4.3611, 4.7258, 4.1035, 3.5771], device='cuda:0'), covar=tensor([0.0545, 0.1042, 0.3276, 0.0546, 0.0887, 0.1066, 0.0963, 0.2496], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0293, 0.0260, 0.0284, 0.0323, 0.0299, 0.0253, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:49:50,425 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118524.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 09:49:56,304 INFO [train.py:873] (0/4) Epoch 16, batch 5100, loss[loss=0.1071, simple_loss=0.1438, pruned_loss=0.03523, over 5964.00 frames. ], tot_loss[loss=0.1101, simple_loss=0.1449, pruned_loss=0.03767, over 1948909.45 frames. ], batch size: 100, lr: 4.89e-03, grad_scale: 4.0 2022-12-08 09:50:12,664 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8927, 1.4149, 3.0629, 1.4901, 3.1816, 3.0439, 2.1815, 3.2493], device='cuda:0'), covar=tensor([0.0294, 0.2859, 0.0383, 0.2112, 0.0330, 0.0426, 0.1011, 0.0245], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0155, 0.0160, 0.0168, 0.0165, 0.0179, 0.0132, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:51:01,826 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.028e+02 2.469e+02 3.033e+02 6.307e+02, threshold=4.938e+02, percent-clipped=1.0 2022-12-08 09:51:05,542 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7904, 1.6999, 2.9606, 2.1717, 2.7931, 1.7711, 2.2417, 2.7774], device='cuda:0'), covar=tensor([0.1363, 0.4438, 0.0700, 0.4341, 0.0997, 0.3441, 0.1513, 0.1126], device='cuda:0'), in_proj_covar=tensor([0.0257, 0.0205, 0.0218, 0.0278, 0.0238, 0.0207, 0.0205, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 09:51:18,452 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118625.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:51:23,925 INFO [train.py:873] (0/4) Epoch 16, batch 5200, loss[loss=0.09682, simple_loss=0.1296, pruned_loss=0.03204, over 6958.00 frames. ], tot_loss[loss=0.1103, simple_loss=0.1449, pruned_loss=0.03784, over 1934958.59 frames. ], batch size: 100, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:51:24,126 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118631.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:51:36,598 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118645.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:12,031 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:18,073 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:21,201 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118696.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:52:29,670 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.100e+02 1.977e+02 2.482e+02 3.057e+02 6.335e+02, threshold=4.964e+02, percent-clipped=4.0 2022-12-08 09:52:51,981 INFO [train.py:873] (0/4) Epoch 16, batch 5300, loss[loss=0.1172, simple_loss=0.1416, pruned_loss=0.04641, over 3934.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1446, pruned_loss=0.03707, over 1989274.90 frames. ], batch size: 100, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:53:03,338 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118744.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:53:03,486 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6603, 1.9885, 2.4976, 2.1656, 2.5952, 2.5747, 2.3684, 2.3614], device='cuda:0'), covar=tensor([0.0761, 0.2736, 0.0978, 0.1574, 0.0575, 0.1016, 0.0939, 0.1306], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0311, 0.0396, 0.0300, 0.0371, 0.0325, 0.0360, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:53:58,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 1.964e+02 2.394e+02 2.893e+02 4.670e+02, threshold=4.788e+02, percent-clipped=0.0 2022-12-08 09:54:00,833 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5061, 1.0814, 2.0347, 1.7831, 1.8499, 2.0963, 1.4422, 2.0526], device='cuda:0'), covar=tensor([0.0839, 0.1451, 0.0262, 0.0550, 0.0665, 0.0291, 0.0707, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0161, 0.0132, 0.0172, 0.0150, 0.0143, 0.0125, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 09:54:01,646 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3452, 1.8496, 2.2783, 1.9724, 2.3395, 2.1582, 2.0718, 2.1368], device='cuda:0'), covar=tensor([0.0672, 0.2027, 0.0613, 0.1249, 0.0479, 0.1003, 0.0655, 0.0948], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0310, 0.0393, 0.0299, 0.0369, 0.0322, 0.0358, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:54:10,956 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8087, 1.6625, 1.8978, 1.6250, 2.0251, 1.7817, 1.6059, 1.8603], device='cuda:0'), covar=tensor([0.0712, 0.1303, 0.0485, 0.0539, 0.0430, 0.0631, 0.0309, 0.0411], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0310, 0.0392, 0.0298, 0.0368, 0.0322, 0.0357, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:54:14,775 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:54:20,970 INFO [train.py:873] (0/4) Epoch 16, batch 5400, loss[loss=0.1039, simple_loss=0.1456, pruned_loss=0.03111, over 14328.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1453, pruned_loss=0.03794, over 1951699.30 frames. ], batch size: 55, lr: 4.89e-03, grad_scale: 8.0 2022-12-08 09:54:57,497 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:55:12,398 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6053, 4.4153, 4.1610, 4.6591, 4.2235, 3.9719, 4.6996, 4.4692], device='cuda:0'), covar=tensor([0.0621, 0.0740, 0.0876, 0.0542, 0.0782, 0.0612, 0.0571, 0.0680], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0142, 0.0145, 0.0158, 0.0146, 0.0123, 0.0167, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 09:55:27,015 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.618e+01 2.141e+02 2.825e+02 3.414e+02 6.875e+02, threshold=5.649e+02, percent-clipped=6.0 2022-12-08 09:55:44,014 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118925.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:55:49,198 INFO [train.py:873] (0/4) Epoch 16, batch 5500, loss[loss=0.09149, simple_loss=0.1324, pruned_loss=0.0253, over 14058.00 frames. ], tot_loss[loss=0.1107, simple_loss=0.1454, pruned_loss=0.03797, over 1942751.23 frames. ], batch size: 22, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:56:01,702 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:18,818 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.34 vs. limit=5.0 2022-12-08 09:56:19,887 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 09:56:21,384 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 09:56:26,065 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118973.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:36,622 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:38,431 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118987.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:43,691 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118993.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:56:43,820 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5201, 2.9843, 3.0011, 2.1636, 2.9659, 3.2085, 3.6311, 2.6773], device='cuda:0'), covar=tensor([0.0651, 0.1176, 0.1084, 0.1678, 0.0998, 0.0800, 0.0661, 0.1424], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0171, 0.0139, 0.0125, 0.0141, 0.0153, 0.0131, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:56:44,240 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2022-12-08 09:56:55,520 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.181e+02 2.740e+02 3.306e+02 6.665e+02, threshold=5.480e+02, percent-clipped=3.0 2022-12-08 09:57:06,129 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8946, 1.9119, 1.8689, 1.8293, 1.8861, 1.2522, 1.5875, 1.6964], device='cuda:0'), covar=tensor([0.0733, 0.0786, 0.0658, 0.0977, 0.0724, 0.0924, 0.1004, 0.1080], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0047, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 09:57:17,516 INFO [train.py:873] (0/4) Epoch 16, batch 5600, loss[loss=0.1028, simple_loss=0.1468, pruned_loss=0.02941, over 14532.00 frames. ], tot_loss[loss=0.1109, simple_loss=0.1456, pruned_loss=0.0381, over 2022486.90 frames. ], batch size: 43, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:57:19,243 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:57:47,877 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2022-12-08 09:57:55,356 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:58:11,194 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119093.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:58:21,971 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.186e+02 2.774e+02 3.458e+02 6.793e+02, threshold=5.547e+02, percent-clipped=4.0 2022-12-08 09:58:27,542 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6192, 2.3104, 2.5199, 1.7498, 2.1789, 2.5169, 2.6153, 2.2447], device='cuda:0'), covar=tensor([0.0795, 0.0797, 0.0902, 0.1343, 0.1122, 0.0821, 0.0732, 0.1213], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0172, 0.0140, 0.0126, 0.0143, 0.0154, 0.0132, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 09:58:43,943 INFO [train.py:873] (0/4) Epoch 16, batch 5700, loss[loss=0.1027, simple_loss=0.1444, pruned_loss=0.03047, over 14288.00 frames. ], tot_loss[loss=0.1122, simple_loss=0.1465, pruned_loss=0.03894, over 1989875.26 frames. ], batch size: 44, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 09:58:48,243 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 09:59:04,282 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119154.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 09:59:39,220 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5837, 3.3936, 3.1485, 3.3056, 3.5412, 3.4830, 3.5901, 3.5674], device='cuda:0'), covar=tensor([0.1009, 0.0590, 0.2051, 0.2610, 0.0743, 0.1082, 0.0952, 0.0889], device='cuda:0'), in_proj_covar=tensor([0.0392, 0.0269, 0.0451, 0.0566, 0.0350, 0.0448, 0.0392, 0.0389], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 09:59:50,092 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.314e+02 2.766e+02 3.252e+02 7.674e+02, threshold=5.531e+02, percent-clipped=3.0 2022-12-08 10:00:11,370 INFO [train.py:873] (0/4) Epoch 16, batch 5800, loss[loss=0.1143, simple_loss=0.1531, pruned_loss=0.03772, over 14285.00 frames. ], tot_loss[loss=0.1114, simple_loss=0.1459, pruned_loss=0.03847, over 2000914.89 frames. ], batch size: 31, lr: 4.88e-03, grad_scale: 8.0 2022-12-08 10:00:48,124 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 10:01:00,882 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:01:06,921 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8561, 1.7499, 2.9720, 2.2621, 2.8591, 1.8830, 2.3649, 2.8262], device='cuda:0'), covar=tensor([0.1143, 0.3786, 0.0710, 0.3415, 0.0926, 0.2962, 0.1187, 0.0954], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0203, 0.0217, 0.0275, 0.0236, 0.0207, 0.0202, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 10:01:17,095 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.269e+02 2.799e+02 3.367e+02 7.852e+02, threshold=5.598e+02, percent-clipped=4.0 2022-12-08 10:01:22,344 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2665, 2.3581, 2.0802, 1.9317, 1.8410, 1.8495, 1.9049, 1.9940], device='cuda:0'), covar=tensor([0.0393, 0.0359, 0.0488, 0.0626, 0.0443, 0.0588, 0.0710, 0.0697], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:01:25,076 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2057, 4.0802, 3.8958, 4.2287, 3.8549, 3.5472, 4.2552, 4.0139], device='cuda:0'), covar=tensor([0.0607, 0.0858, 0.0900, 0.0648, 0.0830, 0.0762, 0.0573, 0.0854], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0143, 0.0145, 0.0158, 0.0146, 0.0122, 0.0167, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:01:38,987 INFO [train.py:873] (0/4) Epoch 16, batch 5900, loss[loss=0.1104, simple_loss=0.137, pruned_loss=0.04196, over 6915.00 frames. ], tot_loss[loss=0.1094, simple_loss=0.1446, pruned_loss=0.03708, over 1996126.29 frames. ], batch size: 100, lr: 4.88e-03, grad_scale: 4.0 2022-12-08 10:01:42,504 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:01:44,038 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2564, 5.1563, 4.5770, 4.8968, 5.0664, 5.2728, 5.3699, 5.3294], device='cuda:0'), covar=tensor([0.1168, 0.0528, 0.2686, 0.2843, 0.0847, 0.0928, 0.0996, 0.0877], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0269, 0.0451, 0.0566, 0.0350, 0.0449, 0.0393, 0.0388], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:02:01,477 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0204, 2.4431, 4.0646, 4.1539, 4.0256, 2.5181, 4.1762, 3.2285], device='cuda:0'), covar=tensor([0.0459, 0.1268, 0.0821, 0.0479, 0.0451, 0.1780, 0.0431, 0.0946], device='cuda:0'), in_proj_covar=tensor([0.0292, 0.0259, 0.0375, 0.0328, 0.0272, 0.0303, 0.0310, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:02:45,457 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 10:02:45,650 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.147e+02 2.622e+02 3.177e+02 5.866e+02, threshold=5.244e+02, percent-clipped=1.0 2022-12-08 10:02:51,981 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9524, 0.9070, 0.9850, 0.8930, 0.8691, 0.7746, 0.8692, 0.8923], device='cuda:0'), covar=tensor([0.0239, 0.0207, 0.0180, 0.0221, 0.0218, 0.0391, 0.0265, 0.0309], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:03:05,590 INFO [train.py:873] (0/4) Epoch 16, batch 6000, loss[loss=0.1199, simple_loss=0.1303, pruned_loss=0.05474, over 3868.00 frames. ], tot_loss[loss=0.1099, simple_loss=0.1446, pruned_loss=0.03756, over 1982600.67 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:03:05,591 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 10:03:14,045 INFO [train.py:905] (0/4) Epoch 16, validation: loss=0.1378, simple_loss=0.175, pruned_loss=0.05031, over 857387.00 frames. 2022-12-08 10:03:14,045 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 10:03:14,142 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119431.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:03:29,676 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119449.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 10:03:33,937 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8319, 0.8463, 0.7615, 0.8750, 0.8837, 0.5010, 0.7258, 0.8752], device='cuda:0'), covar=tensor([0.0400, 0.0375, 0.0410, 0.0421, 0.0268, 0.0369, 0.1101, 0.0721], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:03:58,569 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-08 10:04:20,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.903e+01 2.074e+02 2.533e+02 3.058e+02 7.007e+02, threshold=5.066e+02, percent-clipped=2.0 2022-12-08 10:04:26,261 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119514.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:04:41,253 INFO [train.py:873] (0/4) Epoch 16, batch 6100, loss[loss=0.1408, simple_loss=0.1459, pruned_loss=0.06786, over 3867.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1441, pruned_loss=0.0373, over 1923948.56 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:05:13,136 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2022-12-08 10:05:19,859 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:05:46,240 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8380, 1.8300, 4.0058, 3.7511, 3.7513, 4.0671, 3.4848, 4.0642], device='cuda:0'), covar=tensor([0.1683, 0.1547, 0.0126, 0.0241, 0.0267, 0.0155, 0.0235, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0160, 0.0130, 0.0169, 0.0149, 0.0143, 0.0124, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:05:48,893 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 1.899e+02 2.518e+02 3.457e+02 7.704e+02, threshold=5.035e+02, percent-clipped=3.0 2022-12-08 10:05:49,975 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1809, 2.9262, 2.3221, 3.2433, 3.1079, 3.1970, 2.7784, 2.3129], device='cuda:0'), covar=tensor([0.0984, 0.1550, 0.3359, 0.0739, 0.1063, 0.0944, 0.1540, 0.3070], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0292, 0.0262, 0.0284, 0.0323, 0.0299, 0.0254, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:06:09,183 INFO [train.py:873] (0/4) Epoch 16, batch 6200, loss[loss=0.1226, simple_loss=0.149, pruned_loss=0.04811, over 6947.00 frames. ], tot_loss[loss=0.1094, simple_loss=0.1443, pruned_loss=0.03724, over 1882883.58 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:06:38,644 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119665.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:06:54,522 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2022-12-08 10:07:13,559 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5976, 4.4675, 4.2956, 4.6008, 4.2188, 3.9265, 4.7116, 4.4032], device='cuda:0'), covar=tensor([0.0610, 0.0798, 0.0837, 0.0591, 0.0809, 0.0584, 0.0562, 0.0754], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0142, 0.0146, 0.0157, 0.0146, 0.0121, 0.0167, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:07:16,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.045e+02 2.165e+02 2.712e+02 3.301e+02 6.973e+02, threshold=5.424e+02, percent-clipped=2.0 2022-12-08 10:07:27,880 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7522, 1.2717, 1.6531, 1.1830, 1.5296, 1.7593, 1.5392, 1.4853], device='cuda:0'), covar=tensor([0.0675, 0.0820, 0.0749, 0.0827, 0.1526, 0.0853, 0.0648, 0.1675], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0173, 0.0141, 0.0125, 0.0143, 0.0155, 0.0133, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:07:32,850 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119726.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:07:37,350 INFO [train.py:873] (0/4) Epoch 16, batch 6300, loss[loss=0.1565, simple_loss=0.1755, pruned_loss=0.06871, over 9426.00 frames. ], tot_loss[loss=0.109, simple_loss=0.144, pruned_loss=0.03701, over 1873521.93 frames. ], batch size: 100, lr: 4.87e-03, grad_scale: 4.0 2022-12-08 10:07:37,509 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:07:53,293 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119749.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:08:19,729 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119779.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:08:35,917 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:08:37,896 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4186, 2.1928, 4.4392, 3.0887, 4.2065, 2.2046, 3.3744, 4.2317], device='cuda:0'), covar=tensor([0.0499, 0.3712, 0.0318, 0.5452, 0.0579, 0.2875, 0.1224, 0.0551], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0201, 0.0215, 0.0271, 0.0235, 0.0205, 0.0200, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:08:46,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.194e+02 2.675e+02 3.528e+02 1.237e+03, threshold=5.350e+02, percent-clipped=2.0 2022-12-08 10:09:05,867 INFO [train.py:873] (0/4) Epoch 16, batch 6400, loss[loss=0.1014, simple_loss=0.1451, pruned_loss=0.02882, over 14254.00 frames. ], tot_loss[loss=0.1089, simple_loss=0.144, pruned_loss=0.03685, over 1944997.93 frames. ], batch size: 80, lr: 4.87e-03, grad_scale: 8.0 2022-12-08 10:09:14,276 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8546, 1.8933, 1.7935, 1.8044, 1.6893, 1.1848, 1.7718, 1.8719], device='cuda:0'), covar=tensor([0.1070, 0.0774, 0.0953, 0.1221, 0.2029, 0.0969, 0.1195, 0.0691], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0034, 0.0038, 0.0033, 0.0034, 0.0047, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:09:40,257 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:10:06,692 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8775, 1.5978, 1.8058, 1.6191, 1.9906, 1.7861, 1.5547, 1.8466], device='cuda:0'), covar=tensor([0.0684, 0.1445, 0.0522, 0.0594, 0.0491, 0.0753, 0.0355, 0.0458], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0313, 0.0398, 0.0303, 0.0372, 0.0327, 0.0365, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:10:13,249 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.000e+02 2.417e+02 3.320e+02 6.714e+02, threshold=4.834e+02, percent-clipped=5.0 2022-12-08 10:10:33,939 INFO [train.py:873] (0/4) Epoch 16, batch 6500, loss[loss=0.09695, simple_loss=0.14, pruned_loss=0.02697, over 14190.00 frames. ], tot_loss[loss=0.1099, simple_loss=0.145, pruned_loss=0.0374, over 1941144.78 frames. ], batch size: 37, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:10:39,297 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 10:10:56,701 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1643, 3.6530, 2.8106, 4.3647, 4.0073, 4.1185, 3.5851, 3.0030], device='cuda:0'), covar=tensor([0.0760, 0.1264, 0.3196, 0.0573, 0.0892, 0.1519, 0.1168, 0.2809], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0291, 0.0259, 0.0283, 0.0321, 0.0298, 0.0252, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:10:57,742 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6282, 1.7086, 1.7852, 1.3584, 1.3125, 1.6363, 1.1695, 1.6648], device='cuda:0'), covar=tensor([0.1339, 0.2373, 0.0878, 0.2029, 0.2626, 0.1046, 0.2613, 0.1133], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0101, 0.0095, 0.0100, 0.0116, 0.0090, 0.0120, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 10:11:32,640 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9221, 3.4799, 4.2789, 3.1681, 2.5028, 3.5821, 2.0314, 3.8327], device='cuda:0'), covar=tensor([0.0785, 0.0859, 0.0496, 0.1764, 0.2035, 0.0749, 0.3063, 0.0675], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0095, 0.0100, 0.0116, 0.0090, 0.0119, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 10:11:34,533 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-120000.pt 2022-12-08 10:11:44,555 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 1.947e+02 2.582e+02 3.066e+02 4.807e+02, threshold=5.164e+02, percent-clipped=0.0 2022-12-08 10:11:44,674 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0433, 3.7963, 3.6248, 4.0532, 3.8418, 3.5189, 4.1100, 3.4300], device='cuda:0'), covar=tensor([0.0528, 0.0947, 0.0503, 0.0475, 0.0758, 0.1265, 0.0560, 0.0557], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0275, 0.0198, 0.0192, 0.0183, 0.0155, 0.0285, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:11:55,925 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=120021.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:12:01,247 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8098, 2.5614, 4.7744, 3.2523, 4.5383, 2.2906, 3.4189, 4.5887], device='cuda:0'), covar=tensor([0.0438, 0.3624, 0.0348, 0.5933, 0.0521, 0.3197, 0.1362, 0.0307], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0202, 0.0217, 0.0272, 0.0236, 0.0206, 0.0201, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 10:12:04,535 INFO [train.py:873] (0/4) Epoch 16, batch 6600, loss[loss=0.08986, simple_loss=0.1252, pruned_loss=0.02725, over 5985.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1435, pruned_loss=0.03619, over 1972989.50 frames. ], batch size: 100, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:12:25,221 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2022-12-08 10:12:25,700 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9251, 1.6940, 1.9701, 1.8601, 1.8381, 1.5581, 1.2273, 1.2006], device='cuda:0'), covar=tensor([0.0202, 0.0295, 0.0250, 0.0229, 0.0335, 0.0327, 0.0288, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0033, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:13:12,086 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.190e+01 2.035e+02 2.583e+02 3.251e+02 5.143e+02, threshold=5.166e+02, percent-clipped=0.0 2022-12-08 10:13:14,066 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6963, 1.7026, 1.8179, 1.6856, 1.7740, 1.5029, 1.2750, 1.1910], device='cuda:0'), covar=tensor([0.0196, 0.0279, 0.0200, 0.0266, 0.0194, 0.0352, 0.0305, 0.0410], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0022, 0.0019, 0.0021, 0.0020, 0.0033, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:13:32,125 INFO [train.py:873] (0/4) Epoch 16, batch 6700, loss[loss=0.1168, simple_loss=0.1397, pruned_loss=0.04691, over 6050.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.144, pruned_loss=0.03611, over 2035698.07 frames. ], batch size: 100, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:14:06,003 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:14:36,372 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.97 vs. limit=5.0 2022-12-08 10:14:39,089 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.095e+02 2.165e+02 2.458e+02 2.899e+02 9.707e+02, threshold=4.916e+02, percent-clipped=4.0 2022-12-08 10:14:48,289 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=120218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:14:59,583 INFO [train.py:873] (0/4) Epoch 16, batch 6800, loss[loss=0.1216, simple_loss=0.154, pruned_loss=0.04455, over 14526.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1442, pruned_loss=0.03654, over 2056900.62 frames. ], batch size: 51, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:15:23,023 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 10:16:06,648 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.118e+02 2.648e+02 3.259e+02 6.407e+02, threshold=5.296e+02, percent-clipped=3.0 2022-12-08 10:16:06,762 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6080, 2.4988, 2.2881, 2.3954, 2.5901, 2.5799, 2.5848, 2.5701], device='cuda:0'), covar=tensor([0.1105, 0.0801, 0.2228, 0.2481, 0.1016, 0.1209, 0.1394, 0.1011], device='cuda:0'), in_proj_covar=tensor([0.0395, 0.0273, 0.0456, 0.0568, 0.0350, 0.0450, 0.0394, 0.0391], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:16:18,504 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120321.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:16:27,162 INFO [train.py:873] (0/4) Epoch 16, batch 6900, loss[loss=0.09707, simple_loss=0.1482, pruned_loss=0.02299, over 14466.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1441, pruned_loss=0.03671, over 2014155.73 frames. ], batch size: 51, lr: 4.86e-03, grad_scale: 8.0 2022-12-08 10:17:00,366 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=120369.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:17:10,679 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2022-12-08 10:17:34,669 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.039e+02 2.083e+02 2.449e+02 3.128e+02 5.961e+02, threshold=4.898e+02, percent-clipped=1.0 2022-12-08 10:17:55,017 INFO [train.py:873] (0/4) Epoch 16, batch 7000, loss[loss=0.1196, simple_loss=0.1345, pruned_loss=0.05233, over 3876.00 frames. ], tot_loss[loss=0.109, simple_loss=0.1443, pruned_loss=0.03682, over 2068359.01 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:19:01,657 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.060e+01 2.199e+02 2.820e+02 3.385e+02 6.936e+02, threshold=5.640e+02, percent-clipped=6.0 2022-12-08 10:19:14,966 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2029, 1.8819, 4.7173, 4.2625, 4.1737, 4.8217, 4.4076, 4.8209], device='cuda:0'), covar=tensor([0.1488, 0.1581, 0.0104, 0.0234, 0.0243, 0.0123, 0.0176, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0130, 0.0170, 0.0148, 0.0143, 0.0126, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:19:21,716 INFO [train.py:873] (0/4) Epoch 16, batch 7100, loss[loss=0.1079, simple_loss=0.1498, pruned_loss=0.03305, over 14234.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1443, pruned_loss=0.03693, over 2023880.25 frames. ], batch size: 60, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:20:28,194 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.979e+01 2.056e+02 2.495e+02 3.157e+02 6.004e+02, threshold=4.990e+02, percent-clipped=1.0 2022-12-08 10:20:44,326 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7220, 2.3625, 2.5558, 2.5015, 2.1269, 1.4355, 2.1735, 2.5499], device='cuda:0'), covar=tensor([0.0749, 0.0703, 0.0625, 0.0872, 0.1371, 0.0789, 0.1346, 0.0711], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0034, 0.0047, 0.0036, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:20:48,803 INFO [train.py:873] (0/4) Epoch 16, batch 7200, loss[loss=0.1308, simple_loss=0.1585, pruned_loss=0.05158, over 10352.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1438, pruned_loss=0.03634, over 2017799.60 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:20:57,696 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3425, 1.8345, 2.2460, 1.9995, 2.4178, 2.1986, 2.1860, 2.1644], device='cuda:0'), covar=tensor([0.0724, 0.2433, 0.0885, 0.1334, 0.0675, 0.1032, 0.0884, 0.1368], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0314, 0.0396, 0.0303, 0.0372, 0.0326, 0.0363, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:21:11,402 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5482, 1.1628, 1.9982, 1.7275, 1.8985, 2.0774, 1.3674, 2.0747], device='cuda:0'), covar=tensor([0.0799, 0.1360, 0.0332, 0.0572, 0.0695, 0.0322, 0.0883, 0.0311], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0130, 0.0170, 0.0148, 0.0142, 0.0125, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:21:27,868 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=120676.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:21:55,119 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.056e+02 2.553e+02 2.947e+02 6.025e+02, threshold=5.105e+02, percent-clipped=4.0 2022-12-08 10:22:15,426 INFO [train.py:873] (0/4) Epoch 16, batch 7300, loss[loss=0.1039, simple_loss=0.1355, pruned_loss=0.03609, over 6908.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1436, pruned_loss=0.03641, over 1965978.15 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:22:20,691 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=120737.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:22:54,525 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1190, 2.0997, 2.4594, 2.2785, 2.3089, 1.8693, 1.9503, 1.5320], device='cuda:0'), covar=tensor([0.0299, 0.0601, 0.0188, 0.0199, 0.0186, 0.0427, 0.0316, 0.0509], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:23:22,447 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.181e+02 2.578e+02 3.347e+02 7.648e+02, threshold=5.157e+02, percent-clipped=5.0 2022-12-08 10:23:42,645 INFO [train.py:873] (0/4) Epoch 16, batch 7400, loss[loss=0.1277, simple_loss=0.1574, pruned_loss=0.04897, over 12802.00 frames. ], tot_loss[loss=0.1077, simple_loss=0.1434, pruned_loss=0.03601, over 2002760.00 frames. ], batch size: 100, lr: 4.85e-03, grad_scale: 8.0 2022-12-08 10:23:59,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 10:24:02,758 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4874, 5.2776, 4.8666, 5.1365, 5.0873, 5.4177, 5.4575, 5.4484], device='cuda:0'), covar=tensor([0.0693, 0.0343, 0.1764, 0.2265, 0.0606, 0.0692, 0.0650, 0.0750], device='cuda:0'), in_proj_covar=tensor([0.0392, 0.0273, 0.0449, 0.0568, 0.0349, 0.0445, 0.0391, 0.0391], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:24:09,105 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5856, 3.6652, 4.2303, 3.2088, 2.7865, 3.5802, 2.0697, 3.5279], device='cuda:0'), covar=tensor([0.1077, 0.0681, 0.0690, 0.1212, 0.1711, 0.1406, 0.3009, 0.1147], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0095, 0.0100, 0.0116, 0.0090, 0.0118, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 10:24:19,975 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 10:24:49,561 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 2.203e+02 2.672e+02 3.215e+02 8.256e+02, threshold=5.345e+02, percent-clipped=4.0 2022-12-08 10:25:09,741 INFO [train.py:873] (0/4) Epoch 16, batch 7500, loss[loss=0.1352, simple_loss=0.1586, pruned_loss=0.05592, over 10383.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.1433, pruned_loss=0.03616, over 1989771.32 frames. ], batch size: 100, lr: 4.84e-03, grad_scale: 8.0 2022-12-08 10:25:09,920 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8302, 1.8877, 2.1374, 1.9243, 1.8835, 1.6442, 1.5661, 1.2897], device='cuda:0'), covar=tensor([0.0243, 0.0682, 0.0198, 0.0226, 0.0206, 0.0276, 0.0269, 0.0380], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0031], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:25:35,941 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8855, 3.6577, 3.3697, 3.5683, 3.7587, 3.7694, 3.8452, 3.8482], device='cuda:0'), covar=tensor([0.0843, 0.0608, 0.1992, 0.2646, 0.0765, 0.0919, 0.0975, 0.0803], device='cuda:0'), in_proj_covar=tensor([0.0395, 0.0275, 0.0456, 0.0576, 0.0353, 0.0451, 0.0397, 0.0396], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:25:56,013 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-16.pt 2022-12-08 10:26:35,448 INFO [train.py:873] (0/4) Epoch 17, batch 0, loss[loss=0.1613, simple_loss=0.1847, pruned_loss=0.069, over 14266.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.1847, pruned_loss=0.069, over 14266.00 frames. ], batch size: 63, lr: 4.70e-03, grad_scale: 8.0 2022-12-08 10:26:35,449 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 10:26:39,056 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5954, 2.5473, 4.3684, 4.6800, 4.3331, 2.6710, 4.6424, 3.4989], device='cuda:0'), covar=tensor([0.0321, 0.1249, 0.0643, 0.0345, 0.0449, 0.1791, 0.0298, 0.0887], device='cuda:0'), in_proj_covar=tensor([0.0294, 0.0260, 0.0376, 0.0333, 0.0275, 0.0307, 0.0313, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:26:39,144 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0109, 3.7236, 3.7753, 3.9738, 3.6879, 3.9793, 3.9840, 3.6898], device='cuda:0'), covar=tensor([0.0305, 0.0653, 0.0420, 0.0345, 0.0733, 0.0246, 0.0512, 0.0410], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0279, 0.0200, 0.0194, 0.0187, 0.0157, 0.0291, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:26:40,156 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5116, 2.6565, 2.5454, 2.3815, 2.4817, 1.4107, 2.8660, 2.6805], device='cuda:0'), covar=tensor([0.0910, 0.0397, 0.0766, 0.2473, 0.1493, 0.0835, 0.0421, 0.0637], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:26:42,840 INFO [train.py:905] (0/4) Epoch 17, validation: loss=0.1441, simple_loss=0.1813, pruned_loss=0.05348, over 857387.00 frames. 2022-12-08 10:26:42,840 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 10:26:46,443 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2863, 2.9407, 2.3487, 3.2729, 3.1588, 3.1837, 2.7579, 2.2460], device='cuda:0'), covar=tensor([0.0873, 0.1437, 0.3134, 0.0747, 0.1008, 0.1128, 0.1481, 0.3322], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0291, 0.0259, 0.0284, 0.0324, 0.0299, 0.0254, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:26:52,026 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121003.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:26:56,394 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.236e+01 1.830e+02 2.606e+02 3.490e+02 1.014e+03, threshold=5.212e+02, percent-clipped=7.0 2022-12-08 10:27:17,527 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:27:24,964 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0709, 4.7154, 4.6334, 5.1164, 4.7432, 4.4479, 5.0749, 4.3093], device='cuda:0'), covar=tensor([0.0424, 0.0980, 0.0428, 0.0394, 0.0796, 0.0548, 0.0601, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0281, 0.0202, 0.0195, 0.0188, 0.0158, 0.0292, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:27:45,950 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121064.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:27:48,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9819, 1.6483, 3.5072, 3.2222, 3.2951, 3.5532, 2.8514, 3.5475], device='cuda:0'), covar=tensor([0.1634, 0.1675, 0.0148, 0.0325, 0.0304, 0.0173, 0.0383, 0.0149], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0161, 0.0131, 0.0172, 0.0150, 0.0144, 0.0127, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:28:11,823 INFO [train.py:873] (0/4) Epoch 17, batch 100, loss[loss=0.1119, simple_loss=0.1466, pruned_loss=0.03862, over 12043.00 frames. ], tot_loss[loss=0.1106, simple_loss=0.1459, pruned_loss=0.03768, over 869856.87 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:28:13,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 10:28:24,821 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.352e+02 2.943e+02 3.420e+02 5.915e+02, threshold=5.886e+02, percent-clipped=2.0 2022-12-08 10:29:04,806 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7930, 1.6632, 1.8076, 1.7064, 1.6881, 1.5324, 1.2948, 1.1416], device='cuda:0'), covar=tensor([0.0173, 0.0244, 0.0224, 0.0194, 0.0177, 0.0343, 0.0297, 0.0478], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0021, 0.0019, 0.0020, 0.0020, 0.0032, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:29:13,766 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:29:39,131 INFO [train.py:873] (0/4) Epoch 17, batch 200, loss[loss=0.1037, simple_loss=0.1188, pruned_loss=0.04433, over 2596.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1451, pruned_loss=0.03658, over 1396786.87 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:29:52,501 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.132e+02 2.532e+02 3.029e+02 6.779e+02, threshold=5.065e+02, percent-clipped=1.0 2022-12-08 10:29:52,670 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8133, 0.8665, 0.6690, 0.8352, 0.8445, 0.4144, 0.7145, 0.8605], device='cuda:0'), covar=tensor([0.0440, 0.0362, 0.0515, 0.0360, 0.0273, 0.0367, 0.1273, 0.0750], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:30:07,069 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121225.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:30:21,606 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 10:30:30,777 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0824, 4.6690, 4.5856, 5.0452, 4.6923, 4.5046, 5.0345, 4.2731], device='cuda:0'), covar=tensor([0.0284, 0.0832, 0.0372, 0.0388, 0.0757, 0.0461, 0.0490, 0.0514], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0276, 0.0200, 0.0193, 0.0186, 0.0156, 0.0290, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:30:58,620 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8392, 3.7087, 3.2931, 2.7115, 3.4051, 3.6932, 4.0302, 3.2811], device='cuda:0'), covar=tensor([0.0563, 0.1022, 0.0878, 0.1132, 0.0676, 0.0499, 0.0566, 0.0958], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0172, 0.0142, 0.0127, 0.0143, 0.0154, 0.0133, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:31:06,417 INFO [train.py:873] (0/4) Epoch 17, batch 300, loss[loss=0.11, simple_loss=0.1402, pruned_loss=0.03984, over 6908.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1441, pruned_loss=0.03616, over 1641540.88 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:31:19,308 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.023e+02 2.561e+02 3.068e+02 6.366e+02, threshold=5.121e+02, percent-clipped=2.0 2022-12-08 10:31:22,049 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121311.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:31:40,114 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121332.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:03,767 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121359.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:14,976 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121372.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:32:21,996 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:22,663 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2022-12-08 10:32:33,193 INFO [train.py:873] (0/4) Epoch 17, batch 400, loss[loss=0.1216, simple_loss=0.1577, pruned_loss=0.04277, over 12746.00 frames. ], tot_loss[loss=0.1083, simple_loss=0.1438, pruned_loss=0.03638, over 1786480.52 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:32:45,742 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:32:47,376 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.084e+02 2.168e+02 2.580e+02 3.311e+02 5.353e+02, threshold=5.159e+02, percent-clipped=2.0 2022-12-08 10:33:28,653 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2022-12-08 10:33:35,789 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.97 vs. limit=5.0 2022-12-08 10:33:38,822 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:34:00,287 INFO [train.py:873] (0/4) Epoch 17, batch 500, loss[loss=0.1321, simple_loss=0.1539, pruned_loss=0.05517, over 6928.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.144, pruned_loss=0.03672, over 1824043.19 frames. ], batch size: 100, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:34:14,595 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 2.218e+02 2.633e+02 3.024e+02 9.780e+02, threshold=5.266e+02, percent-clipped=2.0 2022-12-08 10:34:21,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8241, 1.8677, 1.6650, 1.8675, 1.7293, 1.8003, 1.8216, 1.6966], device='cuda:0'), covar=tensor([0.1181, 0.0997, 0.1987, 0.0855, 0.1232, 0.0719, 0.1482, 0.1146], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0292, 0.0260, 0.0285, 0.0324, 0.0301, 0.0254, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:34:23,989 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:34:35,339 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6473, 1.6534, 2.8497, 2.1837, 2.7027, 1.7310, 2.2750, 2.7119], device='cuda:0'), covar=tensor([0.1452, 0.3875, 0.0800, 0.3345, 0.1212, 0.2943, 0.1189, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0203, 0.0219, 0.0273, 0.0237, 0.0208, 0.0202, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 10:35:04,669 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8176, 1.3608, 1.7208, 1.2101, 1.5632, 1.8371, 1.5500, 1.5144], device='cuda:0'), covar=tensor([0.0754, 0.0817, 0.0622, 0.0797, 0.1450, 0.0754, 0.0704, 0.1616], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0170, 0.0141, 0.0126, 0.0143, 0.0153, 0.0132, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:35:27,461 INFO [train.py:873] (0/4) Epoch 17, batch 600, loss[loss=0.1157, simple_loss=0.1535, pruned_loss=0.03901, over 14301.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1437, pruned_loss=0.03635, over 1936140.04 frames. ], batch size: 35, lr: 4.69e-03, grad_scale: 8.0 2022-12-08 10:35:29,306 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0564, 1.2734, 1.3214, 0.8756, 0.8188, 1.0706, 0.9059, 1.2563], device='cuda:0'), covar=tensor([0.1803, 0.2529, 0.1207, 0.3175, 0.3807, 0.1435, 0.1924, 0.1403], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0096, 0.0100, 0.0117, 0.0091, 0.0120, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 10:35:41,062 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.288e+02 2.719e+02 3.282e+02 7.294e+02, threshold=5.437e+02, percent-clipped=3.0 2022-12-08 10:35:50,458 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2016, 2.0317, 2.2097, 2.2729, 1.9757, 2.0455, 2.2500, 2.2348], device='cuda:0'), covar=tensor([0.0276, 0.0497, 0.0292, 0.0281, 0.0473, 0.0707, 0.0374, 0.0340], device='cuda:0'), in_proj_covar=tensor([0.0289, 0.0255, 0.0371, 0.0327, 0.0269, 0.0303, 0.0308, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:36:21,052 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7734, 2.7149, 2.1232, 2.8535, 2.6414, 2.7239, 2.3829, 2.1886], device='cuda:0'), covar=tensor([0.0897, 0.1056, 0.2932, 0.0839, 0.1159, 0.0865, 0.1463, 0.2584], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0291, 0.0259, 0.0283, 0.0323, 0.0298, 0.0253, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:36:23,929 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8235, 1.8638, 1.9907, 1.9224, 1.8906, 1.6438, 1.6590, 1.1803], device='cuda:0'), covar=tensor([0.0234, 0.0265, 0.0211, 0.0204, 0.0263, 0.0302, 0.0238, 0.0426], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0021, 0.0019, 0.0021, 0.0020, 0.0032, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:36:25,627 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:36:32,547 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121667.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:36:55,473 INFO [train.py:873] (0/4) Epoch 17, batch 700, loss[loss=0.09515, simple_loss=0.1365, pruned_loss=0.02689, over 14233.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1436, pruned_loss=0.03607, over 1962636.50 frames. ], batch size: 69, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:37:08,191 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121707.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:37:09,978 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.050e+02 2.051e+02 2.543e+02 3.291e+02 5.675e+02, threshold=5.087e+02, percent-clipped=1.0 2022-12-08 10:37:21,986 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8576, 1.6990, 2.0852, 2.0952, 2.1875, 1.2877, 2.1966, 2.2800], device='cuda:0'), covar=tensor([0.1779, 0.0963, 0.0985, 0.1073, 0.1065, 0.1071, 0.1352, 0.0756], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0038, 0.0032, 0.0033, 0.0046, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:37:57,062 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:38:08,287 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3324, 0.9961, 1.2044, 0.7888, 1.1017, 1.3518, 1.0085, 1.0657], device='cuda:0'), covar=tensor([0.0514, 0.0946, 0.0840, 0.0548, 0.1248, 0.0810, 0.0680, 0.1423], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0170, 0.0141, 0.0126, 0.0143, 0.0154, 0.0133, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:38:14,543 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4058, 2.4437, 2.5661, 2.5336, 2.4890, 2.1706, 1.4492, 2.2973], device='cuda:0'), covar=tensor([0.0542, 0.0520, 0.0391, 0.0400, 0.0403, 0.1185, 0.2304, 0.0421], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0174, 0.0146, 0.0146, 0.0206, 0.0141, 0.0158, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 10:38:20,989 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2022-12-08 10:38:23,111 INFO [train.py:873] (0/4) Epoch 17, batch 800, loss[loss=0.1016, simple_loss=0.1432, pruned_loss=0.03003, over 13988.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1436, pruned_loss=0.03642, over 1993194.35 frames. ], batch size: 22, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:38:24,276 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.15 vs. limit=5.0 2022-12-08 10:38:37,354 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.062e+02 2.469e+02 3.248e+02 9.190e+02, threshold=4.938e+02, percent-clipped=3.0 2022-12-08 10:38:47,574 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:39:29,515 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=121868.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:39:36,421 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8461, 1.8586, 1.6294, 1.8979, 1.6740, 1.8033, 1.8289, 1.6901], device='cuda:0'), covar=tensor([0.1506, 0.0776, 0.2226, 0.0972, 0.1250, 0.0825, 0.1458, 0.1303], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0294, 0.0262, 0.0286, 0.0327, 0.0302, 0.0255, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 10:39:51,404 INFO [train.py:873] (0/4) Epoch 17, batch 900, loss[loss=0.1142, simple_loss=0.1297, pruned_loss=0.04932, over 3879.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1435, pruned_loss=0.03633, over 1981077.56 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:39:52,391 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8284, 1.7977, 2.1036, 1.8108, 1.9128, 1.7374, 1.4292, 1.1273], device='cuda:0'), covar=tensor([0.0196, 0.0330, 0.0216, 0.0284, 0.0247, 0.0297, 0.0275, 0.0472], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0021, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:40:04,968 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.025e+02 2.484e+02 3.273e+02 4.983e+02, threshold=4.968e+02, percent-clipped=1.0 2022-12-08 10:40:22,358 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7473, 1.0441, 1.2366, 1.2163, 0.9090, 1.2690, 1.0696, 0.8503], device='cuda:0'), covar=tensor([0.1867, 0.1232, 0.0491, 0.0507, 0.2144, 0.0988, 0.1176, 0.1803], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0089, 0.0069, 0.0075, 0.0100, 0.0088, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:40:40,925 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121950.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:40:55,373 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121967.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 10:41:17,348 INFO [train.py:873] (0/4) Epoch 17, batch 1000, loss[loss=0.1649, simple_loss=0.1518, pruned_loss=0.089, over 1266.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1434, pruned_loss=0.03636, over 1931774.32 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:41:31,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 2.007e+02 2.510e+02 3.013e+02 6.170e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 10:41:33,430 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122011.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:41:36,892 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122015.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 10:41:40,836 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 10:41:44,483 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 10:41:56,207 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0765, 2.1680, 1.9940, 2.2082, 1.8686, 1.9521, 2.1483, 2.0704], device='cuda:0'), covar=tensor([0.0947, 0.1384, 0.1175, 0.1023, 0.1562, 0.1147, 0.1243, 0.1034], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0142, 0.0145, 0.0158, 0.0146, 0.0121, 0.0166, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:42:07,570 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1350, 1.1564, 0.9868, 1.0949, 1.1918, 0.7511, 0.9442, 1.1408], device='cuda:0'), covar=tensor([0.0455, 0.0591, 0.0456, 0.0440, 0.0351, 0.0474, 0.0832, 0.0738], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0037, 0.0032, 0.0033, 0.0046, 0.0035, 0.0037], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 10:42:18,907 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:42:44,897 INFO [train.py:873] (0/4) Epoch 17, batch 1100, loss[loss=0.1113, simple_loss=0.14, pruned_loss=0.04132, over 6006.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1434, pruned_loss=0.03647, over 1916516.07 frames. ], batch size: 100, lr: 4.68e-03, grad_scale: 8.0 2022-12-08 10:42:59,064 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.098e+02 2.553e+02 3.146e+02 5.520e+02, threshold=5.106e+02, percent-clipped=3.0 2022-12-08 10:43:00,802 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122111.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:43:02,295 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 10:43:05,844 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:43:11,118 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.7603, 5.6144, 5.1388, 5.8958, 5.2871, 5.1154, 5.9510, 5.4723], device='cuda:0'), covar=tensor([0.0633, 0.0660, 0.0809, 0.0444, 0.0753, 0.0370, 0.0436, 0.0738], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0141, 0.0144, 0.0158, 0.0145, 0.0120, 0.0166, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 10:43:52,414 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2022-12-08 10:43:59,277 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122178.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:44:12,205 INFO [train.py:873] (0/4) Epoch 17, batch 1200, loss[loss=0.1047, simple_loss=0.1486, pruned_loss=0.03043, over 14294.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1438, pruned_loss=0.03662, over 1990849.53 frames. ], batch size: 63, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:44:13,776 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122195.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:44:26,232 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.994e+02 2.399e+02 3.114e+02 6.684e+02, threshold=4.798e+02, percent-clipped=5.0 2022-12-08 10:45:07,661 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122256.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:45:39,706 INFO [train.py:873] (0/4) Epoch 17, batch 1300, loss[loss=0.1022, simple_loss=0.1367, pruned_loss=0.03385, over 5960.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.1435, pruned_loss=0.03602, over 2037148.87 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:45:51,721 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122306.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:45:54,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.064e+02 2.522e+02 2.867e+02 6.773e+02, threshold=5.044e+02, percent-clipped=2.0 2022-12-08 10:46:02,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:46:40,011 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2022-12-08 10:46:55,933 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:46:56,791 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7586, 3.9194, 3.9143, 3.7156, 3.7828, 4.0659, 1.5058, 3.5807], device='cuda:0'), covar=tensor([0.0533, 0.0406, 0.0552, 0.0570, 0.0588, 0.0390, 0.3967, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0173, 0.0144, 0.0145, 0.0204, 0.0140, 0.0157, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 10:47:01,568 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9871, 1.8501, 4.4337, 4.0875, 4.0111, 4.5248, 4.1591, 4.5438], device='cuda:0'), covar=tensor([0.1556, 0.1497, 0.0114, 0.0225, 0.0252, 0.0145, 0.0150, 0.0114], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0157, 0.0130, 0.0167, 0.0147, 0.0143, 0.0124, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:47:07,174 INFO [train.py:873] (0/4) Epoch 17, batch 1400, loss[loss=0.1268, simple_loss=0.1575, pruned_loss=0.04799, over 8605.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1435, pruned_loss=0.03616, over 2004001.34 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:47:21,305 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.812e+01 2.172e+02 2.630e+02 3.463e+02 6.435e+02, threshold=5.259e+02, percent-clipped=4.0 2022-12-08 10:48:13,704 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:48:17,801 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122473.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 10:48:32,875 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9265, 1.7769, 4.3807, 4.0346, 3.9942, 4.4579, 3.8195, 4.4323], device='cuda:0'), covar=tensor([0.1532, 0.1533, 0.0109, 0.0228, 0.0229, 0.0130, 0.0304, 0.0121], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0159, 0.0131, 0.0168, 0.0148, 0.0143, 0.0124, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 10:48:35,289 INFO [train.py:873] (0/4) Epoch 17, batch 1500, loss[loss=0.1064, simple_loss=0.1175, pruned_loss=0.04764, over 2599.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1433, pruned_loss=0.0362, over 1975234.97 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:48:49,379 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.116e+02 2.482e+02 3.540e+02 1.353e+03, threshold=4.965e+02, percent-clipped=3.0 2022-12-08 10:48:55,506 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.47 vs. limit=5.0 2022-12-08 10:49:06,551 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:49:23,863 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7996, 1.7122, 2.9479, 2.1681, 2.8016, 1.7543, 2.3717, 2.7952], device='cuda:0'), covar=tensor([0.1260, 0.4195, 0.0842, 0.3483, 0.1325, 0.3197, 0.1231, 0.0828], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0203, 0.0217, 0.0269, 0.0236, 0.0205, 0.0200, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 10:49:26,271 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:49:54,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2022-12-08 10:50:03,166 INFO [train.py:873] (0/4) Epoch 17, batch 1600, loss[loss=0.1307, simple_loss=0.1383, pruned_loss=0.06149, over 1160.00 frames. ], tot_loss[loss=0.1071, simple_loss=0.143, pruned_loss=0.03559, over 2001068.50 frames. ], batch size: 100, lr: 4.67e-03, grad_scale: 8.0 2022-12-08 10:50:14,897 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122606.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:50:17,232 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.386e+01 1.886e+02 2.427e+02 3.060e+02 1.117e+03, threshold=4.854e+02, percent-clipped=4.0 2022-12-08 10:50:21,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 10:50:56,453 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122654.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:51:15,255 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122675.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:51:23,744 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8365, 3.5565, 3.4262, 3.7941, 3.5588, 3.7747, 3.8622, 3.1205], device='cuda:0'), covar=tensor([0.0445, 0.0920, 0.0516, 0.0494, 0.0822, 0.0326, 0.0527, 0.0683], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0276, 0.0200, 0.0194, 0.0186, 0.0156, 0.0289, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:51:30,440 INFO [train.py:873] (0/4) Epoch 17, batch 1700, loss[loss=0.1128, simple_loss=0.147, pruned_loss=0.03929, over 14266.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.144, pruned_loss=0.03662, over 1966371.25 frames. ], batch size: 69, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:51:45,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 2.028e+02 2.462e+02 2.959e+02 4.889e+02, threshold=4.923e+02, percent-clipped=1.0 2022-12-08 10:52:26,126 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 10:52:30,672 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8860, 4.5321, 4.3505, 4.8597, 4.4562, 4.3119, 4.8926, 4.1108], device='cuda:0'), covar=tensor([0.0361, 0.0820, 0.0432, 0.0392, 0.0839, 0.0646, 0.0459, 0.0523], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0277, 0.0201, 0.0195, 0.0187, 0.0156, 0.0290, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:52:40,033 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122773.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 10:52:57,840 INFO [train.py:873] (0/4) Epoch 17, batch 1800, loss[loss=0.1124, simple_loss=0.1465, pruned_loss=0.03911, over 14221.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.1439, pruned_loss=0.03624, over 2017386.44 frames. ], batch size: 35, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:53:12,808 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 2.194e+02 2.593e+02 3.400e+02 7.475e+02, threshold=5.185e+02, percent-clipped=4.0 2022-12-08 10:53:22,455 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122821.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:53:25,026 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122824.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:53:31,300 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4758, 1.6848, 1.9200, 1.9317, 1.8121, 1.8839, 1.5188, 1.4060], device='cuda:0'), covar=tensor([0.1268, 0.1105, 0.0516, 0.0581, 0.1282, 0.0786, 0.2068, 0.1712], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0088, 0.0069, 0.0074, 0.0099, 0.0088, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:53:48,729 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122851.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:53:54,073 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7921, 3.5810, 3.3273, 2.6255, 3.2639, 3.5398, 3.8679, 3.0231], device='cuda:0'), covar=tensor([0.0527, 0.0985, 0.0757, 0.1247, 0.0776, 0.0651, 0.0623, 0.1049], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0172, 0.0139, 0.0126, 0.0143, 0.0154, 0.0134, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 10:54:07,187 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:54:25,552 INFO [train.py:873] (0/4) Epoch 17, batch 1900, loss[loss=0.1808, simple_loss=0.1661, pruned_loss=0.09775, over 1340.00 frames. ], tot_loss[loss=0.1082, simple_loss=0.144, pruned_loss=0.03617, over 2020798.15 frames. ], batch size: 100, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:54:30,922 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=122899.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:54:40,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.015e+02 2.576e+02 3.102e+02 5.463e+02, threshold=5.151e+02, percent-clipped=2.0 2022-12-08 10:55:00,443 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:55:07,631 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 10:55:11,387 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9511, 2.7492, 2.7314, 2.9047, 2.7903, 2.8743, 2.9878, 2.5091], device='cuda:0'), covar=tensor([0.0677, 0.1118, 0.0666, 0.0649, 0.0908, 0.0540, 0.0690, 0.0691], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0278, 0.0202, 0.0196, 0.0186, 0.0157, 0.0290, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:55:37,056 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122975.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:55:52,710 INFO [train.py:873] (0/4) Epoch 17, batch 2000, loss[loss=0.1318, simple_loss=0.1696, pruned_loss=0.04705, over 14324.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1445, pruned_loss=0.03681, over 2021885.34 frames. ], batch size: 28, lr: 4.66e-03, grad_scale: 8.0 2022-12-08 10:56:08,074 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 2.103e+02 2.584e+02 3.281e+02 5.069e+02, threshold=5.168e+02, percent-clipped=0.0 2022-12-08 10:56:19,435 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123023.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:56:31,946 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9288, 2.7265, 2.7413, 2.9004, 2.7476, 2.8397, 2.9725, 2.4588], device='cuda:0'), covar=tensor([0.0647, 0.1168, 0.0656, 0.0650, 0.0983, 0.0545, 0.0721, 0.0742], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0279, 0.0202, 0.0196, 0.0186, 0.0157, 0.0291, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 10:57:20,549 INFO [train.py:873] (0/4) Epoch 17, batch 2100, loss[loss=0.1192, simple_loss=0.1537, pruned_loss=0.04231, over 14313.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1446, pruned_loss=0.03677, over 2060035.98 frames. ], batch size: 66, lr: 4.66e-03, grad_scale: 4.0 2022-12-08 10:57:30,135 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2022-12-08 10:57:36,150 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 2.207e+02 2.592e+02 3.533e+02 7.488e+02, threshold=5.184e+02, percent-clipped=8.0 2022-12-08 10:57:47,113 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123124.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:58:05,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2022-12-08 10:58:29,181 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 10:58:46,036 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2022-12-08 10:58:46,984 INFO [train.py:873] (0/4) Epoch 17, batch 2200, loss[loss=0.111, simple_loss=0.1291, pruned_loss=0.04649, over 3898.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1443, pruned_loss=0.03659, over 1996688.86 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 4.0 2022-12-08 10:59:03,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.073e+02 2.495e+02 2.880e+02 5.648e+02, threshold=4.991e+02, percent-clipped=1.0 2022-12-08 10:59:17,535 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:00:06,678 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 11:00:13,630 INFO [train.py:873] (0/4) Epoch 17, batch 2300, loss[loss=0.1082, simple_loss=0.147, pruned_loss=0.03471, over 14408.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1436, pruned_loss=0.03621, over 1969384.27 frames. ], batch size: 73, lr: 4.65e-03, grad_scale: 4.0 2022-12-08 11:00:29,910 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 2.209e+02 2.627e+02 3.184e+02 5.736e+02, threshold=5.255e+02, percent-clipped=2.0 2022-12-08 11:00:32,599 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123314.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:00:36,456 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 11:00:59,294 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-08 11:01:05,904 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0939, 4.9502, 4.6346, 5.0521, 4.6206, 4.4251, 5.1629, 4.9014], device='cuda:0'), covar=tensor([0.0538, 0.0685, 0.0724, 0.0533, 0.0658, 0.0586, 0.0490, 0.0581], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0143, 0.0146, 0.0161, 0.0147, 0.0123, 0.0170, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:01:26,354 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123375.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:01:41,811 INFO [train.py:873] (0/4) Epoch 17, batch 2400, loss[loss=0.1031, simple_loss=0.1439, pruned_loss=0.03109, over 14064.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1439, pruned_loss=0.03651, over 1987134.26 frames. ], batch size: 29, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:01:51,927 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9803, 2.1659, 1.7947, 2.4126, 2.0188, 1.9626, 1.9157, 1.9461], device='cuda:0'), covar=tensor([0.0385, 0.0533, 0.0380, 0.0294, 0.0450, 0.0518, 0.0491, 0.0507], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:01:58,032 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.088e+02 2.563e+02 3.123e+02 4.912e+02, threshold=5.125e+02, percent-clipped=0.0 2022-12-08 11:02:10,158 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3205, 2.1780, 4.3781, 3.0527, 4.1955, 2.1112, 3.2861, 4.2085], device='cuda:0'), covar=tensor([0.0609, 0.4118, 0.0391, 0.5673, 0.0620, 0.3374, 0.1390, 0.0530], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0203, 0.0219, 0.0271, 0.0238, 0.0205, 0.0204, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:02:12,604 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:03:07,171 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:03:10,657 INFO [train.py:873] (0/4) Epoch 17, batch 2500, loss[loss=0.128, simple_loss=0.1325, pruned_loss=0.06178, over 2564.00 frames. ], tot_loss[loss=0.1074, simple_loss=0.1431, pruned_loss=0.03589, over 1968951.44 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:03:26,843 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 2.132e+02 2.678e+02 3.392e+02 6.335e+02, threshold=5.356e+02, percent-clipped=3.0 2022-12-08 11:03:42,075 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:03:44,771 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1304, 1.4164, 3.9787, 1.7337, 4.0197, 4.2001, 3.2099, 4.4591], device='cuda:0'), covar=tensor([0.0205, 0.3186, 0.0423, 0.2212, 0.0389, 0.0376, 0.0702, 0.0153], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0153, 0.0157, 0.0166, 0.0164, 0.0176, 0.0131, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:04:23,955 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=123576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:04:30,133 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.77 vs. limit=5.0 2022-12-08 11:04:32,393 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1545, 2.3183, 4.1023, 4.3445, 4.1692, 2.4244, 4.2650, 3.2425], device='cuda:0'), covar=tensor([0.0405, 0.1324, 0.0753, 0.0396, 0.0491, 0.1989, 0.0467, 0.1026], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0259, 0.0377, 0.0333, 0.0273, 0.0308, 0.0314, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:04:39,204 INFO [train.py:873] (0/4) Epoch 17, batch 2600, loss[loss=0.1156, simple_loss=0.1491, pruned_loss=0.04105, over 11181.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.1431, pruned_loss=0.0362, over 1961969.74 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:04:55,081 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 1.985e+02 2.407e+02 3.072e+02 5.323e+02, threshold=4.814e+02, percent-clipped=0.0 2022-12-08 11:05:46,077 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123670.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:06:06,727 INFO [train.py:873] (0/4) Epoch 17, batch 2700, loss[loss=0.1473, simple_loss=0.1468, pruned_loss=0.07389, over 1230.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1435, pruned_loss=0.03627, over 1980059.55 frames. ], batch size: 100, lr: 4.65e-03, grad_scale: 8.0 2022-12-08 11:06:22,182 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 2.020e+02 2.571e+02 3.169e+02 5.832e+02, threshold=5.143e+02, percent-clipped=4.0 2022-12-08 11:07:25,920 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:07:33,391 INFO [train.py:873] (0/4) Epoch 17, batch 2800, loss[loss=0.1066, simple_loss=0.1433, pruned_loss=0.03498, over 13507.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1438, pruned_loss=0.0368, over 1935386.66 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:07:49,826 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.052e+02 2.657e+02 3.438e+02 8.382e+02, threshold=5.313e+02, percent-clipped=4.0 2022-12-08 11:08:10,576 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9737, 2.6291, 2.3383, 2.4062, 2.0552, 2.0268, 1.6931, 2.0430], device='cuda:0'), covar=tensor([0.0408, 0.0296, 0.0337, 0.0264, 0.0296, 0.0500, 0.0484, 0.0505], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0021, 0.0019, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:08:42,980 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:09:01,387 INFO [train.py:873] (0/4) Epoch 17, batch 2900, loss[loss=0.1193, simple_loss=0.1202, pruned_loss=0.05921, over 1252.00 frames. ], tot_loss[loss=0.1087, simple_loss=0.1435, pruned_loss=0.03694, over 1895191.13 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:09:06,051 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2022-12-08 11:09:16,751 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.204e+02 2.576e+02 3.173e+02 6.433e+02, threshold=5.152e+02, percent-clipped=4.0 2022-12-08 11:09:36,625 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:09:50,402 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9378, 1.4848, 3.1545, 2.8400, 2.9909, 3.1755, 2.5272, 3.1379], device='cuda:0'), covar=tensor([0.1331, 0.1510, 0.0168, 0.0382, 0.0357, 0.0173, 0.0425, 0.0189], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0131, 0.0169, 0.0148, 0.0144, 0.0124, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 11:10:00,477 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.67 vs. limit=5.0 2022-12-08 11:10:03,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8753, 0.7731, 0.8569, 0.8759, 0.8039, 0.5723, 0.6192, 0.7607], device='cuda:0'), covar=tensor([0.0191, 0.0172, 0.0157, 0.0157, 0.0165, 0.0302, 0.0194, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:10:08,165 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123970.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:10:27,939 INFO [train.py:873] (0/4) Epoch 17, batch 3000, loss[loss=0.08104, simple_loss=0.1243, pruned_loss=0.01889, over 14128.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1436, pruned_loss=0.03732, over 1857943.15 frames. ], batch size: 19, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:10:27,940 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 11:10:36,501 INFO [train.py:905] (0/4) Epoch 17, validation: loss=0.1392, simple_loss=0.1759, pruned_loss=0.05127, over 857387.00 frames. 2022-12-08 11:10:36,502 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 11:10:51,160 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0836, 1.2818, 1.3365, 0.9978, 0.8222, 1.1392, 0.8362, 1.2180], device='cuda:0'), covar=tensor([0.2457, 0.2717, 0.1413, 0.2953, 0.3210, 0.1446, 0.2212, 0.1454], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0099, 0.0094, 0.0099, 0.0115, 0.0089, 0.0117, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:10:52,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.049e+02 2.779e+02 3.287e+02 7.415e+02, threshold=5.559e+02, percent-clipped=4.0 2022-12-08 11:10:58,410 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:10,299 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7944, 2.5410, 3.2544, 2.1628, 2.0821, 2.7352, 1.5410, 2.8586], device='cuda:0'), covar=tensor([0.0979, 0.1157, 0.0574, 0.1841, 0.2211, 0.0986, 0.3183, 0.0826], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0100, 0.0094, 0.0099, 0.0115, 0.0089, 0.0117, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:11:14,493 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:37,605 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124062.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:11:57,166 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:04,346 INFO [train.py:873] (0/4) Epoch 17, batch 3100, loss[loss=0.1043, simple_loss=0.1448, pruned_loss=0.0319, over 14434.00 frames. ], tot_loss[loss=0.1096, simple_loss=0.1442, pruned_loss=0.0375, over 1870152.45 frames. ], batch size: 41, lr: 4.64e-03, grad_scale: 4.0 2022-12-08 11:12:08,231 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:18,234 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4208, 2.4860, 2.5629, 2.5065, 2.5064, 2.1591, 1.4799, 2.2264], device='cuda:0'), covar=tensor([0.0565, 0.0514, 0.0381, 0.0381, 0.0419, 0.1355, 0.2338, 0.0431], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0173, 0.0145, 0.0146, 0.0205, 0.0141, 0.0157, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 11:12:20,720 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.198e+02 2.691e+02 3.318e+02 1.640e+03, threshold=5.381e+02, percent-clipped=4.0 2022-12-08 11:12:31,032 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124123.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:12:35,480 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6568, 1.7078, 4.4337, 2.1721, 4.3968, 4.6670, 4.2317, 5.1011], device='cuda:0'), covar=tensor([0.0191, 0.3194, 0.0365, 0.2157, 0.0262, 0.0332, 0.0273, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0155, 0.0158, 0.0168, 0.0165, 0.0178, 0.0132, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:12:38,777 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124132.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:13:31,920 INFO [train.py:873] (0/4) Epoch 17, batch 3200, loss[loss=0.1048, simple_loss=0.119, pruned_loss=0.04535, over 2721.00 frames. ], tot_loss[loss=0.1091, simple_loss=0.1439, pruned_loss=0.03718, over 1843607.88 frames. ], batch size: 100, lr: 4.64e-03, grad_scale: 8.0 2022-12-08 11:13:48,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 1.975e+02 2.521e+02 3.067e+02 5.500e+02, threshold=5.042e+02, percent-clipped=1.0 2022-12-08 11:14:02,434 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:14:40,343 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:14:58,460 INFO [train.py:873] (0/4) Epoch 17, batch 3300, loss[loss=0.08, simple_loss=0.1085, pruned_loss=0.02574, over 2578.00 frames. ], tot_loss[loss=0.1085, simple_loss=0.1436, pruned_loss=0.0367, over 1866344.76 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:15:14,505 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.029e+02 2.735e+02 3.344e+02 8.067e+02, threshold=5.469e+02, percent-clipped=3.0 2022-12-08 11:15:33,696 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124333.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:15:47,163 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124348.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:15:55,456 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 11:16:02,140 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0491, 2.5637, 5.0565, 3.4432, 4.8005, 2.4217, 3.8104, 4.8459], device='cuda:0'), covar=tensor([0.0426, 0.3580, 0.0330, 0.5795, 0.0819, 0.2945, 0.1117, 0.0277], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0201, 0.0218, 0.0271, 0.0236, 0.0205, 0.0203, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:16:14,102 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8068, 1.4011, 2.5168, 2.2410, 2.4100, 2.5489, 1.6315, 2.5459], device='cuda:0'), covar=tensor([0.1021, 0.1334, 0.0254, 0.0511, 0.0517, 0.0241, 0.0898, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0131, 0.0168, 0.0147, 0.0143, 0.0124, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 11:16:25,278 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124392.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:16:26,083 INFO [train.py:873] (0/4) Epoch 17, batch 3400, loss[loss=0.152, simple_loss=0.1472, pruned_loss=0.07841, over 1168.00 frames. ], tot_loss[loss=0.108, simple_loss=0.1429, pruned_loss=0.03656, over 1827586.11 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 4.0 2022-12-08 11:16:40,657 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124409.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 11:16:43,979 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 2.119e+02 2.705e+02 3.387e+02 6.415e+02, threshold=5.409e+02, percent-clipped=2.0 2022-12-08 11:16:48,651 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124418.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:17:00,900 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:17:54,879 INFO [train.py:873] (0/4) Epoch 17, batch 3500, loss[loss=0.1667, simple_loss=0.1776, pruned_loss=0.07786, over 8609.00 frames. ], tot_loss[loss=0.1093, simple_loss=0.1436, pruned_loss=0.03747, over 1835339.29 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 4.0 2022-12-08 11:17:55,086 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:18:08,007 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3494, 2.3717, 2.4529, 2.4795, 2.4372, 2.0838, 1.3375, 2.1527], device='cuda:0'), covar=tensor([0.0671, 0.0600, 0.0498, 0.0463, 0.0499, 0.1245, 0.2514, 0.0592], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0173, 0.0146, 0.0146, 0.0206, 0.0143, 0.0157, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 11:18:12,216 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.179e+02 2.842e+02 3.919e+02 2.207e+03, threshold=5.685e+02, percent-clipped=7.0 2022-12-08 11:18:25,903 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:18:47,857 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0525, 1.2748, 1.2909, 1.0133, 0.8713, 1.1268, 0.8882, 1.2786], device='cuda:0'), covar=tensor([0.1882, 0.2245, 0.1142, 0.2175, 0.2861, 0.1268, 0.1852, 0.1222], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0095, 0.0100, 0.0115, 0.0090, 0.0118, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:19:08,269 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:19:11,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 11:19:22,685 INFO [train.py:873] (0/4) Epoch 17, batch 3600, loss[loss=0.09003, simple_loss=0.1311, pruned_loss=0.02447, over 14266.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1432, pruned_loss=0.03647, over 1923555.89 frames. ], batch size: 25, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:19:40,801 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 1.900e+02 2.307e+02 2.789e+02 7.825e+02, threshold=4.613e+02, percent-clipped=1.0 2022-12-08 11:19:41,890 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5810, 2.1402, 4.6061, 2.9950, 4.3892, 1.9590, 3.3150, 4.3464], device='cuda:0'), covar=tensor([0.0619, 0.4234, 0.0426, 0.6231, 0.0630, 0.3754, 0.1440, 0.0526], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0202, 0.0218, 0.0271, 0.0237, 0.0205, 0.0202, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:19:51,442 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8150, 1.7121, 4.5949, 2.4262, 4.3907, 4.7905, 4.4382, 5.1978], device='cuda:0'), covar=tensor([0.0221, 0.3257, 0.0430, 0.1959, 0.0320, 0.0360, 0.0291, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0156, 0.0159, 0.0168, 0.0166, 0.0180, 0.0132, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:19:53,695 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124628.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:19:56,450 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6847, 3.3076, 2.5986, 3.8218, 3.7227, 3.6724, 3.1748, 2.6995], device='cuda:0'), covar=tensor([0.0724, 0.1320, 0.2913, 0.0591, 0.0778, 0.1152, 0.1266, 0.2984], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0286, 0.0254, 0.0283, 0.0317, 0.0297, 0.0251, 0.0239], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:20:02,900 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:24,762 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124663.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:36,027 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8871, 1.3233, 2.0180, 1.3438, 1.9739, 2.0607, 1.6887, 2.1589], device='cuda:0'), covar=tensor([0.0296, 0.2142, 0.0563, 0.1920, 0.0606, 0.0657, 0.1214, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0160, 0.0168, 0.0166, 0.0180, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:20:50,216 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124692.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:20:50,939 INFO [train.py:873] (0/4) Epoch 17, batch 3700, loss[loss=0.1254, simple_loss=0.1582, pruned_loss=0.04632, over 12739.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1433, pruned_loss=0.03626, over 1919262.19 frames. ], batch size: 100, lr: 4.63e-03, grad_scale: 8.0 2022-12-08 11:20:56,544 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124699.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:00,475 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124704.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 11:21:07,859 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.074e+02 2.026e+02 2.516e+02 3.153e+02 6.578e+02, threshold=5.031e+02, percent-clipped=7.0 2022-12-08 11:21:12,547 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124718.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:17,807 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124724.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:31,752 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:47,187 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4501, 5.2117, 4.7590, 5.0059, 5.1043, 5.2731, 5.4040, 5.4564], device='cuda:0'), covar=tensor([0.0713, 0.0462, 0.2016, 0.2966, 0.0696, 0.0810, 0.0864, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0394, 0.0272, 0.0459, 0.0572, 0.0349, 0.0450, 0.0392, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:21:50,357 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2022-12-08 11:21:54,336 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:21:58,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.6830, 1.0718, 1.2876, 1.2026, 1.0460, 1.2254, 1.0508, 0.8149], device='cuda:0'), covar=tensor([0.2121, 0.1067, 0.0504, 0.0480, 0.1758, 0.1229, 0.1626, 0.1469], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0090, 0.0071, 0.0076, 0.0100, 0.0090, 0.0102, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 11:22:13,565 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:22:17,561 INFO [train.py:873] (0/4) Epoch 17, batch 3800, loss[loss=0.1245, simple_loss=0.1365, pruned_loss=0.05623, over 3883.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1441, pruned_loss=0.03677, over 1941469.53 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 8.0 2022-12-08 11:22:26,373 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7761, 1.8122, 1.5707, 1.8423, 1.7212, 1.8355, 1.7994, 1.7087], device='cuda:0'), covar=tensor([0.1187, 0.1043, 0.1712, 0.0916, 0.1101, 0.0674, 0.1737, 0.1183], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0289, 0.0258, 0.0287, 0.0322, 0.0301, 0.0255, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:22:35,562 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.150e+02 2.611e+02 3.247e+02 6.025e+02, threshold=5.221e+02, percent-clipped=4.0 2022-12-08 11:22:41,987 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124820.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:23:09,011 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5041, 1.1130, 2.0114, 1.7806, 1.8134, 2.1004, 1.3649, 2.0509], device='cuda:0'), covar=tensor([0.0894, 0.1664, 0.0349, 0.0586, 0.0718, 0.0303, 0.0934, 0.0348], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0160, 0.0132, 0.0170, 0.0149, 0.0145, 0.0126, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 11:23:34,531 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0562, 2.1328, 1.9746, 2.1657, 1.8668, 1.9842, 2.1188, 2.0309], device='cuda:0'), covar=tensor([0.1235, 0.1498, 0.1350, 0.1080, 0.1829, 0.1095, 0.1211, 0.1216], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0164, 0.0149, 0.0124, 0.0171, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:23:35,524 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124881.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:23:45,831 INFO [train.py:873] (0/4) Epoch 17, batch 3900, loss[loss=0.0984, simple_loss=0.1421, pruned_loss=0.02735, over 14576.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1433, pruned_loss=0.0359, over 1960274.59 frames. ], batch size: 34, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:24:04,347 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.040e+02 2.475e+02 3.105e+02 8.016e+02, threshold=4.950e+02, percent-clipped=1.0 2022-12-08 11:24:14,316 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8917, 3.6614, 3.4042, 3.5369, 3.7944, 3.8168, 3.8658, 3.8991], device='cuda:0'), covar=tensor([0.1043, 0.0559, 0.1927, 0.2652, 0.0758, 0.0931, 0.0892, 0.0838], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0272, 0.0457, 0.0571, 0.0348, 0.0449, 0.0391, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:24:17,288 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124928.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:24:59,255 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=124976.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:25:13,841 INFO [train.py:873] (0/4) Epoch 17, batch 4000, loss[loss=0.1155, simple_loss=0.1215, pruned_loss=0.05473, over 1193.00 frames. ], tot_loss[loss=0.1075, simple_loss=0.143, pruned_loss=0.03603, over 1946977.76 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 8.0 2022-12-08 11:25:14,772 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124994.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:25:20,454 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-125000.pt 2022-12-08 11:25:27,917 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125004.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 11:25:36,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.040e+02 2.496e+02 2.950e+02 4.930e+02, threshold=4.992e+02, percent-clipped=0.0 2022-12-08 11:25:40,977 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125019.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:26:09,632 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125052.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:26:28,270 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1837, 2.6628, 3.5800, 2.5073, 2.2323, 3.1489, 1.7038, 3.1997], device='cuda:0'), covar=tensor([0.0672, 0.1338, 0.0592, 0.1701, 0.1986, 0.1000, 0.2972, 0.1024], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0096, 0.0100, 0.0115, 0.0091, 0.0119, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:26:40,489 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:26:44,718 INFO [train.py:873] (0/4) Epoch 17, batch 4100, loss[loss=0.1478, simple_loss=0.1587, pruned_loss=0.06847, over 5936.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1431, pruned_loss=0.03609, over 1948287.49 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:27:03,766 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.146e+02 2.604e+02 3.171e+02 6.787e+02, threshold=5.207e+02, percent-clipped=4.0 2022-12-08 11:27:15,916 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125128.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:27:22,533 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:27:44,268 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9678, 1.9907, 2.2000, 2.0764, 2.0893, 1.1479, 1.8036, 2.0506], device='cuda:0'), covar=tensor([0.1017, 0.0698, 0.0580, 0.1462, 0.0925, 0.0904, 0.1101, 0.1028], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0039, 0.0032, 0.0034, 0.0047, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:27:56,084 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 11:27:57,293 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125176.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:28:08,778 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125189.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:28:11,906 INFO [train.py:873] (0/4) Epoch 17, batch 4200, loss[loss=0.09341, simple_loss=0.1369, pruned_loss=0.02497, over 14266.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1428, pruned_loss=0.03525, over 1975745.75 frames. ], batch size: 60, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:28:17,486 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4946, 1.6509, 2.7811, 2.1010, 2.7083, 1.7065, 2.1622, 2.6116], device='cuda:0'), covar=tensor([0.1371, 0.4094, 0.0860, 0.3414, 0.1049, 0.3152, 0.1285, 0.1166], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0200, 0.0216, 0.0272, 0.0236, 0.0204, 0.0202, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:28:31,908 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.000e+02 2.432e+02 3.049e+02 5.744e+02, threshold=4.863e+02, percent-clipped=3.0 2022-12-08 11:29:40,187 INFO [train.py:873] (0/4) Epoch 17, batch 4300, loss[loss=0.1181, simple_loss=0.1434, pruned_loss=0.04635, over 9524.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1435, pruned_loss=0.03616, over 1978717.11 frames. ], batch size: 100, lr: 4.62e-03, grad_scale: 4.0 2022-12-08 11:29:41,227 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125294.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:29:58,839 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.107e+02 2.137e+02 2.626e+02 3.159e+02 7.655e+02, threshold=5.251e+02, percent-clipped=3.0 2022-12-08 11:30:02,813 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125319.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:11,867 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4153, 2.4355, 2.5137, 2.5038, 2.5387, 2.1764, 1.5104, 2.2058], device='cuda:0'), covar=tensor([0.0622, 0.0544, 0.0488, 0.0444, 0.0435, 0.1323, 0.2481, 0.0477], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0148, 0.0208, 0.0144, 0.0160, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 11:30:13,643 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9053, 1.8576, 1.5869, 1.8790, 1.6972, 1.8432, 1.7923, 1.7215], device='cuda:0'), covar=tensor([0.1085, 0.0856, 0.1816, 0.0805, 0.1301, 0.0691, 0.1341, 0.1149], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0292, 0.0261, 0.0287, 0.0325, 0.0302, 0.0255, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:30:15,132 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8696, 0.8268, 0.6974, 0.8680, 0.8862, 0.3859, 0.8131, 0.8429], device='cuda:0'), covar=tensor([0.0409, 0.0500, 0.0544, 0.0475, 0.0389, 0.0413, 0.0965, 0.0777], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0035, 0.0039, 0.0032, 0.0034, 0.0047, 0.0036, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:30:23,029 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:34,877 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:44,400 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125367.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:30:52,493 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:31:07,129 INFO [train.py:873] (0/4) Epoch 17, batch 4400, loss[loss=0.09511, simple_loss=0.141, pruned_loss=0.0246, over 14061.00 frames. ], tot_loss[loss=0.1083, simple_loss=0.1437, pruned_loss=0.0364, over 1984251.62 frames. ], batch size: 29, lr: 4.61e-03, grad_scale: 8.0 2022-12-08 11:31:26,863 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.784e+01 2.190e+02 2.587e+02 3.207e+02 7.157e+02, threshold=5.175e+02, percent-clipped=2.0 2022-12-08 11:31:28,826 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125417.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 11:31:32,444 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125421.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:31:46,348 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:31:54,279 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1649, 3.5753, 2.6497, 4.3370, 4.1509, 4.1592, 3.5312, 2.9952], device='cuda:0'), covar=tensor([0.0729, 0.1316, 0.3682, 0.0566, 0.0744, 0.1016, 0.1357, 0.2937], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0290, 0.0260, 0.0286, 0.0322, 0.0301, 0.0254, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:32:20,019 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125476.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:25,274 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125482.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:26,983 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125484.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:32:35,101 INFO [train.py:873] (0/4) Epoch 17, batch 4500, loss[loss=0.08142, simple_loss=0.1223, pruned_loss=0.02028, over 13999.00 frames. ], tot_loss[loss=0.1068, simple_loss=0.1427, pruned_loss=0.03541, over 1960643.73 frames. ], batch size: 19, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:32:54,649 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 2.002e+02 2.492e+02 3.086e+02 6.444e+02, threshold=4.984e+02, percent-clipped=2.0 2022-12-08 11:33:01,483 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125524.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:33:07,215 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6443, 2.4016, 2.2193, 2.3870, 2.5653, 2.5727, 2.5711, 2.5706], device='cuda:0'), covar=tensor([0.1170, 0.0913, 0.2766, 0.2698, 0.1217, 0.1274, 0.1488, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0392, 0.0272, 0.0454, 0.0568, 0.0350, 0.0447, 0.0391, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:33:13,294 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5020, 3.1491, 3.8592, 2.9715, 2.5969, 3.4538, 1.8268, 3.6678], device='cuda:0'), covar=tensor([0.1082, 0.1204, 0.0716, 0.1488, 0.1657, 0.0948, 0.3255, 0.0669], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0096, 0.0100, 0.0116, 0.0091, 0.0119, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:34:01,914 INFO [train.py:873] (0/4) Epoch 17, batch 4600, loss[loss=0.1061, simple_loss=0.1471, pruned_loss=0.03252, over 14028.00 frames. ], tot_loss[loss=0.1079, simple_loss=0.1438, pruned_loss=0.03604, over 1983429.27 frames. ], batch size: 26, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:34:11,406 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:34:22,483 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.136e+02 2.137e+02 2.692e+02 3.129e+02 4.972e+02, threshold=5.384e+02, percent-clipped=0.0 2022-12-08 11:34:55,097 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2022-12-08 11:35:05,222 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125664.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:35:30,460 INFO [train.py:873] (0/4) Epoch 17, batch 4700, loss[loss=0.1134, simple_loss=0.1443, pruned_loss=0.04122, over 6964.00 frames. ], tot_loss[loss=0.1088, simple_loss=0.1442, pruned_loss=0.03671, over 1972626.31 frames. ], batch size: 100, lr: 4.61e-03, grad_scale: 4.0 2022-12-08 11:35:37,696 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0910, 1.0868, 0.8773, 1.1278, 1.0973, 0.7758, 1.0062, 1.0618], device='cuda:0'), covar=tensor([0.0488, 0.0712, 0.0512, 0.0438, 0.0460, 0.0508, 0.0674, 0.0691], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0034, 0.0039, 0.0032, 0.0034, 0.0047, 0.0035, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:35:47,197 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125712.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 11:35:50,738 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.065e+02 2.672e+02 3.737e+02 1.296e+03, threshold=5.344e+02, percent-clipped=9.0 2022-12-08 11:35:57,896 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125725.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:04,467 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:13,534 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7695, 3.4574, 3.2422, 2.4132, 3.2369, 3.4713, 3.8860, 3.0456], device='cuda:0'), covar=tensor([0.0592, 0.1126, 0.0860, 0.1334, 0.0869, 0.0681, 0.0659, 0.1013], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0171, 0.0140, 0.0125, 0.0141, 0.0155, 0.0133, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 11:36:24,883 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3959, 3.1127, 2.4192, 3.4416, 3.3353, 3.3552, 2.8611, 2.4573], device='cuda:0'), covar=tensor([0.0764, 0.1264, 0.2978, 0.0722, 0.0974, 0.1061, 0.1475, 0.3018], device='cuda:0'), in_proj_covar=tensor([0.0286, 0.0293, 0.0262, 0.0288, 0.0324, 0.0304, 0.0258, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:36:27,157 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.38 vs. limit=5.0 2022-12-08 11:36:44,463 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:50,818 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:52,523 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125786.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:36:58,259 INFO [train.py:873] (0/4) Epoch 17, batch 4800, loss[loss=0.1291, simple_loss=0.1347, pruned_loss=0.06169, over 2636.00 frames. ], tot_loss[loss=0.1081, simple_loss=0.1438, pruned_loss=0.03625, over 1971092.89 frames. ], batch size: 100, lr: 4.61e-03, grad_scale: 8.0 2022-12-08 11:37:08,185 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3804, 1.4163, 3.4011, 1.5127, 3.2258, 3.4230, 2.4418, 3.6828], device='cuda:0'), covar=tensor([0.0259, 0.3237, 0.0480, 0.2278, 0.0918, 0.0465, 0.1054, 0.0222], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0156, 0.0158, 0.0166, 0.0166, 0.0178, 0.0131, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:37:18,851 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0983, 4.9101, 4.7840, 5.1743, 4.7478, 4.2559, 5.2237, 4.8801], device='cuda:0'), covar=tensor([0.0623, 0.0734, 0.0735, 0.0510, 0.0671, 0.0677, 0.0570, 0.0725], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0147, 0.0150, 0.0166, 0.0151, 0.0126, 0.0173, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:37:19,643 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 1.900e+02 2.405e+02 3.076e+02 6.689e+02, threshold=4.810e+02, percent-clipped=2.0 2022-12-08 11:37:27,521 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5739, 3.2747, 3.1469, 2.4478, 3.0483, 3.2830, 3.6051, 2.8895], device='cuda:0'), covar=tensor([0.0581, 0.0884, 0.0782, 0.1195, 0.1017, 0.0660, 0.0652, 0.1126], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0169, 0.0139, 0.0124, 0.0141, 0.0154, 0.0132, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 11:37:32,553 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=125832.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:38:26,384 INFO [train.py:873] (0/4) Epoch 17, batch 4900, loss[loss=0.09607, simple_loss=0.1373, pruned_loss=0.02741, over 14615.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1431, pruned_loss=0.03601, over 1930395.33 frames. ], batch size: 22, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:38:34,391 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9264, 2.7585, 2.7147, 2.9162, 2.7540, 2.8413, 2.9984, 2.5544], device='cuda:0'), covar=tensor([0.0626, 0.1055, 0.0639, 0.0584, 0.0892, 0.0529, 0.0632, 0.0646], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0276, 0.0200, 0.0193, 0.0184, 0.0157, 0.0286, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 11:38:37,665 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 11:38:47,414 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.229e+02 2.578e+02 3.069e+02 1.052e+03, threshold=5.155e+02, percent-clipped=5.0 2022-12-08 11:38:52,206 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 11:39:03,344 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9946, 2.6049, 3.4855, 2.3494, 2.1927, 2.9458, 1.6268, 3.0416], device='cuda:0'), covar=tensor([0.1380, 0.1103, 0.0602, 0.1876, 0.2175, 0.0853, 0.3034, 0.0902], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0096, 0.0100, 0.0115, 0.0091, 0.0118, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:39:24,577 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125959.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:39:49,035 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2022-12-08 11:39:54,283 INFO [train.py:873] (0/4) Epoch 17, batch 5000, loss[loss=0.09535, simple_loss=0.1357, pruned_loss=0.02749, over 14262.00 frames. ], tot_loss[loss=0.1077, simple_loss=0.1431, pruned_loss=0.03618, over 1913628.27 frames. ], batch size: 63, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:39:58,750 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125998.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:11,506 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126012.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 11:40:15,916 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.233e+02 2.614e+02 3.180e+02 5.706e+02, threshold=5.227e+02, percent-clipped=5.0 2022-12-08 11:40:29,010 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:52,401 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126059.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:40:53,522 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126060.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:08,259 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126077.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:11,122 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:12,004 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126081.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:13,216 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2022-12-08 11:41:22,061 INFO [train.py:873] (0/4) Epoch 17, batch 5100, loss[loss=0.121, simple_loss=0.1215, pruned_loss=0.06027, over 1227.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1427, pruned_loss=0.03527, over 1919959.63 frames. ], batch size: 100, lr: 4.60e-03, grad_scale: 4.0 2022-12-08 11:41:34,950 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7831, 3.4136, 3.2770, 2.8119, 3.1801, 3.4841, 3.8488, 3.2129], device='cuda:0'), covar=tensor([0.0575, 0.1239, 0.0816, 0.1084, 0.1250, 0.0758, 0.0693, 0.0898], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0173, 0.0141, 0.0126, 0.0144, 0.0157, 0.0134, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 11:41:42,935 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.125e+02 2.682e+02 3.243e+02 7.763e+02, threshold=5.364e+02, percent-clipped=5.0 2022-12-08 11:41:47,577 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0265, 3.5722, 3.7035, 4.0076, 3.7833, 3.6671, 3.9845, 3.2487], device='cuda:0'), covar=tensor([0.0865, 0.1681, 0.0849, 0.0881, 0.1152, 0.1524, 0.1071, 0.1122], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0276, 0.0200, 0.0195, 0.0184, 0.0158, 0.0287, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 11:41:50,076 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126125.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:41:56,455 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8562, 1.7191, 3.0820, 2.2355, 2.9445, 1.7750, 2.3805, 2.9023], device='cuda:0'), covar=tensor([0.1150, 0.4295, 0.0669, 0.3842, 0.1077, 0.3227, 0.1391, 0.0752], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0202, 0.0221, 0.0274, 0.0238, 0.0206, 0.0204, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:42:17,952 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4714, 1.6738, 2.6735, 2.1471, 2.5667, 1.7305, 2.1673, 2.4820], device='cuda:0'), covar=tensor([0.1498, 0.3933, 0.0809, 0.2802, 0.1255, 0.2382, 0.1138, 0.0900], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0201, 0.0220, 0.0272, 0.0237, 0.0205, 0.0203, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:42:23,857 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2022-12-08 11:42:49,533 INFO [train.py:873] (0/4) Epoch 17, batch 5200, loss[loss=0.09144, simple_loss=0.1368, pruned_loss=0.02301, over 14457.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.142, pruned_loss=0.03464, over 1890535.16 frames. ], batch size: 51, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:43:10,803 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.118e+02 2.718e+02 3.349e+02 1.527e+03, threshold=5.437e+02, percent-clipped=4.0 2022-12-08 11:43:41,560 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8561, 0.8082, 0.8592, 0.8064, 0.8198, 0.5521, 0.5281, 0.7209], device='cuda:0'), covar=tensor([0.0169, 0.0173, 0.0135, 0.0155, 0.0166, 0.0286, 0.0248, 0.0286], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0019, 0.0021, 0.0020, 0.0033, 0.0027, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:43:48,255 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:43:52,565 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 11:44:02,252 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1158, 3.8913, 3.6270, 3.8036, 4.0061, 4.0536, 4.0714, 4.1091], device='cuda:0'), covar=tensor([0.0843, 0.0533, 0.2040, 0.2239, 0.0736, 0.0804, 0.0943, 0.0710], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0271, 0.0453, 0.0564, 0.0351, 0.0450, 0.0390, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:44:09,503 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5279, 2.8382, 4.4043, 3.4709, 4.4127, 4.3961, 4.2044, 3.9035], device='cuda:0'), covar=tensor([0.0808, 0.2611, 0.0899, 0.1338, 0.0607, 0.0798, 0.1214, 0.1356], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0312, 0.0395, 0.0299, 0.0365, 0.0324, 0.0361, 0.0297], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:44:17,889 INFO [train.py:873] (0/4) Epoch 17, batch 5300, loss[loss=0.08143, simple_loss=0.1234, pruned_loss=0.01975, over 13912.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1423, pruned_loss=0.03519, over 1963497.98 frames. ], batch size: 20, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:44:20,079 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4309, 2.1102, 2.6082, 1.6235, 1.8393, 2.2727, 1.3945, 2.4169], device='cuda:0'), covar=tensor([0.0937, 0.1819, 0.0839, 0.2468, 0.2340, 0.1022, 0.3231, 0.1013], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0101, 0.0095, 0.0100, 0.0115, 0.0090, 0.0117, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 11:44:30,677 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:44:39,315 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 2.097e+02 2.584e+02 3.197e+02 6.466e+02, threshold=5.169e+02, percent-clipped=5.0 2022-12-08 11:45:12,340 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126354.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:35,101 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:36,313 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126381.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:45:46,520 INFO [train.py:873] (0/4) Epoch 17, batch 5400, loss[loss=0.11, simple_loss=0.1523, pruned_loss=0.03388, over 14192.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1427, pruned_loss=0.03535, over 1998058.72 frames. ], batch size: 35, lr: 4.60e-03, grad_scale: 8.0 2022-12-08 11:46:08,053 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.193e+02 2.737e+02 3.423e+02 8.531e+02, threshold=5.474e+02, percent-clipped=2.0 2022-12-08 11:46:19,137 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:46:29,878 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:46:37,587 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:47:14,891 INFO [train.py:873] (0/4) Epoch 17, batch 5500, loss[loss=0.1567, simple_loss=0.1743, pruned_loss=0.06953, over 10347.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1425, pruned_loss=0.03504, over 1964930.28 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:47:31,272 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126511.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:47:36,283 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.086e+02 2.565e+02 2.972e+02 5.597e+02, threshold=5.130e+02, percent-clipped=1.0 2022-12-08 11:47:50,050 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 11:48:32,812 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3085, 3.6954, 2.9471, 4.6805, 4.2330, 4.3903, 3.8321, 3.1226], device='cuda:0'), covar=tensor([0.0819, 0.1343, 0.3283, 0.0408, 0.0991, 0.1033, 0.1214, 0.3155], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0289, 0.0261, 0.0287, 0.0322, 0.0301, 0.0253, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:48:42,924 INFO [train.py:873] (0/4) Epoch 17, batch 5600, loss[loss=0.1715, simple_loss=0.1607, pruned_loss=0.09112, over 1215.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.143, pruned_loss=0.03576, over 1945146.28 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:49:04,887 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 1.973e+02 2.464e+02 3.253e+02 5.344e+02, threshold=4.927e+02, percent-clipped=2.0 2022-12-08 11:49:36,572 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126654.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:49:54,960 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9671, 1.8418, 2.0201, 2.0474, 1.8031, 1.2414, 1.6481, 1.8163], device='cuda:0'), covar=tensor([0.0627, 0.0725, 0.0434, 0.0633, 0.0799, 0.0948, 0.0703, 0.0590], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0035, 0.0040, 0.0033, 0.0035, 0.0048, 0.0036, 0.0039], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:50:10,196 INFO [train.py:873] (0/4) Epoch 17, batch 5700, loss[loss=0.1211, simple_loss=0.1376, pruned_loss=0.05231, over 2665.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1431, pruned_loss=0.03569, over 1977398.44 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 8.0 2022-12-08 11:50:15,940 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1766, 2.5284, 3.9970, 2.8953, 4.0079, 3.9936, 3.8301, 3.2901], device='cuda:0'), covar=tensor([0.0754, 0.3123, 0.0893, 0.1907, 0.0750, 0.0762, 0.1342, 0.2333], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0313, 0.0397, 0.0300, 0.0369, 0.0326, 0.0364, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:50:18,642 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=126702.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:50:32,276 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.662e+01 2.224e+02 2.546e+02 3.152e+02 6.937e+02, threshold=5.092e+02, percent-clipped=3.0 2022-12-08 11:50:48,056 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126736.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:50:54,906 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4910, 2.0757, 3.5333, 2.6385, 3.4239, 1.9425, 2.8202, 3.3586], device='cuda:0'), covar=tensor([0.0862, 0.3840, 0.0602, 0.4829, 0.0754, 0.3396, 0.1393, 0.0865], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0203, 0.0224, 0.0275, 0.0241, 0.0208, 0.0205, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:51:33,018 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-08 11:51:37,724 INFO [train.py:873] (0/4) Epoch 17, batch 5800, loss[loss=0.108, simple_loss=0.1421, pruned_loss=0.03691, over 6906.00 frames. ], tot_loss[loss=0.1075, simple_loss=0.1431, pruned_loss=0.0359, over 1952901.92 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 4.0 2022-12-08 11:51:49,396 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:51:58,029 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126816.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:52:01,107 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.070e+02 2.153e+02 2.501e+02 2.856e+02 4.579e+02, threshold=5.001e+02, percent-clipped=0.0 2022-12-08 11:52:51,056 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126877.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:53:04,763 INFO [train.py:873] (0/4) Epoch 17, batch 5900, loss[loss=0.1216, simple_loss=0.152, pruned_loss=0.0456, over 9494.00 frames. ], tot_loss[loss=0.1071, simple_loss=0.1429, pruned_loss=0.03563, over 1916454.61 frames. ], batch size: 100, lr: 4.59e-03, grad_scale: 2.0 2022-12-08 11:53:29,086 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.272e+02 2.688e+02 3.245e+02 5.334e+02, threshold=5.377e+02, percent-clipped=3.0 2022-12-08 11:53:39,508 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2022-12-08 11:53:48,717 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2772, 1.7341, 2.4743, 2.0206, 2.3165, 1.6368, 2.0636, 2.2016], device='cuda:0'), covar=tensor([0.2137, 0.2896, 0.0736, 0.1989, 0.1297, 0.1979, 0.0935, 0.1076], device='cuda:0'), in_proj_covar=tensor([0.0257, 0.0204, 0.0223, 0.0275, 0.0241, 0.0207, 0.0205, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:54:08,302 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 11:54:13,418 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8600, 2.1535, 2.5471, 2.5398, 2.2579, 1.4947, 2.3200, 2.7031], device='cuda:0'), covar=tensor([0.0571, 0.0647, 0.0756, 0.0720, 0.1393, 0.0819, 0.1090, 0.1232], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0035, 0.0039, 0.0032, 0.0034, 0.0047, 0.0036, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 11:54:32,807 INFO [train.py:873] (0/4) Epoch 17, batch 6000, loss[loss=0.1104, simple_loss=0.144, pruned_loss=0.03842, over 11989.00 frames. ], tot_loss[loss=0.1074, simple_loss=0.143, pruned_loss=0.03595, over 1933670.32 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:54:32,808 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 11:54:41,593 INFO [train.py:905] (0/4) Epoch 17, validation: loss=0.1381, simple_loss=0.176, pruned_loss=0.05009, over 857387.00 frames. 2022-12-08 11:54:41,594 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18009MB 2022-12-08 11:54:53,463 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5786, 2.2497, 3.5016, 3.7423, 3.5146, 2.2711, 3.6099, 2.7675], device='cuda:0'), covar=tensor([0.0454, 0.1286, 0.0851, 0.0453, 0.0548, 0.1871, 0.0395, 0.1041], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0259, 0.0376, 0.0329, 0.0271, 0.0307, 0.0311, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:55:05,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 1.999e+02 2.582e+02 3.270e+02 6.669e+02, threshold=5.164e+02, percent-clipped=4.0 2022-12-08 11:55:20,272 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127036.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:56:01,543 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127084.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:56:06,665 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2022-12-08 11:56:09,389 INFO [train.py:873] (0/4) Epoch 17, batch 6100, loss[loss=0.1312, simple_loss=0.1584, pruned_loss=0.05202, over 14169.00 frames. ], tot_loss[loss=0.1084, simple_loss=0.1435, pruned_loss=0.03661, over 1911966.28 frames. ], batch size: 84, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:56:20,834 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127106.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:56:33,263 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.062e+02 2.398e+02 3.031e+02 5.398e+02, threshold=4.796e+02, percent-clipped=2.0 2022-12-08 11:56:48,319 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6128, 2.3047, 3.5932, 3.7417, 3.5603, 2.2426, 3.6321, 2.7003], device='cuda:0'), covar=tensor([0.0510, 0.1224, 0.0899, 0.0489, 0.0572, 0.1894, 0.0500, 0.1104], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0259, 0.0377, 0.0330, 0.0271, 0.0308, 0.0312, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:57:02,829 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127154.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:57:12,603 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5516, 1.1375, 2.0227, 1.7323, 1.8628, 2.0846, 1.4379, 2.0658], device='cuda:0'), covar=tensor([0.0900, 0.1485, 0.0274, 0.0609, 0.0685, 0.0359, 0.0773, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0158, 0.0130, 0.0168, 0.0147, 0.0142, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 11:57:18,826 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127172.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:57:21,158 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 11:57:25,092 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9093, 2.4716, 2.6313, 1.6604, 2.4375, 2.7609, 2.9537, 2.2902], device='cuda:0'), covar=tensor([0.0711, 0.0966, 0.0967, 0.1694, 0.0988, 0.0823, 0.0670, 0.1391], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0173, 0.0141, 0.0126, 0.0144, 0.0156, 0.0135, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 11:57:37,193 INFO [train.py:873] (0/4) Epoch 17, batch 6200, loss[loss=0.1191, simple_loss=0.1507, pruned_loss=0.04374, over 11165.00 frames. ], tot_loss[loss=0.1086, simple_loss=0.1435, pruned_loss=0.03684, over 1930083.41 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:57:51,795 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-08 11:58:01,694 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.117e+02 2.626e+02 3.289e+02 8.585e+02, threshold=5.253e+02, percent-clipped=4.0 2022-12-08 11:58:27,567 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127249.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:58:57,973 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2673, 1.5031, 4.0618, 1.9508, 4.1444, 4.2856, 3.4540, 4.6843], device='cuda:0'), covar=tensor([0.0225, 0.3166, 0.0431, 0.2049, 0.0410, 0.0407, 0.0582, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0158, 0.0165, 0.0165, 0.0178, 0.0131, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 11:59:02,414 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2500, 1.7552, 2.4430, 2.0979, 2.2914, 1.6873, 2.0807, 2.2311], device='cuda:0'), covar=tensor([0.2574, 0.3286, 0.0634, 0.1485, 0.1436, 0.1701, 0.0915, 0.1220], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0201, 0.0220, 0.0272, 0.0238, 0.0204, 0.0202, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 11:59:05,809 INFO [train.py:873] (0/4) Epoch 17, batch 6300, loss[loss=0.11, simple_loss=0.136, pruned_loss=0.04195, over 4929.00 frames. ], tot_loss[loss=0.1083, simple_loss=0.1435, pruned_loss=0.03653, over 1936908.57 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 4.0 2022-12-08 11:59:19,309 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0118, 4.7962, 4.6190, 5.0191, 4.5825, 4.2647, 5.1186, 4.7953], device='cuda:0'), covar=tensor([0.0611, 0.0854, 0.0789, 0.0527, 0.0710, 0.0662, 0.0563, 0.0716], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0145, 0.0148, 0.0163, 0.0149, 0.0124, 0.0170, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 11:59:21,199 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127310.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 11:59:29,649 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.074e+02 2.479e+02 3.103e+02 6.398e+02, threshold=4.958e+02, percent-clipped=2.0 2022-12-08 12:00:33,559 INFO [train.py:873] (0/4) Epoch 17, batch 6400, loss[loss=0.1087, simple_loss=0.1434, pruned_loss=0.03702, over 14607.00 frames. ], tot_loss[loss=0.1076, simple_loss=0.1431, pruned_loss=0.03603, over 1933786.26 frames. ], batch size: 22, lr: 4.58e-03, grad_scale: 8.0 2022-12-08 12:00:44,264 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=8.04 vs. limit=5.0 2022-12-08 12:00:58,034 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.677e+01 2.186e+02 2.674e+02 3.273e+02 5.591e+02, threshold=5.347e+02, percent-clipped=3.0 2022-12-08 12:01:04,708 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127427.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:01:08,344 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2022-12-08 12:01:43,657 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:01:57,900 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127488.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 12:02:00,418 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2811, 2.5423, 2.6007, 2.6126, 2.1861, 2.6155, 2.5253, 1.4593], device='cuda:0'), covar=tensor([0.1112, 0.0860, 0.0810, 0.0687, 0.0981, 0.0644, 0.1094, 0.2021], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0089, 0.0070, 0.0075, 0.0099, 0.0090, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:02:01,913 INFO [train.py:873] (0/4) Epoch 17, batch 6500, loss[loss=0.1606, simple_loss=0.1439, pruned_loss=0.08861, over 1249.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.143, pruned_loss=0.03571, over 2002468.75 frames. ], batch size: 100, lr: 4.58e-03, grad_scale: 8.0 2022-12-08 12:02:25,881 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.139e+02 2.589e+02 3.452e+02 6.629e+02, threshold=5.179e+02, percent-clipped=2.0 2022-12-08 12:02:25,967 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127520.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:02:34,151 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127529.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:02,939 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9349, 5.8898, 5.4921, 6.0494, 5.4003, 5.4114, 6.1357, 5.7606], device='cuda:0'), covar=tensor([0.0617, 0.0559, 0.0640, 0.0497, 0.0786, 0.0414, 0.0460, 0.0673], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0143, 0.0146, 0.0162, 0.0148, 0.0123, 0.0168, 0.0148], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 12:03:27,641 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127590.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:29,961 INFO [train.py:873] (0/4) Epoch 17, batch 6600, loss[loss=0.1306, simple_loss=0.1616, pruned_loss=0.04977, over 13542.00 frames. ], tot_loss[loss=0.107, simple_loss=0.1426, pruned_loss=0.03568, over 1925790.51 frames. ], batch size: 100, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:03:40,620 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127605.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:03:54,225 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 2.073e+02 2.578e+02 3.202e+02 5.753e+02, threshold=5.156e+02, percent-clipped=2.0 2022-12-08 12:04:19,896 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5406, 3.7517, 3.7184, 3.4049, 2.7804, 3.7942, 3.6079, 2.1459], device='cuda:0'), covar=tensor([0.1254, 0.0984, 0.0806, 0.0764, 0.0906, 0.0482, 0.0796, 0.1793], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0089, 0.0070, 0.0075, 0.0098, 0.0089, 0.0100, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:04:30,731 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127662.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:04:57,439 INFO [train.py:873] (0/4) Epoch 17, batch 6700, loss[loss=0.1036, simple_loss=0.1433, pruned_loss=0.03191, over 14275.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1427, pruned_loss=0.03555, over 1932151.45 frames. ], batch size: 44, lr: 4.57e-03, grad_scale: 4.0 2022-12-08 12:05:06,303 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8961, 1.7239, 1.8469, 1.7637, 1.7039, 1.6196, 1.4225, 1.1188], device='cuda:0'), covar=tensor([0.0155, 0.0260, 0.0205, 0.0194, 0.0217, 0.0316, 0.0282, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0022, 0.0021, 0.0033, 0.0028, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:05:21,963 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.055e+02 2.568e+02 3.169e+02 6.326e+02, threshold=5.135e+02, percent-clipped=1.0 2022-12-08 12:05:23,955 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127723.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:05:49,495 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2022-12-08 12:06:13,182 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127779.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:06:16,558 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127783.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:06:25,290 INFO [train.py:873] (0/4) Epoch 17, batch 6800, loss[loss=0.1409, simple_loss=0.1622, pruned_loss=0.05985, over 7781.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1426, pruned_loss=0.03535, over 1926412.84 frames. ], batch size: 100, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:06:50,326 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.046e+02 2.510e+02 3.119e+02 5.930e+02, threshold=5.019e+02, percent-clipped=2.0 2022-12-08 12:07:07,246 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127840.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:16,143 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:33,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2022-12-08 12:07:35,191 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127872.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:46,420 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127885.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:07:53,422 INFO [train.py:873] (0/4) Epoch 17, batch 6900, loss[loss=0.1068, simple_loss=0.1532, pruned_loss=0.03016, over 14466.00 frames. ], tot_loss[loss=0.1065, simple_loss=0.1422, pruned_loss=0.0354, over 1941051.00 frames. ], batch size: 51, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:08:03,986 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127905.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:09,309 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:17,922 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.924e+01 2.068e+02 2.418e+02 3.139e+02 1.064e+03, threshold=4.836e+02, percent-clipped=7.0 2022-12-08 12:08:28,448 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127933.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:08:46,272 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=127953.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:09:18,451 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8745, 5.0059, 4.9758, 4.6601, 4.8532, 5.2612, 2.1111, 4.6004], device='cuda:0'), covar=tensor([0.0361, 0.0331, 0.0606, 0.0433, 0.0397, 0.0190, 0.3619, 0.0342], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0177, 0.0148, 0.0149, 0.0207, 0.0144, 0.0158, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:09:20,963 INFO [train.py:873] (0/4) Epoch 17, batch 7000, loss[loss=0.08926, simple_loss=0.1346, pruned_loss=0.02194, over 14047.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.142, pruned_loss=0.03519, over 1958963.52 frames. ], batch size: 29, lr: 4.57e-03, grad_scale: 8.0 2022-12-08 12:09:37,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8589, 1.3030, 2.0264, 1.3305, 1.9915, 2.0713, 1.7316, 2.1469], device='cuda:0'), covar=tensor([0.0300, 0.2177, 0.0530, 0.1798, 0.0559, 0.0603, 0.1241, 0.0453], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0158, 0.0166, 0.0165, 0.0177, 0.0131, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:09:43,958 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128018.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:09:47,264 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.259e+02 2.605e+02 3.114e+02 4.629e+02, threshold=5.210e+02, percent-clipped=0.0 2022-12-08 12:10:23,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2022-12-08 12:10:34,350 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7424, 4.1884, 3.1056, 5.0448, 4.4653, 4.7963, 4.1184, 3.3162], device='cuda:0'), covar=tensor([0.0496, 0.0859, 0.2923, 0.0261, 0.0730, 0.0868, 0.0971, 0.2686], device='cuda:0'), in_proj_covar=tensor([0.0280, 0.0287, 0.0257, 0.0284, 0.0319, 0.0298, 0.0250, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:10:36,275 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8560, 2.4096, 3.6659, 2.7756, 3.7230, 3.5445, 3.5214, 3.0960], device='cuda:0'), covar=tensor([0.0855, 0.3116, 0.1158, 0.2005, 0.0971, 0.1168, 0.1365, 0.1869], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0311, 0.0394, 0.0299, 0.0366, 0.0323, 0.0361, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:10:41,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128083.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 12:10:41,396 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 12:10:49,317 INFO [train.py:873] (0/4) Epoch 17, batch 7100, loss[loss=0.1043, simple_loss=0.14, pruned_loss=0.03433, over 14217.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.1422, pruned_loss=0.035, over 2021978.61 frames. ], batch size: 35, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:11:14,255 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.219e+02 2.800e+02 3.533e+02 6.463e+02, threshold=5.601e+02, percent-clipped=4.0 2022-12-08 12:11:16,094 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8462, 3.6407, 3.5321, 3.8670, 3.4648, 3.3409, 3.9319, 3.7032], device='cuda:0'), covar=tensor([0.0639, 0.0909, 0.0953, 0.0579, 0.0988, 0.0754, 0.0597, 0.0813], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0144, 0.0148, 0.0164, 0.0149, 0.0124, 0.0170, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 12:11:22,275 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128131.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:11:25,718 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128135.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:11:36,051 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2022-12-08 12:12:09,589 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128185.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:16,537 INFO [train.py:873] (0/4) Epoch 17, batch 7200, loss[loss=0.09215, simple_loss=0.1364, pruned_loss=0.02397, over 13871.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.1425, pruned_loss=0.03484, over 2049309.61 frames. ], batch size: 23, lr: 4.56e-03, grad_scale: 8.0 2022-12-08 12:12:28,338 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:42,993 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.064e+02 2.461e+02 3.111e+02 6.198e+02, threshold=4.923e+02, percent-clipped=2.0 2022-12-08 12:12:47,617 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128228.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:12:52,153 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128233.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:13:44,321 INFO [train.py:873] (0/4) Epoch 17, batch 7300, loss[loss=0.1328, simple_loss=0.1408, pruned_loss=0.06243, over 3918.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1415, pruned_loss=0.0346, over 2003243.43 frames. ], batch size: 100, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:14:06,025 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128318.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:14:08,137 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128320.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:14:10,504 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.143e+02 2.591e+02 3.352e+02 1.032e+03, threshold=5.183e+02, percent-clipped=4.0 2022-12-08 12:14:47,736 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128366.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:15:00,853 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128381.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:15:11,325 INFO [train.py:873] (0/4) Epoch 17, batch 7400, loss[loss=0.0939, simple_loss=0.1135, pruned_loss=0.03715, over 2639.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1426, pruned_loss=0.03503, over 2080449.47 frames. ], batch size: 100, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:15:13,566 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 12:15:17,916 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1399, 2.0205, 1.8507, 1.9140, 2.0612, 2.1104, 2.1155, 2.0953], device='cuda:0'), covar=tensor([0.1113, 0.0827, 0.2336, 0.2074, 0.1377, 0.1083, 0.1222, 0.0984], device='cuda:0'), in_proj_covar=tensor([0.0397, 0.0278, 0.0457, 0.0573, 0.0357, 0.0459, 0.0397, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:15:23,760 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2511, 1.9583, 2.0834, 2.4974, 2.4265, 2.1411, 2.2132, 2.2493], device='cuda:0'), covar=tensor([0.0371, 0.0862, 0.0320, 0.0226, 0.0269, 0.0412, 0.0405, 0.0451], device='cuda:0'), in_proj_covar=tensor([0.0022, 0.0022, 0.0020, 0.0021, 0.0021, 0.0033, 0.0028, 0.0032], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:15:29,135 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128413.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:15:37,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.104e+02 2.629e+02 3.153e+02 1.005e+03, threshold=5.259e+02, percent-clipped=2.0 2022-12-08 12:15:48,087 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:06,144 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.4457, 5.2698, 5.0166, 5.4767, 5.0762, 4.8203, 5.5579, 5.2644], device='cuda:0'), covar=tensor([0.0553, 0.0704, 0.0838, 0.0482, 0.0609, 0.0524, 0.0537, 0.0751], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0147, 0.0150, 0.0165, 0.0151, 0.0126, 0.0173, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 12:16:14,579 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6215, 2.0531, 2.5284, 2.6299, 2.5167, 2.0410, 2.6449, 2.1992], device='cuda:0'), covar=tensor([0.0589, 0.1223, 0.0814, 0.0578, 0.0806, 0.1701, 0.0577, 0.0996], device='cuda:0'), in_proj_covar=tensor([0.0297, 0.0262, 0.0380, 0.0335, 0.0273, 0.0310, 0.0315, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 12:16:21,992 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:28,249 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 12:16:29,522 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128483.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:16:30,386 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4649, 1.1206, 2.0122, 1.7825, 1.8199, 2.0530, 1.4192, 2.0365], device='cuda:0'), covar=tensor([0.0961, 0.1666, 0.0356, 0.0648, 0.0796, 0.0385, 0.0854, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0159, 0.0132, 0.0169, 0.0150, 0.0144, 0.0127, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:16:38,247 INFO [train.py:873] (0/4) Epoch 17, batch 7500, loss[loss=0.09701, simple_loss=0.1396, pruned_loss=0.02722, over 14192.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1423, pruned_loss=0.03555, over 1960339.28 frames. ], batch size: 69, lr: 4.56e-03, grad_scale: 4.0 2022-12-08 12:16:42,995 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.15 vs. limit=2.0 2022-12-08 12:16:49,210 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:17:04,009 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.021e+02 2.073e+02 2.518e+02 3.139e+02 6.346e+02, threshold=5.035e+02, percent-clipped=2.0 2022-12-08 12:17:08,372 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128528.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:17:11,398 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 12:17:20,488 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 12:17:25,062 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-17.pt 2022-12-08 12:18:06,084 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128554.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:18:06,901 INFO [train.py:873] (0/4) Epoch 18, batch 0, loss[loss=0.09219, simple_loss=0.1462, pruned_loss=0.01907, over 14408.00 frames. ], tot_loss[loss=0.09219, simple_loss=0.1462, pruned_loss=0.01907, over 14408.00 frames. ], batch size: 41, lr: 4.43e-03, grad_scale: 8.0 2022-12-08 12:18:06,902 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 12:18:14,462 INFO [train.py:905] (0/4) Epoch 18, validation: loss=0.1457, simple_loss=0.1856, pruned_loss=0.05295, over 857387.00 frames. 2022-12-08 12:18:14,463 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 12:18:33,522 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=128576.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:19:16,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.261e+01 1.855e+02 2.639e+02 3.708e+02 1.096e+03, threshold=5.279e+02, percent-clipped=10.0 2022-12-08 12:19:17,928 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 12:19:37,188 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2234, 3.0037, 2.8203, 2.9541, 3.1792, 3.1853, 3.2030, 3.2158], device='cuda:0'), covar=tensor([0.1120, 0.0759, 0.2387, 0.2913, 0.0936, 0.1071, 0.1351, 0.0912], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0276, 0.0453, 0.0570, 0.0353, 0.0454, 0.0391, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:19:43,879 INFO [train.py:873] (0/4) Epoch 18, batch 100, loss[loss=0.1242, simple_loss=0.1528, pruned_loss=0.04779, over 13528.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1417, pruned_loss=0.03308, over 916252.59 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:19:58,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.08 vs. limit=5.0 2022-12-08 12:20:01,868 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128676.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:20:27,009 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6814, 4.3199, 4.2630, 4.7446, 4.3908, 4.0977, 4.7026, 4.0073], device='cuda:0'), covar=tensor([0.0394, 0.0950, 0.0449, 0.0390, 0.0771, 0.0787, 0.0541, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0276, 0.0201, 0.0193, 0.0184, 0.0157, 0.0286, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 12:20:35,332 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7469, 3.8636, 4.0238, 3.5641, 3.8934, 3.8864, 1.5601, 3.7079], device='cuda:0'), covar=tensor([0.0414, 0.0431, 0.0352, 0.0574, 0.0367, 0.0406, 0.3183, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0176, 0.0146, 0.0148, 0.0207, 0.0143, 0.0157, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:20:42,730 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.050e+02 2.470e+02 3.144e+02 6.150e+02, threshold=4.940e+02, percent-clipped=2.0 2022-12-08 12:20:55,475 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2022-12-08 12:20:57,861 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2022-12-08 12:21:09,630 INFO [train.py:873] (0/4) Epoch 18, batch 200, loss[loss=0.123, simple_loss=0.1512, pruned_loss=0.04742, over 6988.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1424, pruned_loss=0.03451, over 1340927.83 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:21:22,383 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128769.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:21:32,279 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128781.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:22:09,532 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 1.948e+02 2.392e+02 2.813e+02 5.621e+02, threshold=4.784e+02, percent-clipped=1.0 2022-12-08 12:22:25,604 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128842.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:22:37,492 INFO [train.py:873] (0/4) Epoch 18, batch 300, loss[loss=0.1007, simple_loss=0.142, pruned_loss=0.02974, over 14227.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1423, pruned_loss=0.03504, over 1589180.82 frames. ], batch size: 60, lr: 4.42e-03, grad_scale: 4.0 2022-12-08 12:22:50,350 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2022-12-08 12:22:58,978 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8549, 3.6323, 3.5951, 3.9598, 3.5043, 3.3513, 3.9646, 3.7905], device='cuda:0'), covar=tensor([0.0703, 0.1130, 0.1026, 0.0653, 0.1022, 0.0893, 0.0623, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0164, 0.0150, 0.0126, 0.0172, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 12:23:08,178 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128890.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:23:37,893 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 2.277e+02 2.735e+02 3.513e+02 6.767e+02, threshold=5.469e+02, percent-clipped=7.0 2022-12-08 12:24:02,504 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128951.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:24:05,631 INFO [train.py:873] (0/4) Epoch 18, batch 400, loss[loss=0.08451, simple_loss=0.1253, pruned_loss=0.02187, over 13987.00 frames. ], tot_loss[loss=0.1065, simple_loss=0.1421, pruned_loss=0.03542, over 1703997.00 frames. ], batch size: 20, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:24:24,458 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128976.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:24:53,006 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1357, 2.1359, 5.1028, 4.6359, 4.3249, 5.1738, 4.9579, 5.2300], device='cuda:0'), covar=tensor([0.1600, 0.1529, 0.0081, 0.0202, 0.0248, 0.0120, 0.0096, 0.0100], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0158, 0.0130, 0.0169, 0.0148, 0.0143, 0.0126, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:25:06,979 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.168e+02 2.627e+02 3.251e+02 6.252e+02, threshold=5.254e+02, percent-clipped=2.0 2022-12-08 12:25:07,066 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129024.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:25:34,885 INFO [train.py:873] (0/4) Epoch 18, batch 500, loss[loss=0.1016, simple_loss=0.1404, pruned_loss=0.03143, over 14651.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1419, pruned_loss=0.03523, over 1761942.60 frames. ], batch size: 33, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:25:47,398 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129069.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:25:51,186 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2022-12-08 12:26:29,823 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129117.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:26:35,673 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.085e+02 2.575e+02 3.232e+02 8.773e+02, threshold=5.149e+02, percent-clipped=4.0 2022-12-08 12:26:46,163 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7166, 2.3729, 2.5792, 1.6993, 2.2569, 2.5540, 2.6865, 2.2866], device='cuda:0'), covar=tensor([0.0712, 0.0689, 0.0962, 0.1384, 0.1091, 0.0816, 0.0773, 0.1365], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0172, 0.0142, 0.0126, 0.0144, 0.0156, 0.0137, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:26:47,886 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129137.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:27:03,408 INFO [train.py:873] (0/4) Epoch 18, batch 600, loss[loss=0.1547, simple_loss=0.1498, pruned_loss=0.07975, over 1228.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1426, pruned_loss=0.036, over 1801678.77 frames. ], batch size: 100, lr: 4.42e-03, grad_scale: 8.0 2022-12-08 12:28:04,833 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.209e+01 2.130e+02 2.657e+02 3.284e+02 6.073e+02, threshold=5.313e+02, percent-clipped=6.0 2022-12-08 12:28:17,568 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2022-12-08 12:28:24,079 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129246.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 12:28:32,539 INFO [train.py:873] (0/4) Epoch 18, batch 700, loss[loss=0.1016, simple_loss=0.1405, pruned_loss=0.03136, over 14279.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.1417, pruned_loss=0.03508, over 1840292.28 frames. ], batch size: 66, lr: 4.41e-03, grad_scale: 4.0 2022-12-08 12:29:07,608 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2863, 4.0104, 3.7448, 3.9102, 4.1252, 4.1895, 4.2469, 4.2493], device='cuda:0'), covar=tensor([0.0714, 0.0527, 0.1999, 0.2552, 0.0692, 0.0757, 0.0796, 0.0791], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0277, 0.0453, 0.0572, 0.0351, 0.0455, 0.0391, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:29:29,470 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2022-12-08 12:29:33,261 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.087e+02 1.934e+02 2.414e+02 2.791e+02 7.591e+02, threshold=4.829e+02, percent-clipped=2.0 2022-12-08 12:29:40,575 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2022-12-08 12:29:59,462 INFO [train.py:873] (0/4) Epoch 18, batch 800, loss[loss=0.1113, simple_loss=0.1475, pruned_loss=0.03758, over 14255.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.142, pruned_loss=0.03526, over 1925687.33 frames. ], batch size: 46, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:30:19,753 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 12:30:31,611 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0510, 1.9584, 1.5286, 1.6538, 1.9228, 2.0116, 2.0046, 1.9791], device='cuda:0'), covar=tensor([0.1223, 0.1235, 0.4196, 0.3907, 0.2001, 0.1723, 0.2133, 0.1565], device='cuda:0'), in_proj_covar=tensor([0.0391, 0.0277, 0.0452, 0.0571, 0.0351, 0.0453, 0.0391, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:30:51,482 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129413.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:31:01,566 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.286e+02 2.776e+02 3.751e+02 1.143e+03, threshold=5.551e+02, percent-clipped=10.0 2022-12-08 12:31:11,906 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:31:18,140 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 12:31:27,670 INFO [train.py:873] (0/4) Epoch 18, batch 900, loss[loss=0.1122, simple_loss=0.1475, pruned_loss=0.03846, over 14215.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1424, pruned_loss=0.03518, over 1947671.52 frames. ], batch size: 69, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:31:30,563 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0564, 2.1811, 2.3463, 1.4358, 1.7239, 2.1431, 1.3652, 2.1242], device='cuda:0'), covar=tensor([0.0957, 0.1310, 0.0707, 0.2355, 0.2279, 0.0854, 0.2728, 0.0984], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0101, 0.0095, 0.0099, 0.0114, 0.0090, 0.0117, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 12:31:44,487 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129474.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:31:53,760 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129485.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:32:28,737 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 2.020e+02 2.482e+02 3.043e+02 9.483e+02, threshold=4.965e+02, percent-clipped=2.0 2022-12-08 12:32:48,052 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129546.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:32:55,375 INFO [train.py:873] (0/4) Epoch 18, batch 1000, loss[loss=0.1028, simple_loss=0.1427, pruned_loss=0.0315, over 14390.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1425, pruned_loss=0.03547, over 1907536.21 frames. ], batch size: 53, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:33:27,765 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3896, 1.8141, 2.3379, 2.0134, 2.4352, 2.2621, 2.1822, 2.2440], device='cuda:0'), covar=tensor([0.0708, 0.2513, 0.0789, 0.1525, 0.0618, 0.1214, 0.0842, 0.1327], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0311, 0.0394, 0.0298, 0.0365, 0.0325, 0.0361, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:33:29,291 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=129594.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:33:45,402 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129612.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:33:54,510 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8494, 1.4520, 2.8161, 2.4818, 2.7119, 2.8246, 2.0261, 2.8292], device='cuda:0'), covar=tensor([0.1156, 0.1453, 0.0187, 0.0445, 0.0402, 0.0234, 0.0620, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0156, 0.0130, 0.0167, 0.0146, 0.0142, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:33:56,802 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 2.171e+02 2.641e+02 3.416e+02 7.976e+02, threshold=5.283e+02, percent-clipped=6.0 2022-12-08 12:34:19,206 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.60 vs. limit=5.0 2022-12-08 12:34:23,014 INFO [train.py:873] (0/4) Epoch 18, batch 1100, loss[loss=0.1124, simple_loss=0.1266, pruned_loss=0.04907, over 2619.00 frames. ], tot_loss[loss=0.106, simple_loss=0.142, pruned_loss=0.035, over 1899545.70 frames. ], batch size: 100, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:34:39,228 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129673.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:35:23,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.155e+02 2.731e+02 3.358e+02 7.442e+02, threshold=5.463e+02, percent-clipped=7.0 2022-12-08 12:35:50,416 INFO [train.py:873] (0/4) Epoch 18, batch 1200, loss[loss=0.1011, simple_loss=0.1435, pruned_loss=0.02931, over 14294.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1428, pruned_loss=0.03532, over 1918936.21 frames. ], batch size: 63, lr: 4.41e-03, grad_scale: 8.0 2022-12-08 12:36:02,202 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129769.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:36:08,361 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6627, 1.6696, 1.7491, 1.6678, 1.5565, 1.5565, 1.1852, 1.2519], device='cuda:0'), covar=tensor([0.0215, 0.0316, 0.0199, 0.0206, 0.0234, 0.0318, 0.0284, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:36:12,732 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1050, 1.9711, 2.2472, 2.0228, 1.9472, 1.8799, 1.8242, 1.4894], device='cuda:0'), covar=tensor([0.0175, 0.0320, 0.0184, 0.0270, 0.0241, 0.0295, 0.0256, 0.0349], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:36:21,549 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129791.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:36:22,386 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8070, 4.8307, 5.1951, 4.3262, 5.0154, 5.2555, 2.0686, 4.6921], device='cuda:0'), covar=tensor([0.0260, 0.0237, 0.0309, 0.0410, 0.0231, 0.0139, 0.3010, 0.0247], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0179, 0.0148, 0.0151, 0.0210, 0.0145, 0.0160, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:36:24,220 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1809, 2.5855, 5.2300, 3.6311, 4.8706, 2.5237, 3.9938, 4.9026], device='cuda:0'), covar=tensor([0.0329, 0.3552, 0.0247, 0.5146, 0.0470, 0.2957, 0.1142, 0.0317], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0196, 0.0218, 0.0267, 0.0236, 0.0202, 0.0199, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 12:36:51,110 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.078e+02 2.653e+02 3.269e+02 7.697e+02, threshold=5.307e+02, percent-clipped=3.0 2022-12-08 12:37:08,148 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 12:37:14,901 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129852.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:37:17,147 INFO [train.py:873] (0/4) Epoch 18, batch 1300, loss[loss=0.1044, simple_loss=0.1152, pruned_loss=0.04677, over 2625.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.142, pruned_loss=0.03504, over 1902410.33 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:37:28,571 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 12:37:34,189 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2927, 1.7355, 2.4623, 2.0406, 2.2992, 1.6655, 2.0316, 2.2812], device='cuda:0'), covar=tensor([0.2497, 0.2923, 0.0736, 0.2165, 0.1395, 0.1734, 0.1006, 0.1021], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0197, 0.0220, 0.0267, 0.0236, 0.0202, 0.0200, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 12:37:49,818 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3250, 2.1172, 2.3175, 1.6080, 1.8881, 2.3340, 2.2844, 2.0223], device='cuda:0'), covar=tensor([0.0790, 0.0591, 0.0821, 0.1207, 0.1217, 0.0847, 0.0819, 0.1270], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0171, 0.0141, 0.0125, 0.0143, 0.0155, 0.0136, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:37:50,654 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6680, 1.4907, 3.6588, 1.7376, 3.5555, 3.7382, 2.7157, 3.9942], device='cuda:0'), covar=tensor([0.0222, 0.3054, 0.0403, 0.2197, 0.0677, 0.0441, 0.0865, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0162, 0.0170, 0.0170, 0.0182, 0.0135, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:38:18,569 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.020e+02 2.586e+02 3.074e+02 6.619e+02, threshold=5.173e+02, percent-clipped=1.0 2022-12-08 12:38:44,884 INFO [train.py:873] (0/4) Epoch 18, batch 1400, loss[loss=0.1283, simple_loss=0.156, pruned_loss=0.0503, over 8605.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1424, pruned_loss=0.03541, over 1944430.58 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:38:55,869 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129968.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:03,400 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 12:39:04,899 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7162, 1.0334, 1.2756, 1.1850, 0.9815, 1.2629, 1.0855, 0.7942], device='cuda:0'), covar=tensor([0.1719, 0.1345, 0.0463, 0.0674, 0.1775, 0.1182, 0.1597, 0.1560], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0091, 0.0071, 0.0077, 0.0101, 0.0091, 0.0102, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:39:19,197 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7117, 1.7522, 1.8767, 1.9746, 2.0309, 1.8271, 1.6385, 1.3647], device='cuda:0'), covar=tensor([0.0307, 0.0618, 0.0265, 0.0311, 0.0224, 0.0328, 0.0347, 0.0714], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:39:24,236 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-130000.pt 2022-12-08 12:39:28,322 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130000.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:30,043 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130002.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:39:51,054 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 2.021e+02 2.494e+02 2.988e+02 5.995e+02, threshold=4.988e+02, percent-clipped=3.0 2022-12-08 12:40:13,307 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1798, 4.2706, 4.3300, 3.9336, 4.1428, 4.5742, 1.6581, 3.8891], device='cuda:0'), covar=tensor([0.0459, 0.0414, 0.0565, 0.0600, 0.0474, 0.0271, 0.3820, 0.0430], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0178, 0.0148, 0.0150, 0.0209, 0.0144, 0.0159, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:40:15,733 INFO [train.py:873] (0/4) Epoch 18, batch 1500, loss[loss=0.1306, simple_loss=0.135, pruned_loss=0.06308, over 1241.00 frames. ], tot_loss[loss=0.1065, simple_loss=0.1423, pruned_loss=0.03541, over 1986071.45 frames. ], batch size: 100, lr: 4.40e-03, grad_scale: 4.0 2022-12-08 12:40:21,011 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130061.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:40:23,156 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130063.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:40:28,223 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130069.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:40:57,144 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8545, 2.6592, 2.4787, 2.5966, 2.8080, 2.8073, 2.8338, 2.8440], device='cuda:0'), covar=tensor([0.1114, 0.0908, 0.2265, 0.2644, 0.1055, 0.1127, 0.1258, 0.0946], device='cuda:0'), in_proj_covar=tensor([0.0391, 0.0279, 0.0454, 0.0572, 0.0354, 0.0457, 0.0394, 0.0403], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:41:09,684 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130117.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:41:17,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.103e+02 2.601e+02 2.993e+02 6.866e+02, threshold=5.203e+02, percent-clipped=5.0 2022-12-08 12:41:22,430 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.58 vs. limit=5.0 2022-12-08 12:41:35,564 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130147.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:41:37,838 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 12:41:42,391 INFO [train.py:873] (0/4) Epoch 18, batch 1600, loss[loss=0.09574, simple_loss=0.1346, pruned_loss=0.02846, over 14244.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.142, pruned_loss=0.03493, over 2017239.59 frames. ], batch size: 25, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:41:58,198 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0178, 1.9914, 2.0565, 2.0386, 1.9983, 1.5714, 1.3104, 1.7370], device='cuda:0'), covar=tensor([0.0773, 0.0702, 0.0543, 0.0519, 0.0468, 0.1849, 0.2520, 0.0589], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0176, 0.0146, 0.0148, 0.0207, 0.0143, 0.0157, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:42:44,998 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.040e+02 2.513e+02 3.158e+02 8.589e+02, threshold=5.027e+02, percent-clipped=3.0 2022-12-08 12:42:46,830 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5600, 3.2606, 4.0061, 2.9016, 2.4049, 3.5043, 1.8969, 3.4966], device='cuda:0'), covar=tensor([0.1076, 0.0896, 0.0559, 0.1969, 0.1986, 0.0941, 0.2850, 0.0890], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0104, 0.0096, 0.0101, 0.0116, 0.0092, 0.0118, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 12:43:03,437 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2022-12-08 12:43:06,013 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 12:43:10,261 INFO [train.py:873] (0/4) Epoch 18, batch 1700, loss[loss=0.08728, simple_loss=0.1286, pruned_loss=0.02297, over 14498.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.1424, pruned_loss=0.03496, over 2024559.66 frames. ], batch size: 51, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:43:22,137 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130268.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:43:37,559 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8241, 1.8130, 1.6442, 1.8962, 1.7929, 1.8137, 1.7931, 1.6492], device='cuda:0'), covar=tensor([0.1196, 0.0925, 0.2217, 0.1027, 0.1232, 0.0740, 0.1584, 0.1172], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0285, 0.0256, 0.0286, 0.0319, 0.0300, 0.0250, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:44:04,000 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130316.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:44:12,325 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.118e+02 2.048e+02 2.469e+02 3.168e+02 7.190e+02, threshold=4.938e+02, percent-clipped=8.0 2022-12-08 12:44:21,923 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3605, 2.6906, 2.6767, 2.6737, 2.2347, 2.7538, 2.6322, 1.5096], device='cuda:0'), covar=tensor([0.1145, 0.0781, 0.0592, 0.0646, 0.1054, 0.0581, 0.1051, 0.2100], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0090, 0.0070, 0.0076, 0.0100, 0.0090, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:44:33,768 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2022-12-08 12:44:37,342 INFO [train.py:873] (0/4) Epoch 18, batch 1800, loss[loss=0.1136, simple_loss=0.1485, pruned_loss=0.03932, over 14380.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.142, pruned_loss=0.0347, over 2050124.99 frames. ], batch size: 73, lr: 4.40e-03, grad_scale: 8.0 2022-12-08 12:44:38,379 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:44:40,093 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130358.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:44:45,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 12:44:58,168 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0268, 2.1066, 2.4143, 2.1982, 2.0465, 1.7944, 1.7031, 1.5905], device='cuda:0'), covar=tensor([0.0338, 0.0518, 0.0223, 0.0437, 0.0267, 0.0461, 0.0444, 0.0522], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:45:03,075 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130384.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:39,183 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.710e+01 2.001e+02 2.479e+02 3.133e+02 6.689e+02, threshold=4.957e+02, percent-clipped=4.0 2022-12-08 12:45:44,515 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:49,391 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130438.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:54,631 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6172, 2.2900, 2.3444, 2.4159, 1.9089, 1.8757, 1.7799, 1.9833], device='cuda:0'), covar=tensor([0.0211, 0.0436, 0.0341, 0.0365, 0.0362, 0.0549, 0.0615, 0.0563], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:45:55,548 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:56,295 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1547, 2.0488, 2.0935, 2.2063, 2.0958, 2.1089, 2.2684, 1.9334], device='cuda:0'), covar=tensor([0.1154, 0.1548, 0.0914, 0.0868, 0.1235, 0.0847, 0.0938, 0.0759], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0279, 0.0201, 0.0195, 0.0186, 0.0159, 0.0289, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 12:45:57,284 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130447.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:45:57,343 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9046, 2.0947, 2.0432, 2.5760, 2.2916, 2.0008, 2.0832, 2.2200], device='cuda:0'), covar=tensor([0.0457, 0.0547, 0.0424, 0.0210, 0.0238, 0.0395, 0.0389, 0.0489], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:46:04,198 INFO [train.py:873] (0/4) Epoch 18, batch 1900, loss[loss=0.1057, simple_loss=0.1214, pruned_loss=0.04493, over 3837.00 frames. ], tot_loss[loss=0.1065, simple_loss=0.1425, pruned_loss=0.03522, over 2044782.36 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:46:13,710 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8457, 1.8616, 2.0807, 2.0078, 1.7508, 1.6150, 1.4843, 1.2795], device='cuda:0'), covar=tensor([0.0223, 0.0438, 0.0178, 0.0184, 0.0256, 0.0307, 0.0276, 0.0445], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 12:46:37,718 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:46:39,399 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:46:43,022 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130499.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:47:06,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.089e+02 2.666e+02 3.206e+02 5.999e+02, threshold=5.333e+02, percent-clipped=5.0 2022-12-08 12:47:08,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 12:47:12,245 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5050, 1.0872, 2.0138, 1.7432, 1.8621, 2.0507, 1.3320, 2.0967], device='cuda:0'), covar=tensor([0.0907, 0.1645, 0.0302, 0.0598, 0.0701, 0.0376, 0.0914, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0132, 0.0169, 0.0148, 0.0144, 0.0127, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:47:25,640 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:47:31,376 INFO [train.py:873] (0/4) Epoch 18, batch 2000, loss[loss=0.1317, simple_loss=0.1498, pruned_loss=0.05678, over 7820.00 frames. ], tot_loss[loss=0.1066, simple_loss=0.1427, pruned_loss=0.03523, over 2014253.26 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:47:40,356 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130565.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:47:51,514 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2021, 4.9539, 4.6260, 4.8255, 4.8656, 5.1267, 5.1947, 5.2043], device='cuda:0'), covar=tensor([0.0863, 0.0441, 0.2341, 0.2983, 0.0802, 0.0821, 0.0940, 0.0834], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0277, 0.0454, 0.0569, 0.0353, 0.0451, 0.0392, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:48:10,926 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6457, 4.6363, 5.0683, 4.1879, 4.9116, 5.0964, 1.8489, 4.5847], device='cuda:0'), covar=tensor([0.0323, 0.0397, 0.0402, 0.0558, 0.0351, 0.0179, 0.3238, 0.0336], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0176, 0.0146, 0.0149, 0.0208, 0.0143, 0.0158, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 12:48:18,854 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130609.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 12:48:28,752 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5284, 4.0008, 3.0757, 4.7682, 4.3601, 4.5790, 4.0201, 3.3826], device='cuda:0'), covar=tensor([0.0565, 0.1030, 0.2945, 0.0494, 0.0750, 0.1188, 0.0961, 0.2552], device='cuda:0'), in_proj_covar=tensor([0.0279, 0.0285, 0.0257, 0.0287, 0.0320, 0.0299, 0.0250, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:48:34,514 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:48:35,244 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 2.007e+02 2.517e+02 3.001e+02 4.503e+02, threshold=5.034e+02, percent-clipped=0.0 2022-12-08 12:48:59,205 INFO [train.py:873] (0/4) Epoch 18, batch 2100, loss[loss=0.1366, simple_loss=0.1641, pruned_loss=0.05452, over 9501.00 frames. ], tot_loss[loss=0.1069, simple_loss=0.1425, pruned_loss=0.03566, over 1992383.44 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:49:00,403 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130656.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:49:02,058 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130658.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:49:41,915 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:49:43,623 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=130706.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:50:01,668 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.249e+02 2.636e+02 3.244e+02 1.045e+03, threshold=5.271e+02, percent-clipped=2.0 2022-12-08 12:50:13,051 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130740.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:50:26,107 INFO [train.py:873] (0/4) Epoch 18, batch 2200, loss[loss=0.09841, simple_loss=0.1329, pruned_loss=0.03197, over 5997.00 frames. ], tot_loss[loss=0.1078, simple_loss=0.143, pruned_loss=0.03632, over 2011641.84 frames. ], batch size: 100, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:50:36,903 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2022-12-08 12:50:55,162 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130788.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:51:00,278 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130794.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:51:22,889 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 12:51:29,862 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.113e+02 2.522e+02 3.160e+02 5.458e+02, threshold=5.044e+02, percent-clipped=1.0 2022-12-08 12:51:53,881 INFO [train.py:873] (0/4) Epoch 18, batch 2300, loss[loss=0.09774, simple_loss=0.1414, pruned_loss=0.02702, over 14287.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.1428, pruned_loss=0.03581, over 1959163.01 frames. ], batch size: 44, lr: 4.39e-03, grad_scale: 4.0 2022-12-08 12:51:58,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2022-12-08 12:51:58,430 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6291, 1.4139, 3.5865, 1.6234, 3.5687, 3.7390, 2.5633, 4.0101], device='cuda:0'), covar=tensor([0.0239, 0.3280, 0.0484, 0.2301, 0.0648, 0.0465, 0.0990, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0159, 0.0166, 0.0166, 0.0177, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:52:37,092 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130904.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:52:51,938 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130921.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:52:56,900 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.034e+02 2.487e+02 3.081e+02 6.323e+02, threshold=4.975e+02, percent-clipped=2.0 2022-12-08 12:53:22,074 INFO [train.py:873] (0/4) Epoch 18, batch 2400, loss[loss=0.09377, simple_loss=0.1383, pruned_loss=0.0246, over 14064.00 frames. ], tot_loss[loss=0.1073, simple_loss=0.1431, pruned_loss=0.03571, over 2053661.32 frames. ], batch size: 29, lr: 4.39e-03, grad_scale: 8.0 2022-12-08 12:53:42,148 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4862, 4.2276, 3.9361, 4.0993, 4.3133, 4.4569, 4.4524, 4.4675], device='cuda:0'), covar=tensor([0.0906, 0.0481, 0.2022, 0.2822, 0.0742, 0.0784, 0.0903, 0.0813], device='cuda:0'), in_proj_covar=tensor([0.0394, 0.0280, 0.0455, 0.0571, 0.0355, 0.0455, 0.0398, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 12:54:00,404 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 12:54:19,822 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8284, 1.6241, 4.3082, 4.0466, 3.9939, 4.6289, 4.1565, 4.4230], device='cuda:0'), covar=tensor([0.2425, 0.2512, 0.0299, 0.0412, 0.0419, 0.0259, 0.0293, 0.0316], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0132, 0.0169, 0.0148, 0.0144, 0.0127, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:54:25,831 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.032e+02 2.544e+02 3.351e+02 8.852e+02, threshold=5.088e+02, percent-clipped=2.0 2022-12-08 12:54:37,301 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131040.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:54:50,162 INFO [train.py:873] (0/4) Epoch 18, batch 2500, loss[loss=0.1056, simple_loss=0.1296, pruned_loss=0.04087, over 3850.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1423, pruned_loss=0.03498, over 2065741.15 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:55:14,845 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3953, 2.2763, 5.2130, 4.7476, 4.5300, 5.3678, 5.0800, 5.3559], device='cuda:0'), covar=tensor([0.1410, 0.1346, 0.0100, 0.0184, 0.0219, 0.0109, 0.0128, 0.0127], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0159, 0.0132, 0.0169, 0.0148, 0.0144, 0.0127, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 12:55:19,755 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:19,853 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:25,062 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131094.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:55:35,414 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5913, 2.9308, 2.7921, 3.0179, 2.2926, 2.9758, 2.8200, 1.6380], device='cuda:0'), covar=tensor([0.0976, 0.0608, 0.1245, 0.0465, 0.0965, 0.0546, 0.0830, 0.1862], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0090, 0.0070, 0.0076, 0.0100, 0.0090, 0.0101, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:55:43,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0838, 3.5584, 3.2456, 2.4153, 3.4658, 3.7255, 4.1440, 3.1443], device='cuda:0'), covar=tensor([0.0491, 0.1226, 0.0903, 0.1494, 0.0757, 0.0561, 0.0892, 0.1039], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0171, 0.0141, 0.0126, 0.0143, 0.0156, 0.0136, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:55:54,048 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.009e+02 2.192e+02 2.699e+02 3.352e+02 8.933e+02, threshold=5.398e+02, percent-clipped=5.0 2022-12-08 12:55:56,764 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2221, 2.8779, 2.9430, 2.0459, 2.7222, 2.8933, 3.1833, 2.6361], device='cuda:0'), covar=tensor([0.0662, 0.0802, 0.0842, 0.1276, 0.0917, 0.0685, 0.0707, 0.1214], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0171, 0.0140, 0.0126, 0.0142, 0.0155, 0.0136, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 12:56:01,758 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131136.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:56:07,943 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:56:19,377 INFO [train.py:873] (0/4) Epoch 18, batch 2600, loss[loss=0.124, simple_loss=0.1415, pruned_loss=0.05323, over 4938.00 frames. ], tot_loss[loss=0.1071, simple_loss=0.1428, pruned_loss=0.03569, over 2011058.60 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:56:35,611 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131174.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:57:02,569 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131204.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 12:57:17,449 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:57:22,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 2.143e+02 2.636e+02 3.180e+02 5.935e+02, threshold=5.272e+02, percent-clipped=1.0 2022-12-08 12:57:29,991 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131235.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 12:57:44,012 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131252.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:57:46,567 INFO [train.py:873] (0/4) Epoch 18, batch 2700, loss[loss=0.1561, simple_loss=0.1433, pruned_loss=0.08449, over 1326.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1421, pruned_loss=0.03533, over 1926308.08 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 4.0 2022-12-08 12:57:59,031 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 12:58:50,256 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.346e+01 2.091e+02 2.717e+02 3.519e+02 6.308e+02, threshold=5.434e+02, percent-clipped=4.0 2022-12-08 12:59:14,298 INFO [train.py:873] (0/4) Epoch 18, batch 2800, loss[loss=0.09046, simple_loss=0.1334, pruned_loss=0.02378, over 14420.00 frames. ], tot_loss[loss=0.1067, simple_loss=0.1425, pruned_loss=0.03549, over 1973065.61 frames. ], batch size: 73, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 12:59:44,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.04 vs. limit=5.0 2022-12-08 12:59:46,315 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2138, 4.9187, 4.5684, 4.7994, 4.8909, 5.1358, 5.1154, 5.1756], device='cuda:0'), covar=tensor([0.0668, 0.0356, 0.1717, 0.2150, 0.0596, 0.0598, 0.0639, 0.0695], device='cuda:0'), in_proj_covar=tensor([0.0391, 0.0276, 0.0449, 0.0568, 0.0350, 0.0450, 0.0391, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:00:02,515 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131409.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:00:08,441 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131416.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:00:18,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.304e+02 1.962e+02 2.375e+02 2.873e+02 4.439e+02, threshold=4.749e+02, percent-clipped=0.0 2022-12-08 13:00:42,310 INFO [train.py:873] (0/4) Epoch 18, batch 2900, loss[loss=0.09569, simple_loss=0.11, pruned_loss=0.04067, over 2593.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1414, pruned_loss=0.03472, over 1978281.77 frames. ], batch size: 100, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 13:00:54,416 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.04 vs. limit=5.0 2022-12-08 13:00:55,750 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131470.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:02,105 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131477.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:15,146 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131492.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:01:29,427 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3657, 2.2654, 5.1644, 4.6613, 4.4514, 5.2161, 5.0219, 5.2765], device='cuda:0'), covar=tensor([0.1400, 0.1392, 0.0083, 0.0196, 0.0214, 0.0110, 0.0108, 0.0081], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0159, 0.0132, 0.0169, 0.0148, 0.0145, 0.0127, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 13:01:44,925 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1015, 3.2370, 3.3552, 3.2056, 3.2245, 2.9172, 1.5519, 3.0324], device='cuda:0'), covar=tensor([0.0472, 0.0340, 0.0349, 0.0399, 0.0319, 0.0821, 0.2874, 0.0306], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0175, 0.0145, 0.0149, 0.0207, 0.0142, 0.0157, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:01:44,990 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4700, 1.4763, 1.4099, 1.5763, 1.5414, 1.0099, 1.3427, 1.3632], device='cuda:0'), covar=tensor([0.0558, 0.0712, 0.0532, 0.0517, 0.0450, 0.0734, 0.0767, 0.0600], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0035, 0.0039, 0.0033, 0.0034, 0.0048, 0.0036, 0.0039], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:01:46,458 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 2.257e+02 2.729e+02 3.510e+02 6.783e+02, threshold=5.458e+02, percent-clipped=7.0 2022-12-08 13:01:48,375 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131530.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 13:02:08,700 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131553.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:02:10,258 INFO [train.py:873] (0/4) Epoch 18, batch 3000, loss[loss=0.108, simple_loss=0.1444, pruned_loss=0.03573, over 14247.00 frames. ], tot_loss[loss=0.1062, simple_loss=0.1423, pruned_loss=0.03511, over 2031755.66 frames. ], batch size: 99, lr: 4.38e-03, grad_scale: 8.0 2022-12-08 13:02:10,259 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 13:02:18,731 INFO [train.py:905] (0/4) Epoch 18, validation: loss=0.1388, simple_loss=0.176, pruned_loss=0.05082, over 857387.00 frames. 2022-12-08 13:02:18,732 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 13:03:22,346 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 2.067e+02 2.535e+02 3.249e+02 7.779e+02, threshold=5.070e+02, percent-clipped=4.0 2022-12-08 13:03:25,813 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.9903, 1.2897, 1.2907, 0.9872, 0.8164, 1.0205, 0.8857, 1.1205], device='cuda:0'), covar=tensor([0.2338, 0.2961, 0.1291, 0.2492, 0.3171, 0.1617, 0.2320, 0.1605], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0103, 0.0095, 0.0101, 0.0116, 0.0092, 0.0117, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 13:03:44,986 INFO [train.py:873] (0/4) Epoch 18, batch 3100, loss[loss=0.08801, simple_loss=0.1204, pruned_loss=0.02779, over 3897.00 frames. ], tot_loss[loss=0.1072, simple_loss=0.1426, pruned_loss=0.03587, over 1988562.06 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:04:06,418 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5046, 5.3935, 4.9576, 5.5401, 5.1135, 5.0699, 5.5406, 5.3537], device='cuda:0'), covar=tensor([0.0605, 0.0563, 0.0841, 0.0497, 0.0704, 0.0477, 0.0520, 0.0697], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0146, 0.0150, 0.0164, 0.0149, 0.0125, 0.0172, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:04:44,668 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2022-12-08 13:04:49,757 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 2.137e+02 2.592e+02 3.013e+02 4.856e+02, threshold=5.183e+02, percent-clipped=0.0 2022-12-08 13:05:12,289 INFO [train.py:873] (0/4) Epoch 18, batch 3200, loss[loss=0.1386, simple_loss=0.1358, pruned_loss=0.0707, over 1239.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1421, pruned_loss=0.0346, over 1990995.58 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:05:20,639 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131765.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:05:27,297 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131772.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:05:42,984 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9597, 3.3890, 2.7334, 4.1635, 3.9617, 4.0563, 3.4825, 2.8647], device='cuda:0'), covar=tensor([0.0784, 0.1504, 0.3339, 0.0565, 0.0796, 0.1046, 0.1321, 0.3055], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0291, 0.0259, 0.0293, 0.0324, 0.0304, 0.0254, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:05:44,716 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2583, 2.9100, 2.9180, 1.9551, 2.7396, 2.9685, 3.3127, 2.6041], device='cuda:0'), covar=tensor([0.0549, 0.0604, 0.0841, 0.1368, 0.0866, 0.0785, 0.0507, 0.1134], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0172, 0.0142, 0.0127, 0.0144, 0.0157, 0.0137, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:06:17,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.113e+02 2.491e+02 2.980e+02 5.007e+02, threshold=4.982e+02, percent-clipped=0.0 2022-12-08 13:06:18,127 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:06:34,205 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131848.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:06:40,047 INFO [train.py:873] (0/4) Epoch 18, batch 3300, loss[loss=0.1234, simple_loss=0.1562, pruned_loss=0.04523, over 14127.00 frames. ], tot_loss[loss=0.106, simple_loss=0.1421, pruned_loss=0.03497, over 1972009.33 frames. ], batch size: 99, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:06:41,389 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.58 vs. limit=2.0 2022-12-08 13:06:41,877 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131857.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:00,046 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=131878.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:34,784 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:07:44,548 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.221e+02 2.033e+02 2.408e+02 3.060e+02 5.712e+02, threshold=4.815e+02, percent-clipped=3.0 2022-12-08 13:07:50,991 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 13:07:55,404 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 13:08:06,596 INFO [train.py:873] (0/4) Epoch 18, batch 3400, loss[loss=0.1237, simple_loss=0.1497, pruned_loss=0.04887, over 10332.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1424, pruned_loss=0.03507, over 1970127.53 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:08:50,583 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 13:08:59,495 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 13:09:10,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.058e+02 2.441e+02 2.998e+02 5.403e+02, threshold=4.883e+02, percent-clipped=4.0 2022-12-08 13:09:12,593 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132031.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:14,771 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 13:09:32,585 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132054.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:33,283 INFO [train.py:873] (0/4) Epoch 18, batch 3500, loss[loss=0.0902, simple_loss=0.135, pruned_loss=0.02268, over 14649.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1417, pruned_loss=0.0345, over 1984294.82 frames. ], batch size: 23, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:09:42,093 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132065.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:09:47,819 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:05,393 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132092.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:23,473 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132113.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:25,323 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132115.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 13:10:29,246 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132120.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:37,536 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 2.039e+02 2.680e+02 3.337e+02 6.531e+02, threshold=5.359e+02, percent-clipped=3.0 2022-12-08 13:10:54,121 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132148.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:10:59,845 INFO [train.py:873] (0/4) Epoch 18, batch 3600, loss[loss=0.1004, simple_loss=0.1322, pruned_loss=0.03432, over 5988.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.1424, pruned_loss=0.03437, over 2039726.98 frames. ], batch size: 100, lr: 4.37e-03, grad_scale: 8.0 2022-12-08 13:11:01,200 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.22 vs. limit=5.0 2022-12-08 13:11:35,494 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2022-12-08 13:11:35,837 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132196.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:11:41,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2022-12-08 13:11:51,510 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:12:05,413 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.022e+02 2.046e+02 2.523e+02 3.076e+02 9.449e+02, threshold=5.046e+02, percent-clipped=2.0 2022-12-08 13:12:14,860 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2689, 1.7239, 2.4538, 2.0664, 2.3353, 1.6758, 2.0662, 2.2948], device='cuda:0'), covar=tensor([0.2143, 0.3122, 0.0682, 0.1770, 0.1433, 0.2027, 0.0930, 0.0901], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0200, 0.0220, 0.0271, 0.0237, 0.0203, 0.0200, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:12:28,035 INFO [train.py:873] (0/4) Epoch 18, batch 3700, loss[loss=0.1333, simple_loss=0.1544, pruned_loss=0.05609, over 7758.00 frames. ], tot_loss[loss=0.1064, simple_loss=0.1428, pruned_loss=0.03501, over 1982127.63 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:13:32,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.108e+02 2.462e+02 3.226e+02 6.885e+02, threshold=4.924e+02, percent-clipped=3.0 2022-12-08 13:13:39,104 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4027, 2.1880, 4.4229, 3.1683, 4.1803, 2.1719, 3.3412, 4.2187], device='cuda:0'), covar=tensor([0.0571, 0.3734, 0.0365, 0.5052, 0.0736, 0.2991, 0.1289, 0.0426], device='cuda:0'), in_proj_covar=tensor([0.0257, 0.0202, 0.0222, 0.0273, 0.0240, 0.0204, 0.0203, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:13:43,129 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8245, 1.7566, 1.6721, 1.9220, 1.7867, 1.8372, 1.8049, 1.6703], device='cuda:0'), covar=tensor([0.1321, 0.1074, 0.1903, 0.0838, 0.1550, 0.0656, 0.1474, 0.1143], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0291, 0.0261, 0.0293, 0.0325, 0.0305, 0.0253, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:13:54,142 INFO [train.py:873] (0/4) Epoch 18, batch 3800, loss[loss=0.1041, simple_loss=0.1385, pruned_loss=0.0348, over 13544.00 frames. ], tot_loss[loss=0.1065, simple_loss=0.1429, pruned_loss=0.03507, over 1979719.38 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 4.0 2022-12-08 13:14:17,936 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9918, 3.6175, 3.8248, 4.0979, 3.8381, 3.7865, 4.0884, 3.2889], device='cuda:0'), covar=tensor([0.1139, 0.1822, 0.1022, 0.0885, 0.1215, 0.1256, 0.1150, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0278, 0.0203, 0.0197, 0.0185, 0.0161, 0.0291, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 13:14:22,278 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132387.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:14:42,726 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132410.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 13:14:59,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 2.016e+02 2.539e+02 3.051e+02 5.858e+02, threshold=5.078e+02, percent-clipped=3.0 2022-12-08 13:15:21,337 INFO [train.py:873] (0/4) Epoch 18, batch 3900, loss[loss=0.09452, simple_loss=0.1354, pruned_loss=0.02682, over 14249.00 frames. ], tot_loss[loss=0.1059, simple_loss=0.142, pruned_loss=0.03485, over 1932842.28 frames. ], batch size: 39, lr: 4.36e-03, grad_scale: 4.0 2022-12-08 13:15:27,220 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2022-12-08 13:16:12,163 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:16:26,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.073e+02 2.605e+02 3.188e+02 7.778e+02, threshold=5.211e+02, percent-clipped=4.0 2022-12-08 13:16:48,238 INFO [train.py:873] (0/4) Epoch 18, batch 4000, loss[loss=0.1057, simple_loss=0.142, pruned_loss=0.03468, over 12755.00 frames. ], tot_loss[loss=0.1052, simple_loss=0.1417, pruned_loss=0.03431, over 2053840.49 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:16:53,538 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132561.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:17:14,234 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.9434, 5.7980, 5.3757, 5.9375, 5.4192, 5.3340, 5.9865, 5.7410], device='cuda:0'), covar=tensor([0.0503, 0.0690, 0.0784, 0.0428, 0.0695, 0.0389, 0.0456, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0145, 0.0148, 0.0164, 0.0149, 0.0124, 0.0171, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:17:17,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 13:17:54,706 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.137e+02 2.506e+02 3.319e+02 7.426e+02, threshold=5.012e+02, percent-clipped=3.0 2022-12-08 13:18:16,274 INFO [train.py:873] (0/4) Epoch 18, batch 4100, loss[loss=0.1228, simple_loss=0.1582, pruned_loss=0.04367, over 14160.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1413, pruned_loss=0.03423, over 2027772.19 frames. ], batch size: 99, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:18:17,173 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1578, 2.0043, 1.8169, 1.9330, 2.0878, 2.1075, 2.1017, 2.1037], device='cuda:0'), covar=tensor([0.1208, 0.0967, 0.2723, 0.2354, 0.1311, 0.1141, 0.1684, 0.1114], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0275, 0.0449, 0.0569, 0.0349, 0.0450, 0.0390, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:18:20,653 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132660.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:18:33,095 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5421, 1.4994, 3.4996, 1.5441, 3.4662, 3.6376, 2.4563, 3.8680], device='cuda:0'), covar=tensor([0.0239, 0.3157, 0.0457, 0.2399, 0.0747, 0.0398, 0.1025, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0154, 0.0160, 0.0167, 0.0167, 0.0177, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:18:44,629 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132687.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:05,267 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132710.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:19:11,850 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8247, 1.6725, 1.9398, 1.6550, 1.9500, 1.7701, 1.6013, 1.8390], device='cuda:0'), covar=tensor([0.0690, 0.1347, 0.0522, 0.0579, 0.0572, 0.0877, 0.0317, 0.0425], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0306, 0.0389, 0.0293, 0.0363, 0.0315, 0.0357, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:19:13,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 13:19:14,356 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132721.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:16,968 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4440, 1.6479, 2.6490, 2.0583, 2.5016, 1.6392, 2.1142, 2.4580], device='cuda:0'), covar=tensor([0.1757, 0.3667, 0.0611, 0.2367, 0.1232, 0.2603, 0.1289, 0.0913], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0198, 0.0220, 0.0269, 0.0238, 0.0202, 0.0201, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:19:22,530 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.213e+02 2.659e+02 3.609e+02 5.635e+02, threshold=5.318e+02, percent-clipped=5.0 2022-12-08 13:19:27,153 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132735.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:36,735 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132746.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:44,647 INFO [train.py:873] (0/4) Epoch 18, batch 4200, loss[loss=0.0997, simple_loss=0.1306, pruned_loss=0.03441, over 3891.00 frames. ], tot_loss[loss=0.1061, simple_loss=0.1423, pruned_loss=0.03496, over 1979776.02 frames. ], batch size: 100, lr: 4.36e-03, grad_scale: 8.0 2022-12-08 13:19:47,328 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=132758.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:19:54,007 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9341, 4.6566, 4.2897, 4.5405, 4.6582, 4.8141, 4.8807, 4.8265], device='cuda:0'), covar=tensor([0.0662, 0.0488, 0.2044, 0.2680, 0.0675, 0.0757, 0.0672, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0392, 0.0275, 0.0449, 0.0570, 0.0349, 0.0452, 0.0391, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:20:30,587 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132807.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:20:50,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.134e+02 2.571e+02 3.230e+02 5.703e+02, threshold=5.142e+02, percent-clipped=2.0 2022-12-08 13:21:13,200 INFO [train.py:873] (0/4) Epoch 18, batch 4300, loss[loss=0.0898, simple_loss=0.1339, pruned_loss=0.02283, over 14550.00 frames. ], tot_loss[loss=0.1063, simple_loss=0.1425, pruned_loss=0.03505, over 1994864.29 frames. ], batch size: 43, lr: 4.35e-03, grad_scale: 4.0 2022-12-08 13:22:20,489 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 2.027e+02 2.464e+02 3.102e+02 5.439e+02, threshold=4.929e+02, percent-clipped=1.0 2022-12-08 13:22:41,307 INFO [train.py:873] (0/4) Epoch 18, batch 4400, loss[loss=0.07258, simple_loss=0.1195, pruned_loss=0.01283, over 13920.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.142, pruned_loss=0.03456, over 2025069.30 frames. ], batch size: 20, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:23:35,943 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133016.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:23:48,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.037e+02 1.881e+02 2.190e+02 2.815e+02 5.316e+02, threshold=4.380e+02, percent-clipped=1.0 2022-12-08 13:23:56,426 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9151, 1.2496, 2.0039, 1.2713, 1.9583, 2.0541, 1.6458, 2.1391], device='cuda:0'), covar=tensor([0.0295, 0.2415, 0.0601, 0.2131, 0.0673, 0.0710, 0.1296, 0.0407], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0154, 0.0160, 0.0167, 0.0167, 0.0179, 0.0134, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:24:10,122 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0791, 2.6143, 3.3930, 2.4610, 2.2086, 3.0046, 1.6038, 3.0822], device='cuda:0'), covar=tensor([0.0804, 0.1211, 0.0772, 0.2125, 0.2033, 0.0855, 0.3334, 0.0979], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0103, 0.0095, 0.0101, 0.0115, 0.0092, 0.0117, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 13:24:10,912 INFO [train.py:873] (0/4) Epoch 18, batch 4500, loss[loss=0.1037, simple_loss=0.1497, pruned_loss=0.02885, over 14388.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.1412, pruned_loss=0.03388, over 1993181.84 frames. ], batch size: 53, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:24:34,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 13:24:52,351 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133102.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:25:17,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.087e+02 2.404e+02 3.093e+02 4.922e+02, threshold=4.807e+02, percent-clipped=5.0 2022-12-08 13:25:38,355 INFO [train.py:873] (0/4) Epoch 18, batch 4600, loss[loss=0.1078, simple_loss=0.1426, pruned_loss=0.03656, over 14465.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1419, pruned_loss=0.03453, over 1937035.99 frames. ], batch size: 49, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:25:41,141 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4152, 3.5848, 3.7391, 3.4324, 3.6138, 3.5421, 1.5091, 3.3553], device='cuda:0'), covar=tensor([0.0426, 0.0444, 0.0332, 0.0546, 0.0335, 0.0479, 0.3176, 0.0336], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0179, 0.0148, 0.0151, 0.0210, 0.0143, 0.0160, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:26:00,664 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8635, 0.8008, 0.7138, 0.8905, 0.8528, 0.3495, 0.7396, 0.8798], device='cuda:0'), covar=tensor([0.0423, 0.0610, 0.0572, 0.0450, 0.0309, 0.0308, 0.1078, 0.0622], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0035, 0.0040, 0.0034, 0.0034, 0.0048, 0.0036, 0.0038], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:26:44,621 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 2.089e+02 2.539e+02 2.980e+02 5.650e+02, threshold=5.079e+02, percent-clipped=2.0 2022-12-08 13:27:05,956 INFO [train.py:873] (0/4) Epoch 18, batch 4700, loss[loss=0.1106, simple_loss=0.1523, pruned_loss=0.03448, over 14154.00 frames. ], tot_loss[loss=0.1058, simple_loss=0.1419, pruned_loss=0.03487, over 1922704.40 frames. ], batch size: 84, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:27:08,643 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8255, 4.1612, 3.5939, 3.9240, 2.9455, 4.0366, 3.9318, 2.5578], device='cuda:0'), covar=tensor([0.1139, 0.0441, 0.1093, 0.0799, 0.0792, 0.0405, 0.0932, 0.1473], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0091, 0.0071, 0.0077, 0.0100, 0.0092, 0.0103, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:27:59,042 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133316.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:28:12,072 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.140e+02 2.537e+02 3.389e+02 1.353e+03, threshold=5.074e+02, percent-clipped=8.0 2022-12-08 13:28:32,820 INFO [train.py:873] (0/4) Epoch 18, batch 4800, loss[loss=0.155, simple_loss=0.145, pruned_loss=0.0825, over 1322.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1414, pruned_loss=0.03481, over 1896731.77 frames. ], batch size: 100, lr: 4.35e-03, grad_scale: 8.0 2022-12-08 13:28:32,949 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8212, 4.5284, 4.2804, 4.4285, 4.5955, 4.7086, 4.7888, 4.7669], device='cuda:0'), covar=tensor([0.0701, 0.0517, 0.2022, 0.2629, 0.0638, 0.0817, 0.0779, 0.0786], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0278, 0.0453, 0.0574, 0.0351, 0.0453, 0.0391, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:28:40,334 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=133364.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:28:49,086 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133374.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:28:51,163 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133376.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:13,888 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133402.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:20,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4462, 2.2797, 3.3970, 2.5121, 3.3363, 3.2634, 3.1430, 2.7409], device='cuda:0'), covar=tensor([0.1069, 0.3398, 0.1085, 0.2042, 0.0869, 0.1160, 0.1377, 0.2246], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0309, 0.0390, 0.0297, 0.0365, 0.0319, 0.0361, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:29:27,123 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5596, 1.4528, 2.7157, 1.5598, 2.7311, 2.6457, 2.0120, 2.8647], device='cuda:0'), covar=tensor([0.0317, 0.2748, 0.0417, 0.1983, 0.0457, 0.0508, 0.1282, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0154, 0.0160, 0.0167, 0.0167, 0.0178, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:29:33,294 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0050, 3.6294, 3.3775, 2.6354, 3.3690, 3.6617, 4.0258, 3.3031], device='cuda:0'), covar=tensor([0.0589, 0.0965, 0.0896, 0.1231, 0.0847, 0.0633, 0.0825, 0.1023], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0171, 0.0141, 0.0126, 0.0145, 0.0157, 0.0136, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:29:38,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 1.939e+02 2.430e+02 2.952e+02 7.648e+02, threshold=4.861e+02, percent-clipped=3.0 2022-12-08 13:29:42,972 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133435.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:29:44,657 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133437.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:55,374 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=133450.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:29:59,644 INFO [train.py:873] (0/4) Epoch 18, batch 4900, loss[loss=0.1279, simple_loss=0.1473, pruned_loss=0.05423, over 3849.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.141, pruned_loss=0.03412, over 1982190.09 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 8.0 2022-12-08 13:30:16,306 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133474.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:30:32,643 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3181, 3.7616, 2.9511, 4.5375, 4.1821, 4.3723, 3.8494, 3.2186], device='cuda:0'), covar=tensor([0.0666, 0.1125, 0.3170, 0.0498, 0.0844, 0.1126, 0.1069, 0.2557], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0290, 0.0260, 0.0293, 0.0321, 0.0305, 0.0255, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:31:05,364 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 2.069e+02 2.523e+02 3.509e+02 9.363e+02, threshold=5.045e+02, percent-clipped=9.0 2022-12-08 13:31:09,197 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133535.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:31:12,635 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1204, 1.9213, 3.1874, 2.3545, 3.1333, 1.9308, 2.5314, 3.1203], device='cuda:0'), covar=tensor([0.0921, 0.3929, 0.0681, 0.4672, 0.0901, 0.2982, 0.1375, 0.0662], device='cuda:0'), in_proj_covar=tensor([0.0259, 0.0202, 0.0224, 0.0274, 0.0243, 0.0206, 0.0205, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:31:26,063 INFO [train.py:873] (0/4) Epoch 18, batch 5000, loss[loss=0.1047, simple_loss=0.1312, pruned_loss=0.03909, over 5002.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1415, pruned_loss=0.03461, over 1974576.40 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:32:32,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.004e+02 2.618e+02 3.128e+02 4.580e+02, threshold=5.235e+02, percent-clipped=0.0 2022-12-08 13:32:50,849 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133652.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:32:53,176 INFO [train.py:873] (0/4) Epoch 18, batch 5100, loss[loss=0.1054, simple_loss=0.1425, pruned_loss=0.0342, over 14249.00 frames. ], tot_loss[loss=0.1055, simple_loss=0.1413, pruned_loss=0.03485, over 1988987.97 frames. ], batch size: 63, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:33:09,586 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2022-12-08 13:33:11,991 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3716, 2.0806, 4.9188, 4.4345, 4.3220, 4.9991, 4.7857, 5.0726], device='cuda:0'), covar=tensor([0.1459, 0.1499, 0.0095, 0.0203, 0.0240, 0.0119, 0.0096, 0.0087], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0130, 0.0167, 0.0146, 0.0143, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 13:33:44,361 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133713.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:33:45,135 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6044, 1.5784, 1.5978, 1.6658, 1.7358, 1.1579, 1.4650, 1.5459], device='cuda:0'), covar=tensor([0.0857, 0.0833, 0.0728, 0.0967, 0.0640, 0.0908, 0.0807, 0.0636], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0035, 0.0040, 0.0034, 0.0035, 0.0049, 0.0037, 0.0039], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:33:58,864 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133730.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 13:34:00,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.056e+02 2.175e+02 2.651e+02 3.112e+02 1.000e+03, threshold=5.302e+02, percent-clipped=1.0 2022-12-08 13:34:00,908 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:34:01,861 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7245, 4.5985, 4.3643, 4.7589, 4.3390, 4.0515, 4.8178, 4.4945], device='cuda:0'), covar=tensor([0.0645, 0.0691, 0.0900, 0.0620, 0.0833, 0.0699, 0.0548, 0.0849], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0146, 0.0150, 0.0165, 0.0150, 0.0126, 0.0172, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:34:21,229 INFO [train.py:873] (0/4) Epoch 18, batch 5200, loss[loss=0.1012, simple_loss=0.1457, pruned_loss=0.02834, over 14301.00 frames. ], tot_loss[loss=0.1051, simple_loss=0.1413, pruned_loss=0.03444, over 2011634.25 frames. ], batch size: 60, lr: 4.34e-03, grad_scale: 8.0 2022-12-08 13:34:26,533 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9972, 1.9916, 2.1617, 2.1404, 2.1101, 1.8761, 1.7608, 1.2914], device='cuda:0'), covar=tensor([0.0226, 0.0575, 0.0296, 0.0244, 0.0197, 0.0308, 0.0272, 0.0557], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:34:27,605 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.58 vs. limit=5.0 2022-12-08 13:35:27,073 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:35:29,271 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 1.917e+02 2.403e+02 3.012e+02 6.543e+02, threshold=4.806e+02, percent-clipped=3.0 2022-12-08 13:35:35,158 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1000, 3.1064, 2.9293, 3.2089, 2.7981, 2.8477, 3.1511, 3.0786], device='cuda:0'), covar=tensor([0.0821, 0.1022, 0.1010, 0.0695, 0.1351, 0.0789, 0.0845, 0.0886], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0145, 0.0150, 0.0164, 0.0150, 0.0125, 0.0171, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:35:39,637 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1782, 2.1560, 4.1274, 2.9703, 3.9905, 1.9867, 3.1562, 4.0408], device='cuda:0'), covar=tensor([0.0693, 0.3849, 0.0470, 0.4536, 0.0822, 0.3064, 0.1321, 0.0612], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0200, 0.0223, 0.0269, 0.0240, 0.0203, 0.0203, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:35:48,359 INFO [train.py:873] (0/4) Epoch 18, batch 5300, loss[loss=0.1673, simple_loss=0.1592, pruned_loss=0.08773, over 1207.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1414, pruned_loss=0.03401, over 2009665.61 frames. ], batch size: 100, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:36:21,422 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.6084, 1.7338, 1.7969, 1.3083, 1.2258, 1.5736, 1.1228, 1.6959], device='cuda:0'), covar=tensor([0.1551, 0.2272, 0.1035, 0.2540, 0.3290, 0.1270, 0.3011, 0.1233], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0104, 0.0096, 0.0101, 0.0117, 0.0092, 0.0118, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 13:36:33,185 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 13:36:49,476 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4098, 2.2212, 2.0784, 2.2346, 2.2718, 1.4321, 2.0806, 2.2869], device='cuda:0'), covar=tensor([0.0705, 0.0769, 0.1148, 0.0885, 0.0812, 0.0778, 0.1111, 0.0603], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0035, 0.0050, 0.0037, 0.0040], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:36:55,585 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.086e+02 2.076e+02 2.589e+02 3.060e+02 6.026e+02, threshold=5.178e+02, percent-clipped=1.0 2022-12-08 13:37:14,384 INFO [train.py:873] (0/4) Epoch 18, batch 5400, loss[loss=0.09756, simple_loss=0.1409, pruned_loss=0.02709, over 14264.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1417, pruned_loss=0.03445, over 1980097.58 frames. ], batch size: 57, lr: 4.34e-03, grad_scale: 4.0 2022-12-08 13:37:51,041 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133997.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:00,648 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:13,625 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7378, 1.6833, 1.7847, 1.7690, 1.7067, 1.7806, 1.5206, 1.3019], device='cuda:0'), covar=tensor([0.0226, 0.0264, 0.0194, 0.0218, 0.0234, 0.0299, 0.0254, 0.0383], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0022, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:38:20,345 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134030.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:38:21,987 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134032.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:38:22,721 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.308e+02 1.858e+02 2.390e+02 3.308e+02 6.761e+02, threshold=4.781e+02, percent-clipped=5.0 2022-12-08 13:38:42,049 INFO [train.py:873] (0/4) Epoch 18, batch 5500, loss[loss=0.09761, simple_loss=0.1385, pruned_loss=0.02836, over 13938.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1408, pruned_loss=0.0334, over 1979220.91 frames. ], batch size: 20, lr: 4.33e-03, grad_scale: 4.0 2022-12-08 13:38:44,888 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134058.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:01,649 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2022-12-08 13:39:02,231 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134078.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 13:39:03,894 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:47,400 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134130.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:39:50,149 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 2.212e+02 2.722e+02 3.305e+02 6.043e+02, threshold=5.445e+02, percent-clipped=3.0 2022-12-08 13:40:09,463 INFO [train.py:873] (0/4) Epoch 18, batch 5600, loss[loss=0.08488, simple_loss=0.1013, pruned_loss=0.03424, over 2658.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1416, pruned_loss=0.03392, over 2003291.19 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:40:29,737 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134178.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:40:55,445 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7409, 1.4853, 3.6282, 3.3712, 3.5337, 3.7571, 2.8734, 3.6766], device='cuda:0'), covar=tensor([0.2392, 0.2506, 0.0256, 0.0439, 0.0422, 0.0266, 0.0658, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0157, 0.0130, 0.0167, 0.0146, 0.0142, 0.0125, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 13:40:56,674 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2022-12-08 13:41:14,341 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0831, 1.9654, 2.0469, 2.1394, 2.0466, 2.0584, 2.1569, 1.8665], device='cuda:0'), covar=tensor([0.0924, 0.1288, 0.0819, 0.0839, 0.0956, 0.0668, 0.0920, 0.0718], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0275, 0.0200, 0.0195, 0.0184, 0.0157, 0.0289, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 13:41:18,750 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 1.953e+02 2.384e+02 3.110e+02 5.351e+02, threshold=4.768e+02, percent-clipped=0.0 2022-12-08 13:41:38,335 INFO [train.py:873] (0/4) Epoch 18, batch 5700, loss[loss=0.09003, simple_loss=0.1306, pruned_loss=0.02473, over 14182.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1415, pruned_loss=0.03409, over 1997560.80 frames. ], batch size: 84, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:42:08,745 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.2693, 1.2083, 1.2373, 1.0568, 1.1080, 1.0075, 1.1213, 1.0182], device='cuda:0'), covar=tensor([0.0270, 0.0330, 0.0221, 0.0313, 0.0276, 0.0454, 0.0315, 0.0427], device='cuda:0'), in_proj_covar=tensor([0.0023, 0.0023, 0.0020, 0.0022, 0.0021, 0.0034, 0.0028, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:42:21,765 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134304.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:25,189 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134308.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:26,033 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134309.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:42:40,193 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0796, 3.0764, 2.9094, 3.2233, 2.7551, 2.8443, 3.1662, 3.0776], device='cuda:0'), covar=tensor([0.0765, 0.1149, 0.1134, 0.0763, 0.1391, 0.0841, 0.0842, 0.0922], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0152, 0.0166, 0.0152, 0.0127, 0.0174, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:42:47,524 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.380e+02 2.136e+02 2.600e+02 3.296e+02 5.182e+02, threshold=5.200e+02, percent-clipped=3.0 2022-12-08 13:43:05,220 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:06,868 INFO [train.py:873] (0/4) Epoch 18, batch 5800, loss[loss=0.1335, simple_loss=0.1364, pruned_loss=0.0653, over 1210.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1404, pruned_loss=0.03379, over 1939149.54 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:43:07,807 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134356.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:16,135 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134365.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:43:20,261 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134370.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:44:15,976 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6738, 2.3951, 3.5810, 2.5697, 3.4762, 3.4445, 3.3807, 2.8601], device='cuda:0'), covar=tensor([0.0728, 0.3010, 0.0815, 0.1965, 0.0830, 0.0938, 0.1193, 0.2305], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0309, 0.0388, 0.0300, 0.0364, 0.0320, 0.0359, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:44:17,489 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.136e+02 2.508e+02 3.176e+02 6.308e+02, threshold=5.017e+02, percent-clipped=2.0 2022-12-08 13:44:33,881 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0456, 2.0379, 4.0936, 2.8671, 3.9512, 2.0452, 2.9160, 3.9293], device='cuda:0'), covar=tensor([0.0652, 0.4003, 0.0484, 0.5141, 0.0646, 0.3247, 0.1532, 0.0573], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0200, 0.0222, 0.0271, 0.0239, 0.0202, 0.0203, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:44:37,062 INFO [train.py:873] (0/4) Epoch 18, batch 5900, loss[loss=0.1101, simple_loss=0.1517, pruned_loss=0.03423, over 14324.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1402, pruned_loss=0.03389, over 1906814.96 frames. ], batch size: 55, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:44:52,873 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8541, 3.4734, 2.8069, 4.0324, 3.9021, 3.8925, 3.4942, 2.7918], device='cuda:0'), covar=tensor([0.0870, 0.1288, 0.3109, 0.0715, 0.0890, 0.1151, 0.1226, 0.3055], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0290, 0.0260, 0.0293, 0.0323, 0.0304, 0.0254, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:44:57,306 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2795, 1.6230, 2.4255, 2.0453, 2.3277, 1.6564, 2.0412, 2.3492], device='cuda:0'), covar=tensor([0.2284, 0.3589, 0.0713, 0.1895, 0.1567, 0.2319, 0.1085, 0.0950], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0199, 0.0222, 0.0271, 0.0238, 0.0202, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 13:45:12,392 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4241, 2.4611, 2.5505, 2.4935, 2.4795, 2.1389, 1.4855, 2.2602], device='cuda:0'), covar=tensor([0.0647, 0.0558, 0.0382, 0.0390, 0.0415, 0.1231, 0.2419, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0175, 0.0147, 0.0149, 0.0207, 0.0141, 0.0157, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:45:45,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.002e+02 2.502e+02 3.052e+02 5.386e+02, threshold=5.003e+02, percent-clipped=2.0 2022-12-08 13:45:59,924 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3326, 5.3230, 5.0334, 5.3523, 5.0297, 4.9622, 5.4480, 5.0290], device='cuda:0'), covar=tensor([0.0631, 0.0743, 0.0800, 0.0683, 0.0708, 0.0581, 0.0618, 0.0946], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0151, 0.0153, 0.0168, 0.0154, 0.0129, 0.0175, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2022-12-08 13:46:04,785 INFO [train.py:873] (0/4) Epoch 18, batch 6000, loss[loss=0.1146, simple_loss=0.1261, pruned_loss=0.05157, over 3868.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.1409, pruned_loss=0.03436, over 1903808.36 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:46:04,786 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 13:46:14,283 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4509, 1.7394, 3.3161, 2.0339, 3.5741, 3.4713, 2.4259, 3.6616], device='cuda:0'), covar=tensor([0.0242, 0.2920, 0.0434, 0.1778, 0.0293, 0.0313, 0.0737, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0155, 0.0162, 0.0168, 0.0169, 0.0179, 0.0132, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:46:17,879 INFO [train.py:905] (0/4) Epoch 18, validation: loss=0.1412, simple_loss=0.1783, pruned_loss=0.05203, over 857387.00 frames. 2022-12-08 13:46:17,880 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 13:46:22,802 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2022-12-08 13:46:49,569 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7265, 3.3537, 2.6377, 3.9220, 3.8006, 3.7570, 3.4009, 2.6866], device='cuda:0'), covar=tensor([0.1012, 0.1338, 0.3126, 0.0660, 0.0814, 0.1142, 0.1083, 0.2942], device='cuda:0'), in_proj_covar=tensor([0.0288, 0.0294, 0.0264, 0.0297, 0.0327, 0.0309, 0.0257, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:47:26,071 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 2.089e+02 2.569e+02 3.442e+02 1.078e+03, threshold=5.138e+02, percent-clipped=4.0 2022-12-08 13:47:44,020 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134653.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:47:45,553 INFO [train.py:873] (0/4) Epoch 18, batch 6100, loss[loss=0.1121, simple_loss=0.1469, pruned_loss=0.03864, over 13530.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1407, pruned_loss=0.03374, over 1902395.70 frames. ], batch size: 100, lr: 4.33e-03, grad_scale: 8.0 2022-12-08 13:47:49,992 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134660.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:47:54,208 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134665.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:48:05,097 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 13:48:25,747 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=134701.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:48:41,486 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 13:48:48,544 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2116, 4.3516, 4.5561, 3.8755, 4.4032, 4.5926, 1.7876, 4.1433], device='cuda:0'), covar=tensor([0.0376, 0.0445, 0.0325, 0.0510, 0.0328, 0.0243, 0.2980, 0.0315], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0175, 0.0147, 0.0148, 0.0206, 0.0141, 0.0157, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:48:53,032 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4157, 2.4991, 4.2143, 4.5000, 4.0969, 2.5776, 4.4816, 3.3191], device='cuda:0'), covar=tensor([0.0390, 0.1337, 0.0966, 0.0465, 0.0544, 0.2036, 0.0424, 0.1050], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0261, 0.0377, 0.0334, 0.0274, 0.0309, 0.0312, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:48:53,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.283e+02 2.174e+02 2.543e+02 3.201e+02 1.223e+03, threshold=5.086e+02, percent-clipped=2.0 2022-12-08 13:49:12,509 INFO [train.py:873] (0/4) Epoch 18, batch 6200, loss[loss=0.103, simple_loss=0.1354, pruned_loss=0.03529, over 6947.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.141, pruned_loss=0.03436, over 1906986.87 frames. ], batch size: 100, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:49:17,938 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2022-12-08 13:49:28,426 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0744, 3.7882, 3.5997, 3.7563, 3.9582, 3.9686, 4.0621, 4.0299], device='cuda:0'), covar=tensor([0.0874, 0.0700, 0.2140, 0.2730, 0.0766, 0.0950, 0.0908, 0.0885], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0278, 0.0452, 0.0576, 0.0355, 0.0458, 0.0389, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:50:03,150 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0556, 2.6792, 3.3466, 2.1269, 2.1754, 2.8146, 1.6701, 2.8903], device='cuda:0'), covar=tensor([0.1016, 0.1060, 0.0630, 0.1790, 0.1856, 0.1002, 0.2874, 0.0978], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0101, 0.0095, 0.0099, 0.0114, 0.0092, 0.0115, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 13:50:20,437 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.042e+02 2.445e+02 3.055e+02 6.497e+02, threshold=4.889e+02, percent-clipped=2.0 2022-12-08 13:50:35,209 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1676, 1.4085, 1.6557, 1.6313, 1.5496, 1.6486, 1.3853, 1.2626], device='cuda:0'), covar=tensor([0.1470, 0.1332, 0.0389, 0.0581, 0.1387, 0.1112, 0.1889, 0.1533], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0090, 0.0071, 0.0077, 0.0101, 0.0092, 0.0103, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:50:40,210 INFO [train.py:873] (0/4) Epoch 18, batch 6300, loss[loss=0.1057, simple_loss=0.1491, pruned_loss=0.03109, over 14269.00 frames. ], tot_loss[loss=0.1052, simple_loss=0.1413, pruned_loss=0.03455, over 1937392.01 frames. ], batch size: 76, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:50:40,473 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5211, 4.0857, 2.9883, 4.9328, 4.3867, 4.6849, 4.0889, 3.2932], device='cuda:0'), covar=tensor([0.0742, 0.0948, 0.3230, 0.0456, 0.0583, 0.0850, 0.0985, 0.2723], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0289, 0.0260, 0.0293, 0.0323, 0.0304, 0.0253, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:50:46,269 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134862.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:50:58,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2188, 2.4252, 4.2152, 4.4131, 4.1352, 2.4261, 4.3379, 3.1282], device='cuda:0'), covar=tensor([0.0441, 0.1417, 0.0860, 0.0383, 0.0529, 0.2161, 0.0391, 0.1085], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0259, 0.0374, 0.0331, 0.0271, 0.0306, 0.0310, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:51:39,640 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134923.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:51:44,653 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7263, 1.5078, 2.8947, 1.5328, 2.9844, 2.8589, 2.1838, 3.0942], device='cuda:0'), covar=tensor([0.0277, 0.2941, 0.0504, 0.2110, 0.0406, 0.0543, 0.1011, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0156, 0.0162, 0.0168, 0.0169, 0.0180, 0.0133, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:51:46,849 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2022-12-08 13:51:47,984 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.107e+02 2.599e+02 3.257e+02 7.432e+02, threshold=5.197e+02, percent-clipped=1.0 2022-12-08 13:52:07,216 INFO [train.py:873] (0/4) Epoch 18, batch 6400, loss[loss=0.0886, simple_loss=0.135, pruned_loss=0.0211, over 14254.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.141, pruned_loss=0.03404, over 1934439.50 frames. ], batch size: 80, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:52:12,158 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134960.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:52:16,341 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134965.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:52:47,092 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-135000.pt 2022-12-08 13:52:57,646 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:53:01,788 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135013.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:53:18,809 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.100e+02 2.565e+02 3.370e+02 5.862e+02, threshold=5.131e+02, percent-clipped=3.0 2022-12-08 13:53:38,283 INFO [train.py:873] (0/4) Epoch 18, batch 6500, loss[loss=0.1039, simple_loss=0.1407, pruned_loss=0.03354, over 14576.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.141, pruned_loss=0.03425, over 1931901.95 frames. ], batch size: 49, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:53:54,184 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8580, 1.7634, 2.0640, 1.9041, 1.7744, 1.7881, 1.7479, 1.2318], device='cuda:0'), covar=tensor([0.0218, 0.0398, 0.0172, 0.0195, 0.0219, 0.0303, 0.0265, 0.0328], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 13:53:59,671 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0199, 1.9787, 1.9582, 2.0146, 1.9014, 1.6321, 1.2962, 1.7294], device='cuda:0'), covar=tensor([0.0897, 0.0817, 0.0707, 0.0617, 0.0734, 0.1538, 0.2839, 0.0821], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0177, 0.0148, 0.0150, 0.0208, 0.0141, 0.0158, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:54:00,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2996, 4.8147, 4.7789, 5.2575, 4.9655, 4.4477, 5.2272, 4.4585], device='cuda:0'), covar=tensor([0.0360, 0.0963, 0.0414, 0.0400, 0.0741, 0.0587, 0.0569, 0.0489], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0279, 0.0203, 0.0200, 0.0186, 0.0161, 0.0293, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 13:54:11,813 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2022-12-08 13:54:15,572 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4041, 2.0387, 2.4088, 2.4496, 2.2339, 2.0017, 2.4248, 2.1496], device='cuda:0'), covar=tensor([0.0436, 0.1097, 0.0506, 0.0466, 0.0808, 0.1404, 0.0612, 0.0716], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0260, 0.0374, 0.0332, 0.0272, 0.0307, 0.0310, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 13:54:19,508 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([6.0117, 5.4830, 5.3562, 5.9458, 5.5508, 4.8783, 5.8864, 4.7681], device='cuda:0'), covar=tensor([0.0309, 0.0877, 0.0404, 0.0351, 0.0761, 0.0350, 0.0463, 0.0561], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0281, 0.0204, 0.0201, 0.0188, 0.0162, 0.0295, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 13:54:36,812 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2703, 2.6168, 2.4731, 2.7028, 2.1808, 2.6742, 2.5306, 1.6141], device='cuda:0'), covar=tensor([0.1049, 0.0702, 0.0951, 0.0432, 0.0994, 0.0579, 0.0841, 0.1833], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0091, 0.0071, 0.0077, 0.0101, 0.0093, 0.0103, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:54:36,875 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3751, 2.2188, 3.2527, 2.4999, 3.1980, 3.1087, 3.0387, 2.6329], device='cuda:0'), covar=tensor([0.0892, 0.2887, 0.0992, 0.1832, 0.0746, 0.1141, 0.1343, 0.1864], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0308, 0.0389, 0.0299, 0.0361, 0.0321, 0.0360, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 13:54:46,810 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.038e+02 2.650e+02 3.158e+02 7.242e+02, threshold=5.300e+02, percent-clipped=3.0 2022-12-08 13:55:05,522 INFO [train.py:873] (0/4) Epoch 18, batch 6600, loss[loss=0.1036, simple_loss=0.1451, pruned_loss=0.03103, over 13986.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1411, pruned_loss=0.03421, over 1927431.71 frames. ], batch size: 26, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:55:20,610 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:55:29,588 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8434, 0.7971, 0.8274, 0.8042, 0.7671, 0.5367, 0.5903, 0.7045], device='cuda:0'), covar=tensor([0.0238, 0.0188, 0.0185, 0.0187, 0.0196, 0.0356, 0.0244, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 13:56:01,489 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:56:13,599 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135232.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 13:56:14,313 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.150e+02 2.615e+02 3.167e+02 5.492e+02, threshold=5.230e+02, percent-clipped=1.0 2022-12-08 13:56:34,471 INFO [train.py:873] (0/4) Epoch 18, batch 6700, loss[loss=0.1452, simple_loss=0.1365, pruned_loss=0.07693, over 1207.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.1413, pruned_loss=0.03402, over 1950390.57 frames. ], batch size: 100, lr: 4.32e-03, grad_scale: 8.0 2022-12-08 13:57:11,841 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9838, 2.0134, 2.0721, 2.0901, 2.0480, 1.6712, 1.3286, 1.8196], device='cuda:0'), covar=tensor([0.0792, 0.0689, 0.0504, 0.0415, 0.0526, 0.1488, 0.2236, 0.0604], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0177, 0.0148, 0.0150, 0.0208, 0.0142, 0.0158, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 13:57:22,570 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3181, 2.9128, 3.0057, 2.0150, 2.7847, 3.0691, 3.3308, 2.6271], device='cuda:0'), covar=tensor([0.0665, 0.0897, 0.0841, 0.1385, 0.0933, 0.0756, 0.0646, 0.1165], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0170, 0.0139, 0.0126, 0.0146, 0.0155, 0.0137, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 13:57:37,121 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 13:57:41,809 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.031e+02 2.097e+02 2.445e+02 2.909e+02 5.288e+02, threshold=4.890e+02, percent-clipped=1.0 2022-12-08 13:58:00,104 INFO [train.py:873] (0/4) Epoch 18, batch 6800, loss[loss=0.108, simple_loss=0.1393, pruned_loss=0.03829, over 4942.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.1412, pruned_loss=0.03385, over 1958788.66 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 13:58:35,997 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0720, 2.9221, 3.6893, 2.7703, 2.2096, 3.2486, 1.5959, 3.2867], device='cuda:0'), covar=tensor([0.1314, 0.1188, 0.0736, 0.1684, 0.2223, 0.0966, 0.3342, 0.1265], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0102, 0.0095, 0.0099, 0.0116, 0.0092, 0.0116, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 13:58:49,251 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2022-12-08 13:59:07,783 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.213e+02 2.677e+02 3.514e+02 1.198e+03, threshold=5.353e+02, percent-clipped=7.0 2022-12-08 13:59:27,059 INFO [train.py:873] (0/4) Epoch 18, batch 6900, loss[loss=0.1068, simple_loss=0.1496, pruned_loss=0.03195, over 14283.00 frames. ], tot_loss[loss=0.105, simple_loss=0.1413, pruned_loss=0.03435, over 1981574.57 frames. ], batch size: 69, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 13:59:52,906 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2022-12-08 14:00:17,055 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3823, 3.2446, 2.9564, 3.1011, 3.3435, 3.3770, 3.3962, 3.4093], device='cuda:0'), covar=tensor([0.1018, 0.0606, 0.2204, 0.2531, 0.0767, 0.0917, 0.1018, 0.0860], device='cuda:0'), in_proj_covar=tensor([0.0395, 0.0277, 0.0452, 0.0574, 0.0353, 0.0456, 0.0386, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:00:21,763 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135518.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:00:29,503 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135527.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:00:35,294 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.207e+02 2.585e+02 3.410e+02 8.303e+02, threshold=5.169e+02, percent-clipped=6.0 2022-12-08 14:00:53,012 INFO [train.py:873] (0/4) Epoch 18, batch 7000, loss[loss=0.1829, simple_loss=0.1597, pruned_loss=0.1031, over 1220.00 frames. ], tot_loss[loss=0.1051, simple_loss=0.1412, pruned_loss=0.03454, over 1956834.02 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 4.0 2022-12-08 14:00:58,104 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7773, 4.6305, 4.3549, 4.8057, 4.3453, 4.2444, 4.8789, 4.5804], device='cuda:0'), covar=tensor([0.0601, 0.0824, 0.0864, 0.0548, 0.0890, 0.0633, 0.0555, 0.0791], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0167, 0.0151, 0.0127, 0.0174, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:01:03,620 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135566.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:01:59,468 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:02:02,411 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 2.039e+02 2.453e+02 3.005e+02 1.669e+03, threshold=4.906e+02, percent-clipped=2.0 2022-12-08 14:02:21,584 INFO [train.py:873] (0/4) Epoch 18, batch 7100, loss[loss=0.1096, simple_loss=0.1516, pruned_loss=0.03376, over 14487.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1413, pruned_loss=0.03427, over 1972053.65 frames. ], batch size: 49, lr: 4.31e-03, grad_scale: 4.0 2022-12-08 14:02:53,421 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:03:12,077 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2360, 4.0193, 3.7521, 3.8958, 4.1127, 4.1872, 4.2263, 4.2382], device='cuda:0'), covar=tensor([0.0902, 0.0557, 0.2235, 0.2560, 0.0727, 0.0840, 0.0891, 0.0773], device='cuda:0'), in_proj_covar=tensor([0.0394, 0.0274, 0.0451, 0.0571, 0.0351, 0.0454, 0.0384, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:03:29,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([6.0500, 5.9058, 5.4804, 6.1471, 5.5737, 5.4685, 6.1422, 5.8739], device='cuda:0'), covar=tensor([0.0625, 0.0645, 0.0768, 0.0391, 0.0684, 0.0347, 0.0545, 0.0620], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0167, 0.0152, 0.0127, 0.0174, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:03:30,725 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.202e+02 1.933e+02 2.466e+02 3.185e+02 5.052e+02, threshold=4.932e+02, percent-clipped=2.0 2022-12-08 14:03:48,867 INFO [train.py:873] (0/4) Epoch 18, batch 7200, loss[loss=0.1438, simple_loss=0.1393, pruned_loss=0.07411, over 2576.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1417, pruned_loss=0.03482, over 1994851.17 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 14:04:18,260 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3389, 2.0829, 4.2963, 2.9126, 4.1529, 1.9933, 3.1897, 4.1932], device='cuda:0'), covar=tensor([0.0613, 0.4145, 0.0479, 0.5707, 0.0635, 0.3540, 0.1494, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0256, 0.0201, 0.0221, 0.0270, 0.0239, 0.0204, 0.0202, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:04:53,014 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135827.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:04:58,959 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 2.075e+02 2.523e+02 3.199e+02 6.288e+02, threshold=5.045e+02, percent-clipped=2.0 2022-12-08 14:05:18,066 INFO [train.py:873] (0/4) Epoch 18, batch 7300, loss[loss=0.1076, simple_loss=0.1412, pruned_loss=0.03703, over 5991.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.1405, pruned_loss=0.03409, over 1979857.70 frames. ], batch size: 100, lr: 4.31e-03, grad_scale: 8.0 2022-12-08 14:05:35,220 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=135875.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:06:16,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.12 vs. limit=5.0 2022-12-08 14:06:27,399 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.731e+01 2.137e+02 2.530e+02 3.235e+02 6.685e+02, threshold=5.060e+02, percent-clipped=2.0 2022-12-08 14:06:44,772 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 14:06:45,611 INFO [train.py:873] (0/4) Epoch 18, batch 7400, loss[loss=0.1434, simple_loss=0.166, pruned_loss=0.06039, over 9485.00 frames. ], tot_loss[loss=0.1048, simple_loss=0.1412, pruned_loss=0.03419, over 2020508.95 frames. ], batch size: 100, lr: 4.30e-03, grad_scale: 8.0 2022-12-08 14:07:09,847 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135982.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:13,162 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135986.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:20,295 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135994.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:07:23,761 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1485, 2.0225, 4.4292, 4.1219, 4.0158, 4.5435, 3.9884, 4.5319], device='cuda:0'), covar=tensor([0.1387, 0.1411, 0.0103, 0.0225, 0.0242, 0.0122, 0.0182, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0155, 0.0130, 0.0168, 0.0147, 0.0142, 0.0126, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 14:07:34,172 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4835, 2.2491, 4.3994, 3.0439, 4.2574, 2.0679, 3.2459, 4.3074], device='cuda:0'), covar=tensor([0.0447, 0.3569, 0.0381, 0.4644, 0.0543, 0.3055, 0.1300, 0.0422], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0200, 0.0220, 0.0269, 0.0239, 0.0204, 0.0201, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:07:55,412 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.076e+02 2.483e+02 2.988e+02 6.746e+02, threshold=4.966e+02, percent-clipped=1.0 2022-12-08 14:08:04,513 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:08:14,774 INFO [train.py:873] (0/4) Epoch 18, batch 7500, loss[loss=0.1125, simple_loss=0.1496, pruned_loss=0.03773, over 14265.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1407, pruned_loss=0.03358, over 2043040.28 frames. ], batch size: 80, lr: 4.30e-03, grad_scale: 8.0 2022-12-08 14:08:14,910 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136055.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:08:16,689 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7378, 1.5450, 1.6977, 1.7956, 1.8216, 1.2212, 1.4959, 1.5852], device='cuda:0'), covar=tensor([0.0738, 0.0950, 0.0668, 0.0635, 0.0506, 0.0877, 0.0653, 0.0796], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0038, 0.0040], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:08:21,682 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6982, 4.9426, 5.3587, 5.6993, 5.2845, 4.6012, 5.5801, 4.5040], device='cuda:0'), covar=tensor([0.0692, 0.1832, 0.0764, 0.0969, 0.1136, 0.0598, 0.1033, 0.1114], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0278, 0.0202, 0.0199, 0.0185, 0.0159, 0.0289, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 14:09:01,854 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-18.pt 2022-12-08 14:09:45,357 INFO [train.py:873] (0/4) Epoch 19, batch 0, loss[loss=0.1384, simple_loss=0.1688, pruned_loss=0.05404, over 12035.00 frames. ], tot_loss[loss=0.1384, simple_loss=0.1688, pruned_loss=0.05404, over 12035.00 frames. ], batch size: 100, lr: 4.19e-03, grad_scale: 8.0 2022-12-08 14:09:45,357 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 14:09:53,120 INFO [train.py:905] (0/4) Epoch 19, validation: loss=0.1445, simple_loss=0.1825, pruned_loss=0.05324, over 857387.00 frames. 2022-12-08 14:09:53,122 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 14:10:08,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.063e+01 1.495e+02 2.536e+02 3.389e+02 8.096e+02, threshold=5.072e+02, percent-clipped=9.0 2022-12-08 14:10:15,602 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2022-12-08 14:10:16,254 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136142.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:10:17,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9090, 2.0300, 3.8488, 2.7068, 3.8077, 1.9564, 2.9271, 3.8443], device='cuda:0'), covar=tensor([0.0732, 0.3971, 0.0604, 0.5468, 0.0715, 0.3312, 0.1421, 0.0611], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0199, 0.0219, 0.0270, 0.0238, 0.0203, 0.0201, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:10:32,479 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5250, 1.4057, 3.5676, 1.7021, 3.4041, 3.6001, 2.6267, 3.8710], device='cuda:0'), covar=tensor([0.0259, 0.3324, 0.0436, 0.2250, 0.0806, 0.0443, 0.0930, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0156, 0.0162, 0.0169, 0.0169, 0.0179, 0.0134, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:11:10,107 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136203.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:11:22,945 INFO [train.py:873] (0/4) Epoch 19, batch 100, loss[loss=0.09408, simple_loss=0.138, pruned_loss=0.02508, over 14430.00 frames. ], tot_loss[loss=0.1057, simple_loss=0.1426, pruned_loss=0.03442, over 875976.18 frames. ], batch size: 53, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:11:37,511 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.463e+02 2.960e+02 3.618e+02 1.073e+03, threshold=5.920e+02, percent-clipped=4.0 2022-12-08 14:12:22,973 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:12:23,499 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2022-12-08 14:12:50,168 INFO [train.py:873] (0/4) Epoch 19, batch 200, loss[loss=0.1026, simple_loss=0.1449, pruned_loss=0.03017, over 14138.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.1413, pruned_loss=0.03395, over 1270817.35 frames. ], batch size: 84, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:13:05,485 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136334.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:06,200 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.129e+02 2.646e+02 3.149e+02 5.242e+02, threshold=5.292e+02, percent-clipped=0.0 2022-12-08 14:13:06,444 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:09,320 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136338.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:19,472 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:13:24,167 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 14:13:59,564 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136396.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:14:18,703 INFO [train.py:873] (0/4) Epoch 19, batch 300, loss[loss=0.09432, simple_loss=0.1383, pruned_loss=0.02516, over 14227.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1406, pruned_loss=0.03336, over 1604109.79 frames. ], batch size: 60, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:14:22,475 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1526, 1.4619, 1.7226, 1.6324, 1.5826, 1.6598, 1.3464, 1.2592], device='cuda:0'), covar=tensor([0.1357, 0.1357, 0.0395, 0.0611, 0.1477, 0.1025, 0.1914, 0.1433], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0091, 0.0070, 0.0077, 0.0101, 0.0091, 0.0102, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:14:33,863 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 1.988e+02 2.460e+02 2.943e+02 5.876e+02, threshold=4.921e+02, percent-clipped=3.0 2022-12-08 14:15:29,525 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136498.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:15:45,665 INFO [train.py:873] (0/4) Epoch 19, batch 400, loss[loss=0.1422, simple_loss=0.14, pruned_loss=0.07218, over 1159.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1398, pruned_loss=0.03273, over 1743347.06 frames. ], batch size: 100, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:15:46,672 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8063, 0.8041, 0.7155, 0.8812, 0.8871, 0.4799, 0.7810, 0.8337], device='cuda:0'), covar=tensor([0.0483, 0.0679, 0.0528, 0.0471, 0.0393, 0.0357, 0.1050, 0.0734], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0037, 0.0040], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:15:48,321 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8972, 2.3232, 2.4760, 2.7843, 2.6119, 1.7567, 2.4449, 2.5799], device='cuda:0'), covar=tensor([0.0334, 0.0740, 0.1018, 0.1012, 0.1156, 0.0740, 0.0717, 0.0770], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0037, 0.0040], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:15:50,290 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 14:16:01,299 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.033e+02 2.479e+02 3.164e+02 1.103e+03, threshold=4.959e+02, percent-clipped=2.0 2022-12-08 14:16:14,231 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8646, 3.5496, 2.7224, 4.1397, 3.9053, 3.9667, 3.4235, 2.9358], device='cuda:0'), covar=tensor([0.0874, 0.1153, 0.3008, 0.0530, 0.0783, 0.1028, 0.1212, 0.2633], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0286, 0.0256, 0.0289, 0.0321, 0.0301, 0.0253, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:16:18,856 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2022-12-08 14:16:41,691 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6177, 2.3687, 2.5753, 2.4570, 2.5700, 1.6547, 2.3708, 2.5388], device='cuda:0'), covar=tensor([0.0427, 0.0731, 0.0386, 0.1380, 0.1133, 0.0735, 0.0928, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0036, 0.0041, 0.0034, 0.0036, 0.0051, 0.0038, 0.0040], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:17:14,358 INFO [train.py:873] (0/4) Epoch 19, batch 500, loss[loss=0.1359, simple_loss=0.1602, pruned_loss=0.05585, over 9428.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1406, pruned_loss=0.03356, over 1796285.30 frames. ], batch size: 100, lr: 4.18e-03, grad_scale: 8.0 2022-12-08 14:17:30,014 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 1.937e+02 2.457e+02 2.957e+02 6.738e+02, threshold=4.914e+02, percent-clipped=4.0 2022-12-08 14:17:32,433 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136638.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:36,769 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:42,863 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136650.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:17:45,903 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7969, 1.4703, 2.0230, 1.5952, 1.8847, 1.3575, 1.6582, 1.9136], device='cuda:0'), covar=tensor([0.3208, 0.3148, 0.0581, 0.1997, 0.1637, 0.1438, 0.1437, 0.1122], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0197, 0.0217, 0.0269, 0.0237, 0.0201, 0.0199, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:18:00,137 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.48 vs. limit=5.0 2022-12-08 14:18:13,558 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:13,787 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0153, 2.3885, 3.9707, 4.1429, 3.8523, 2.3920, 4.0471, 3.0538], device='cuda:0'), covar=tensor([0.0490, 0.1385, 0.1093, 0.0500, 0.0595, 0.2134, 0.0555, 0.1125], device='cuda:0'), in_proj_covar=tensor([0.0297, 0.0263, 0.0380, 0.0335, 0.0275, 0.0310, 0.0313, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:18:17,713 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136691.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:24,490 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136698.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:29,888 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:18:40,752 INFO [train.py:873] (0/4) Epoch 19, batch 600, loss[loss=0.101, simple_loss=0.1458, pruned_loss=0.02806, over 14568.00 frames. ], tot_loss[loss=0.105, simple_loss=0.1414, pruned_loss=0.03431, over 1878519.13 frames. ], batch size: 22, lr: 4.18e-03, grad_scale: 4.0 2022-12-08 14:18:57,079 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.012e+02 2.028e+02 2.404e+02 2.918e+02 4.938e+02, threshold=4.808e+02, percent-clipped=1.0 2022-12-08 14:19:10,219 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0490, 2.1973, 2.0346, 2.2076, 1.8985, 2.0505, 2.1580, 2.0916], device='cuda:0'), covar=tensor([0.1049, 0.1247, 0.1121, 0.1017, 0.1462, 0.0982, 0.1185, 0.1105], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0150, 0.0151, 0.0166, 0.0153, 0.0128, 0.0175, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:19:40,317 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3969, 3.6984, 3.6616, 3.4168, 2.7919, 3.7783, 3.6314, 1.8951], device='cuda:0'), covar=tensor([0.1172, 0.0855, 0.0734, 0.0825, 0.0859, 0.0496, 0.0746, 0.1819], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0091, 0.0071, 0.0077, 0.0101, 0.0092, 0.0102, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:19:51,215 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136798.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:19:58,208 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:20:03,724 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0722, 2.6516, 5.0953, 3.4854, 4.7893, 2.4295, 3.7371, 4.8688], device='cuda:0'), covar=tensor([0.0403, 0.3318, 0.0266, 0.5328, 0.0644, 0.2885, 0.1205, 0.0314], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0196, 0.0217, 0.0268, 0.0237, 0.0201, 0.0199, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:20:08,084 INFO [train.py:873] (0/4) Epoch 19, batch 700, loss[loss=0.1201, simple_loss=0.1524, pruned_loss=0.04389, over 10376.00 frames. ], tot_loss[loss=0.1049, simple_loss=0.1411, pruned_loss=0.03435, over 1861393.58 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:20:24,140 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.847e+01 1.962e+02 2.462e+02 3.266e+02 7.518e+02, threshold=4.924e+02, percent-clipped=4.0 2022-12-08 14:20:32,582 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=136846.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:20:51,841 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136867.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:21:15,680 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.75 vs. limit=5.0 2022-12-08 14:21:20,581 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.11 vs. limit=5.0 2022-12-08 14:21:35,621 INFO [train.py:873] (0/4) Epoch 19, batch 800, loss[loss=0.1809, simple_loss=0.1614, pruned_loss=0.1001, over 1201.00 frames. ], tot_loss[loss=0.1051, simple_loss=0.1409, pruned_loss=0.0347, over 1833014.76 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:21:52,113 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.241e+02 2.727e+02 3.173e+02 1.328e+03, threshold=5.453e+02, percent-clipped=3.0 2022-12-08 14:22:25,625 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8748, 3.6856, 3.6042, 3.9055, 3.5524, 3.2795, 3.9549, 3.7454], device='cuda:0'), covar=tensor([0.0654, 0.1004, 0.0852, 0.0623, 0.0901, 0.0788, 0.0658, 0.0771], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0150, 0.0165, 0.0152, 0.0127, 0.0174, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:22:40,659 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136991.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:22:47,250 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136999.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:23:02,978 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2022-12-08 14:23:03,205 INFO [train.py:873] (0/4) Epoch 19, batch 900, loss[loss=0.08861, simple_loss=0.1375, pruned_loss=0.01985, over 13966.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1404, pruned_loss=0.03356, over 1916161.38 frames. ], batch size: 23, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:23:20,504 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.006e+02 2.396e+02 3.046e+02 6.535e+02, threshold=4.793e+02, percent-clipped=3.0 2022-12-08 14:23:22,039 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137039.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:23:22,152 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4960, 1.4956, 1.5309, 1.3998, 1.3119, 1.2304, 1.3063, 1.1663], device='cuda:0'), covar=tensor([0.0226, 0.0212, 0.0214, 0.0230, 0.0229, 0.0370, 0.0240, 0.0402], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0022, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:23:35,508 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5470, 4.3516, 4.1661, 4.5420, 4.1588, 3.8024, 4.5995, 4.4024], device='cuda:0'), covar=tensor([0.0585, 0.0890, 0.0784, 0.0535, 0.0705, 0.0657, 0.0589, 0.0668], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0150, 0.0165, 0.0152, 0.0127, 0.0174, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:24:07,983 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2446, 3.0079, 2.7642, 2.9460, 3.1349, 3.1631, 3.1727, 3.2127], device='cuda:0'), covar=tensor([0.0889, 0.0796, 0.2257, 0.2292, 0.0960, 0.1056, 0.1213, 0.0955], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0282, 0.0460, 0.0579, 0.0357, 0.0463, 0.0394, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:24:08,594 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.18 vs. limit=5.0 2022-12-08 14:24:29,763 INFO [train.py:873] (0/4) Epoch 19, batch 1000, loss[loss=0.1235, simple_loss=0.1469, pruned_loss=0.04999, over 4955.00 frames. ], tot_loss[loss=0.1047, simple_loss=0.141, pruned_loss=0.03418, over 1915008.96 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:24:47,893 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.038e+02 2.060e+02 2.475e+02 3.449e+02 7.129e+02, threshold=4.949e+02, percent-clipped=6.0 2022-12-08 14:24:54,866 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2081, 4.7163, 3.8350, 5.4386, 4.7682, 5.1486, 4.8602, 4.3307], device='cuda:0'), covar=tensor([0.0555, 0.0851, 0.2110, 0.0583, 0.0870, 0.1307, 0.0729, 0.1622], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0287, 0.0260, 0.0291, 0.0321, 0.0303, 0.0255, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:25:04,507 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2606, 2.2955, 2.2602, 2.0119, 2.1576, 1.8663, 1.8201, 1.7064], device='cuda:0'), covar=tensor([0.0239, 0.0315, 0.0293, 0.0381, 0.0321, 0.0397, 0.0329, 0.0375], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0022, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:25:09,580 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=137162.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:25:24,779 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0225, 1.5375, 3.9468, 1.8467, 3.9813, 4.1647, 3.0324, 4.4127], device='cuda:0'), covar=tensor([0.0228, 0.2962, 0.0389, 0.2049, 0.0449, 0.0388, 0.0737, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0159, 0.0167, 0.0167, 0.0177, 0.0131, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:25:57,946 INFO [train.py:873] (0/4) Epoch 19, batch 1100, loss[loss=0.09042, simple_loss=0.136, pruned_loss=0.02241, over 14331.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.1407, pruned_loss=0.03407, over 1898930.28 frames. ], batch size: 31, lr: 4.17e-03, grad_scale: 4.0 2022-12-08 14:26:15,916 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.041e+02 1.998e+02 2.599e+02 3.181e+02 8.198e+02, threshold=5.197e+02, percent-clipped=8.0 2022-12-08 14:26:43,218 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5064, 3.2946, 3.2267, 3.5198, 3.3608, 3.4887, 3.5662, 3.0119], device='cuda:0'), covar=tensor([0.0556, 0.1096, 0.0545, 0.0561, 0.0792, 0.0428, 0.0586, 0.0602], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0278, 0.0202, 0.0200, 0.0187, 0.0160, 0.0292, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 14:27:09,980 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:27:25,930 INFO [train.py:873] (0/4) Epoch 19, batch 1200, loss[loss=0.1411, simple_loss=0.161, pruned_loss=0.06058, over 7794.00 frames. ], tot_loss[loss=0.1056, simple_loss=0.1413, pruned_loss=0.03498, over 1857741.58 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:27:43,344 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.257e+02 2.745e+02 3.542e+02 9.944e+02, threshold=5.490e+02, percent-clipped=8.0 2022-12-08 14:27:51,879 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:27:52,204 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.70 vs. limit=5.0 2022-12-08 14:28:54,280 INFO [train.py:873] (0/4) Epoch 19, batch 1300, loss[loss=0.1204, simple_loss=0.1303, pruned_loss=0.05531, over 2659.00 frames. ], tot_loss[loss=0.1053, simple_loss=0.1411, pruned_loss=0.03475, over 1867578.76 frames. ], batch size: 100, lr: 4.17e-03, grad_scale: 8.0 2022-12-08 14:29:08,748 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1398, 2.5323, 2.4934, 2.5484, 2.1083, 2.5379, 2.5005, 1.4242], device='cuda:0'), covar=tensor([0.1047, 0.0843, 0.0700, 0.0596, 0.1022, 0.0662, 0.0900, 0.1966], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0091, 0.0070, 0.0077, 0.0100, 0.0092, 0.0102, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:29:12,653 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.901e+02 2.314e+02 2.897e+02 7.672e+02, threshold=4.628e+02, percent-clipped=3.0 2022-12-08 14:29:33,437 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2022-12-08 14:29:34,020 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:29:38,413 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8402, 1.4772, 2.0419, 1.6316, 1.9297, 1.4447, 1.7121, 1.9418], device='cuda:0'), covar=tensor([0.2560, 0.3087, 0.0549, 0.1569, 0.1231, 0.1230, 0.1150, 0.0985], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0197, 0.0217, 0.0269, 0.0238, 0.0201, 0.0200, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:29:47,508 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.93 vs. limit=5.0 2022-12-08 14:30:15,556 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=137510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:30:21,587 INFO [train.py:873] (0/4) Epoch 19, batch 1400, loss[loss=0.1354, simple_loss=0.1575, pruned_loss=0.05671, over 7746.00 frames. ], tot_loss[loss=0.1054, simple_loss=0.1417, pruned_loss=0.03455, over 1907267.66 frames. ], batch size: 100, lr: 4.16e-03, grad_scale: 4.0 2022-12-08 14:30:40,257 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.087e+02 2.410e+02 2.920e+02 5.806e+02, threshold=4.820e+02, percent-clipped=3.0 2022-12-08 14:31:27,305 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2822, 4.8482, 4.7690, 5.2839, 4.9796, 4.5112, 5.2576, 4.4129], device='cuda:0'), covar=tensor([0.0288, 0.0877, 0.0359, 0.0360, 0.0679, 0.0520, 0.0454, 0.0435], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0277, 0.0202, 0.0200, 0.0186, 0.0160, 0.0292, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 14:31:37,981 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4208, 2.7597, 2.5855, 2.7452, 2.3373, 2.8307, 2.6398, 1.6159], device='cuda:0'), covar=tensor([0.1137, 0.0762, 0.1196, 0.0629, 0.0925, 0.0599, 0.0917, 0.1911], device='cuda:0'), in_proj_covar=tensor([0.0138, 0.0092, 0.0071, 0.0078, 0.0102, 0.0093, 0.0103, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:31:40,617 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1441, 1.9939, 2.0149, 1.9806, 2.0672, 1.7053, 1.7150, 1.3938], device='cuda:0'), covar=tensor([0.0184, 0.0311, 0.0299, 0.0333, 0.0231, 0.0339, 0.0238, 0.0349], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 14:31:49,006 INFO [train.py:873] (0/4) Epoch 19, batch 1500, loss[loss=0.09809, simple_loss=0.1403, pruned_loss=0.02793, over 14219.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.1407, pruned_loss=0.0339, over 1897557.56 frames. ], batch size: 80, lr: 4.16e-03, grad_scale: 4.0 2022-12-08 14:32:07,872 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 1.988e+02 2.718e+02 3.293e+02 9.468e+02, threshold=5.435e+02, percent-clipped=6.0 2022-12-08 14:32:34,704 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5974, 1.5641, 1.5420, 1.6660, 1.7299, 1.1604, 1.4637, 1.4765], device='cuda:0'), covar=tensor([0.0611, 0.0643, 0.0745, 0.0676, 0.0400, 0.0855, 0.0704, 0.0662], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0050, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:32:44,406 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1942, 2.0533, 2.1089, 2.2817, 2.1912, 1.3929, 1.8830, 2.1855], device='cuda:0'), covar=tensor([0.1107, 0.0976, 0.0860, 0.0657, 0.0564, 0.0835, 0.0807, 0.0760], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0034, 0.0036, 0.0050, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:33:18,036 INFO [train.py:873] (0/4) Epoch 19, batch 1600, loss[loss=0.1014, simple_loss=0.1391, pruned_loss=0.03184, over 14150.00 frames. ], tot_loss[loss=0.1045, simple_loss=0.141, pruned_loss=0.03403, over 1985694.70 frames. ], batch size: 35, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:33:36,639 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.099e+02 2.503e+02 3.045e+02 5.451e+02, threshold=5.005e+02, percent-clipped=1.0 2022-12-08 14:33:44,564 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6638, 1.4083, 3.7185, 1.8585, 3.6429, 3.8455, 2.7537, 4.1010], device='cuda:0'), covar=tensor([0.0281, 0.3239, 0.0430, 0.2069, 0.0596, 0.0403, 0.0818, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0154, 0.0162, 0.0168, 0.0168, 0.0179, 0.0132, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:34:41,240 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8915, 2.4903, 3.6458, 2.8208, 3.7557, 3.6019, 3.6226, 3.0832], device='cuda:0'), covar=tensor([0.0997, 0.2978, 0.1325, 0.1834, 0.0965, 0.1069, 0.1325, 0.1856], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0311, 0.0391, 0.0300, 0.0366, 0.0323, 0.0362, 0.0297], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:34:44,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1498, 2.0237, 1.8203, 1.9173, 2.0826, 2.1454, 2.0809, 2.0896], device='cuda:0'), covar=tensor([0.1209, 0.1050, 0.3060, 0.3046, 0.1462, 0.1382, 0.1747, 0.1341], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0281, 0.0460, 0.0580, 0.0361, 0.0464, 0.0396, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:34:46,399 INFO [train.py:873] (0/4) Epoch 19, batch 1700, loss[loss=0.1205, simple_loss=0.1504, pruned_loss=0.04527, over 14257.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.1406, pruned_loss=0.03321, over 2025635.65 frames. ], batch size: 80, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:35:05,299 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.077e+02 2.557e+02 3.257e+02 5.458e+02, threshold=5.115e+02, percent-clipped=1.0 2022-12-08 14:35:07,195 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7402, 1.0292, 1.2361, 1.2091, 0.9596, 1.2878, 1.0171, 0.7857], device='cuda:0'), covar=tensor([0.1846, 0.0998, 0.0394, 0.0543, 0.1929, 0.1089, 0.1410, 0.1389], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0092, 0.0071, 0.0078, 0.0101, 0.0093, 0.0102, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:35:15,797 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9537, 1.9585, 2.1096, 1.8848, 1.7325, 1.7451, 1.7594, 1.3609], device='cuda:0'), covar=tensor([0.0204, 0.0457, 0.0250, 0.0299, 0.0316, 0.0288, 0.0256, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 14:35:53,031 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137892.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:35:55,474 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137895.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:36:14,637 INFO [train.py:873] (0/4) Epoch 19, batch 1800, loss[loss=0.1076, simple_loss=0.1498, pruned_loss=0.03264, over 14267.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1404, pruned_loss=0.03345, over 1963762.96 frames. ], batch size: 76, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:36:33,146 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.219e+02 2.138e+02 2.501e+02 2.995e+02 6.998e+02, threshold=5.002e+02, percent-clipped=2.0 2022-12-08 14:36:46,835 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2524, 3.2299, 2.7214, 2.8735, 3.2248, 3.2715, 3.3776, 3.3508], device='cuda:0'), covar=tensor([0.1526, 0.0922, 0.2936, 0.3449, 0.1449, 0.1474, 0.1386, 0.1267], device='cuda:0'), in_proj_covar=tensor([0.0404, 0.0282, 0.0461, 0.0582, 0.0362, 0.0468, 0.0398, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:36:46,915 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137953.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:36:49,256 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137956.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:37:04,563 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5331, 1.9495, 2.0761, 2.0945, 1.8543, 2.1715, 1.7540, 1.3150], device='cuda:0'), covar=tensor([0.0734, 0.0822, 0.0531, 0.0632, 0.1200, 0.0775, 0.1533, 0.1902], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0091, 0.0070, 0.0078, 0.0101, 0.0093, 0.0102, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:37:42,664 INFO [train.py:873] (0/4) Epoch 19, batch 1900, loss[loss=0.1464, simple_loss=0.1387, pruned_loss=0.07702, over 1276.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.1403, pruned_loss=0.03321, over 1951947.80 frames. ], batch size: 100, lr: 4.16e-03, grad_scale: 8.0 2022-12-08 14:38:01,168 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.140e+02 2.687e+02 3.163e+02 5.638e+02, threshold=5.373e+02, percent-clipped=1.0 2022-12-08 14:38:03,083 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0675, 1.8845, 4.6179, 4.1704, 4.1428, 4.7178, 4.1882, 4.6683], device='cuda:0'), covar=tensor([0.1610, 0.1626, 0.0106, 0.0266, 0.0274, 0.0131, 0.0194, 0.0117], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0156, 0.0131, 0.0170, 0.0148, 0.0143, 0.0127, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 14:38:45,850 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4360, 2.2557, 1.8964, 2.0242, 2.3788, 2.4153, 2.4486, 2.3961], device='cuda:0'), covar=tensor([0.1610, 0.1302, 0.3847, 0.3942, 0.1764, 0.1737, 0.1809, 0.1599], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0278, 0.0456, 0.0579, 0.0360, 0.0465, 0.0394, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:39:09,900 INFO [train.py:873] (0/4) Epoch 19, batch 2000, loss[loss=0.1253, simple_loss=0.1527, pruned_loss=0.04893, over 14238.00 frames. ], tot_loss[loss=0.1046, simple_loss=0.1408, pruned_loss=0.0342, over 1913214.12 frames. ], batch size: 89, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:39:28,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 1.998e+02 2.546e+02 3.326e+02 7.451e+02, threshold=5.092e+02, percent-clipped=3.0 2022-12-08 14:39:34,873 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2022-12-08 14:40:20,675 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2022-12-08 14:40:22,003 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2572, 4.0569, 3.7703, 3.9262, 4.1622, 4.1941, 4.2574, 4.2567], device='cuda:0'), covar=tensor([0.0831, 0.0503, 0.1953, 0.2413, 0.0709, 0.0912, 0.0898, 0.0788], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0278, 0.0456, 0.0577, 0.0361, 0.0466, 0.0394, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:40:37,634 INFO [train.py:873] (0/4) Epoch 19, batch 2100, loss[loss=0.09493, simple_loss=0.1379, pruned_loss=0.02596, over 14476.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1402, pruned_loss=0.03349, over 1947345.64 frames. ], batch size: 51, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:40:55,804 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4557, 2.4724, 4.4210, 4.6555, 4.2334, 2.6666, 4.6066, 3.3284], device='cuda:0'), covar=tensor([0.0404, 0.1384, 0.0731, 0.0386, 0.0524, 0.2019, 0.0412, 0.1007], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0261, 0.0376, 0.0332, 0.0272, 0.0306, 0.0312, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:40:56,378 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.080e+02 1.995e+02 2.366e+02 2.965e+02 6.942e+02, threshold=4.733e+02, percent-clipped=2.0 2022-12-08 14:41:04,829 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138248.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:41:05,981 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2022-12-08 14:41:07,331 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138251.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:41:11,702 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5283, 3.2020, 5.5573, 4.2635, 5.1997, 3.0007, 4.2755, 5.3132], device='cuda:0'), covar=tensor([0.0338, 0.2620, 0.0258, 0.4002, 0.0327, 0.2228, 0.0970, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0195, 0.0217, 0.0264, 0.0236, 0.0199, 0.0199, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:41:24,477 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138270.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:41:59,925 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1421, 2.0188, 4.7472, 4.3053, 4.1910, 4.8744, 4.5526, 4.8512], device='cuda:0'), covar=tensor([0.1481, 0.1455, 0.0100, 0.0235, 0.0231, 0.0118, 0.0138, 0.0107], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0155, 0.0130, 0.0169, 0.0147, 0.0142, 0.0126, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 14:42:05,013 INFO [train.py:873] (0/4) Epoch 19, batch 2200, loss[loss=0.1117, simple_loss=0.1465, pruned_loss=0.03843, over 14218.00 frames. ], tot_loss[loss=0.1035, simple_loss=0.1401, pruned_loss=0.03346, over 1953403.30 frames. ], batch size: 94, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:42:09,665 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138322.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:42:17,245 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138331.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:42:22,959 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.147e+02 2.085e+02 2.687e+02 3.331e+02 7.282e+02, threshold=5.373e+02, percent-clipped=5.0 2022-12-08 14:43:02,755 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:43:07,517 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2022-12-08 14:43:24,799 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4061, 2.6752, 4.2325, 4.5315, 4.2250, 2.5600, 4.4818, 3.3878], device='cuda:0'), covar=tensor([0.0429, 0.1296, 0.0972, 0.0457, 0.0555, 0.2199, 0.0442, 0.1018], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0264, 0.0379, 0.0335, 0.0275, 0.0310, 0.0315, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:43:28,945 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3916, 3.0461, 3.0229, 2.2357, 2.8078, 3.0298, 3.4211, 2.7711], device='cuda:0'), covar=tensor([0.0656, 0.0999, 0.0902, 0.1239, 0.1058, 0.0833, 0.0718, 0.1174], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0167, 0.0139, 0.0124, 0.0143, 0.0154, 0.0137, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:43:32,132 INFO [train.py:873] (0/4) Epoch 19, batch 2300, loss[loss=0.1367, simple_loss=0.149, pruned_loss=0.06223, over 5011.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.14, pruned_loss=0.03333, over 1959924.52 frames. ], batch size: 100, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:43:50,854 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.988e+01 2.098e+02 2.453e+02 3.240e+02 6.762e+02, threshold=4.906e+02, percent-clipped=1.0 2022-12-08 14:44:28,915 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2022-12-08 14:44:32,934 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9258, 2.6347, 3.4649, 2.3269, 2.2781, 2.9041, 1.8442, 2.8264], device='cuda:0'), covar=tensor([0.0940, 0.1314, 0.0607, 0.2065, 0.1957, 0.0845, 0.2808, 0.1037], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0103, 0.0097, 0.0101, 0.0116, 0.0093, 0.0117, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 14:45:01,195 INFO [train.py:873] (0/4) Epoch 19, batch 2400, loss[loss=0.1269, simple_loss=0.1227, pruned_loss=0.06558, over 1297.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1405, pruned_loss=0.0339, over 1915496.24 frames. ], batch size: 100, lr: 4.15e-03, grad_scale: 8.0 2022-12-08 14:45:18,700 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.048e+02 2.033e+02 2.636e+02 3.505e+02 6.447e+02, threshold=5.271e+02, percent-clipped=6.0 2022-12-08 14:45:28,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:45:30,867 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138551.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:04,147 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2449, 2.3784, 2.2698, 2.3393, 2.2111, 1.9902, 1.9372, 1.9181], device='cuda:0'), covar=tensor([0.0231, 0.0321, 0.0231, 0.0211, 0.0295, 0.0473, 0.0442, 0.0402], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:46:09,975 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138596.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:12,442 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138599.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:17,074 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2022-12-08 14:46:28,221 INFO [train.py:873] (0/4) Epoch 19, batch 2500, loss[loss=0.08414, simple_loss=0.1204, pruned_loss=0.02393, over 14422.00 frames. ], tot_loss[loss=0.1043, simple_loss=0.1405, pruned_loss=0.03403, over 1901622.86 frames. ], batch size: 18, lr: 4.15e-03, grad_scale: 4.0 2022-12-08 14:46:36,688 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:46:47,812 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.139e+02 2.554e+02 3.094e+02 5.985e+02, threshold=5.107e+02, percent-clipped=2.0 2022-12-08 14:47:04,115 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0848, 1.9653, 2.2442, 2.3712, 1.9970, 1.9481, 2.3108, 1.9941], device='cuda:0'), covar=tensor([0.0416, 0.0914, 0.0415, 0.0391, 0.0726, 0.1220, 0.0460, 0.0558], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0266, 0.0380, 0.0337, 0.0277, 0.0312, 0.0316, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:47:21,853 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138678.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:24,870 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138681.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:26,438 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2063, 3.1036, 2.9739, 3.2369, 2.8834, 2.8956, 3.2537, 3.1505], device='cuda:0'), covar=tensor([0.0679, 0.1025, 0.0885, 0.0652, 0.1072, 0.0819, 0.0797, 0.0811], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0148, 0.0150, 0.0166, 0.0152, 0.0127, 0.0175, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:47:29,047 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138686.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:29,068 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3121, 2.3675, 2.6089, 2.4248, 1.8164, 2.1189, 2.1658, 2.1165], device='cuda:0'), covar=tensor([0.0387, 0.0663, 0.0303, 0.0328, 0.0561, 0.0520, 0.0488, 0.0571], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 14:47:38,296 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138697.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:47:45,104 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8054, 2.7942, 2.6321, 2.8854, 2.5174, 2.6333, 2.8692, 2.7774], device='cuda:0'), covar=tensor([0.0817, 0.1133, 0.1086, 0.0731, 0.1333, 0.0859, 0.0914, 0.0907], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0149, 0.0151, 0.0167, 0.0153, 0.0127, 0.0175, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:47:55,442 INFO [train.py:873] (0/4) Epoch 19, batch 2600, loss[loss=0.1036, simple_loss=0.1425, pruned_loss=0.03231, over 14239.00 frames. ], tot_loss[loss=0.104, simple_loss=0.1404, pruned_loss=0.03377, over 1922580.44 frames. ], batch size: 94, lr: 4.15e-03, grad_scale: 2.0 2022-12-08 14:48:15,369 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.052e+02 2.043e+02 2.512e+02 3.155e+02 8.164e+02, threshold=5.025e+02, percent-clipped=4.0 2022-12-08 14:48:17,481 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138742.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:21,814 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9187, 1.8269, 3.7343, 3.4543, 3.5067, 3.7788, 3.1696, 3.7817], device='cuda:0'), covar=tensor([0.1608, 0.1500, 0.0134, 0.0294, 0.0305, 0.0152, 0.0319, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0155, 0.0130, 0.0169, 0.0147, 0.0142, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 14:48:21,877 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138747.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:31,387 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138758.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:48:50,519 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8186, 0.8426, 0.7215, 0.8181, 0.8606, 0.4282, 0.7933, 0.8297], device='cuda:0'), covar=tensor([0.0395, 0.0518, 0.0622, 0.0517, 0.0339, 0.0432, 0.0956, 0.0798], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0051, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:48:51,246 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2773, 4.9598, 4.6343, 4.7925, 4.8810, 5.1288, 5.1610, 5.1928], device='cuda:0'), covar=tensor([0.0571, 0.0408, 0.2036, 0.2532, 0.0671, 0.0692, 0.0837, 0.0698], device='cuda:0'), in_proj_covar=tensor([0.0400, 0.0276, 0.0451, 0.0571, 0.0359, 0.0462, 0.0391, 0.0404], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:49:08,958 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8715, 0.7496, 0.8335, 0.7963, 0.7679, 0.5072, 0.6311, 0.6809], device='cuda:0'), covar=tensor([0.0133, 0.0154, 0.0150, 0.0145, 0.0132, 0.0252, 0.0152, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0023, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0033], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:49:11,763 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0946, 2.9568, 2.3491, 3.2651, 3.0997, 3.1272, 2.8589, 2.3117], device='cuda:0'), covar=tensor([0.1221, 0.1181, 0.2840, 0.0668, 0.1025, 0.0930, 0.1282, 0.2628], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0288, 0.0260, 0.0293, 0.0324, 0.0303, 0.0256, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:49:20,172 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5478, 1.8073, 4.4569, 2.4224, 4.2550, 4.7140, 4.1889, 4.9124], device='cuda:0'), covar=tensor([0.0262, 0.3112, 0.0457, 0.2074, 0.0363, 0.0381, 0.0377, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0153, 0.0157, 0.0165, 0.0165, 0.0176, 0.0131, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:49:23,531 INFO [train.py:873] (0/4) Epoch 19, batch 2700, loss[loss=0.1007, simple_loss=0.139, pruned_loss=0.03116, over 10318.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1402, pruned_loss=0.0335, over 1943398.47 frames. ], batch size: 100, lr: 4.14e-03, grad_scale: 2.0 2022-12-08 14:49:43,635 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 2.157e+02 2.518e+02 3.000e+02 5.718e+02, threshold=5.036e+02, percent-clipped=2.0 2022-12-08 14:49:44,061 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2022-12-08 14:50:08,351 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0033, 3.8187, 3.6636, 4.0470, 3.6107, 3.3929, 4.0445, 3.8904], device='cuda:0'), covar=tensor([0.0616, 0.0727, 0.0911, 0.0596, 0.0901, 0.0676, 0.0609, 0.0785], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0148, 0.0150, 0.0165, 0.0151, 0.0126, 0.0173, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:50:23,684 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7619, 3.2483, 3.1673, 3.1727, 2.3453, 3.2578, 3.0722, 1.6215], device='cuda:0'), covar=tensor([0.1307, 0.1478, 0.1221, 0.0944, 0.1064, 0.0627, 0.1120, 0.2182], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0090, 0.0070, 0.0077, 0.0100, 0.0092, 0.0101, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:50:44,107 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138909.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:50:51,208 INFO [train.py:873] (0/4) Epoch 19, batch 2800, loss[loss=0.1018, simple_loss=0.1399, pruned_loss=0.03185, over 14148.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1399, pruned_loss=0.03273, over 1996311.53 frames. ], batch size: 99, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:50:59,207 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138926.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:51:10,963 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 2.185e+02 2.764e+02 3.570e+02 5.389e+02, threshold=5.528e+02, percent-clipped=1.0 2022-12-08 14:51:34,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4075, 2.6901, 4.1963, 3.0757, 4.1600, 4.1497, 3.9984, 3.5869], device='cuda:0'), covar=tensor([0.0698, 0.3287, 0.1234, 0.1899, 0.0924, 0.1000, 0.1830, 0.1650], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0310, 0.0389, 0.0297, 0.0364, 0.0323, 0.0361, 0.0297], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:51:37,425 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138970.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 14:51:40,706 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=138974.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:51:44,222 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:18,361 INFO [train.py:873] (0/4) Epoch 19, batch 2900, loss[loss=0.08761, simple_loss=0.1273, pruned_loss=0.02398, over 6947.00 frames. ], tot_loss[loss=0.1032, simple_loss=0.1401, pruned_loss=0.03312, over 1998468.75 frames. ], batch size: 100, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:52:26,197 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139026.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:30,376 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8103, 1.5188, 2.0617, 1.5829, 1.9083, 1.4144, 1.6654, 1.9086], device='cuda:0'), covar=tensor([0.2991, 0.2965, 0.0623, 0.1664, 0.1431, 0.1224, 0.1271, 0.1191], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0196, 0.0217, 0.0266, 0.0239, 0.0200, 0.0200, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 14:52:35,639 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139037.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:38,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 1.948e+02 2.422e+02 2.806e+02 5.439e+02, threshold=4.845e+02, percent-clipped=0.0 2022-12-08 14:52:39,982 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139042.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:52:49,572 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139053.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:53:22,080 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.8577, 2.8681, 4.6027, 3.4821, 4.6428, 4.4440, 4.4267, 4.0870], device='cuda:0'), covar=tensor([0.0807, 0.3317, 0.1041, 0.1708, 0.0690, 0.1028, 0.1462, 0.1354], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0308, 0.0388, 0.0296, 0.0362, 0.0322, 0.0359, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:53:45,725 INFO [train.py:873] (0/4) Epoch 19, batch 3000, loss[loss=0.1011, simple_loss=0.1389, pruned_loss=0.03169, over 13957.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1398, pruned_loss=0.03279, over 1993255.64 frames. ], batch size: 26, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:53:45,726 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 14:53:50,600 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6350, 2.4635, 2.6540, 1.8412, 1.9637, 2.6531, 2.5224, 2.2657], device='cuda:0'), covar=tensor([0.0869, 0.0691, 0.0809, 0.1126, 0.1835, 0.0669, 0.0858, 0.1311], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0168, 0.0140, 0.0125, 0.0144, 0.0155, 0.0138, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:53:54,216 INFO [train.py:905] (0/4) Epoch 19, validation: loss=0.142, simple_loss=0.1782, pruned_loss=0.05288, over 857387.00 frames. 2022-12-08 14:53:54,216 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 14:54:03,737 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.3027, 4.1419, 3.9444, 4.3952, 4.0625, 3.8777, 4.4431, 3.6476], device='cuda:0'), covar=tensor([0.0487, 0.0881, 0.0443, 0.0425, 0.0786, 0.1094, 0.0511, 0.0566], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0279, 0.0204, 0.0202, 0.0187, 0.0162, 0.0294, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 14:54:05,914 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9011, 2.1793, 2.8777, 2.9533, 2.8714, 2.1727, 2.8474, 2.3694], device='cuda:0'), covar=tensor([0.0539, 0.1200, 0.0887, 0.0665, 0.0668, 0.1681, 0.0590, 0.1074], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0260, 0.0374, 0.0331, 0.0271, 0.0306, 0.0310, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:54:09,948 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 14:54:14,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.026e+02 2.406e+02 3.345e+02 1.533e+03, threshold=4.812e+02, percent-clipped=5.0 2022-12-08 14:54:47,438 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9337, 1.3453, 2.0023, 1.3047, 1.9494, 2.0759, 1.6451, 2.1472], device='cuda:0'), covar=tensor([0.0326, 0.1926, 0.0533, 0.1876, 0.0675, 0.0639, 0.1179, 0.0441], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0161, 0.0168, 0.0168, 0.0178, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:55:01,979 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2197, 1.8530, 2.2121, 1.5493, 1.9074, 2.2673, 2.1588, 1.9855], device='cuda:0'), covar=tensor([0.1007, 0.0825, 0.0976, 0.1399, 0.1613, 0.0931, 0.0822, 0.1473], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0168, 0.0140, 0.0125, 0.0144, 0.0155, 0.0138, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 14:55:03,874 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2022-12-08 14:55:04,877 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 14:55:17,110 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2022-12-08 14:55:22,155 INFO [train.py:873] (0/4) Epoch 19, batch 3100, loss[loss=0.08877, simple_loss=0.1321, pruned_loss=0.02274, over 14240.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.1402, pruned_loss=0.03326, over 1966944.70 frames. ], batch size: 57, lr: 4.14e-03, grad_scale: 4.0 2022-12-08 14:55:34,508 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3109, 2.3143, 2.3550, 2.6375, 2.4669, 1.5562, 2.4174, 2.4399], device='cuda:0'), covar=tensor([0.0777, 0.0864, 0.0493, 0.0624, 0.0780, 0.0788, 0.0680, 0.0932], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0050, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 14:55:40,042 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:55:41,531 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.036e+02 2.564e+02 2.982e+02 6.918e+02, threshold=5.128e+02, percent-clipped=2.0 2022-12-08 14:56:03,653 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139265.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 14:56:07,075 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2022-12-08 14:56:12,974 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139276.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:24,478 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139289.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:33,153 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:56:48,454 INFO [train.py:873] (0/4) Epoch 19, batch 3200, loss[loss=0.102, simple_loss=0.136, pruned_loss=0.03401, over 5956.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1403, pruned_loss=0.03361, over 1962168.88 frames. ], batch size: 100, lr: 4.14e-03, grad_scale: 8.0 2022-12-08 14:57:06,563 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139337.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:06,603 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139337.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 14:57:08,976 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.774e+01 1.930e+02 2.439e+02 3.004e+02 8.294e+02, threshold=4.879e+02, percent-clipped=2.0 2022-12-08 14:57:10,871 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139342.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:17,635 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139350.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:19,977 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139353.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:47,840 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139385.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:57:52,088 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:58:01,921 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139401.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 14:58:16,250 INFO [train.py:873] (0/4) Epoch 19, batch 3300, loss[loss=0.1123, simple_loss=0.1465, pruned_loss=0.03904, over 14588.00 frames. ], tot_loss[loss=0.1042, simple_loss=0.1404, pruned_loss=0.03396, over 1941451.53 frames. ], batch size: 21, lr: 4.14e-03, grad_scale: 8.0 2022-12-08 14:58:33,043 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2022-12-08 14:58:35,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.023e+02 2.414e+02 3.086e+02 5.603e+02, threshold=4.828e+02, percent-clipped=3.0 2022-12-08 14:58:36,789 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7253, 3.6375, 3.4408, 3.7740, 3.3334, 3.2598, 3.7800, 3.6112], device='cuda:0'), covar=tensor([0.0639, 0.0926, 0.0884, 0.0650, 0.0966, 0.0671, 0.0652, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0148, 0.0149, 0.0166, 0.0152, 0.0125, 0.0173, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 14:59:12,397 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9269, 1.5288, 3.5338, 3.2380, 3.3029, 3.5487, 2.9296, 3.5536], device='cuda:0'), covar=tensor([0.1662, 0.1843, 0.0146, 0.0332, 0.0346, 0.0176, 0.0316, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0157, 0.0132, 0.0170, 0.0149, 0.0143, 0.0126, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 14:59:23,534 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2022-12-08 14:59:36,109 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2022-12-08 14:59:39,961 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6593, 1.9492, 2.5802, 2.1720, 2.6084, 2.5376, 2.4683, 2.3738], device='cuda:0'), covar=tensor([0.0865, 0.2478, 0.0977, 0.1519, 0.0541, 0.1038, 0.0883, 0.1304], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0305, 0.0384, 0.0293, 0.0357, 0.0317, 0.0357, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 14:59:41,883 INFO [train.py:873] (0/4) Epoch 19, batch 3400, loss[loss=0.09827, simple_loss=0.1346, pruned_loss=0.03099, over 14557.00 frames. ], tot_loss[loss=0.1041, simple_loss=0.1404, pruned_loss=0.03387, over 1948841.37 frames. ], batch size: 34, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:00:02,110 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 2.047e+02 2.601e+02 3.313e+02 6.688e+02, threshold=5.202e+02, percent-clipped=6.0 2022-12-08 15:00:08,613 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2022-12-08 15:00:24,224 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139565.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:00:24,539 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.01 vs. limit=5.0 2022-12-08 15:00:29,770 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139571.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:00:39,036 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139582.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:00:49,352 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:06,700 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139613.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:10,079 INFO [train.py:873] (0/4) Epoch 19, batch 3500, loss[loss=0.1013, simple_loss=0.1143, pruned_loss=0.04419, over 2674.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.14, pruned_loss=0.03367, over 1902871.71 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:01:14,701 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2351, 2.4807, 2.5031, 2.6195, 2.1627, 2.5414, 2.3764, 1.5459], device='cuda:0'), covar=tensor([0.0871, 0.0832, 0.0911, 0.0488, 0.1023, 0.0801, 0.1009, 0.1839], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0092, 0.0072, 0.0077, 0.0101, 0.0093, 0.0103, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 15:01:21,847 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2022-12-08 15:01:22,758 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2022-12-08 15:01:23,109 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139632.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 15:01:23,197 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139632.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:30,130 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.003e+02 2.440e+02 2.964e+02 5.520e+02, threshold=4.880e+02, percent-clipped=2.0 2022-12-08 15:01:32,832 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:34,852 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139645.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:01:57,520 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1175, 1.1823, 1.0272, 1.1756, 1.2241, 0.7870, 1.0405, 1.1377], device='cuda:0'), covar=tensor([0.0766, 0.0636, 0.0726, 0.0560, 0.0450, 0.0656, 0.1190, 0.0869], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0036, 0.0041, 0.0035, 0.0036, 0.0051, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:02:37,619 INFO [train.py:873] (0/4) Epoch 19, batch 3600, loss[loss=0.09524, simple_loss=0.1383, pruned_loss=0.02611, over 11977.00 frames. ], tot_loss[loss=0.1033, simple_loss=0.1402, pruned_loss=0.03318, over 1931009.59 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:02:57,975 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.754e+01 2.074e+02 2.504e+02 3.273e+02 8.629e+02, threshold=5.008e+02, percent-clipped=8.0 2022-12-08 15:03:07,398 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-08 15:03:22,612 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 15:03:24,041 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139769.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:36,002 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2022-12-08 15:03:41,867 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139789.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:55,323 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139804.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:03:57,305 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139806.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:07,189 INFO [train.py:873] (0/4) Epoch 19, batch 3700, loss[loss=0.1041, simple_loss=0.1427, pruned_loss=0.03277, over 14145.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.14, pruned_loss=0.03257, over 1976905.73 frames. ], batch size: 84, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:04:18,396 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139830.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:26,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.254e+02 1.959e+02 2.384e+02 2.847e+02 4.025e+02, threshold=4.768e+02, percent-clipped=0.0 2022-12-08 15:04:36,819 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:36,841 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:49,237 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:04:50,998 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139867.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:14,772 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139894.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:16,531 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4263, 1.8381, 2.3708, 1.9901, 2.4084, 2.3072, 2.1907, 2.2130], device='cuda:0'), covar=tensor([0.0628, 0.2322, 0.0846, 0.1350, 0.0533, 0.1115, 0.0768, 0.0980], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0306, 0.0384, 0.0294, 0.0359, 0.0318, 0.0358, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:05:29,244 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:31,727 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8234, 0.8203, 0.6675, 0.8196, 0.8043, 0.4589, 0.7836, 0.8211], device='cuda:0'), covar=tensor([0.0559, 0.0593, 0.0543, 0.0565, 0.0446, 0.0437, 0.1056, 0.0930], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0036, 0.0041, 0.0035, 0.0036, 0.0051, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:05:34,120 INFO [train.py:873] (0/4) Epoch 19, batch 3800, loss[loss=0.1412, simple_loss=0.1423, pruned_loss=0.07002, over 1230.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1402, pruned_loss=0.03267, over 1983567.15 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:05:42,643 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0905, 2.1154, 4.9835, 4.4489, 4.3118, 5.1111, 4.8861, 5.1357], device='cuda:0'), covar=tensor([0.1580, 0.1485, 0.0088, 0.0188, 0.0219, 0.0115, 0.0084, 0.0090], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0156, 0.0131, 0.0169, 0.0148, 0.0142, 0.0126, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:05:43,467 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139927.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:48,200 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139932.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 15:05:53,128 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139938.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:54,736 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.144e+02 2.015e+02 2.567e+02 3.063e+02 4.403e+02, threshold=5.135e+02, percent-clipped=0.0 2022-12-08 15:05:56,571 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139942.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:05:59,166 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:06:29,963 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139980.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:06:41,144 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=139993.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:06:47,628 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-140000.pt 2022-12-08 15:07:06,100 INFO [train.py:873] (0/4) Epoch 19, batch 3900, loss[loss=0.11, simple_loss=0.1524, pruned_loss=0.03381, over 14254.00 frames. ], tot_loss[loss=0.1038, simple_loss=0.1406, pruned_loss=0.03344, over 1989415.44 frames. ], batch size: 57, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:07:25,634 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.144e+02 2.574e+02 3.333e+02 6.487e+02, threshold=5.147e+02, percent-clipped=6.0 2022-12-08 15:08:07,701 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3267, 2.2089, 3.3596, 3.5407, 3.3321, 2.2478, 3.3847, 2.5894], device='cuda:0'), covar=tensor([0.0531, 0.1340, 0.0854, 0.0514, 0.0614, 0.1923, 0.0496, 0.1150], device='cuda:0'), in_proj_covar=tensor([0.0295, 0.0263, 0.0380, 0.0334, 0.0274, 0.0312, 0.0316, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:08:08,417 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2299, 2.2068, 2.4146, 1.5541, 1.7695, 2.2914, 1.3295, 2.3521], device='cuda:0'), covar=tensor([0.1212, 0.1379, 0.0835, 0.2260, 0.2410, 0.0919, 0.3179, 0.0897], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0103, 0.0097, 0.0101, 0.0116, 0.0092, 0.0117, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 15:08:10,229 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1772, 1.2071, 1.3350, 1.1517, 1.0504, 0.9700, 1.1481, 1.0638], device='cuda:0'), covar=tensor([0.0280, 0.0266, 0.0239, 0.0309, 0.0295, 0.0466, 0.0326, 0.0469], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 15:08:21,732 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140103.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:27,922 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140110.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:33,683 INFO [train.py:873] (0/4) Epoch 19, batch 4000, loss[loss=0.1505, simple_loss=0.1407, pruned_loss=0.0802, over 2663.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1402, pruned_loss=0.03347, over 1955724.81 frames. ], batch size: 100, lr: 4.13e-03, grad_scale: 8.0 2022-12-08 15:08:41,227 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140125.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:08:54,674 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.130e+02 2.579e+02 3.097e+02 6.234e+02, threshold=5.158e+02, percent-clipped=1.0 2022-12-08 15:08:59,203 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:12,095 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140160.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:14,025 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140162.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:14,989 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4830, 2.2681, 2.4765, 2.2204, 2.4676, 1.8748, 1.8047, 1.6836], device='cuda:0'), covar=tensor([0.0259, 0.0299, 0.0204, 0.0365, 0.0210, 0.0508, 0.0436, 0.0612], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0021, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 15:09:15,874 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140164.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:22,111 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:09:52,863 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140206.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:02,907 INFO [train.py:873] (0/4) Epoch 19, batch 4100, loss[loss=0.08939, simple_loss=0.1324, pruned_loss=0.02316, over 14377.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1394, pruned_loss=0.03272, over 2003112.10 frames. ], batch size: 31, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:10:11,498 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140227.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:21,042 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140238.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:22,567 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.026e+02 2.683e+02 3.306e+02 6.239e+02, threshold=5.366e+02, percent-clipped=2.0 2022-12-08 15:10:23,444 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.96 vs. limit=5.0 2022-12-08 15:10:39,358 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140259.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:45,894 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2022-12-08 15:10:52,806 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140275.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:10:52,851 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8998, 1.3470, 1.9954, 1.2798, 1.9566, 2.0476, 1.6169, 2.1449], device='cuda:0'), covar=tensor([0.0349, 0.2380, 0.0601, 0.2124, 0.0656, 0.0719, 0.1449, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0162, 0.0171, 0.0169, 0.0181, 0.0135, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:11:02,828 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140286.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:11:29,478 INFO [train.py:873] (0/4) Epoch 19, batch 4200, loss[loss=0.1102, simple_loss=0.1337, pruned_loss=0.04333, over 2648.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.1397, pruned_loss=0.03305, over 1952602.88 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:11:31,986 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140320.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:11:49,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.052e+02 2.429e+02 3.115e+02 5.934e+02, threshold=4.859e+02, percent-clipped=1.0 2022-12-08 15:12:29,839 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140387.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:12:56,648 INFO [train.py:873] (0/4) Epoch 19, batch 4300, loss[loss=0.1323, simple_loss=0.1597, pruned_loss=0.05248, over 11178.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1406, pruned_loss=0.03354, over 1985469.18 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:13:03,518 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140425.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:16,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.481e+01 2.038e+02 2.405e+02 2.912e+02 6.757e+02, threshold=4.809e+02, percent-clipped=4.0 2022-12-08 15:13:17,905 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2022-12-08 15:13:21,028 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140445.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:23,426 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140448.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:33,425 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140459.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:34,410 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140460.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:36,108 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140462.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:39,483 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140466.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:45,307 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140473.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:13:58,900 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.68 vs. limit=5.0 2022-12-08 15:13:59,353 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140489.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:02,824 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140493.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:04,806 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.95 vs. limit=5.0 2022-12-08 15:14:14,423 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140506.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:16,039 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140508.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:17,039 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140509.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:17,789 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140510.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:23,506 INFO [train.py:873] (0/4) Epoch 19, batch 4400, loss[loss=0.13, simple_loss=0.1463, pruned_loss=0.05685, over 3864.00 frames. ], tot_loss[loss=0.1036, simple_loss=0.1405, pruned_loss=0.03332, over 2015938.42 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 8.0 2022-12-08 15:14:30,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5209, 2.2953, 3.4300, 2.5041, 3.3727, 3.2412, 3.2059, 2.7593], device='cuda:0'), covar=tensor([0.1137, 0.3128, 0.1134, 0.2045, 0.0961, 0.1340, 0.1556, 0.2001], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0306, 0.0383, 0.0294, 0.0360, 0.0319, 0.0358, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:14:42,637 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8085, 1.9490, 2.7164, 2.2093, 2.7374, 2.6742, 2.4686, 2.3978], device='cuda:0'), covar=tensor([0.1072, 0.2955, 0.1077, 0.1816, 0.0761, 0.1358, 0.1116, 0.1513], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0305, 0.0383, 0.0294, 0.0360, 0.0318, 0.0358, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:14:43,958 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.073e+02 2.032e+02 2.425e+02 2.812e+02 5.381e+02, threshold=4.850e+02, percent-clipped=2.0 2022-12-08 15:14:48,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2022-12-08 15:14:52,688 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140550.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:14:56,016 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140554.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:09,699 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140570.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:49,843 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140615.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:15:51,859 INFO [train.py:873] (0/4) Epoch 19, batch 4500, loss[loss=0.07677, simple_loss=0.125, pruned_loss=0.01425, over 14279.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.14, pruned_loss=0.0328, over 2000457.33 frames. ], batch size: 44, lr: 4.12e-03, grad_scale: 4.0 2022-12-08 15:16:01,014 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2022-12-08 15:16:12,049 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 2.100e+02 2.600e+02 3.124e+02 5.170e+02, threshold=5.201e+02, percent-clipped=3.0 2022-12-08 15:17:17,737 INFO [train.py:873] (0/4) Epoch 19, batch 4600, loss[loss=0.1023, simple_loss=0.1332, pruned_loss=0.03565, over 5953.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1402, pruned_loss=0.0333, over 1992684.21 frames. ], batch size: 100, lr: 4.12e-03, grad_scale: 4.0 2022-12-08 15:17:23,432 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140723.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:17:36,393 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8874, 1.5071, 3.0306, 1.6785, 3.1539, 3.0446, 2.2563, 3.2230], device='cuda:0'), covar=tensor([0.0272, 0.2726, 0.0463, 0.1923, 0.0356, 0.0477, 0.0940, 0.0268], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0156, 0.0161, 0.0169, 0.0168, 0.0180, 0.0135, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:17:39,454 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 2.073e+02 2.575e+02 3.535e+02 1.946e+03, threshold=5.149e+02, percent-clipped=6.0 2022-12-08 15:17:41,345 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140743.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:17:54,999 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140759.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:00,892 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140766.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:16,712 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2022-12-08 15:18:17,130 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140784.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:29,498 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7347, 1.4314, 3.0620, 2.7555, 2.9368, 3.0695, 2.2148, 3.0207], device='cuda:0'), covar=tensor([0.1891, 0.2170, 0.0264, 0.0542, 0.0532, 0.0292, 0.0671, 0.0271], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0158, 0.0132, 0.0169, 0.0148, 0.0143, 0.0127, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:18:36,902 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140807.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:38,886 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140809.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:43,015 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140814.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:18:45,758 INFO [train.py:873] (0/4) Epoch 19, batch 4700, loss[loss=0.1029, simple_loss=0.1429, pruned_loss=0.03142, over 13945.00 frames. ], tot_loss[loss=0.1027, simple_loss=0.1396, pruned_loss=0.03286, over 2036005.47 frames. ], batch size: 23, lr: 4.11e-03, grad_scale: 4.0 2022-12-08 15:19:06,570 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 2.144e+02 2.562e+02 3.057e+02 6.222e+02, threshold=5.124e+02, percent-clipped=4.0 2022-12-08 15:19:10,105 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140845.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:19:28,288 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140865.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:19:32,434 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:11,769 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140915.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:13,317 INFO [train.py:873] (0/4) Epoch 19, batch 4800, loss[loss=0.1045, simple_loss=0.1394, pruned_loss=0.03478, over 4977.00 frames. ], tot_loss[loss=0.102, simple_loss=0.139, pruned_loss=0.03248, over 1988014.35 frames. ], batch size: 100, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:20:13,448 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140917.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:34,741 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.115e+02 2.464e+02 2.978e+02 7.209e+02, threshold=4.928e+02, percent-clipped=2.0 2022-12-08 15:20:35,501 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2022-12-08 15:20:40,045 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140947.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:53,891 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=140963.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:20:54,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 15:21:07,387 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:21:34,071 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141008.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:21:36,638 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9477, 1.6656, 3.9893, 3.7263, 3.7165, 4.0743, 3.5312, 4.0683], device='cuda:0'), covar=tensor([0.1605, 0.1672, 0.0135, 0.0254, 0.0288, 0.0163, 0.0225, 0.0127], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0131, 0.0169, 0.0148, 0.0143, 0.0126, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:21:41,896 INFO [train.py:873] (0/4) Epoch 19, batch 4900, loss[loss=0.08606, simple_loss=0.135, pruned_loss=0.01857, over 14615.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1391, pruned_loss=0.03189, over 2041753.15 frames. ], batch size: 22, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:22:02,360 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 2.070e+02 2.764e+02 3.531e+02 1.295e+03, threshold=5.527e+02, percent-clipped=7.0 2022-12-08 15:22:04,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141043.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:22:34,903 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141079.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:22:36,150 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2022-12-08 15:22:45,510 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141091.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:08,040 INFO [train.py:873] (0/4) Epoch 19, batch 5000, loss[loss=0.1267, simple_loss=0.1577, pruned_loss=0.0479, over 11940.00 frames. ], tot_loss[loss=0.1037, simple_loss=0.1406, pruned_loss=0.0334, over 1976918.35 frames. ], batch size: 100, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:23:28,878 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.058e+02 2.477e+02 3.160e+02 7.208e+02, threshold=4.953e+02, percent-clipped=1.0 2022-12-08 15:23:30,780 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141143.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:32,322 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:49,070 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:23:49,128 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:13,405 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:22,796 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141204.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:30,563 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:24:34,008 INFO [train.py:873] (0/4) Epoch 19, batch 5100, loss[loss=0.1209, simple_loss=0.1553, pruned_loss=0.04332, over 14278.00 frames. ], tot_loss[loss=0.1044, simple_loss=0.1406, pruned_loss=0.03409, over 1916769.05 frames. ], batch size: 31, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:24:54,061 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 2.078e+02 2.511e+02 3.062e+02 5.602e+02, threshold=5.021e+02, percent-clipped=1.0 2022-12-08 15:25:21,744 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141273.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:25:21,894 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6764, 2.2893, 3.5106, 2.6017, 3.5363, 3.3761, 3.2915, 2.9266], device='cuda:0'), covar=tensor([0.0891, 0.3241, 0.1112, 0.1967, 0.0948, 0.1198, 0.1228, 0.1869], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0308, 0.0385, 0.0298, 0.0361, 0.0323, 0.0360, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:25:23,871 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2022-12-08 15:25:31,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8970, 3.8458, 3.1658, 3.4236, 3.7285, 3.8950, 4.0151, 3.9303], device='cuda:0'), covar=tensor([0.1312, 0.0725, 0.3021, 0.3503, 0.1308, 0.1358, 0.1161, 0.1271], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0275, 0.0452, 0.0567, 0.0356, 0.0460, 0.0394, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:25:47,980 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141303.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:25:59,794 INFO [train.py:873] (0/4) Epoch 19, batch 5200, loss[loss=0.09643, simple_loss=0.141, pruned_loss=0.02592, over 14523.00 frames. ], tot_loss[loss=0.1034, simple_loss=0.1399, pruned_loss=0.03342, over 1921242.76 frames. ], batch size: 34, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:26:07,360 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2031, 2.2013, 4.7659, 4.3462, 4.2152, 4.8298, 4.4902, 4.8704], device='cuda:0'), covar=tensor([0.1510, 0.1383, 0.0101, 0.0211, 0.0268, 0.0138, 0.0137, 0.0099], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0131, 0.0169, 0.0148, 0.0143, 0.0126, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:26:10,262 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141329.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:26:20,994 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.019e+02 2.014e+02 2.549e+02 3.340e+02 7.347e+02, threshold=5.099e+02, percent-clipped=2.0 2022-12-08 15:26:38,346 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.5020, 5.3499, 4.9759, 5.4504, 5.1711, 4.9902, 5.5918, 5.1944], device='cuda:0'), covar=tensor([0.0502, 0.0761, 0.0819, 0.0565, 0.0716, 0.0380, 0.0440, 0.0670], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0148, 0.0150, 0.0165, 0.0151, 0.0126, 0.0171, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:26:54,172 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141379.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:04,091 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141390.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:28,119 INFO [train.py:873] (0/4) Epoch 19, batch 5300, loss[loss=0.07262, simple_loss=0.1143, pruned_loss=0.01548, over 13912.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1397, pruned_loss=0.03259, over 1995296.97 frames. ], batch size: 20, lr: 4.11e-03, grad_scale: 8.0 2022-12-08 15:27:28,246 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141417.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:27:36,871 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141427.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:37,889 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3948, 2.3924, 2.1423, 2.2433, 2.1947, 1.4629, 2.2553, 2.3210], device='cuda:0'), covar=tensor([0.0963, 0.0756, 0.0861, 0.1051, 0.1238, 0.0939, 0.0862, 0.0862], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0036, 0.0042, 0.0035, 0.0037, 0.0051, 0.0038, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:27:48,828 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.160e+02 2.222e+02 2.556e+02 2.994e+02 5.858e+02, threshold=5.112e+02, percent-clipped=4.0 2022-12-08 15:27:49,062 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141441.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:27:49,072 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8984, 1.7780, 1.8535, 1.8898, 1.8540, 1.2834, 1.4778, 1.6848], device='cuda:0'), covar=tensor([0.0624, 0.0589, 0.0633, 0.0509, 0.0540, 0.0814, 0.0922, 0.0724], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0036, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:28:07,268 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 15:28:10,354 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141465.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:21,411 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141478.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:28:39,752 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141499.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:42,512 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141502.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:51,515 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:28:53,941 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2022-12-08 15:28:54,901 INFO [train.py:873] (0/4) Epoch 19, batch 5400, loss[loss=0.1033, simple_loss=0.14, pruned_loss=0.03334, over 11961.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1394, pruned_loss=0.03191, over 2048745.13 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:29:16,144 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.017e+02 2.549e+02 3.465e+02 7.895e+02, threshold=5.098e+02, percent-clipped=4.0 2022-12-08 15:29:43,876 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141573.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:29:49,940 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2006, 2.7031, 4.0038, 3.1863, 4.0882, 3.9607, 3.8490, 3.4391], device='cuda:0'), covar=tensor([0.0808, 0.2839, 0.0881, 0.1625, 0.0814, 0.0961, 0.1499, 0.1616], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0308, 0.0386, 0.0296, 0.0360, 0.0322, 0.0359, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:30:03,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9885, 2.6155, 3.9290, 4.1082, 3.9473, 2.4379, 4.0144, 3.0929], device='cuda:0'), covar=tensor([0.0478, 0.1303, 0.1163, 0.0555, 0.0568, 0.2063, 0.0530, 0.1181], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0263, 0.0380, 0.0334, 0.0273, 0.0310, 0.0316, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:30:10,171 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141603.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:30:22,125 INFO [train.py:873] (0/4) Epoch 19, batch 5500, loss[loss=0.1073, simple_loss=0.1356, pruned_loss=0.03948, over 5973.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1393, pruned_loss=0.03184, over 2007020.91 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:30:25,594 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141621.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:30:25,729 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0400, 2.7027, 2.8099, 2.0080, 2.5476, 2.7889, 3.0480, 2.4580], device='cuda:0'), covar=tensor([0.0751, 0.0741, 0.0847, 0.1294, 0.0913, 0.0751, 0.0576, 0.1184], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0170, 0.0141, 0.0126, 0.0145, 0.0156, 0.0139, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 15:30:43,271 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.230e+02 2.074e+02 2.608e+02 3.444e+02 6.357e+02, threshold=5.216e+02, percent-clipped=6.0 2022-12-08 15:30:52,122 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141651.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:31:04,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2022-12-08 15:31:13,740 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1767, 1.5259, 4.0918, 2.0339, 4.0713, 4.2670, 3.4939, 4.6404], device='cuda:0'), covar=tensor([0.0258, 0.3252, 0.0390, 0.2079, 0.0388, 0.0418, 0.0585, 0.0174], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0155, 0.0161, 0.0168, 0.0167, 0.0179, 0.0134, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:31:21,358 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141685.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:31:22,724 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3709, 5.1504, 4.8348, 5.3387, 4.9412, 4.6916, 5.4380, 5.0884], device='cuda:0'), covar=tensor([0.0479, 0.0771, 0.0809, 0.0457, 0.0722, 0.0473, 0.0470, 0.0658], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0151, 0.0165, 0.0152, 0.0127, 0.0172, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:31:48,910 INFO [train.py:873] (0/4) Epoch 19, batch 5600, loss[loss=0.09889, simple_loss=0.1389, pruned_loss=0.02947, over 14335.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1395, pruned_loss=0.03229, over 1996659.55 frames. ], batch size: 73, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:32:09,418 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.697e+01 2.227e+02 2.853e+02 3.557e+02 6.540e+02, threshold=5.705e+02, percent-clipped=5.0 2022-12-08 15:32:37,262 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141773.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:32:58,663 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141797.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:33:00,366 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141799.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:33:16,296 INFO [train.py:873] (0/4) Epoch 19, batch 5700, loss[loss=0.1443, simple_loss=0.1393, pruned_loss=0.07464, over 1244.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1399, pruned_loss=0.03259, over 2011217.12 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:33:36,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.084e+02 2.066e+02 2.688e+02 3.167e+02 5.846e+02, threshold=5.376e+02, percent-clipped=2.0 2022-12-08 15:33:41,755 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=141847.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:34:26,406 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2022-12-08 15:34:41,495 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141916.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:34:42,167 INFO [train.py:873] (0/4) Epoch 19, batch 5800, loss[loss=0.1078, simple_loss=0.1498, pruned_loss=0.0329, over 14285.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.1401, pruned_loss=0.03283, over 2001306.69 frames. ], batch size: 31, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:34:44,082 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.2085, 2.0588, 2.1441, 2.2388, 2.1689, 2.1169, 2.2655, 1.9466], device='cuda:0'), covar=tensor([0.0904, 0.1241, 0.0728, 0.0766, 0.0904, 0.0743, 0.0947, 0.0675], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0279, 0.0204, 0.0202, 0.0187, 0.0162, 0.0294, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 15:34:47,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2022-12-08 15:34:51,873 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141928.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:03,273 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.310e+01 2.008e+02 2.595e+02 3.223e+02 5.261e+02, threshold=5.190e+02, percent-clipped=0.0 2022-12-08 15:35:07,191 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1432, 3.1086, 2.9895, 3.1936, 2.8621, 2.9526, 3.2265, 3.0694], device='cuda:0'), covar=tensor([0.0694, 0.0924, 0.0927, 0.0676, 0.1151, 0.0670, 0.0692, 0.0824], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0163, 0.0149, 0.0125, 0.0169, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:35:10,401 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2022-12-08 15:35:35,489 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141977.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:43,161 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141985.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:35:46,520 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141989.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:36:11,182 INFO [train.py:873] (0/4) Epoch 19, batch 5900, loss[loss=0.08923, simple_loss=0.1316, pruned_loss=0.02344, over 14231.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.1391, pruned_loss=0.0328, over 1938898.70 frames. ], batch size: 89, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:36:24,791 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142033.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:36:31,706 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 2.065e+02 2.424e+02 3.179e+02 4.807e+02, threshold=4.849e+02, percent-clipped=0.0 2022-12-08 15:37:00,273 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142073.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 15:37:21,225 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142097.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:37:23,464 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2022-12-08 15:37:28,698 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1787, 4.8782, 4.6734, 5.1695, 4.7843, 4.4930, 5.2414, 4.9794], device='cuda:0'), covar=tensor([0.0520, 0.0731, 0.0869, 0.0408, 0.0661, 0.0488, 0.0513, 0.0618], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0147, 0.0149, 0.0163, 0.0150, 0.0126, 0.0169, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:37:38,673 INFO [train.py:873] (0/4) Epoch 19, batch 6000, loss[loss=0.1107, simple_loss=0.1389, pruned_loss=0.04121, over 5973.00 frames. ], tot_loss[loss=0.1028, simple_loss=0.1396, pruned_loss=0.03301, over 1963072.34 frames. ], batch size: 100, lr: 4.10e-03, grad_scale: 8.0 2022-12-08 15:37:38,675 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 15:37:47,304 INFO [train.py:905] (0/4) Epoch 19, validation: loss=0.1418, simple_loss=0.1782, pruned_loss=0.05266, over 857387.00 frames. 2022-12-08 15:37:47,304 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 15:37:50,880 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142121.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:38:08,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.241e+02 1.928e+02 2.466e+02 3.087e+02 6.019e+02, threshold=4.931e+02, percent-clipped=3.0 2022-12-08 15:38:11,914 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142145.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:38:15,364 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3466, 2.2064, 2.5979, 1.6527, 1.8075, 2.3514, 1.4250, 2.3533], device='cuda:0'), covar=tensor([0.1018, 0.1397, 0.0823, 0.2449, 0.2437, 0.1138, 0.3303, 0.1085], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0106, 0.0099, 0.0103, 0.0117, 0.0094, 0.0119, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 15:38:24,962 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 15:38:38,414 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2575, 1.8420, 3.3604, 2.3738, 3.2662, 1.7959, 2.5652, 3.2129], device='cuda:0'), covar=tensor([0.0773, 0.4081, 0.0619, 0.4075, 0.0730, 0.3263, 0.1423, 0.0681], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0200, 0.0221, 0.0269, 0.0242, 0.0202, 0.0203, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 15:38:53,760 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1397, 1.8025, 2.1748, 1.8498, 2.2384, 2.0480, 1.9767, 2.0744], device='cuda:0'), covar=tensor([0.0726, 0.2260, 0.0806, 0.0956, 0.0652, 0.1253, 0.0638, 0.0843], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0309, 0.0387, 0.0297, 0.0363, 0.0321, 0.0361, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:39:03,537 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2022-12-08 15:39:15,264 INFO [train.py:873] (0/4) Epoch 19, batch 6100, loss[loss=0.0977, simple_loss=0.1376, pruned_loss=0.02892, over 13530.00 frames. ], tot_loss[loss=0.1039, simple_loss=0.1404, pruned_loss=0.0337, over 1972364.09 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:39:36,596 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 2.000e+02 2.533e+02 3.049e+02 1.114e+03, threshold=5.065e+02, percent-clipped=3.0 2022-12-08 15:39:39,536 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.35 vs. limit=5.0 2022-12-08 15:39:51,553 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0108, 1.8639, 4.8542, 2.3885, 4.5147, 5.0002, 4.5884, 5.4183], device='cuda:0'), covar=tensor([0.0170, 0.2850, 0.0270, 0.1827, 0.0244, 0.0264, 0.0210, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0162, 0.0169, 0.0167, 0.0179, 0.0133, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:40:03,700 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142272.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:40:14,732 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142284.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:40:42,866 INFO [train.py:873] (0/4) Epoch 19, batch 6200, loss[loss=0.1148, simple_loss=0.1479, pruned_loss=0.04088, over 14268.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.1397, pruned_loss=0.03307, over 1987360.37 frames. ], batch size: 37, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:41:04,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.036e+02 2.651e+02 3.192e+02 8.704e+02, threshold=5.303e+02, percent-clipped=4.0 2022-12-08 15:42:01,419 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8670, 1.5959, 1.8771, 1.5587, 1.9572, 1.7128, 1.6065, 1.8257], device='cuda:0'), covar=tensor([0.0596, 0.1167, 0.0533, 0.0528, 0.0525, 0.0886, 0.0447, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0309, 0.0386, 0.0298, 0.0364, 0.0323, 0.0360, 0.0297], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:42:11,862 INFO [train.py:873] (0/4) Epoch 19, batch 6300, loss[loss=0.1068, simple_loss=0.148, pruned_loss=0.03273, over 12752.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.1399, pruned_loss=0.03258, over 1994264.49 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:42:12,058 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8030, 2.4483, 2.6590, 1.7955, 2.3541, 2.6179, 2.8043, 2.3834], device='cuda:0'), covar=tensor([0.0763, 0.0698, 0.0843, 0.1266, 0.0976, 0.0738, 0.0703, 0.1225], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0169, 0.0141, 0.0126, 0.0145, 0.0156, 0.0139, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 15:42:29,780 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0208, 1.4260, 1.5907, 1.5985, 1.4874, 1.5771, 1.2732, 1.1428], device='cuda:0'), covar=tensor([0.1414, 0.1252, 0.0466, 0.0601, 0.1504, 0.1205, 0.1605, 0.1901], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0091, 0.0071, 0.0076, 0.0101, 0.0092, 0.0102, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0006, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 15:42:32,398 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.018e+02 2.111e+02 2.568e+02 3.235e+02 6.197e+02, threshold=5.136e+02, percent-clipped=2.0 2022-12-08 15:42:45,525 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6610, 4.1956, 3.1299, 5.0108, 4.4701, 4.7222, 4.1735, 3.2577], device='cuda:0'), covar=tensor([0.0594, 0.1053, 0.3245, 0.0375, 0.0758, 0.1128, 0.0946, 0.2897], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0289, 0.0259, 0.0293, 0.0322, 0.0304, 0.0256, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:43:03,901 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6490, 4.6866, 5.0824, 4.3323, 4.8521, 5.0520, 2.0195, 4.5416], device='cuda:0'), covar=tensor([0.0294, 0.0331, 0.0240, 0.0406, 0.0269, 0.0190, 0.2719, 0.0259], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0176, 0.0147, 0.0150, 0.0210, 0.0144, 0.0158, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 15:43:18,874 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0402, 4.8831, 4.5100, 5.0546, 4.6381, 4.2801, 5.1122, 4.8405], device='cuda:0'), covar=tensor([0.0526, 0.0774, 0.0841, 0.0506, 0.0741, 0.0624, 0.0514, 0.0616], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0147, 0.0150, 0.0163, 0.0151, 0.0126, 0.0170, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 15:43:38,808 INFO [train.py:873] (0/4) Epoch 19, batch 6400, loss[loss=0.08118, simple_loss=0.1235, pruned_loss=0.01941, over 13883.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.1395, pruned_loss=0.03258, over 1962913.93 frames. ], batch size: 23, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:44:00,397 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 1.982e+02 2.595e+02 3.310e+02 8.116e+02, threshold=5.191e+02, percent-clipped=3.0 2022-12-08 15:44:13,974 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0215, 1.8159, 4.7927, 4.3524, 4.2291, 5.0011, 4.6721, 4.9468], device='cuda:0'), covar=tensor([0.2113, 0.2329, 0.0190, 0.0359, 0.0354, 0.0205, 0.0197, 0.0162], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0169, 0.0149, 0.0143, 0.0127, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:44:26,077 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0490, 2.0354, 4.0886, 2.9835, 3.9871, 1.9521, 2.9835, 3.9229], device='cuda:0'), covar=tensor([0.0704, 0.3860, 0.0507, 0.5009, 0.0754, 0.3261, 0.1482, 0.0769], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0197, 0.0219, 0.0267, 0.0239, 0.0200, 0.0201, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 15:44:27,745 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142572.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:44:28,709 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142573.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:44:38,071 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142584.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:44:50,459 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2022-12-08 15:45:07,274 INFO [train.py:873] (0/4) Epoch 19, batch 6500, loss[loss=0.1111, simple_loss=0.1445, pruned_loss=0.03886, over 13941.00 frames. ], tot_loss[loss=0.1029, simple_loss=0.1399, pruned_loss=0.03288, over 1988538.62 frames. ], batch size: 23, lr: 4.09e-03, grad_scale: 16.0 2022-12-08 15:45:09,238 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6133, 2.4100, 2.4273, 2.5448, 2.1846, 1.6061, 2.5708, 2.6698], device='cuda:0'), covar=tensor([0.0872, 0.0565, 0.0611, 0.1074, 0.0925, 0.0748, 0.0573, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:45:09,994 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142620.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:19,976 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=142632.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:20,061 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.8148, 0.8441, 0.6649, 0.8788, 0.8128, 0.5065, 0.7178, 0.8639], device='cuda:0'), covar=tensor([0.0375, 0.0499, 0.0564, 0.0477, 0.0392, 0.0354, 0.0916, 0.0636], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0037, 0.0041, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 15:45:21,835 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142634.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:45:27,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.114e+02 2.216e+02 2.673e+02 3.413e+02 8.329e+02, threshold=5.346e+02, percent-clipped=5.0 2022-12-08 15:45:47,778 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0600, 2.0888, 1.9131, 1.8902, 1.8944, 1.8156, 1.5353, 1.5111], device='cuda:0'), covar=tensor([0.0244, 0.0265, 0.0415, 0.0438, 0.0350, 0.0324, 0.0358, 0.0476], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0023, 0.0035, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 15:46:34,426 INFO [train.py:873] (0/4) Epoch 19, batch 6600, loss[loss=0.08418, simple_loss=0.1331, pruned_loss=0.01765, over 14235.00 frames. ], tot_loss[loss=0.1023, simple_loss=0.1395, pruned_loss=0.03253, over 1992175.44 frames. ], batch size: 35, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:46:40,658 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142724.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:46:56,546 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.141e+02 2.630e+02 3.221e+02 9.486e+02, threshold=5.260e+02, percent-clipped=4.0 2022-12-08 15:47:06,365 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 15:47:34,364 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142785.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:48:02,762 INFO [train.py:873] (0/4) Epoch 19, batch 6700, loss[loss=0.136, simple_loss=0.129, pruned_loss=0.07146, over 1272.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1395, pruned_loss=0.03241, over 1953342.25 frames. ], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2022-12-08 15:48:19,964 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9384, 1.8393, 4.3872, 4.0131, 3.9524, 4.5007, 4.1204, 4.5039], device='cuda:0'), covar=tensor([0.1893, 0.1821, 0.0167, 0.0314, 0.0312, 0.0222, 0.0241, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0155, 0.0131, 0.0168, 0.0147, 0.0142, 0.0126, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:48:23,750 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.068e+02 1.995e+02 2.478e+02 2.934e+02 5.263e+02, threshold=4.957e+02, percent-clipped=1.0 2022-12-08 15:48:48,067 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9962, 1.6070, 4.0689, 1.7119, 3.8482, 4.1746, 3.2378, 4.3819], device='cuda:0'), covar=tensor([0.0253, 0.3329, 0.0399, 0.2435, 0.0473, 0.0403, 0.0703, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0155, 0.0160, 0.0167, 0.0166, 0.0178, 0.0133, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:49:28,806 INFO [train.py:873] (0/4) Epoch 19, batch 6800, loss[loss=0.0933, simple_loss=0.1359, pruned_loss=0.02534, over 13949.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1385, pruned_loss=0.03153, over 1932732.40 frames. ], batch size: 23, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:49:39,771 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142929.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:49:41,565 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9117, 1.4119, 1.9797, 1.2909, 1.9871, 2.0621, 1.6913, 2.1509], device='cuda:0'), covar=tensor([0.0302, 0.2139, 0.0663, 0.1972, 0.0669, 0.0710, 0.1337, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0160, 0.0166, 0.0166, 0.0177, 0.0133, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:49:50,756 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.071e+02 1.868e+02 2.461e+02 3.045e+02 8.260e+02, threshold=4.922e+02, percent-clipped=6.0 2022-12-08 15:50:00,304 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0553, 2.5544, 4.9497, 3.3531, 4.8730, 2.3909, 3.5771, 4.8518], device='cuda:0'), covar=tensor([0.0423, 0.3342, 0.0494, 0.5697, 0.0454, 0.2972, 0.1368, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0198, 0.0218, 0.0265, 0.0238, 0.0199, 0.0201, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 15:50:42,265 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1486, 1.6470, 4.0012, 2.0161, 4.0733, 4.2069, 3.3365, 4.5676], device='cuda:0'), covar=tensor([0.0226, 0.3081, 0.0444, 0.2023, 0.0365, 0.0392, 0.0609, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0154, 0.0159, 0.0166, 0.0165, 0.0177, 0.0132, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:50:56,727 INFO [train.py:873] (0/4) Epoch 19, batch 6900, loss[loss=0.1092, simple_loss=0.1454, pruned_loss=0.03648, over 14281.00 frames. ], tot_loss[loss=0.1009, simple_loss=0.1385, pruned_loss=0.03172, over 1977999.60 frames. ], batch size: 60, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:51:16,116 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9914, 2.7626, 3.5501, 2.4752, 2.3098, 2.9839, 1.7787, 2.9471], device='cuda:0'), covar=tensor([0.1259, 0.1046, 0.0648, 0.2002, 0.1880, 0.0929, 0.2712, 0.1118], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0103, 0.0097, 0.0101, 0.0115, 0.0092, 0.0116, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 15:51:18,510 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 2.146e+02 2.489e+02 3.023e+02 7.661e+02, threshold=4.978e+02, percent-clipped=3.0 2022-12-08 15:51:24,044 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2022-12-08 15:51:51,237 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143080.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:52:04,370 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1460, 3.4992, 3.3095, 3.3989, 2.5812, 3.3692, 3.3043, 1.8111], device='cuda:0'), covar=tensor([0.0909, 0.0558, 0.0629, 0.0508, 0.0785, 0.0655, 0.0592, 0.1701], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0092, 0.0071, 0.0077, 0.0101, 0.0092, 0.0103, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 15:52:23,546 INFO [train.py:873] (0/4) Epoch 19, batch 7000, loss[loss=0.08071, simple_loss=0.1275, pruned_loss=0.01697, over 14045.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1391, pruned_loss=0.03226, over 1987925.71 frames. ], batch size: 19, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:52:46,400 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 1.976e+02 2.529e+02 3.068e+02 1.002e+03, threshold=5.057e+02, percent-clipped=4.0 2022-12-08 15:53:20,843 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4923, 1.7324, 2.6118, 2.0857, 2.5414, 1.6694, 2.1856, 2.4798], device='cuda:0'), covar=tensor([0.1648, 0.3519, 0.0954, 0.2630, 0.1164, 0.2873, 0.1110, 0.1048], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0197, 0.0218, 0.0263, 0.0237, 0.0199, 0.0200, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 15:53:50,607 INFO [train.py:873] (0/4) Epoch 19, batch 7100, loss[loss=0.1278, simple_loss=0.1307, pruned_loss=0.06245, over 2671.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1394, pruned_loss=0.03234, over 1988445.54 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 4.0 2022-12-08 15:54:00,945 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143229.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:54:12,484 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.837e+01 2.042e+02 2.415e+02 3.007e+02 1.079e+03, threshold=4.829e+02, percent-clipped=5.0 2022-12-08 15:54:42,660 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=143277.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:54:55,069 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2022-12-08 15:55:17,482 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.11 vs. limit=5.0 2022-12-08 15:55:17,809 INFO [train.py:873] (0/4) Epoch 19, batch 7200, loss[loss=0.08832, simple_loss=0.1174, pruned_loss=0.02961, over 3902.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1392, pruned_loss=0.03239, over 1989560.54 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:55:40,829 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.274e+02 2.664e+02 3.314e+02 6.390e+02, threshold=5.328e+02, percent-clipped=7.0 2022-12-08 15:56:13,759 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143380.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:56:30,725 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143400.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:56:38,630 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9773, 1.6988, 3.5340, 3.2113, 3.3128, 3.4957, 2.7556, 3.5417], device='cuda:0'), covar=tensor([0.1665, 0.1712, 0.0157, 0.0336, 0.0328, 0.0186, 0.0388, 0.0159], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0156, 0.0131, 0.0168, 0.0148, 0.0142, 0.0126, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 15:56:46,182 INFO [train.py:873] (0/4) Epoch 19, batch 7300, loss[loss=0.1147, simple_loss=0.146, pruned_loss=0.0417, over 14008.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1394, pruned_loss=0.03291, over 1964724.05 frames. ], batch size: 26, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:56:55,449 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=143428.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 15:57:08,204 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.099e+02 2.101e+02 2.553e+02 3.059e+02 5.821e+02, threshold=5.106e+02, percent-clipped=2.0 2022-12-08 15:57:25,015 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143461.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 15:57:35,056 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143473.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:57:50,618 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2022-12-08 15:58:13,338 INFO [train.py:873] (0/4) Epoch 19, batch 7400, loss[loss=0.1091, simple_loss=0.1135, pruned_loss=0.05233, over 1234.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.139, pruned_loss=0.03272, over 1952953.98 frames. ], batch size: 100, lr: 4.08e-03, grad_scale: 8.0 2022-12-08 15:58:29,095 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143534.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 15:58:36,573 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 2.037e+02 2.578e+02 3.136e+02 8.501e+02, threshold=5.155e+02, percent-clipped=4.0 2022-12-08 15:58:42,563 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5263, 3.2221, 2.5135, 3.6977, 3.5740, 3.5887, 3.1076, 2.4839], device='cuda:0'), covar=tensor([0.0841, 0.1306, 0.2999, 0.0623, 0.0865, 0.0888, 0.1243, 0.3053], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0287, 0.0257, 0.0291, 0.0320, 0.0302, 0.0256, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 15:59:41,170 INFO [train.py:873] (0/4) Epoch 19, batch 7500, loss[loss=0.08556, simple_loss=0.1297, pruned_loss=0.02072, over 14154.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1392, pruned_loss=0.03282, over 1966406.38 frames. ], batch size: 35, lr: 4.07e-03, grad_scale: 8.0 2022-12-08 16:00:03,425 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.124e+02 2.640e+02 3.469e+02 7.311e+02, threshold=5.280e+02, percent-clipped=6.0 2022-12-08 16:00:07,856 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8307, 2.1571, 3.7671, 3.9615, 3.6767, 2.2303, 3.9178, 2.7802], device='cuda:0'), covar=tensor([0.0573, 0.1609, 0.1091, 0.0504, 0.0684, 0.2326, 0.0551, 0.1317], device='cuda:0'), in_proj_covar=tensor([0.0298, 0.0264, 0.0381, 0.0334, 0.0274, 0.0312, 0.0316, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:00:27,737 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-19.pt 2022-12-08 16:01:11,359 INFO [train.py:873] (0/4) Epoch 20, batch 0, loss[loss=0.1235, simple_loss=0.1546, pruned_loss=0.04615, over 14222.00 frames. ], tot_loss[loss=0.1235, simple_loss=0.1546, pruned_loss=0.04615, over 14222.00 frames. ], batch size: 25, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:01:11,360 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 16:01:18,885 INFO [train.py:905] (0/4) Epoch 20, validation: loss=0.1452, simple_loss=0.1824, pruned_loss=0.05396, over 857387.00 frames. 2022-12-08 16:01:18,886 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 16:01:30,591 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0737, 2.2105, 2.1965, 2.3842, 1.9883, 2.3783, 2.1166, 1.4554], device='cuda:0'), covar=tensor([0.0764, 0.0732, 0.0775, 0.0583, 0.0949, 0.0590, 0.1246, 0.1549], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0092, 0.0071, 0.0077, 0.0102, 0.0092, 0.0102, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006, 0.0006], device='cuda:0') 2022-12-08 16:01:38,100 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1644, 4.0184, 3.9171, 4.1506, 3.8800, 3.7205, 4.2812, 3.9643], device='cuda:0'), covar=tensor([0.0691, 0.1081, 0.0870, 0.0712, 0.0972, 0.0792, 0.0615, 0.0957], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0146, 0.0149, 0.0163, 0.0150, 0.0126, 0.0169, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:02:15,749 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.236e+01 1.946e+02 2.466e+02 3.510e+02 7.205e+02, threshold=4.933e+02, percent-clipped=3.0 2022-12-08 16:02:27,725 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0901, 1.9944, 2.0877, 2.1821, 2.0695, 2.0583, 2.1943, 1.9422], device='cuda:0'), covar=tensor([0.1233, 0.1319, 0.0799, 0.0780, 0.1009, 0.0674, 0.0877, 0.0738], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0278, 0.0204, 0.0201, 0.0186, 0.0161, 0.0292, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 16:02:27,730 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143756.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 16:02:47,466 INFO [train.py:873] (0/4) Epoch 20, batch 100, loss[loss=0.1139, simple_loss=0.1465, pruned_loss=0.04061, over 13528.00 frames. ], tot_loss[loss=0.1017, simple_loss=0.1395, pruned_loss=0.03201, over 884785.14 frames. ], batch size: 100, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:03:25,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.53 vs. limit=5.0 2022-12-08 16:03:31,998 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143829.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:03:43,924 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 2.111e+02 2.727e+02 3.354e+02 8.934e+02, threshold=5.454e+02, percent-clipped=6.0 2022-12-08 16:03:45,784 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9708, 4.0970, 4.4006, 3.6740, 4.2225, 4.3150, 1.6043, 4.0367], device='cuda:0'), covar=tensor([0.0389, 0.0385, 0.0322, 0.0558, 0.0338, 0.0302, 0.3185, 0.0313], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0177, 0.0148, 0.0150, 0.0210, 0.0144, 0.0158, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:04:15,259 INFO [train.py:873] (0/4) Epoch 20, batch 200, loss[loss=0.07266, simple_loss=0.1183, pruned_loss=0.01354, over 14184.00 frames. ], tot_loss[loss=0.1004, simple_loss=0.1381, pruned_loss=0.03136, over 1307826.77 frames. ], batch size: 37, lr: 3.97e-03, grad_scale: 8.0 2022-12-08 16:04:55,036 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9524, 2.0538, 2.1595, 1.9749, 1.8817, 1.5814, 1.6932, 1.3423], device='cuda:0'), covar=tensor([0.0234, 0.0441, 0.0199, 0.0261, 0.0268, 0.0342, 0.0308, 0.0450], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:05:04,870 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8942, 3.7564, 3.6144, 3.8678, 3.5807, 3.4706, 3.9879, 3.7756], device='cuda:0'), covar=tensor([0.0663, 0.0916, 0.0873, 0.0709, 0.0853, 0.0713, 0.0518, 0.0827], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0146, 0.0149, 0.0164, 0.0151, 0.0126, 0.0170, 0.0150], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:05:11,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.198e+02 1.866e+02 2.261e+02 3.101e+02 5.918e+02, threshold=4.522e+02, percent-clipped=1.0 2022-12-08 16:05:13,338 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2022-12-08 16:05:34,715 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7862, 2.3483, 3.7433, 3.9494, 3.5685, 2.2770, 3.9006, 2.8297], device='cuda:0'), covar=tensor([0.0510, 0.1433, 0.0921, 0.0462, 0.0710, 0.2170, 0.0422, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0264, 0.0378, 0.0334, 0.0274, 0.0311, 0.0315, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:05:39,040 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6876, 3.4201, 3.3882, 3.7060, 3.5017, 3.6592, 3.7298, 3.1272], device='cuda:0'), covar=tensor([0.0445, 0.1021, 0.0516, 0.0484, 0.0732, 0.0382, 0.0579, 0.0658], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0278, 0.0204, 0.0202, 0.0187, 0.0162, 0.0293, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 16:05:43,233 INFO [train.py:873] (0/4) Epoch 20, batch 300, loss[loss=0.09543, simple_loss=0.1389, pruned_loss=0.02599, over 14227.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1375, pruned_loss=0.0313, over 1571992.51 frames. ], batch size: 60, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:06:39,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.977e+01 2.046e+02 2.429e+02 3.312e+02 9.305e+02, threshold=4.857e+02, percent-clipped=8.0 2022-12-08 16:06:50,851 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144056.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:07:10,842 INFO [train.py:873] (0/4) Epoch 20, batch 400, loss[loss=0.1113, simple_loss=0.1487, pruned_loss=0.037, over 12746.00 frames. ], tot_loss[loss=0.1007, simple_loss=0.1383, pruned_loss=0.03157, over 1735236.55 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:07:12,921 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2022-12-08 16:07:32,234 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=144104.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:07:54,096 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144129.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:08:07,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 2.224e+02 2.595e+02 3.172e+02 7.454e+02, threshold=5.190e+02, percent-clipped=6.0 2022-12-08 16:08:35,923 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=144177.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:08:37,519 INFO [train.py:873] (0/4) Epoch 20, batch 500, loss[loss=0.08542, simple_loss=0.1129, pruned_loss=0.029, over 2548.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1393, pruned_loss=0.03249, over 1797946.48 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:09:34,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 2.083e+02 2.620e+02 3.196e+02 5.834e+02, threshold=5.241e+02, percent-clipped=3.0 2022-12-08 16:10:05,212 INFO [train.py:873] (0/4) Epoch 20, batch 600, loss[loss=0.1126, simple_loss=0.1215, pruned_loss=0.05183, over 1335.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1389, pruned_loss=0.03198, over 1870607.33 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:10:34,322 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2022-12-08 16:11:02,325 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.036e+02 2.548e+02 2.967e+02 5.900e+02, threshold=5.097e+02, percent-clipped=2.0 2022-12-08 16:11:33,548 INFO [train.py:873] (0/4) Epoch 20, batch 700, loss[loss=0.1067, simple_loss=0.1225, pruned_loss=0.04547, over 2618.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1386, pruned_loss=0.03215, over 1879755.28 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 4.0 2022-12-08 16:12:02,196 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144411.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:12:27,397 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4774, 1.6913, 1.9303, 1.8856, 1.7785, 1.8052, 1.5414, 1.4267], device='cuda:0'), covar=tensor([0.1104, 0.1312, 0.0509, 0.0600, 0.1176, 0.0907, 0.1559, 0.1608], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0094, 0.0073, 0.0079, 0.0103, 0.0094, 0.0104, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:0') 2022-12-08 16:12:31,474 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.983e+02 2.498e+02 3.051e+02 6.384e+02, threshold=4.995e+02, percent-clipped=1.0 2022-12-08 16:12:37,155 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.63 vs. limit=5.0 2022-12-08 16:12:56,157 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144472.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:13:02,249 INFO [train.py:873] (0/4) Epoch 20, batch 800, loss[loss=0.09964, simple_loss=0.1395, pruned_loss=0.02987, over 13954.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1394, pruned_loss=0.03292, over 1909805.25 frames. ], batch size: 23, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:13:02,353 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8639, 1.3194, 1.9635, 1.2528, 1.9432, 2.0652, 1.6750, 2.1412], device='cuda:0'), covar=tensor([0.0329, 0.2307, 0.0567, 0.2091, 0.0670, 0.0717, 0.1322, 0.0415], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0162, 0.0169, 0.0167, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:13:34,067 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2022-12-08 16:13:58,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 2.092e+02 2.630e+02 3.168e+02 5.138e+02, threshold=5.259e+02, percent-clipped=1.0 2022-12-08 16:14:29,534 INFO [train.py:873] (0/4) Epoch 20, batch 900, loss[loss=0.1062, simple_loss=0.1242, pruned_loss=0.04413, over 3879.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1389, pruned_loss=0.03216, over 1939999.90 frames. ], batch size: 100, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:15:27,787 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 2.044e+02 2.378e+02 2.904e+02 6.661e+02, threshold=4.757e+02, percent-clipped=2.0 2022-12-08 16:15:52,713 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.1795, 1.4855, 3.2896, 1.6268, 3.0708, 3.2889, 2.4271, 3.5077], device='cuda:0'), covar=tensor([0.0308, 0.3089, 0.0407, 0.2314, 0.1107, 0.0447, 0.1032, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0157, 0.0162, 0.0169, 0.0167, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:15:58,943 INFO [train.py:873] (0/4) Epoch 20, batch 1000, loss[loss=0.1014, simple_loss=0.1463, pruned_loss=0.02828, over 14201.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1397, pruned_loss=0.03277, over 1943242.72 frames. ], batch size: 46, lr: 3.96e-03, grad_scale: 8.0 2022-12-08 16:16:03,081 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2022-12-08 16:16:07,730 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4256, 3.9964, 3.0515, 4.7797, 4.1402, 4.5353, 4.0801, 3.4966], device='cuda:0'), covar=tensor([0.0680, 0.1036, 0.3081, 0.0474, 0.1376, 0.1243, 0.0993, 0.2438], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0287, 0.0258, 0.0291, 0.0321, 0.0304, 0.0257, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:16:31,215 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3106, 3.4305, 3.5613, 3.3109, 3.4995, 3.0642, 1.5625, 3.3315], device='cuda:0'), covar=tensor([0.0475, 0.0407, 0.0350, 0.0476, 0.0353, 0.0823, 0.3027, 0.0317], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0150, 0.0210, 0.0144, 0.0159, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:16:56,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 1.902e+02 2.363e+02 2.821e+02 5.781e+02, threshold=4.726e+02, percent-clipped=3.0 2022-12-08 16:17:17,202 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=144767.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:17:27,640 INFO [train.py:873] (0/4) Epoch 20, batch 1100, loss[loss=0.1208, simple_loss=0.1498, pruned_loss=0.04592, over 14203.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1388, pruned_loss=0.03196, over 1959378.11 frames. ], batch size: 89, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:18:18,889 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2022-12-08 16:18:25,976 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.093e+02 2.171e+02 2.584e+02 3.315e+02 6.802e+02, threshold=5.169e+02, percent-clipped=3.0 2022-12-08 16:18:37,587 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9949, 1.8438, 1.9969, 1.8918, 2.0196, 1.3925, 1.6675, 1.9256], device='cuda:0'), covar=tensor([0.0668, 0.0743, 0.0633, 0.1032, 0.0672, 0.0836, 0.0812, 0.0544], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 16:18:47,173 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.8915, 3.2966, 2.9770, 3.2449, 2.3667, 3.3580, 3.1399, 1.6899], device='cuda:0'), covar=tensor([0.1109, 0.0687, 0.1197, 0.0497, 0.1014, 0.0431, 0.0855, 0.1859], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0094, 0.0072, 0.0078, 0.0102, 0.0094, 0.0104, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:0') 2022-12-08 16:18:56,028 INFO [train.py:873] (0/4) Epoch 20, batch 1200, loss[loss=0.1367, simple_loss=0.1266, pruned_loss=0.07347, over 1237.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1393, pruned_loss=0.03214, over 2004439.09 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:19:54,168 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.060e+02 1.946e+02 2.467e+02 2.965e+02 6.907e+02, threshold=4.934e+02, percent-clipped=1.0 2022-12-08 16:20:09,912 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2022-12-08 16:20:23,993 INFO [train.py:873] (0/4) Epoch 20, batch 1300, loss[loss=0.1147, simple_loss=0.1254, pruned_loss=0.05203, over 2659.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.139, pruned_loss=0.03199, over 1974672.57 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:20:24,098 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0127, 4.1437, 4.2404, 3.9381, 4.1194, 4.4507, 1.7149, 3.8958], device='cuda:0'), covar=tensor([0.0466, 0.0392, 0.0473, 0.0499, 0.0525, 0.0290, 0.3658, 0.0445], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0176, 0.0148, 0.0150, 0.0209, 0.0144, 0.0159, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:20:39,337 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144996.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:20:41,666 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.6225, 5.0643, 5.0853, 5.6531, 5.2012, 4.7481, 5.5449, 4.7118], device='cuda:0'), covar=tensor([0.0403, 0.1274, 0.0328, 0.0365, 0.0779, 0.0410, 0.0524, 0.0492], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0273, 0.0201, 0.0200, 0.0184, 0.0160, 0.0290, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 16:20:43,081 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-145000.pt 2022-12-08 16:21:10,366 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:18,699 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5111, 1.4277, 3.5357, 1.6754, 3.3907, 3.5852, 2.5722, 3.8458], device='cuda:0'), covar=tensor([0.0254, 0.3034, 0.0428, 0.2182, 0.0756, 0.0438, 0.0879, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0156, 0.0162, 0.0168, 0.0166, 0.0181, 0.0133, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:21:25,301 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 16:21:26,337 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.036e+02 2.136e+02 2.595e+02 3.096e+02 6.124e+02, threshold=5.190e+02, percent-clipped=3.0 2022-12-08 16:21:36,841 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145057.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 16:21:37,563 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0263, 3.8487, 3.5458, 3.7331, 3.9226, 3.9853, 4.0169, 4.0082], device='cuda:0'), covar=tensor([0.0865, 0.0557, 0.1984, 0.2499, 0.0787, 0.0834, 0.0949, 0.0823], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0280, 0.0455, 0.0574, 0.0356, 0.0463, 0.0395, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:21:41,827 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3482, 2.0351, 2.2817, 2.1027, 2.1739, 1.9741, 1.8620, 1.4587], device='cuda:0'), covar=tensor([0.0174, 0.0286, 0.0216, 0.0323, 0.0188, 0.0308, 0.0325, 0.0462], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:21:44,411 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145066.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:45,156 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145067.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:21:55,279 INFO [train.py:873] (0/4) Epoch 20, batch 1400, loss[loss=0.07532, simple_loss=0.1178, pruned_loss=0.01643, over 14469.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1389, pruned_loss=0.03197, over 1960868.91 frames. ], batch size: 18, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:22:03,078 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145088.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:26,071 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145115.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:27,117 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.9619, 2.0390, 3.9171, 2.8293, 3.8470, 2.0547, 3.0781, 3.8766], device='cuda:0'), covar=tensor([0.0621, 0.3753, 0.0461, 0.4725, 0.0596, 0.2984, 0.1154, 0.0454], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0199, 0.0222, 0.0267, 0.0241, 0.0202, 0.0204, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 16:22:33,142 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2022-12-08 16:22:36,129 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145127.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:22:52,062 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.811e+01 2.056e+02 2.726e+02 3.195e+02 8.712e+02, threshold=5.453e+02, percent-clipped=3.0 2022-12-08 16:22:57,667 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6237, 4.4219, 4.0628, 4.2378, 4.3717, 4.5100, 4.5825, 4.5697], device='cuda:0'), covar=tensor([0.0724, 0.0486, 0.2120, 0.2721, 0.0822, 0.0808, 0.0890, 0.0755], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0281, 0.0456, 0.0577, 0.0358, 0.0463, 0.0397, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:23:20,735 INFO [train.py:873] (0/4) Epoch 20, batch 1500, loss[loss=0.1065, simple_loss=0.1414, pruned_loss=0.03578, over 14359.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.1393, pruned_loss=0.03202, over 2019055.13 frames. ], batch size: 53, lr: 3.95e-03, grad_scale: 4.0 2022-12-08 16:24:19,276 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.332e+01 2.024e+02 2.567e+02 3.376e+02 1.316e+03, threshold=5.133e+02, percent-clipped=5.0 2022-12-08 16:24:44,637 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4040, 1.7262, 4.2809, 2.2411, 4.2618, 4.4647, 3.8074, 4.8087], device='cuda:0'), covar=tensor([0.0226, 0.3032, 0.0370, 0.1935, 0.0327, 0.0403, 0.0510, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0156, 0.0161, 0.0168, 0.0166, 0.0180, 0.0133, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:24:48,111 INFO [train.py:873] (0/4) Epoch 20, batch 1600, loss[loss=0.1062, simple_loss=0.1366, pruned_loss=0.03797, over 7804.00 frames. ], tot_loss[loss=0.1026, simple_loss=0.1398, pruned_loss=0.03266, over 1992505.39 frames. ], batch size: 100, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:25:46,274 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.072e+02 1.921e+02 2.439e+02 2.878e+02 7.278e+02, threshold=4.878e+02, percent-clipped=1.0 2022-12-08 16:25:51,951 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145352.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:26:08,318 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:26:15,270 INFO [train.py:873] (0/4) Epoch 20, batch 1700, loss[loss=0.1088, simple_loss=0.1468, pruned_loss=0.03544, over 14372.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1394, pruned_loss=0.03233, over 1973578.14 frames. ], batch size: 73, lr: 3.95e-03, grad_scale: 8.0 2022-12-08 16:26:18,697 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145383.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:26:47,536 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.4309, 2.2553, 1.9864, 2.3269, 2.0340, 1.4935, 1.7284, 2.1389], device='cuda:0'), covar=tensor([0.0478, 0.0516, 0.0790, 0.0518, 0.0935, 0.0730, 0.1025, 0.0615], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0041], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 16:26:52,723 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145422.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:27:02,010 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145432.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:27:06,234 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9033, 2.8479, 2.1662, 2.9618, 2.7977, 2.8486, 2.6125, 2.3141], device='cuda:0'), covar=tensor([0.1053, 0.1232, 0.2848, 0.0921, 0.1055, 0.0836, 0.1278, 0.2435], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0286, 0.0257, 0.0293, 0.0321, 0.0305, 0.0257, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:27:13,922 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.159e+01 2.091e+02 2.554e+02 3.165e+02 6.798e+02, threshold=5.109e+02, percent-clipped=6.0 2022-12-08 16:27:42,937 INFO [train.py:873] (0/4) Epoch 20, batch 1800, loss[loss=0.09547, simple_loss=0.1348, pruned_loss=0.02808, over 14293.00 frames. ], tot_loss[loss=0.1019, simple_loss=0.1388, pruned_loss=0.03251, over 1911713.88 frames. ], batch size: 39, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:28:15,359 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8158, 3.8974, 4.1066, 3.6151, 3.9820, 4.0556, 1.7422, 3.7772], device='cuda:0'), covar=tensor([0.0393, 0.0392, 0.0329, 0.0529, 0.0321, 0.0360, 0.2978, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0178, 0.0149, 0.0152, 0.0211, 0.0144, 0.0160, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:28:37,344 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.1216, 4.9517, 4.7310, 5.0722, 4.7536, 4.5398, 5.1426, 4.8823], device='cuda:0'), covar=tensor([0.0458, 0.0714, 0.0612, 0.0485, 0.0603, 0.0492, 0.0506, 0.0570], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0148, 0.0150, 0.0166, 0.0152, 0.0128, 0.0172, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:28:41,423 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.270e+02 2.808e+02 3.331e+02 1.195e+03, threshold=5.616e+02, percent-clipped=5.0 2022-12-08 16:29:10,284 INFO [train.py:873] (0/4) Epoch 20, batch 1900, loss[loss=0.1483, simple_loss=0.1618, pruned_loss=0.06742, over 7812.00 frames. ], tot_loss[loss=0.1025, simple_loss=0.139, pruned_loss=0.03304, over 1931088.41 frames. ], batch size: 100, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:29:23,492 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145594.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:30:07,027 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:30:09,685 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 2.010e+02 2.453e+02 2.850e+02 4.730e+02, threshold=4.905e+02, percent-clipped=0.0 2022-12-08 16:30:15,030 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145652.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:30:17,598 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145655.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:30:38,452 INFO [train.py:873] (0/4) Epoch 20, batch 2000, loss[loss=0.1073, simple_loss=0.1461, pruned_loss=0.03427, over 14280.00 frames. ], tot_loss[loss=0.1024, simple_loss=0.1392, pruned_loss=0.03278, over 1932997.80 frames. ], batch size: 80, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:30:41,992 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145683.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:30:48,371 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8921, 1.6712, 1.8069, 1.9534, 1.4684, 1.7607, 1.6980, 1.8922], device='cuda:0'), covar=tensor([0.0241, 0.0433, 0.0267, 0.0242, 0.0403, 0.0477, 0.0378, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0262, 0.0377, 0.0332, 0.0272, 0.0309, 0.0315, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:30:56,733 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145700.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:00,234 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145704.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:16,137 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145722.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:20,720 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145727.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:23,656 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 16:31:24,000 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145731.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:31:37,254 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.155e+02 1.878e+02 2.490e+02 3.077e+02 5.719e+02, threshold=4.981e+02, percent-clipped=2.0 2022-12-08 16:31:58,225 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=145770.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:32:05,944 INFO [train.py:873] (0/4) Epoch 20, batch 2100, loss[loss=0.08555, simple_loss=0.1278, pruned_loss=0.02166, over 14109.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1387, pruned_loss=0.03266, over 1873052.78 frames. ], batch size: 29, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:32:43,263 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145821.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:33:04,938 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 2.012e+02 2.500e+02 3.223e+02 4.888e+02, threshold=5.000e+02, percent-clipped=0.0 2022-12-08 16:33:16,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2022-12-08 16:33:33,924 INFO [train.py:873] (0/4) Epoch 20, batch 2200, loss[loss=0.112, simple_loss=0.121, pruned_loss=0.05149, over 1284.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1391, pruned_loss=0.03248, over 1911735.30 frames. ], batch size: 100, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:33:36,717 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145882.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:34:19,767 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6480, 1.6119, 2.8339, 2.1953, 2.6881, 1.7621, 2.3284, 2.7048], device='cuda:0'), covar=tensor([0.1640, 0.3983, 0.0858, 0.3035, 0.1139, 0.2844, 0.1199, 0.0971], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0198, 0.0220, 0.0263, 0.0238, 0.0199, 0.0200, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 16:34:32,821 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 2.107e+02 2.702e+02 3.321e+02 7.589e+02, threshold=5.404e+02, percent-clipped=7.0 2022-12-08 16:34:36,627 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145950.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:34:59,798 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8999, 1.2578, 1.9980, 1.3098, 1.9294, 2.0875, 1.7054, 2.1335], device='cuda:0'), covar=tensor([0.0287, 0.2525, 0.0598, 0.2176, 0.0742, 0.0719, 0.1290, 0.0464], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0163, 0.0170, 0.0168, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:35:01,468 INFO [train.py:873] (0/4) Epoch 20, batch 2300, loss[loss=0.0966, simple_loss=0.1389, pruned_loss=0.02715, over 14287.00 frames. ], tot_loss[loss=0.1011, simple_loss=0.1382, pruned_loss=0.03203, over 1905740.14 frames. ], batch size: 44, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:35:19,631 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145999.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:35:41,639 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1288, 1.3085, 1.2964, 1.0203, 0.8007, 1.0819, 0.8570, 1.2296], device='cuda:0'), covar=tensor([0.1897, 0.2816, 0.1386, 0.2916, 0.3791, 0.1807, 0.2209, 0.1544], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0103, 0.0096, 0.0101, 0.0115, 0.0093, 0.0117, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 16:35:44,191 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146027.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:36:00,973 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.169e+02 2.442e+02 2.958e+02 6.970e+02, threshold=4.884e+02, percent-clipped=2.0 2022-12-08 16:36:21,063 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7695, 3.4105, 3.6088, 3.8743, 3.6605, 3.5554, 3.8821, 3.2203], device='cuda:0'), covar=tensor([0.1301, 0.1972, 0.1105, 0.0984, 0.1394, 0.2058, 0.1066, 0.1169], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0278, 0.0204, 0.0202, 0.0187, 0.0163, 0.0291, 0.0172], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 16:36:26,104 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0038, 2.1949, 2.9129, 2.3223, 2.9363, 2.8871, 2.8312, 2.5704], device='cuda:0'), covar=tensor([0.0840, 0.2526, 0.0986, 0.1603, 0.0670, 0.1113, 0.1149, 0.1579], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0306, 0.0387, 0.0297, 0.0362, 0.0319, 0.0356, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:36:26,862 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146075.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:36:30,340 INFO [train.py:873] (0/4) Epoch 20, batch 2400, loss[loss=0.08823, simple_loss=0.1357, pruned_loss=0.02039, over 14287.00 frames. ], tot_loss[loss=0.1019, simple_loss=0.1389, pruned_loss=0.03252, over 1897337.27 frames. ], batch size: 39, lr: 3.94e-03, grad_scale: 8.0 2022-12-08 16:37:29,040 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 1.997e+02 2.339e+02 2.913e+02 8.016e+02, threshold=4.679e+02, percent-clipped=1.0 2022-12-08 16:37:31,879 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146149.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:37:41,624 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9048, 4.5604, 4.4380, 4.9494, 4.5665, 4.3496, 4.9227, 4.2001], device='cuda:0'), covar=tensor([0.0425, 0.1003, 0.0484, 0.0390, 0.0824, 0.0578, 0.0514, 0.0519], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0276, 0.0203, 0.0202, 0.0186, 0.0162, 0.0290, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 16:37:56,518 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146177.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:37:58,097 INFO [train.py:873] (0/4) Epoch 20, batch 2500, loss[loss=0.0881, simple_loss=0.1295, pruned_loss=0.02334, over 13542.00 frames. ], tot_loss[loss=0.1015, simple_loss=0.1387, pruned_loss=0.03209, over 1868279.18 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:38:11,923 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2022-12-08 16:38:26,579 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146210.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:38:58,168 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.933e+01 2.087e+02 2.444e+02 3.038e+02 5.934e+02, threshold=4.888e+02, percent-clipped=3.0 2022-12-08 16:39:01,866 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146250.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:39:19,284 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4549, 1.4358, 3.4897, 1.6205, 3.3083, 3.5699, 2.6010, 3.8008], device='cuda:0'), covar=tensor([0.0265, 0.3122, 0.0409, 0.2324, 0.0861, 0.0411, 0.0910, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0158, 0.0163, 0.0170, 0.0168, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:39:25,793 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8072, 1.5931, 1.7585, 1.5288, 1.5833, 1.5072, 1.3259, 1.1715], device='cuda:0'), covar=tensor([0.0161, 0.0294, 0.0199, 0.0217, 0.0220, 0.0311, 0.0254, 0.0343], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0023, 0.0022, 0.0035, 0.0029, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:39:27,231 INFO [train.py:873] (0/4) Epoch 20, batch 2600, loss[loss=0.1079, simple_loss=0.1438, pruned_loss=0.03601, over 11181.00 frames. ], tot_loss[loss=0.1016, simple_loss=0.139, pruned_loss=0.0321, over 1902963.58 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:39:43,713 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146298.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:39:44,629 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146299.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:39:56,924 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2022-12-08 16:40:25,673 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.546e+01 2.007e+02 2.593e+02 2.996e+02 8.778e+02, threshold=5.186e+02, percent-clipped=5.0 2022-12-08 16:40:26,667 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146347.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:40:27,822 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2022-12-08 16:40:54,845 INFO [train.py:873] (0/4) Epoch 20, batch 2700, loss[loss=0.09343, simple_loss=0.1207, pruned_loss=0.03306, over 3849.00 frames. ], tot_loss[loss=0.1021, simple_loss=0.1389, pruned_loss=0.03264, over 1877768.21 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:41:34,968 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2430, 2.6894, 3.6694, 2.5009, 2.3402, 3.0023, 1.8949, 3.1543], device='cuda:0'), covar=tensor([0.0875, 0.1107, 0.0579, 0.2276, 0.1853, 0.0925, 0.2590, 0.1131], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0102, 0.0096, 0.0101, 0.0115, 0.0092, 0.0116, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 16:41:53,288 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 2.024e+02 2.634e+02 3.181e+02 1.190e+03, threshold=5.268e+02, percent-clipped=3.0 2022-12-08 16:42:19,614 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9583, 1.7565, 1.9808, 1.9424, 1.9006, 1.3159, 1.5942, 1.8002], device='cuda:0'), covar=tensor([0.0626, 0.0812, 0.0457, 0.1087, 0.0638, 0.0811, 0.0880, 0.0711], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0038, 0.0042, 0.0035, 0.0037, 0.0052, 0.0040, 0.0042], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 16:42:20,411 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146477.0, num_to_drop=1, layers_to_drop={2} 2022-12-08 16:42:21,925 INFO [train.py:873] (0/4) Epoch 20, batch 2800, loss[loss=0.08982, simple_loss=0.1319, pruned_loss=0.02384, over 14420.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.1392, pruned_loss=0.03266, over 1981664.52 frames. ], batch size: 73, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:42:37,912 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.6006, 2.3577, 2.9715, 1.9096, 1.9826, 2.5569, 1.5597, 2.5480], device='cuda:0'), covar=tensor([0.0979, 0.1359, 0.0736, 0.2613, 0.2095, 0.0917, 0.3041, 0.1096], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0101, 0.0096, 0.0100, 0.0114, 0.0092, 0.0115, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 16:42:44,662 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146505.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:42:55,533 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3297, 5.1665, 4.7719, 4.9822, 5.0108, 5.2520, 5.3628, 5.2977], device='cuda:0'), covar=tensor([0.0714, 0.0414, 0.2075, 0.2632, 0.0643, 0.0757, 0.0658, 0.0743], device='cuda:0'), in_proj_covar=tensor([0.0397, 0.0283, 0.0455, 0.0575, 0.0359, 0.0466, 0.0398, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:43:02,224 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146525.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 16:43:20,289 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 1.986e+02 2.533e+02 3.212e+02 8.104e+02, threshold=5.065e+02, percent-clipped=2.0 2022-12-08 16:43:41,059 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2204, 2.8533, 2.9080, 2.0266, 2.6777, 2.9316, 3.1690, 2.6131], device='cuda:0'), covar=tensor([0.0662, 0.0719, 0.0807, 0.1221, 0.0897, 0.0676, 0.0698, 0.1064], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0167, 0.0140, 0.0125, 0.0144, 0.0156, 0.0139, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 16:43:48,733 INFO [train.py:873] (0/4) Epoch 20, batch 2900, loss[loss=0.08822, simple_loss=0.1269, pruned_loss=0.02475, over 6947.00 frames. ], tot_loss[loss=0.102, simple_loss=0.1389, pruned_loss=0.03257, over 1923026.47 frames. ], batch size: 100, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:43:59,412 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146591.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:44:17,998 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0029, 2.0356, 2.0810, 2.0844, 2.0951, 1.6288, 1.3824, 1.9031], device='cuda:0'), covar=tensor([0.0671, 0.0565, 0.0438, 0.0428, 0.0485, 0.1448, 0.2075, 0.0472], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0177, 0.0148, 0.0150, 0.0209, 0.0143, 0.0157, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:44:30,276 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146626.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:44:38,885 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2022-12-08 16:44:47,463 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.123e+02 2.667e+02 3.296e+02 6.992e+02, threshold=5.334e+02, percent-clipped=2.0 2022-12-08 16:44:52,786 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146652.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:45:15,292 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.61 vs. limit=5.0 2022-12-08 16:45:16,677 INFO [train.py:873] (0/4) Epoch 20, batch 3000, loss[loss=0.08277, simple_loss=0.1234, pruned_loss=0.02105, over 14128.00 frames. ], tot_loss[loss=0.1022, simple_loss=0.1391, pruned_loss=0.03265, over 1842992.82 frames. ], batch size: 19, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:45:16,678 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 16:45:22,951 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.5607, 1.6807, 1.8338, 1.2366, 1.4695, 1.0759, 1.1398, 1.1084], device='cuda:0'), covar=tensor([0.0108, 0.0100, 0.0074, 0.0138, 0.0122, 0.0242, 0.0163, 0.0240], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0023, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:45:23,213 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0021, 4.3786, 3.2953, 5.1344, 4.5628, 5.0812, 4.2808, 3.8120], device='cuda:0'), covar=tensor([0.0361, 0.0676, 0.2748, 0.0253, 0.0507, 0.0590, 0.0924, 0.2090], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0284, 0.0258, 0.0291, 0.0322, 0.0304, 0.0255, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:45:25,136 INFO [train.py:905] (0/4) Epoch 20, validation: loss=0.1444, simple_loss=0.1794, pruned_loss=0.05469, over 857387.00 frames. 2022-12-08 16:45:25,137 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 16:45:25,318 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2970, 3.1980, 3.8639, 2.9268, 2.4262, 3.2116, 1.8725, 3.5406], device='cuda:0'), covar=tensor([0.0932, 0.1181, 0.0621, 0.1683, 0.1956, 0.0910, 0.3067, 0.0971], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0102, 0.0096, 0.0101, 0.0115, 0.0093, 0.0116, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 16:45:30,424 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4461, 2.2435, 4.3810, 2.9410, 4.3044, 2.0835, 3.4029, 4.3405], device='cuda:0'), covar=tensor([0.0642, 0.3582, 0.0507, 0.5781, 0.0795, 0.3363, 0.1240, 0.0459], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0198, 0.0222, 0.0265, 0.0240, 0.0202, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 16:45:32,064 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146687.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 16:45:39,000 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0409, 1.6172, 3.9288, 1.8838, 3.9110, 4.0639, 3.2299, 4.3751], device='cuda:0'), covar=tensor([0.0238, 0.3129, 0.0471, 0.2202, 0.0502, 0.0414, 0.0647, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0163, 0.0170, 0.0168, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:46:23,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.030e+02 2.519e+02 3.160e+02 5.604e+02, threshold=5.039e+02, percent-clipped=1.0 2022-12-08 16:46:52,204 INFO [train.py:873] (0/4) Epoch 20, batch 3100, loss[loss=0.09881, simple_loss=0.1415, pruned_loss=0.02805, over 14434.00 frames. ], tot_loss[loss=0.1019, simple_loss=0.1391, pruned_loss=0.03231, over 1900372.96 frames. ], batch size: 41, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:46:54,035 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.0085, 2.0087, 2.0991, 2.0971, 2.0628, 1.6370, 1.2932, 1.8957], device='cuda:0'), covar=tensor([0.0730, 0.0660, 0.0477, 0.0413, 0.0513, 0.1369, 0.2296, 0.0481], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0178, 0.0148, 0.0151, 0.0210, 0.0143, 0.0158, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:47:14,820 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146805.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:47:28,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5757, 3.1613, 2.4590, 3.6889, 3.5544, 3.5502, 3.0762, 2.4784], device='cuda:0'), covar=tensor([0.0772, 0.1328, 0.3163, 0.0615, 0.0925, 0.1069, 0.1471, 0.3131], device='cuda:0'), in_proj_covar=tensor([0.0281, 0.0283, 0.0256, 0.0290, 0.0321, 0.0303, 0.0254, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:47:48,885 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146844.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:47:50,641 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.017e+02 2.599e+02 3.064e+02 9.305e+02, threshold=5.198e+02, percent-clipped=5.0 2022-12-08 16:47:56,845 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=146853.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:48:19,589 INFO [train.py:873] (0/4) Epoch 20, batch 3200, loss[loss=0.08672, simple_loss=0.1317, pruned_loss=0.02089, over 14597.00 frames. ], tot_loss[loss=0.1013, simple_loss=0.1388, pruned_loss=0.03188, over 1994448.79 frames. ], batch size: 30, lr: 3.93e-03, grad_scale: 8.0 2022-12-08 16:48:43,354 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146905.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:49:08,451 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.97 vs. limit=5.0 2022-12-08 16:49:18,825 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.042e+02 2.442e+02 3.021e+02 7.374e+02, threshold=4.884e+02, percent-clipped=4.0 2022-12-08 16:49:19,729 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146947.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:49:37,773 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.5123, 1.7015, 2.7036, 2.2487, 2.5671, 1.7604, 2.2391, 2.5717], device='cuda:0'), covar=tensor([0.1782, 0.3636, 0.0811, 0.2747, 0.1409, 0.2733, 0.1005, 0.0972], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0198, 0.0222, 0.0266, 0.0241, 0.0202, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 16:49:47,630 INFO [train.py:873] (0/4) Epoch 20, batch 3300, loss[loss=0.09757, simple_loss=0.1394, pruned_loss=0.02785, over 14588.00 frames. ], tot_loss[loss=0.1014, simple_loss=0.1389, pruned_loss=0.03199, over 1992577.16 frames. ], batch size: 21, lr: 3.92e-03, grad_scale: 8.0 2022-12-08 16:49:50,228 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146982.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 16:50:47,402 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.122e+02 2.526e+02 3.271e+02 7.839e+02, threshold=5.052e+02, percent-clipped=6.0 2022-12-08 16:51:09,945 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0559, 3.8871, 3.7347, 4.0986, 3.6225, 3.4581, 4.1285, 3.9091], device='cuda:0'), covar=tensor([0.0534, 0.0785, 0.0824, 0.0519, 0.0858, 0.0705, 0.0478, 0.0700], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0148, 0.0150, 0.0164, 0.0151, 0.0127, 0.0171, 0.0152], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:51:14,963 INFO [train.py:873] (0/4) Epoch 20, batch 3400, loss[loss=0.07992, simple_loss=0.119, pruned_loss=0.02043, over 6937.00 frames. ], tot_loss[loss=0.101, simple_loss=0.1383, pruned_loss=0.03183, over 1955138.53 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:52:16,822 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 1.939e+02 2.278e+02 2.850e+02 6.486e+02, threshold=4.556e+02, percent-clipped=1.0 2022-12-08 16:52:16,941 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.9076, 4.7871, 4.5017, 4.9424, 4.5666, 4.2042, 4.9756, 4.6570], device='cuda:0'), covar=tensor([0.0576, 0.0786, 0.0771, 0.0540, 0.0786, 0.0620, 0.0547, 0.0816], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0149, 0.0150, 0.0166, 0.0152, 0.0128, 0.0172, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:52:45,345 INFO [train.py:873] (0/4) Epoch 20, batch 3500, loss[loss=0.114, simple_loss=0.1487, pruned_loss=0.03961, over 9512.00 frames. ], tot_loss[loss=0.1004, simple_loss=0.138, pruned_loss=0.03135, over 1989337.08 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:52:49,321 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.57 vs. limit=5.0 2022-12-08 16:53:03,638 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147200.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:53:39,363 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147241.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:53:44,458 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2022-12-08 16:53:44,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 2.013e+02 2.574e+02 3.119e+02 7.865e+02, threshold=5.147e+02, percent-clipped=11.0 2022-12-08 16:53:44,835 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147247.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:12,621 INFO [train.py:873] (0/4) Epoch 20, batch 3600, loss[loss=0.09796, simple_loss=0.1316, pruned_loss=0.03214, over 5988.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1384, pruned_loss=0.03156, over 1958662.78 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 8.0 2022-12-08 16:54:15,382 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147282.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:26,927 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147295.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:34,032 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147302.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:40,958 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147310.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:54:48,374 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4963, 4.5829, 4.8688, 4.0864, 4.6799, 4.9795, 2.0947, 4.4038], device='cuda:0'), covar=tensor([0.0320, 0.0319, 0.0336, 0.0440, 0.0304, 0.0169, 0.2784, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0178, 0.0150, 0.0153, 0.0212, 0.0145, 0.0159, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 16:54:58,646 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147330.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:55:03,546 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1678, 1.2078, 1.2449, 0.9527, 0.9696, 0.8288, 1.1204, 0.8900], device='cuda:0'), covar=tensor([0.0310, 0.0284, 0.0267, 0.0328, 0.0308, 0.0525, 0.0329, 0.0488], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:55:08,552 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2489, 3.6602, 2.8972, 4.3854, 4.1004, 4.1775, 3.7041, 2.9824], device='cuda:0'), covar=tensor([0.0605, 0.1095, 0.2975, 0.0471, 0.0822, 0.1251, 0.1079, 0.2765], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0284, 0.0256, 0.0291, 0.0322, 0.0304, 0.0254, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 16:55:14,167 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.052e+02 2.112e+02 2.582e+02 3.106e+02 5.993e+02, threshold=5.164e+02, percent-clipped=4.0 2022-12-08 16:55:34,710 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147371.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:55:41,608 INFO [train.py:873] (0/4) Epoch 20, batch 3700, loss[loss=0.08768, simple_loss=0.1301, pruned_loss=0.02265, over 13539.00 frames. ], tot_loss[loss=0.101, simple_loss=0.1385, pruned_loss=0.0318, over 1989300.92 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:56:06,131 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147407.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:56:29,363 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147434.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:56:39,537 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6203, 3.4805, 3.2932, 3.6911, 3.2842, 3.2541, 3.6826, 3.4787], device='cuda:0'), covar=tensor([0.0592, 0.0950, 0.0968, 0.0538, 0.0990, 0.0766, 0.0637, 0.0750], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0150, 0.0152, 0.0167, 0.0152, 0.0128, 0.0174, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 16:56:41,247 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.266e+02 2.765e+02 3.309e+02 4.888e+02, threshold=5.530e+02, percent-clipped=0.0 2022-12-08 16:56:58,750 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147468.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:57:07,329 INFO [train.py:873] (0/4) Epoch 20, batch 3800, loss[loss=0.106, simple_loss=0.139, pruned_loss=0.03647, over 13544.00 frames. ], tot_loss[loss=0.1005, simple_loss=0.1381, pruned_loss=0.03141, over 2001191.54 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:57:21,589 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147495.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:57:25,952 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147500.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:58:07,480 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.824e+01 2.099e+02 2.660e+02 3.307e+02 5.236e+02, threshold=5.319e+02, percent-clipped=0.0 2022-12-08 16:58:07,570 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147548.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:58:26,132 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2022-12-08 16:58:34,639 INFO [train.py:873] (0/4) Epoch 20, batch 3900, loss[loss=0.1235, simple_loss=0.1267, pruned_loss=0.06018, over 1249.00 frames. ], tot_loss[loss=0.0993, simple_loss=0.1373, pruned_loss=0.03066, over 1980994.24 frames. ], batch size: 100, lr: 3.92e-03, grad_scale: 4.0 2022-12-08 16:58:50,262 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147597.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:59:08,372 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.0743, 1.1415, 1.0210, 1.0782, 1.1944, 0.7533, 0.9908, 1.0872], device='cuda:0'), covar=tensor([0.0672, 0.0689, 0.0831, 0.0631, 0.0492, 0.0464, 0.1005, 0.0857], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0037, 0.0042, 0.0035, 0.0037, 0.0051, 0.0039, 0.0042], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 16:59:35,033 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 1.905e+02 2.298e+02 3.086e+02 1.551e+03, threshold=4.597e+02, percent-clipped=3.0 2022-12-08 16:59:38,102 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7913, 2.4464, 3.2714, 2.1721, 2.1122, 2.7873, 1.5630, 2.8559], device='cuda:0'), covar=tensor([0.1096, 0.1508, 0.0494, 0.1738, 0.1923, 0.0839, 0.3107, 0.0891], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0104, 0.0097, 0.0102, 0.0116, 0.0094, 0.0117, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 16:59:42,147 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1941, 2.1959, 2.3798, 2.2384, 2.1618, 1.9451, 1.7697, 1.7851], device='cuda:0'), covar=tensor([0.0409, 0.0418, 0.0282, 0.0284, 0.0281, 0.0432, 0.0387, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 16:59:44,916 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147659.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 16:59:51,003 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147666.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:00:01,897 INFO [train.py:873] (0/4) Epoch 20, batch 4000, loss[loss=0.09904, simple_loss=0.1424, pruned_loss=0.02783, over 14151.00 frames. ], tot_loss[loss=0.09864, simple_loss=0.1369, pruned_loss=0.03017, over 1947478.60 frames. ], batch size: 84, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:00:37,467 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147720.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:01:01,455 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.046e+02 2.183e+02 2.642e+02 3.333e+02 1.015e+03, threshold=5.284e+02, percent-clipped=7.0 2022-12-08 17:01:10,146 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7225, 2.3570, 4.7444, 3.1616, 4.5815, 2.0686, 3.5827, 4.5423], device='cuda:0'), covar=tensor([0.0530, 0.3515, 0.0357, 0.5275, 0.0501, 0.3229, 0.1174, 0.0543], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0197, 0.0222, 0.0265, 0.0240, 0.0200, 0.0200, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:01:14,059 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147763.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:01:27,732 INFO [train.py:873] (0/4) Epoch 20, batch 4100, loss[loss=0.09896, simple_loss=0.1253, pruned_loss=0.03633, over 3848.00 frames. ], tot_loss[loss=0.09941, simple_loss=0.1374, pruned_loss=0.03073, over 2006547.21 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:01:37,685 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147790.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:01:58,504 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.4064, 4.5245, 4.7953, 4.1915, 4.6324, 4.7990, 1.8920, 4.3699], device='cuda:0'), covar=tensor([0.0316, 0.0347, 0.0278, 0.0378, 0.0239, 0.0211, 0.2907, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0179, 0.0150, 0.0152, 0.0212, 0.0145, 0.0160, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 17:01:59,409 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147815.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:27,883 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 2.007e+02 2.574e+02 3.284e+02 8.087e+02, threshold=5.147e+02, percent-clipped=4.0 2022-12-08 17:02:36,131 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147857.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:52,196 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147876.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:02:54,612 INFO [train.py:873] (0/4) Epoch 20, batch 4200, loss[loss=0.1247, simple_loss=0.1508, pruned_loss=0.04924, over 4961.00 frames. ], tot_loss[loss=0.09969, simple_loss=0.1377, pruned_loss=0.03086, over 2016540.56 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:03:10,394 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147897.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:22,124 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147911.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:28,100 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147918.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:43,905 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6350, 3.7492, 3.9381, 3.5825, 3.8364, 3.7210, 1.6170, 3.6773], device='cuda:0'), covar=tensor([0.0355, 0.0359, 0.0314, 0.0451, 0.0307, 0.0550, 0.3002, 0.0294], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0179, 0.0150, 0.0152, 0.0212, 0.0145, 0.0159, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 17:03:52,190 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=147945.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:03:55,541 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 1.943e+02 2.282e+02 3.039e+02 5.429e+02, threshold=4.564e+02, percent-clipped=1.0 2022-12-08 17:04:09,972 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147966.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:12,102 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([0.7371, 1.0933, 1.2933, 1.1985, 0.8560, 1.2346, 1.0142, 0.8470], device='cuda:0'), covar=tensor([0.1799, 0.0968, 0.0486, 0.0477, 0.2060, 0.1101, 0.1494, 0.1532], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:0') 2022-12-08 17:04:15,528 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147972.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:21,715 INFO [train.py:873] (0/4) Epoch 20, batch 4300, loss[loss=0.1208, simple_loss=0.1467, pruned_loss=0.04746, over 7789.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1386, pruned_loss=0.03152, over 1949875.47 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:04:52,059 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148014.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:04:52,964 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148015.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:05:00,354 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.3466, 2.1856, 2.1485, 2.6230, 2.3318, 1.4077, 1.9305, 2.3214], device='cuda:0'), covar=tensor([0.0797, 0.0811, 0.0701, 0.0352, 0.0735, 0.0896, 0.0884, 0.0614], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0038, 0.0042, 0.0035, 0.0037, 0.0052, 0.0039, 0.0042], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2022-12-08 17:05:22,415 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 2.071e+02 2.502e+02 3.240e+02 4.957e+02, threshold=5.004e+02, percent-clipped=3.0 2022-12-08 17:05:35,029 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148063.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:05:48,146 INFO [train.py:873] (0/4) Epoch 20, batch 4400, loss[loss=0.1111, simple_loss=0.1487, pruned_loss=0.03675, over 14061.00 frames. ], tot_loss[loss=0.1006, simple_loss=0.1384, pruned_loss=0.03145, over 1959946.68 frames. ], batch size: 29, lr: 3.91e-03, grad_scale: 8.0 2022-12-08 17:05:58,137 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148090.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:16,514 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148111.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:39,820 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148138.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:06:49,733 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.267e+02 2.050e+02 2.451e+02 3.050e+02 8.671e+02, threshold=4.901e+02, percent-clipped=4.0 2022-12-08 17:07:08,034 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148171.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:07:15,616 INFO [train.py:873] (0/4) Epoch 20, batch 4500, loss[loss=0.09293, simple_loss=0.14, pruned_loss=0.02291, over 14527.00 frames. ], tot_loss[loss=0.1002, simple_loss=0.1383, pruned_loss=0.03107, over 2003436.54 frames. ], batch size: 34, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:07:45,047 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148213.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:08:17,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.029e+02 2.580e+02 3.112e+02 6.429e+02, threshold=5.160e+02, percent-clipped=2.0 2022-12-08 17:08:32,445 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148267.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:08:42,532 INFO [train.py:873] (0/4) Epoch 20, batch 4600, loss[loss=0.09106, simple_loss=0.1266, pruned_loss=0.02777, over 5953.00 frames. ], tot_loss[loss=0.09997, simple_loss=0.138, pruned_loss=0.03095, over 1996334.61 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:08:49,647 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148287.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:13,743 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148315.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:42,775 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148348.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:44,260 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 2.179e+02 2.688e+02 3.360e+02 6.935e+02, threshold=5.375e+02, percent-clipped=4.0 2022-12-08 17:09:48,067 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2022-12-08 17:09:55,115 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148363.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:09:59,545 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3414, 3.1560, 2.8800, 3.0855, 3.2628, 3.3181, 3.2924, 3.3431], device='cuda:0'), covar=tensor([0.0934, 0.0651, 0.2163, 0.2362, 0.0896, 0.0937, 0.1186, 0.0818], device='cuda:0'), in_proj_covar=tensor([0.0400, 0.0285, 0.0455, 0.0574, 0.0364, 0.0469, 0.0402, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:10:09,363 INFO [train.py:873] (0/4) Epoch 20, batch 4700, loss[loss=0.09521, simple_loss=0.1336, pruned_loss=0.02839, over 9501.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1382, pruned_loss=0.03098, over 2004009.37 frames. ], batch size: 100, lr: 3.91e-03, grad_scale: 4.0 2022-12-08 17:10:15,914 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1514, 2.0156, 2.0819, 2.1692, 2.0605, 1.9966, 2.2086, 1.8878], device='cuda:0'), covar=tensor([0.0831, 0.1296, 0.0843, 0.0834, 0.0974, 0.0698, 0.0889, 0.0728], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0276, 0.0205, 0.0203, 0.0186, 0.0163, 0.0293, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 17:10:52,734 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148429.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:03,277 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.58 vs. limit=2.0 2022-12-08 17:11:10,544 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.097e+02 2.538e+02 3.093e+02 5.864e+02, threshold=5.077e+02, percent-clipped=2.0 2022-12-08 17:11:27,224 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 17:11:28,986 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148471.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:35,537 INFO [train.py:873] (0/4) Epoch 20, batch 4800, loss[loss=0.07562, simple_loss=0.1129, pruned_loss=0.01919, over 14299.00 frames. ], tot_loss[loss=0.09977, simple_loss=0.1378, pruned_loss=0.03089, over 1991021.83 frames. ], batch size: 18, lr: 3.90e-03, grad_scale: 8.0 2022-12-08 17:11:45,455 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148490.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:11:52,397 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148498.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:05,553 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148513.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:10,308 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148519.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:37,249 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.096e+02 2.575e+02 3.014e+02 5.003e+02, threshold=5.149e+02, percent-clipped=0.0 2022-12-08 17:12:40,376 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2022-12-08 17:12:45,255 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148559.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:46,808 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148561.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:12:46,861 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.1365, 1.2837, 1.2443, 1.0784, 0.8171, 1.0811, 0.8261, 1.2129], device='cuda:0'), covar=tensor([0.1989, 0.2583, 0.1558, 0.2345, 0.3402, 0.1509, 0.1714, 0.1403], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0103, 0.0097, 0.0102, 0.0116, 0.0094, 0.0116, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 17:12:52,323 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148567.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:13:02,972 INFO [train.py:873] (0/4) Epoch 20, batch 4900, loss[loss=0.1319, simple_loss=0.1462, pruned_loss=0.05879, over 1267.00 frames. ], tot_loss[loss=0.1005, simple_loss=0.1381, pruned_loss=0.0314, over 1956803.28 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:13:33,586 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148615.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:13:57,690 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148643.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:14:04,643 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.967e+02 2.434e+02 3.108e+02 5.007e+02, threshold=4.868e+02, percent-clipped=0.0 2022-12-08 17:14:04,855 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148651.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:14:10,090 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 17:14:29,318 INFO [train.py:873] (0/4) Epoch 20, batch 5000, loss[loss=0.1114, simple_loss=0.1486, pruned_loss=0.03709, over 14344.00 frames. ], tot_loss[loss=0.1004, simple_loss=0.1377, pruned_loss=0.0315, over 1952328.89 frames. ], batch size: 66, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:14:58,904 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148712.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:02,390 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148716.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:33,632 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.005e+02 2.005e+02 2.534e+02 3.099e+02 5.530e+02, threshold=5.067e+02, percent-clipped=2.0 2022-12-08 17:15:56,330 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148777.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:15:58,225 INFO [train.py:873] (0/4) Epoch 20, batch 5100, loss[loss=0.1141, simple_loss=0.1262, pruned_loss=0.05098, over 2609.00 frames. ], tot_loss[loss=0.1006, simple_loss=0.1379, pruned_loss=0.03165, over 1977980.58 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:16:03,531 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148785.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:16:30,497 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2022-12-08 17:17:01,005 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.285e+02 2.852e+02 3.620e+02 6.279e+02, threshold=5.703e+02, percent-clipped=4.0 2022-12-08 17:17:03,823 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148854.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:17:25,637 INFO [train.py:873] (0/4) Epoch 20, batch 5200, loss[loss=0.1163, simple_loss=0.1509, pruned_loss=0.04082, over 11190.00 frames. ], tot_loss[loss=0.1009, simple_loss=0.1381, pruned_loss=0.03189, over 1979722.13 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 8.0 2022-12-08 17:17:25,728 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4393, 1.0882, 1.9739, 1.7830, 1.7803, 2.0230, 1.3529, 1.9698], device='cuda:0'), covar=tensor([0.1163, 0.2015, 0.0495, 0.0732, 0.1007, 0.0482, 0.1054, 0.0543], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0156, 0.0132, 0.0169, 0.0149, 0.0142, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 17:18:21,914 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148943.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:18:29,634 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2022-12-08 17:18:29,832 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.154e+02 1.963e+02 2.482e+02 3.012e+02 7.254e+02, threshold=4.965e+02, percent-clipped=1.0 2022-12-08 17:18:41,517 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7985, 1.4725, 3.8168, 1.8099, 3.7964, 3.9679, 2.9274, 4.2109], device='cuda:0'), covar=tensor([0.0291, 0.3267, 0.0459, 0.2171, 0.0515, 0.0425, 0.0890, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0162, 0.0169, 0.0167, 0.0181, 0.0133, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:18:41,759 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.41 vs. limit=5.0 2022-12-08 17:18:53,063 INFO [train.py:873] (0/4) Epoch 20, batch 5300, loss[loss=0.1095, simple_loss=0.1453, pruned_loss=0.03682, over 10369.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1379, pruned_loss=0.03114, over 1986952.89 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 4.0 2022-12-08 17:19:03,309 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=148991.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:19:16,985 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149007.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:19:36,341 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2022-12-08 17:19:47,023 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1308, 2.4771, 3.9850, 3.0388, 3.9396, 3.8535, 3.7643, 3.2612], device='cuda:0'), covar=tensor([0.1029, 0.3111, 0.0957, 0.1767, 0.0784, 0.1082, 0.1549, 0.1603], device='cuda:0'), in_proj_covar=tensor([0.0344, 0.0301, 0.0383, 0.0294, 0.0360, 0.0315, 0.0353, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:19:50,361 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5846, 3.5032, 4.1975, 3.2352, 2.4349, 3.7173, 2.1109, 3.6492], device='cuda:0'), covar=tensor([0.1119, 0.0776, 0.0618, 0.1090, 0.1871, 0.0743, 0.2686, 0.0986], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0104, 0.0098, 0.0103, 0.0116, 0.0094, 0.0117, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 17:19:53,279 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.0059, 1.8128, 4.9130, 2.6372, 4.5471, 5.0381, 4.8071, 5.4599], device='cuda:0'), covar=tensor([0.0216, 0.3266, 0.0423, 0.1934, 0.0269, 0.0420, 0.0242, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0157, 0.0162, 0.0170, 0.0167, 0.0181, 0.0134, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:19:57,355 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 2.154e+02 2.538e+02 2.985e+02 9.704e+02, threshold=5.077e+02, percent-clipped=4.0 2022-12-08 17:20:09,321 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.8924, 1.6186, 1.9706, 1.6173, 1.9401, 1.8442, 1.6466, 1.8131], device='cuda:0'), covar=tensor([0.0648, 0.1220, 0.0577, 0.0553, 0.0571, 0.0761, 0.0378, 0.0416], device='cuda:0'), in_proj_covar=tensor([0.0341, 0.0299, 0.0379, 0.0291, 0.0356, 0.0312, 0.0349, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:20:14,515 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149072.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:20:15,382 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.7281, 4.4857, 4.3219, 4.6932, 4.3227, 4.1064, 4.7489, 4.4904], device='cuda:0'), covar=tensor([0.0605, 0.0916, 0.0836, 0.0667, 0.0860, 0.0658, 0.0668, 0.0883], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0153, 0.0154, 0.0171, 0.0155, 0.0132, 0.0178, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 17:20:20,460 INFO [train.py:873] (0/4) Epoch 20, batch 5400, loss[loss=0.1128, simple_loss=0.1431, pruned_loss=0.04128, over 9478.00 frames. ], tot_loss[loss=0.09987, simple_loss=0.1376, pruned_loss=0.03108, over 1938686.74 frames. ], batch size: 100, lr: 3.90e-03, grad_scale: 2.0 2022-12-08 17:20:23,268 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.1817, 4.0255, 3.6543, 3.8652, 4.0868, 4.1395, 4.1955, 4.1821], device='cuda:0'), covar=tensor([0.0946, 0.0606, 0.2300, 0.2880, 0.0839, 0.0977, 0.0954, 0.0920], device='cuda:0'), in_proj_covar=tensor([0.0399, 0.0285, 0.0450, 0.0570, 0.0362, 0.0462, 0.0398, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:20:25,623 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149085.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:20:36,858 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.7336, 3.5677, 4.3158, 3.2397, 2.6058, 3.7397, 2.2732, 3.8673], device='cuda:0'), covar=tensor([0.0743, 0.0666, 0.0425, 0.1674, 0.1785, 0.0688, 0.2480, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0104, 0.0098, 0.0103, 0.0116, 0.0094, 0.0117, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 17:20:53,657 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2022-12-08 17:21:06,246 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2022-12-08 17:21:07,668 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149133.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:21:14,975 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5711, 3.2709, 2.5612, 3.7277, 3.5934, 3.6322, 3.1421, 2.6228], device='cuda:0'), covar=tensor([0.0794, 0.1252, 0.2986, 0.0626, 0.0826, 0.1102, 0.1267, 0.2808], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0286, 0.0258, 0.0292, 0.0323, 0.0305, 0.0257, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:21:25,134 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.118e+02 1.972e+02 2.518e+02 2.987e+02 6.531e+02, threshold=5.036e+02, percent-clipped=3.0 2022-12-08 17:21:26,127 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149154.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:21:37,047 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.2279, 4.8043, 4.6858, 5.2169, 4.8106, 4.4157, 5.1756, 4.3715], device='cuda:0'), covar=tensor([0.0363, 0.0972, 0.0463, 0.0388, 0.0852, 0.0599, 0.0525, 0.0504], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0277, 0.0205, 0.0202, 0.0187, 0.0163, 0.0293, 0.0171], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 17:21:48,144 INFO [train.py:873] (0/4) Epoch 20, batch 5500, loss[loss=0.09923, simple_loss=0.1454, pruned_loss=0.02652, over 14407.00 frames. ], tot_loss[loss=0.09928, simple_loss=0.1373, pruned_loss=0.03063, over 1961578.75 frames. ], batch size: 41, lr: 3.90e-03, grad_scale: 2.0 2022-12-08 17:22:06,461 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.3360, 1.8421, 3.4438, 2.4330, 3.3605, 1.9289, 2.4661, 3.2832], device='cuda:0'), covar=tensor([0.0854, 0.3961, 0.0641, 0.5116, 0.0816, 0.3132, 0.1661, 0.0659], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0197, 0.0222, 0.0267, 0.0239, 0.0200, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:22:07,922 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149202.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:22:14,176 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2694, 3.6747, 3.4146, 3.4712, 2.5539, 3.5874, 3.4036, 1.8734], device='cuda:0'), covar=tensor([0.1244, 0.0628, 0.0769, 0.0586, 0.0946, 0.0464, 0.0739, 0.1707], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0094, 0.0073, 0.0078, 0.0102, 0.0093, 0.0104, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0008, 0.0006, 0.0006, 0.0006, 0.0007, 0.0006, 0.0007, 0.0006], device='cuda:0') 2022-12-08 17:22:52,116 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 2.034e+02 2.619e+02 3.145e+02 6.128e+02, threshold=5.238e+02, percent-clipped=4.0 2022-12-08 17:23:14,998 INFO [train.py:873] (0/4) Epoch 20, batch 5600, loss[loss=0.09646, simple_loss=0.1346, pruned_loss=0.02918, over 14164.00 frames. ], tot_loss[loss=0.1004, simple_loss=0.1384, pruned_loss=0.03116, over 2015332.77 frames. ], batch size: 99, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:23:39,982 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149307.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:24:09,314 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1635, 2.1544, 2.4553, 1.6325, 1.7713, 2.3028, 1.3425, 2.2678], device='cuda:0'), covar=tensor([0.1126, 0.1512, 0.0836, 0.2465, 0.2424, 0.0913, 0.3187, 0.0939], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0104, 0.0098, 0.0103, 0.0116, 0.0094, 0.0118, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0005, 0.0004, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2022-12-08 17:24:19,494 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 1.989e+02 2.324e+02 2.779e+02 6.746e+02, threshold=4.648e+02, percent-clipped=1.0 2022-12-08 17:24:21,361 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149355.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:24:36,169 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149372.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:24:42,617 INFO [train.py:873] (0/4) Epoch 20, batch 5700, loss[loss=0.08227, simple_loss=0.1171, pruned_loss=0.02374, over 3882.00 frames. ], tot_loss[loss=0.0987, simple_loss=0.1371, pruned_loss=0.03015, over 2029390.44 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:25:09,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.4205, 1.4448, 3.4665, 1.5805, 3.2768, 3.5433, 2.4880, 3.7247], device='cuda:0'), covar=tensor([0.0256, 0.3214, 0.0436, 0.2300, 0.0865, 0.0421, 0.0916, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0158, 0.0163, 0.0171, 0.0168, 0.0182, 0.0134, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:25:18,478 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=149420.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:25:31,695 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.5755, 2.0365, 3.5950, 2.6045, 3.5262, 1.9818, 2.7496, 3.4994], device='cuda:0'), covar=tensor([0.0678, 0.3664, 0.0612, 0.4402, 0.0692, 0.3078, 0.1364, 0.0653], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0197, 0.0224, 0.0267, 0.0239, 0.0202, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:25:36,217 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2022-12-08 17:25:37,658 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.0238, 3.5550, 2.8452, 4.2607, 3.9737, 4.1604, 3.6690, 2.9538], device='cuda:0'), covar=tensor([0.0769, 0.1154, 0.2829, 0.0550, 0.0814, 0.1147, 0.1046, 0.2866], device='cuda:0'), in_proj_covar=tensor([0.0282, 0.0286, 0.0257, 0.0291, 0.0322, 0.0304, 0.0256, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:25:47,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 7.719e+01 2.141e+02 2.467e+02 2.944e+02 6.062e+02, threshold=4.935e+02, percent-clipped=6.0 2022-12-08 17:26:10,322 INFO [train.py:873] (0/4) Epoch 20, batch 5800, loss[loss=0.09038, simple_loss=0.1363, pruned_loss=0.02223, over 14098.00 frames. ], tot_loss[loss=0.09895, simple_loss=0.1372, pruned_loss=0.03037, over 1979946.76 frames. ], batch size: 29, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:26:40,936 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149514.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:27:07,380 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6404, 2.3589, 4.6901, 3.1935, 4.4172, 2.1943, 3.5250, 4.5595], device='cuda:0'), covar=tensor([0.0534, 0.3581, 0.0332, 0.5615, 0.0541, 0.3052, 0.1360, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0196, 0.0224, 0.0266, 0.0240, 0.0201, 0.0201, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:27:15,062 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.003e+02 2.561e+02 2.963e+02 6.960e+02, threshold=5.123e+02, percent-clipped=3.0 2022-12-08 17:27:25,701 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149565.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:27:34,508 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149575.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:27:38,142 INFO [train.py:873] (0/4) Epoch 20, batch 5900, loss[loss=0.1437, simple_loss=0.1335, pruned_loss=0.07697, over 1245.00 frames. ], tot_loss[loss=0.09925, simple_loss=0.1371, pruned_loss=0.03071, over 1940629.24 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 4.0 2022-12-08 17:28:20,671 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149626.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:28:43,779 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.112e+02 2.459e+02 3.260e+02 5.779e+02, threshold=4.919e+02, percent-clipped=5.0 2022-12-08 17:29:07,359 INFO [train.py:873] (0/4) Epoch 20, batch 6000, loss[loss=0.1173, simple_loss=0.1479, pruned_loss=0.04339, over 10310.00 frames. ], tot_loss[loss=0.1001, simple_loss=0.1374, pruned_loss=0.03135, over 1870733.45 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:29:07,359 INFO [train.py:896] (0/4) Computing validation loss 2022-12-08 17:29:22,627 INFO [train.py:905] (0/4) Epoch 20, validation: loss=0.1445, simple_loss=0.1806, pruned_loss=0.0542, over 857387.00 frames. 2022-12-08 17:29:22,627 INFO [train.py:906] (0/4) Maximum memory allocated so far is 18037MB 2022-12-08 17:29:40,843 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.2790, 2.2008, 3.2076, 3.3994, 3.1699, 2.1591, 3.2149, 2.5295], device='cuda:0'), covar=tensor([0.0595, 0.1401, 0.0963, 0.0599, 0.0755, 0.2192, 0.0559, 0.1180], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0260, 0.0375, 0.0331, 0.0271, 0.0307, 0.0313, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2022-12-08 17:29:43,998 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2022-12-08 17:30:05,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2022-12-08 17:30:21,962 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149746.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:30:27,889 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 2.065e+02 2.623e+02 3.098e+02 9.406e+02, threshold=5.245e+02, percent-clipped=5.0 2022-12-08 17:30:50,977 INFO [train.py:873] (0/4) Epoch 20, batch 6100, loss[loss=0.1093, simple_loss=0.143, pruned_loss=0.03774, over 13530.00 frames. ], tot_loss[loss=0.0995, simple_loss=0.137, pruned_loss=0.03098, over 1886892.84 frames. ], batch size: 100, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:31:16,080 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149807.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 17:31:21,315 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.6884, 3.2275, 3.1203, 2.1331, 3.1918, 3.4200, 3.7595, 2.7803], device='cuda:0'), covar=tensor([0.0579, 0.1197, 0.0860, 0.1466, 0.0764, 0.0540, 0.0718, 0.1117], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0167, 0.0140, 0.0124, 0.0144, 0.0155, 0.0139, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 17:31:28,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2022-12-08 17:31:28,280 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2022-12-08 17:31:54,671 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.52 vs. limit=5.0 2022-12-08 17:31:56,635 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.144e+02 2.583e+02 3.219e+02 5.785e+02, threshold=5.165e+02, percent-clipped=1.0 2022-12-08 17:32:11,593 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149870.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:32:19,290 INFO [train.py:873] (0/4) Epoch 20, batch 6200, loss[loss=0.1059, simple_loss=0.149, pruned_loss=0.03137, over 14413.00 frames. ], tot_loss[loss=0.1003, simple_loss=0.1375, pruned_loss=0.03154, over 1844956.88 frames. ], batch size: 41, lr: 3.89e-03, grad_scale: 8.0 2022-12-08 17:32:55,078 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.8503, 5.8042, 5.4457, 5.7966, 5.4475, 5.3518, 5.8868, 5.5369], device='cuda:0'), covar=tensor([0.0586, 0.0574, 0.0724, 0.0688, 0.0774, 0.0549, 0.0548, 0.0784], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0149, 0.0150, 0.0168, 0.0152, 0.0129, 0.0174, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:0') 2022-12-08 17:32:56,981 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149921.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:33:03,264 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1626, 1.9969, 2.0746, 2.1665, 2.0638, 2.0771, 2.2060, 1.8371], device='cuda:0'), covar=tensor([0.0942, 0.1268, 0.0796, 0.0860, 0.0980, 0.0698, 0.0975, 0.0735], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0270, 0.0200, 0.0198, 0.0184, 0.0158, 0.0287, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 17:33:24,558 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.165e+02 2.069e+02 2.540e+02 3.004e+02 8.039e+02, threshold=5.080e+02, percent-clipped=2.0 2022-12-08 17:33:47,751 INFO [train.py:873] (0/4) Epoch 20, batch 6300, loss[loss=0.08942, simple_loss=0.1356, pruned_loss=0.0216, over 13995.00 frames. ], tot_loss[loss=0.09988, simple_loss=0.1377, pruned_loss=0.03103, over 1899588.70 frames. ], batch size: 26, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:34:06,628 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/checkpoint-150000.pt 2022-12-08 17:34:20,243 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.6901, 4.4674, 4.1072, 4.3261, 4.4735, 4.5936, 4.6948, 4.6597], device='cuda:0'), covar=tensor([0.0758, 0.0440, 0.1992, 0.2236, 0.0755, 0.0742, 0.0717, 0.0732], device='cuda:0'), in_proj_covar=tensor([0.0395, 0.0282, 0.0448, 0.0571, 0.0360, 0.0461, 0.0396, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:34:55,927 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2022-12-08 17:34:56,193 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 2.076e+02 2.693e+02 3.406e+02 8.300e+02, threshold=5.385e+02, percent-clipped=3.0 2022-12-08 17:35:08,378 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.2690, 2.5651, 4.1323, 3.0771, 4.1279, 4.0013, 3.9271, 3.4607], device='cuda:0'), covar=tensor([0.0757, 0.3206, 0.0810, 0.1880, 0.0748, 0.0954, 0.1435, 0.1699], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0307, 0.0388, 0.0299, 0.0363, 0.0321, 0.0359, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:35:18,742 INFO [train.py:873] (0/4) Epoch 20, batch 6400, loss[loss=0.1118, simple_loss=0.148, pruned_loss=0.03782, over 14326.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1384, pruned_loss=0.03165, over 1903632.49 frames. ], batch size: 55, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:35:18,923 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150079.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:35:39,417 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150102.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 17:35:54,351 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9037, 1.5141, 3.0889, 1.6228, 3.1640, 3.0170, 2.2899, 3.2464], device='cuda:0'), covar=tensor([0.0289, 0.2926, 0.0417, 0.2092, 0.0368, 0.0505, 0.1015, 0.0241], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0158, 0.0162, 0.0171, 0.0168, 0.0182, 0.0135, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:36:12,190 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150140.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:36:23,032 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.042e+02 2.543e+02 3.172e+02 7.027e+02, threshold=5.086e+02, percent-clipped=3.0 2022-12-08 17:36:34,128 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.8629, 3.7329, 3.4142, 2.7189, 3.2934, 3.5948, 3.9859, 3.2965], device='cuda:0'), covar=tensor([0.0618, 0.0879, 0.0795, 0.1043, 0.0840, 0.0567, 0.0645, 0.0866], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0166, 0.0139, 0.0123, 0.0144, 0.0155, 0.0138, 0.0142], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 17:36:38,278 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150170.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:36:46,071 INFO [train.py:873] (0/4) Epoch 20, batch 6500, loss[loss=0.09966, simple_loss=0.1374, pruned_loss=0.03093, over 14323.00 frames. ], tot_loss[loss=0.1004, simple_loss=0.138, pruned_loss=0.0314, over 1896562.02 frames. ], batch size: 60, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:37:20,279 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150218.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:37:22,112 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9687, 1.9631, 2.0825, 2.1064, 2.0522, 1.5665, 1.3850, 1.8240], device='cuda:0'), covar=tensor([0.0810, 0.0691, 0.0501, 0.0411, 0.0492, 0.1555, 0.2192, 0.0566], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0177, 0.0149, 0.0151, 0.0211, 0.0145, 0.0158, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2022-12-08 17:37:23,051 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150221.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:37:33,267 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.7774, 1.3951, 2.9099, 1.6468, 3.0065, 2.8984, 2.1989, 3.1218], device='cuda:0'), covar=tensor([0.0292, 0.2961, 0.0442, 0.1935, 0.0386, 0.0536, 0.1143, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0158, 0.0163, 0.0170, 0.0168, 0.0181, 0.0134, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:37:50,499 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.155e+02 2.479e+02 3.022e+02 7.876e+02, threshold=4.957e+02, percent-clipped=4.0 2022-12-08 17:38:04,353 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150269.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:38:09,014 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150274.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:38:13,122 INFO [train.py:873] (0/4) Epoch 20, batch 6600, loss[loss=0.1102, simple_loss=0.1475, pruned_loss=0.03642, over 14272.00 frames. ], tot_loss[loss=0.09979, simple_loss=0.1373, pruned_loss=0.03113, over 1863345.42 frames. ], batch size: 57, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:38:41,973 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2022-12-08 17:39:02,535 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150335.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:39:18,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.127e+02 2.141e+02 2.452e+02 3.103e+02 5.286e+02, threshold=4.904e+02, percent-clipped=1.0 2022-12-08 17:39:41,069 INFO [train.py:873] (0/4) Epoch 20, batch 6700, loss[loss=0.08184, simple_loss=0.1252, pruned_loss=0.01926, over 14234.00 frames. ], tot_loss[loss=0.09955, simple_loss=0.1375, pruned_loss=0.03078, over 1944980.04 frames. ], batch size: 69, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:40:01,131 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150402.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:40:29,783 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150435.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:40:42,819 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150450.0, num_to_drop=1, layers_to_drop={0} 2022-12-08 17:40:45,137 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.053e+02 2.408e+02 2.928e+02 1.041e+03, threshold=4.817e+02, percent-clipped=2.0 2022-12-08 17:41:07,584 INFO [train.py:873] (0/4) Epoch 20, batch 6800, loss[loss=0.1751, simple_loss=0.1545, pruned_loss=0.09783, over 1333.00 frames. ], tot_loss[loss=0.1003, simple_loss=0.1381, pruned_loss=0.03122, over 1961440.54 frames. ], batch size: 100, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:41:13,141 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.3023, 2.6347, 5.3122, 3.7105, 5.0265, 2.4430, 3.9802, 5.1335], device='cuda:0'), covar=tensor([0.0361, 0.3214, 0.0242, 0.4827, 0.0434, 0.2852, 0.1106, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0196, 0.0223, 0.0264, 0.0240, 0.0199, 0.0202, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:41:57,973 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150537.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:42:12,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 2.195e+02 2.476e+02 3.043e+02 9.162e+02, threshold=4.952e+02, percent-clipped=3.0 2022-12-08 17:42:22,493 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.4700, 1.4107, 1.5368, 1.3280, 1.3193, 1.2568, 1.3143, 1.0708], device='cuda:0'), covar=tensor([0.0213, 0.0298, 0.0209, 0.0229, 0.0227, 0.0359, 0.0237, 0.0375], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0034], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 17:42:34,233 INFO [train.py:873] (0/4) Epoch 20, batch 6900, loss[loss=0.1081, simple_loss=0.1551, pruned_loss=0.03058, over 14466.00 frames. ], tot_loss[loss=0.1009, simple_loss=0.1386, pruned_loss=0.03164, over 1935499.20 frames. ], batch size: 51, lr: 3.88e-03, grad_scale: 8.0 2022-12-08 17:42:36,925 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([5.8988, 5.3746, 5.2677, 5.8480, 5.3698, 4.6731, 5.7832, 4.8556], device='cuda:0'), covar=tensor([0.0248, 0.0761, 0.0343, 0.0313, 0.0862, 0.0337, 0.0383, 0.0436], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0273, 0.0202, 0.0198, 0.0184, 0.0159, 0.0288, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0004, 0.0002], device='cuda:0') 2022-12-08 17:42:51,061 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150598.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:42:57,073 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.15 vs. limit=5.0 2022-12-08 17:43:11,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2022-12-08 17:43:19,283 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150630.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:43:25,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2022-12-08 17:43:40,041 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 1.852e+02 2.409e+02 3.250e+02 6.967e+02, threshold=4.817e+02, percent-clipped=2.0 2022-12-08 17:43:58,214 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2022-12-08 17:44:01,999 INFO [train.py:873] (0/4) Epoch 20, batch 7000, loss[loss=0.1025, simple_loss=0.142, pruned_loss=0.03148, over 14585.00 frames. ], tot_loss[loss=0.1003, simple_loss=0.1381, pruned_loss=0.03128, over 1952779.59 frames. ], batch size: 33, lr: 3.88e-03, grad_scale: 4.0 2022-12-08 17:44:47,777 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.9731, 1.6986, 3.0936, 2.3160, 2.9902, 1.8035, 2.3474, 3.0356], device='cuda:0'), covar=tensor([0.1017, 0.3644, 0.0778, 0.3317, 0.0867, 0.2713, 0.1392, 0.0801], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0197, 0.0225, 0.0265, 0.0240, 0.0199, 0.0202, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:44:48,588 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150732.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:44:50,826 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150735.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:45:06,716 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.298e+02 2.589e+02 3.066e+02 6.150e+02, threshold=5.179e+02, percent-clipped=4.0 2022-12-08 17:45:20,812 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.3160, 0.9521, 1.1921, 0.7876, 1.0730, 1.3305, 1.0461, 1.1063], device='cuda:0'), covar=tensor([0.0544, 0.1239, 0.0900, 0.0711, 0.1318, 0.1066, 0.0970, 0.1566], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0168, 0.0141, 0.0125, 0.0145, 0.0157, 0.0140, 0.0143], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0007, 0.0006, 0.0005, 0.0006, 0.0007, 0.0006, 0.0006], device='cuda:0') 2022-12-08 17:45:28,338 INFO [train.py:873] (0/4) Epoch 20, batch 7100, loss[loss=0.1146, simple_loss=0.1411, pruned_loss=0.04408, over 6946.00 frames. ], tot_loss[loss=0.09982, simple_loss=0.1376, pruned_loss=0.03103, over 1938143.78 frames. ], batch size: 100, lr: 3.87e-03, grad_scale: 4.0 2022-12-08 17:45:31,477 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150783.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:45:40,436 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150793.0, num_to_drop=1, layers_to_drop={1} 2022-12-08 17:46:11,484 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([4.5716, 1.6408, 4.3081, 2.1478, 4.3651, 4.5797, 4.0683, 4.9693], device='cuda:0'), covar=tensor([0.0204, 0.3078, 0.0439, 0.2062, 0.0343, 0.0450, 0.0407, 0.0152], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0158, 0.0163, 0.0170, 0.0167, 0.0182, 0.0134, 0.0155], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2022-12-08 17:46:33,853 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.271e+02 2.646e+02 3.296e+02 2.153e+03, threshold=5.293e+02, percent-clipped=6.0 2022-12-08 17:46:55,668 INFO [train.py:873] (0/4) Epoch 20, batch 7200, loss[loss=0.09556, simple_loss=0.1357, pruned_loss=0.02772, over 14274.00 frames. ], tot_loss[loss=0.1018, simple_loss=0.1388, pruned_loss=0.03247, over 1970650.59 frames. ], batch size: 18, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:47:07,889 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150893.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:47:14,192 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150900.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:47:40,758 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150930.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:48:01,352 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 1.977e+02 2.439e+02 3.091e+02 5.350e+02, threshold=4.878e+02, percent-clipped=1.0 2022-12-08 17:48:07,491 INFO [zipformer.py:626] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150961.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:48:22,740 INFO [zipformer.py:626] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=150978.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:48:23,468 INFO [train.py:873] (0/4) Epoch 20, batch 7300, loss[loss=0.1137, simple_loss=0.1193, pruned_loss=0.05408, over 1246.00 frames. ], tot_loss[loss=0.09984, simple_loss=0.1375, pruned_loss=0.0311, over 1993229.15 frames. ], batch size: 100, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:49:23,623 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.9173, 1.6907, 4.3943, 4.0580, 3.9737, 4.5133, 4.0307, 4.4787], device='cuda:0'), covar=tensor([0.1666, 0.1650, 0.0125, 0.0241, 0.0281, 0.0151, 0.0199, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0168, 0.0149, 0.0143, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 17:49:29,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.047e+02 2.139e+02 2.523e+02 3.177e+02 9.954e+02, threshold=5.046e+02, percent-clipped=7.0 2022-12-08 17:49:37,223 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([2.1821, 2.4121, 2.1701, 2.3330, 2.3159, 2.1584, 1.9592, 1.9271], device='cuda:0'), covar=tensor([0.0269, 0.0313, 0.0353, 0.0218, 0.0301, 0.0388, 0.0372, 0.0345], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0024, 0.0022, 0.0024, 0.0023, 0.0036, 0.0030, 0.0035], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:0') 2022-12-08 17:49:51,031 INFO [train.py:873] (0/4) Epoch 20, batch 7400, loss[loss=0.0798, simple_loss=0.1279, pruned_loss=0.01587, over 14224.00 frames. ], tot_loss[loss=0.1008, simple_loss=0.1378, pruned_loss=0.03188, over 1908828.94 frames. ], batch size: 35, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:49:59,285 INFO [zipformer.py:626] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151088.0, num_to_drop=1, layers_to_drop={3} 2022-12-08 17:50:04,335 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([1.7687, 1.3765, 2.4878, 2.2127, 2.3491, 2.5126, 1.6970, 2.5279], device='cuda:0'), covar=tensor([0.1099, 0.1559, 0.0251, 0.0581, 0.0656, 0.0303, 0.0883, 0.0299], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0157, 0.0132, 0.0168, 0.0149, 0.0142, 0.0125, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2022-12-08 17:50:50,234 INFO [zipformer.py:1414] (0/4) attn_weights_entropy = tensor([3.0039, 1.8011, 3.1345, 2.2669, 3.0434, 1.8301, 2.3728, 3.0986], device='cuda:0'), covar=tensor([0.0863, 0.3674, 0.0690, 0.4218, 0.0893, 0.2722, 0.1467, 0.0674], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0195, 0.0223, 0.0264, 0.0239, 0.0199, 0.0202, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005], device='cuda:0') 2022-12-08 17:50:55,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.036e+02 2.135e+02 2.491e+02 3.146e+02 1.107e+03, threshold=4.982e+02, percent-clipped=6.0 2022-12-08 17:51:11,981 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2022-12-08 17:51:17,606 INFO [train.py:873] (0/4) Epoch 20, batch 7500, loss[loss=0.1026, simple_loss=0.1389, pruned_loss=0.03315, over 14082.00 frames. ], tot_loss[loss=0.101, simple_loss=0.138, pruned_loss=0.03195, over 1943961.92 frames. ], batch size: 29, lr: 3.87e-03, grad_scale: 8.0 2022-12-08 17:51:21,927 INFO [zipformer.py:626] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151184.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:51:29,972 INFO [zipformer.py:626] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151193.0, num_to_drop=0, layers_to_drop=set() 2022-12-08 17:52:04,539 INFO [checkpoint.py:75] (0/4) Saving checkpoint to pruned_transducer_stateless7/exp/v1/epoch-20.pt 2022-12-08 17:52:12,943 INFO [train.py:1091] (0/4) Done!