2023-04-03 17:22:38,516 INFO [decode.py:659] Decoding started
2023-04-03 17:22:38,516 INFO [decode.py:665] Device: cuda:0
2023-04-03 17:22:38,518 INFO [decode.py:675] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '62e404dd3f3a811d73e424199b3408e309c06e1a', 'k2-git-date': 'Mon Jan 30 02:26:16 2023', 'lhotse-version': '1.12.0.dev+git.3ccfeb7.clean', 'torch-version': '1.13.0', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.8', 'icefall-git-branch': 'master', 'icefall-git-sha1': 'd74822d-dirty', 'icefall-git-date': 'Tue Mar 21 21:35:32 2023', 'icefall-path': '/home/lishaojie/icefall', 'k2-path': '/home/lishaojie/.conda/envs/env_lishaojie/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/home/lishaojie/.conda/envs/env_lishaojie/lib/python3.8/site-packages/lhotse/__init__.py', 'hostname': 'cnc533', 'IP address': '127.0.1.1'}, 'epoch': 29, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming/exp1'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'modified_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 64, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 200, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'res_dir': PosixPath('pruned_transducer_stateless7_streaming/exp1/modified_beam_search'), 'suffix': 'epoch-29-avg-9-streaming-chunk-size-64-modified_beam_search-beam-size-4-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
2023-04-03 17:22:38,519 INFO [decode.py:677] About to create model
2023-04-03 17:22:38,918 INFO [zipformer.py:405] 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.
2023-04-03 17:22:38,925 INFO [decode.py:748] Calculating the averaged model over epoch range from 20 (excluded) to 29
2023-04-03 17:22:40,997 INFO [decode.py:782] Number of model parameters: 70369391
2023-04-03 17:22:40,997 INFO [commonvoice_fr.py:406] About to get test cuts
2023-04-03 17:22:44,389 INFO [decode.py:560] batch 0/?, cuts processed until now is 27
2023-04-03 17:23:03,653 INFO [zipformer.py:2441] attn_weights_entropy = tensor([2.4396, 2.2447, 2.4559, 1.5968, 2.3244, 2.4744, 2.4945, 1.9745],
       device='cuda:0'), covar=tensor([0.0486, 0.0604, 0.0566, 0.0803, 0.0805, 0.0564, 0.0533, 0.1100],
       device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0133, 0.0136, 0.0116, 0.0123, 0.0135, 0.0136, 0.0158],
       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')
2023-04-03 17:23:05,745 INFO [decode.py:560] batch 20/?, cuts processed until now is 604
2023-04-03 17:23:23,592 INFO [zipformer.py:2441] attn_weights_entropy = tensor([2.0329, 1.8940, 1.7735, 2.1519, 2.5543, 2.1581, 1.8200, 1.6895],
       device='cuda:0'), covar=tensor([0.2199, 0.2101, 0.1994, 0.1657, 0.1436, 0.1120, 0.2161, 0.2030],
       device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0208, 0.0212, 0.0195, 0.0242, 0.0187, 0.0214, 0.0202],
       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')
2023-04-03 17:23:26,475 INFO [decode.py:560] batch 40/?, cuts processed until now is 1209
2023-04-03 17:23:41,645 INFO [zipformer.py:2441] attn_weights_entropy = tensor([2.7772, 2.7137, 2.1413, 1.0557, 2.3674, 2.2655, 2.0341, 2.4959],
       device='cuda:0'), covar=tensor([0.0948, 0.0634, 0.1572, 0.1977, 0.1178, 0.2220, 0.2033, 0.0825],
       device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0187, 0.0196, 0.0178, 0.0206, 0.0207, 0.0220, 0.0192],
       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')
2023-04-03 17:23:46,667 INFO [decode.py:560] batch 60/?, cuts processed until now is 1866
2023-04-03 17:24:07,713 INFO [decode.py:560] batch 80/?, cuts processed until now is 2422
2023-04-03 17:24:10,930 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.5944, 1.6332, 1.4009, 1.6980, 2.0331, 1.9103, 1.6203, 1.4564],
       device='cuda:0'), covar=tensor([0.0370, 0.0328, 0.0636, 0.0290, 0.0200, 0.0398, 0.0360, 0.0438],
       device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0103, 0.0143, 0.0108, 0.0097, 0.0111, 0.0100, 0.0110],
       device='cuda:0'), out_proj_covar=tensor([7.4944e-05, 7.9098e-05, 1.1173e-04, 8.2734e-05, 7.5248e-05, 8.1783e-05,
        7.3728e-05, 8.3511e-05], device='cuda:0')
2023-04-03 17:24:19,887 INFO [zipformer.py:2441] attn_weights_entropy = tensor([2.5455, 2.5531, 2.0976, 0.9939, 2.3244, 2.0391, 1.9380, 2.4056],
       device='cuda:0'), covar=tensor([0.0949, 0.0639, 0.1516, 0.2032, 0.1282, 0.2361, 0.2408, 0.0816],
       device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0187, 0.0196, 0.0178, 0.0206, 0.0207, 0.0220, 0.0192],
       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')
2023-04-03 17:24:27,881 INFO [decode.py:560] batch 100/?, cuts processed until now is 3088
2023-04-03 17:24:48,430 INFO [decode.py:560] batch 120/?, cuts processed until now is 3672
2023-04-03 17:25:08,496 INFO [decode.py:560] batch 140/?, cuts processed until now is 4348
2023-04-03 17:25:28,540 INFO [decode.py:560] batch 160/?, cuts processed until now is 5035
2023-04-03 17:25:40,611 INFO [zipformer.py:2441] attn_weights_entropy = tensor([2.4775, 2.3269, 2.5159, 1.6390, 2.4735, 2.5963, 2.5525, 2.0379],
       device='cuda:0'), covar=tensor([0.0466, 0.0579, 0.0534, 0.0716, 0.0834, 0.0513, 0.0447, 0.1035],
       device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0133, 0.0136, 0.0116, 0.0123, 0.0135, 0.0136, 0.0158],
       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')
2023-04-03 17:25:48,922 INFO [decode.py:560] batch 180/?, cuts processed until now is 5674
2023-04-03 17:26:09,282 INFO [decode.py:560] batch 200/?, cuts processed until now is 6301
2023-04-03 17:26:29,657 INFO [decode.py:560] batch 220/?, cuts processed until now is 6914
2023-04-03 17:26:50,151 INFO [decode.py:560] batch 240/?, cuts processed until now is 7540
2023-04-03 17:27:10,601 INFO [decode.py:560] batch 260/?, cuts processed until now is 8161
2023-04-03 17:27:30,848 INFO [decode.py:560] batch 280/?, cuts processed until now is 8857
2023-04-03 17:27:50,769 INFO [decode.py:560] batch 300/?, cuts processed until now is 9574
2023-04-03 17:28:11,398 INFO [decode.py:560] batch 320/?, cuts processed until now is 10169
2023-04-03 17:28:25,548 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.8605, 1.7340, 2.4232, 3.4970, 2.3920, 2.5277, 1.1464, 2.9314],
       device='cuda:0'), covar=tensor([0.1522, 0.1223, 0.1074, 0.0445, 0.0713, 0.1245, 0.1648, 0.0431],
       device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0114, 0.0131, 0.0162, 0.0098, 0.0133, 0.0122, 0.0098],
       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0003],
       device='cuda:0')
2023-04-03 17:28:31,696 INFO [decode.py:560] batch 340/?, cuts processed until now is 10810
2023-04-03 17:28:52,180 INFO [decode.py:560] batch 360/?, cuts processed until now is 11452
2023-04-03 17:29:12,247 INFO [decode.py:560] batch 380/?, cuts processed until now is 12133
2023-04-03 17:29:33,066 INFO [decode.py:560] batch 400/?, cuts processed until now is 12706
2023-04-03 17:29:53,675 INFO [decode.py:560] batch 420/?, cuts processed until now is 13299
2023-04-03 17:30:07,997 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.8529, 1.7486, 1.5706, 1.8211, 2.1766, 2.0610, 1.7289, 1.5829],
       device='cuda:0'), covar=tensor([0.0370, 0.0322, 0.0575, 0.0276, 0.0210, 0.0417, 0.0338, 0.0404],
       device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0103, 0.0143, 0.0108, 0.0097, 0.0111, 0.0100, 0.0110],
       device='cuda:0'), out_proj_covar=tensor([7.4944e-05, 7.9098e-05, 1.1173e-04, 8.2734e-05, 7.5248e-05, 8.1783e-05,
        7.3728e-05, 8.3511e-05], device='cuda:0')
2023-04-03 17:30:13,957 INFO [decode.py:560] batch 440/?, cuts processed until now is 13891
2023-04-03 17:30:34,408 INFO [decode.py:560] batch 460/?, cuts processed until now is 14515
2023-04-03 17:30:54,607 INFO [decode.py:560] batch 480/?, cuts processed until now is 15158
2023-04-03 17:31:14,650 INFO [decode.py:560] batch 500/?, cuts processed until now is 15743
2023-04-03 17:31:18,578 INFO [decode.py:576] The transcripts are stored in pruned_transducer_stateless7_streaming/exp1/modified_beam_search/recogs-test-cv-beam_size_4-epoch-29-avg-9-streaming-chunk-size-64-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-03 17:31:18,889 INFO [utils.py:558] [test-cv-beam_size_4] %WER 10.19% [15988 / 156915, 1250 ins, 1549 del, 13189 sub ]
2023-04-03 17:31:19,408 INFO [decode.py:589] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp1/modified_beam_search/errs-test-cv-beam_size_4-epoch-29-avg-9-streaming-chunk-size-64-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-03 17:31:19,408 INFO [decode.py:609] 
For test-cv, WER of different settings are:
beam_size_4	10.19	best for test-cv

2023-04-03 17:31:19,408 INFO [decode.py:808] Done!