Commit
·
82bb052
1
Parent(s):
9e6a862
Upload 9 files
Browse files- config.json +116 -0
- optimizer.pt +3 -0
- preprocessor_config.json +10 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2362 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,116 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "model",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"adapter_kernel_size": 3,
|
| 5 |
+
"adapter_stride": 2,
|
| 6 |
+
"add_adapter": false,
|
| 7 |
+
"apply_spec_augment": true,
|
| 8 |
+
"architectures": [
|
| 9 |
+
"Wav2Vec2ForCTC"
|
| 10 |
+
],
|
| 11 |
+
"attention_dropout": 0.1,
|
| 12 |
+
"bos_token_id": 1,
|
| 13 |
+
"classifier_proj_size": 256,
|
| 14 |
+
"codevector_dim": 768,
|
| 15 |
+
"contrastive_logits_temperature": 0.1,
|
| 16 |
+
"conv_bias": true,
|
| 17 |
+
"conv_dim": [
|
| 18 |
+
512,
|
| 19 |
+
512,
|
| 20 |
+
512,
|
| 21 |
+
512,
|
| 22 |
+
512,
|
| 23 |
+
512,
|
| 24 |
+
512
|
| 25 |
+
],
|
| 26 |
+
"conv_kernel": [
|
| 27 |
+
10,
|
| 28 |
+
3,
|
| 29 |
+
3,
|
| 30 |
+
3,
|
| 31 |
+
3,
|
| 32 |
+
2,
|
| 33 |
+
2
|
| 34 |
+
],
|
| 35 |
+
"conv_stride": [
|
| 36 |
+
5,
|
| 37 |
+
2,
|
| 38 |
+
2,
|
| 39 |
+
2,
|
| 40 |
+
2,
|
| 41 |
+
2,
|
| 42 |
+
2
|
| 43 |
+
],
|
| 44 |
+
"ctc_loss_reduction": "mean",
|
| 45 |
+
"ctc_zero_infinity": false,
|
| 46 |
+
"diversity_loss_weight": 0.1,
|
| 47 |
+
"do_stable_layer_norm": true,
|
| 48 |
+
"eos_token_id": 2,
|
| 49 |
+
"feat_extract_activation": "gelu",
|
| 50 |
+
"feat_extract_dropout": 0.0,
|
| 51 |
+
"feat_extract_norm": "layer",
|
| 52 |
+
"feat_proj_dropout": 0.0,
|
| 53 |
+
"feat_quantizer_dropout": 0.0,
|
| 54 |
+
"final_dropout": 0.0,
|
| 55 |
+
"gradient_checkpointing": false,
|
| 56 |
+
"hidden_act": "gelu",
|
| 57 |
+
"hidden_dropout": 0.1,
|
| 58 |
+
"hidden_size": 1024,
|
| 59 |
+
"initializer_range": 0.02,
|
| 60 |
+
"intermediate_size": 4096,
|
| 61 |
+
"layer_norm_eps": 1e-05,
|
| 62 |
+
"layerdrop": 0.1,
|
| 63 |
+
"mask_channel_length": 10,
|
| 64 |
+
"mask_channel_min_space": 1,
|
| 65 |
+
"mask_channel_other": 0.0,
|
| 66 |
+
"mask_channel_prob": 0.0,
|
| 67 |
+
"mask_channel_selection": "static",
|
| 68 |
+
"mask_feature_length": 10,
|
| 69 |
+
"mask_feature_min_masks": 0,
|
| 70 |
+
"mask_feature_prob": 0.0,
|
| 71 |
+
"mask_time_length": 10,
|
| 72 |
+
"mask_time_min_masks": 2,
|
| 73 |
+
"mask_time_min_space": 1,
|
| 74 |
+
"mask_time_other": 0.0,
|
| 75 |
+
"mask_time_prob": 0.05,
|
| 76 |
+
"mask_time_selection": "static",
|
| 77 |
+
"model_type": "wav2vec2",
|
| 78 |
+
"num_adapter_layers": 3,
|
| 79 |
+
"num_attention_heads": 16,
|
| 80 |
+
"num_codevector_groups": 2,
|
| 81 |
+
"num_codevectors_per_group": 320,
|
| 82 |
+
"num_conv_pos_embedding_groups": 16,
|
| 83 |
+
"num_conv_pos_embeddings": 128,
|
| 84 |
+
"num_feat_extract_layers": 7,
|
| 85 |
+
"num_hidden_layers": 24,
|
| 86 |
+
"num_negatives": 100,
|
| 87 |
+
"output_hidden_size": 1024,
|
| 88 |
+
"pad_token_id": 93,
|
| 89 |
+
"proj_codevector_dim": 768,
|
| 90 |
+
"tdnn_dilation": [
|
| 91 |
+
1,
|
| 92 |
+
2,
|
| 93 |
+
3,
|
| 94 |
+
1,
|
| 95 |
+
1
|
| 96 |
+
],
|
| 97 |
+
"tdnn_dim": [
|
| 98 |
+
512,
|
| 99 |
+
512,
|
| 100 |
+
512,
|
| 101 |
+
512,
|
| 102 |
+
1500
|
| 103 |
+
],
|
| 104 |
+
"tdnn_kernel": [
|
| 105 |
+
5,
|
| 106 |
+
3,
|
| 107 |
+
3,
|
| 108 |
+
1,
|
| 109 |
+
1
|
| 110 |
+
],
|
| 111 |
+
"torch_dtype": "float32",
|
| 112 |
+
"transformers_version": "4.29.2",
|
| 113 |
+
"use_weighted_layer_sum": false,
|
| 114 |
+
"vocab_size": 96,
|
| 115 |
+
"xvector_output_dim": 512
|
| 116 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f68ac63d56a47945a7688752e70762bbeeb447537b3d49037e523351308bfa09
|
| 3 |
+
size 2490846033
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
| 4 |
+
"feature_size": 1,
|
| 5 |
+
"padding_side": "right",
|
| 6 |
+
"padding_value": 0.0,
|
| 7 |
+
"processor_class": "Wav2Vec2Processor",
|
| 8 |
+
"return_attention_mask": true,
|
| 9 |
+
"sampling_rate": 16000
|
| 10 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac0438124578379c07b70504e3079bb3446baa584b0665c2751fc1978beb0866
|
| 3 |
+
size 1262292337
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1386a7d62a2d4585741a3d47d7739851b325e4d48271d2db7f8de2334a6d2169
|
| 3 |
+
size 14567
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:04383622ae6e29f4901cb49db405fc81308549a4f4d9e3998ababef6db2d00c2
|
| 3 |
+
size 559
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd9cb17b6408aeecb1d9c148323209be7a91a20ae52158608cae1b06c4bd28ff
|
| 3 |
+
size 559
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2362 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3557792992347966,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-xlsr53-TH-cmv-ckp-farfield-40epochs/checkpoint-33000",
|
| 4 |
+
"epoch": 40.0,
|
| 5 |
+
"global_step": 34000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.12,
|
| 12 |
+
"learning_rate": 2.8528887336629845e-06,
|
| 13 |
+
"loss": 37.8606,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.24,
|
| 18 |
+
"learning_rate": 5.73518869138435e-06,
|
| 19 |
+
"loss": 36.6664,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.35,
|
| 24 |
+
"learning_rate": 8.646899873164096e-06,
|
| 25 |
+
"loss": 24.0296,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.47,
|
| 30 |
+
"learning_rate": 1.1588022279002226e-05,
|
| 31 |
+
"loss": 11.6622,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.59,
|
| 36 |
+
"learning_rate": 1.4529144684840354e-05,
|
| 37 |
+
"loss": 7.3831,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.71,
|
| 42 |
+
"learning_rate": 1.747026709067848e-05,
|
| 43 |
+
"loss": 5.378,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.82,
|
| 48 |
+
"learning_rate": 2.0411389496516608e-05,
|
| 49 |
+
"loss": 4.2757,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.94,
|
| 54 |
+
"learning_rate": 2.3352511902354736e-05,
|
| 55 |
+
"loss": 3.8626,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.06,
|
| 60 |
+
"learning_rate": 2.6293634308192865e-05,
|
| 61 |
+
"loss": 3.7576,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.18,
|
| 66 |
+
"learning_rate": 2.9234756714030993e-05,
|
| 67 |
+
"loss": 3.694,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.18,
|
| 72 |
+
"eval_loss": 3.646904945373535,
|
| 73 |
+
"eval_runtime": 108.0062,
|
| 74 |
+
"eval_samples_per_second": 39.007,
|
| 75 |
+
"eval_steps_per_second": 0.611,
|
| 76 |
+
"eval_wer": 1.0,
|
| 77 |
+
"step": 1000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.29,
|
| 81 |
+
"learning_rate": 3.217587911986912e-05,
|
| 82 |
+
"loss": 3.6476,
|
| 83 |
+
"step": 1100
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.41,
|
| 87 |
+
"learning_rate": 3.511700152570725e-05,
|
| 88 |
+
"loss": 3.6198,
|
| 89 |
+
"step": 1200
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.53,
|
| 93 |
+
"learning_rate": 3.805812393154537e-05,
|
| 94 |
+
"loss": 3.6033,
|
| 95 |
+
"step": 1300
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.65,
|
| 99 |
+
"learning_rate": 4.099924633738351e-05,
|
| 100 |
+
"loss": 3.5974,
|
| 101 |
+
"step": 1400
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.76,
|
| 105 |
+
"learning_rate": 4.3940368743221636e-05,
|
| 106 |
+
"loss": 3.5822,
|
| 107 |
+
"step": 1500
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.88,
|
| 111 |
+
"learning_rate": 4.6881491149059764e-05,
|
| 112 |
+
"loss": 3.5273,
|
| 113 |
+
"step": 1600
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 2.0,
|
| 117 |
+
"learning_rate": 4.982261355489789e-05,
|
| 118 |
+
"loss": 3.3921,
|
| 119 |
+
"step": 1700
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.12,
|
| 123 |
+
"learning_rate": 5.2763735960736014e-05,
|
| 124 |
+
"loss": 3.0588,
|
| 125 |
+
"step": 1800
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 2.24,
|
| 129 |
+
"learning_rate": 5.570485836657414e-05,
|
| 130 |
+
"loss": 2.592,
|
| 131 |
+
"step": 1900
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 2.35,
|
| 135 |
+
"learning_rate": 5.864598077241228e-05,
|
| 136 |
+
"loss": 2.2453,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.35,
|
| 141 |
+
"eval_loss": 1.3178459405899048,
|
| 142 |
+
"eval_runtime": 106.709,
|
| 143 |
+
"eval_samples_per_second": 39.481,
|
| 144 |
+
"eval_steps_per_second": 0.619,
|
| 145 |
+
"eval_wer": 0.8124043495771245,
|
| 146 |
+
"step": 2000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 2.47,
|
| 150 |
+
"learning_rate": 6.15871031782504e-05,
|
| 151 |
+
"loss": 2.0098,
|
| 152 |
+
"step": 2100
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 2.59,
|
| 156 |
+
"learning_rate": 6.452822558408853e-05,
|
| 157 |
+
"loss": 1.8594,
|
| 158 |
+
"step": 2200
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.71,
|
| 162 |
+
"learning_rate": 6.746934798992666e-05,
|
| 163 |
+
"loss": 1.7491,
|
| 164 |
+
"step": 2300
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.82,
|
| 168 |
+
"learning_rate": 7.041047039576478e-05,
|
| 169 |
+
"loss": 1.6576,
|
| 170 |
+
"step": 2400
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.94,
|
| 174 |
+
"learning_rate": 7.335159280160291e-05,
|
| 175 |
+
"loss": 1.6024,
|
| 176 |
+
"step": 2500
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.06,
|
| 180 |
+
"learning_rate": 7.629271520744105e-05,
|
| 181 |
+
"loss": 1.5301,
|
| 182 |
+
"step": 2600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.18,
|
| 186 |
+
"learning_rate": 7.923383761327917e-05,
|
| 187 |
+
"loss": 1.4785,
|
| 188 |
+
"step": 2700
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.29,
|
| 192 |
+
"learning_rate": 8.217496001911729e-05,
|
| 193 |
+
"loss": 1.4362,
|
| 194 |
+
"step": 2800
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.41,
|
| 198 |
+
"learning_rate": 8.511608242495543e-05,
|
| 199 |
+
"loss": 1.3932,
|
| 200 |
+
"step": 2900
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.53,
|
| 204 |
+
"learning_rate": 8.805720483079356e-05,
|
| 205 |
+
"loss": 1.3807,
|
| 206 |
+
"step": 3000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.53,
|
| 210 |
+
"eval_loss": 0.6274150609970093,
|
| 211 |
+
"eval_runtime": 114.1009,
|
| 212 |
+
"eval_samples_per_second": 36.923,
|
| 213 |
+
"eval_steps_per_second": 0.578,
|
| 214 |
+
"eval_wer": 0.5551751913008458,
|
| 215 |
+
"step": 3000
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.65,
|
| 219 |
+
"learning_rate": 9.099832723663168e-05,
|
| 220 |
+
"loss": 1.3404,
|
| 221 |
+
"step": 3100
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 3.76,
|
| 225 |
+
"learning_rate": 9.39394496424698e-05,
|
| 226 |
+
"loss": 1.314,
|
| 227 |
+
"step": 3200
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 3.88,
|
| 231 |
+
"learning_rate": 9.688057204830794e-05,
|
| 232 |
+
"loss": 1.2968,
|
| 233 |
+
"step": 3300
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 4.0,
|
| 237 |
+
"learning_rate": 9.982169445414606e-05,
|
| 238 |
+
"loss": 1.2514,
|
| 239 |
+
"step": 3400
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 4.12,
|
| 243 |
+
"learning_rate": 9.985126147657674e-05,
|
| 244 |
+
"loss": 1.2271,
|
| 245 |
+
"step": 3500
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 4.24,
|
| 249 |
+
"learning_rate": 9.936705484698494e-05,
|
| 250 |
+
"loss": 1.1943,
|
| 251 |
+
"step": 3600
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 4.35,
|
| 255 |
+
"learning_rate": 9.855002472512436e-05,
|
| 256 |
+
"loss": 1.1604,
|
| 257 |
+
"step": 3700
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.47,
|
| 261 |
+
"learning_rate": 9.740567937595607e-05,
|
| 262 |
+
"loss": 1.1508,
|
| 263 |
+
"step": 3800
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 4.59,
|
| 267 |
+
"learning_rate": 9.59417337627618e-05,
|
| 268 |
+
"loss": 1.1627,
|
| 269 |
+
"step": 3900
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 4.71,
|
| 273 |
+
"learning_rate": 9.41680575342979e-05,
|
| 274 |
+
"loss": 1.1464,
|
| 275 |
+
"step": 4000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 4.71,
|
| 279 |
+
"eval_loss": 0.5025569200515747,
|
| 280 |
+
"eval_runtime": 105.3594,
|
| 281 |
+
"eval_samples_per_second": 39.987,
|
| 282 |
+
"eval_steps_per_second": 0.626,
|
| 283 |
+
"eval_wer": 0.4815142972211035,
|
| 284 |
+
"step": 4000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 4.82,
|
| 288 |
+
"learning_rate": 9.209660848546103e-05,
|
| 289 |
+
"loss": 1.1392,
|
| 290 |
+
"step": 4100
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.94,
|
| 294 |
+
"learning_rate": 8.97413519400613e-05,
|
| 295 |
+
"loss": 1.1173,
|
| 296 |
+
"step": 4200
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 5.06,
|
| 300 |
+
"learning_rate": 8.711816659920938e-05,
|
| 301 |
+
"loss": 1.1,
|
| 302 |
+
"step": 4300
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 5.18,
|
| 306 |
+
"learning_rate": 8.424473749006998e-05,
|
| 307 |
+
"loss": 1.0627,
|
| 308 |
+
"step": 4400
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 5.29,
|
| 312 |
+
"learning_rate": 8.114043673670144e-05,
|
| 313 |
+
"loss": 1.0712,
|
| 314 |
+
"step": 4500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 5.41,
|
| 318 |
+
"learning_rate": 7.782619295680218e-05,
|
| 319 |
+
"loss": 1.0416,
|
| 320 |
+
"step": 4600
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 5.53,
|
| 324 |
+
"learning_rate": 7.432435016486675e-05,
|
| 325 |
+
"loss": 1.0254,
|
| 326 |
+
"step": 4700
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 5.65,
|
| 330 |
+
"learning_rate": 7.06585171330004e-05,
|
| 331 |
+
"loss": 1.0391,
|
| 332 |
+
"step": 4800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 5.76,
|
| 336 |
+
"learning_rate": 6.68534082249731e-05,
|
| 337 |
+
"loss": 1.0147,
|
| 338 |
+
"step": 4900
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.88,
|
| 342 |
+
"learning_rate": 6.29346767765805e-05,
|
| 343 |
+
"loss": 1.002,
|
| 344 |
+
"step": 5000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 5.88,
|
| 348 |
+
"eval_loss": 0.44318458437919617,
|
| 349 |
+
"eval_runtime": 108.9364,
|
| 350 |
+
"eval_samples_per_second": 38.674,
|
| 351 |
+
"eval_steps_per_second": 0.606,
|
| 352 |
+
"eval_wer": 0.45110753121224323,
|
| 353 |
+
"step": 5000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 6.0,
|
| 357 |
+
"learning_rate": 5.892874214563035e-05,
|
| 358 |
+
"loss": 0.984,
|
| 359 |
+
"step": 5100
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 6.12,
|
| 363 |
+
"learning_rate": 5.4862611597551125e-05,
|
| 364 |
+
"loss": 0.9435,
|
| 365 |
+
"step": 5200
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 6.24,
|
| 369 |
+
"learning_rate": 5.076369822743683e-05,
|
| 370 |
+
"loss": 0.9651,
|
| 371 |
+
"step": 5300
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 6.35,
|
| 375 |
+
"learning_rate": 4.665963614606392e-05,
|
| 376 |
+
"loss": 0.9421,
|
| 377 |
+
"step": 5400
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 6.47,
|
| 381 |
+
"learning_rate": 4.257809417586155e-05,
|
| 382 |
+
"loss": 0.9307,
|
| 383 |
+
"step": 5500
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 6.59,
|
| 387 |
+
"learning_rate": 3.854658931286227e-05,
|
| 388 |
+
"loss": 0.925,
|
| 389 |
+
"step": 5600
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 6.71,
|
| 393 |
+
"learning_rate": 3.459230121223818e-05,
|
| 394 |
+
"loss": 0.9263,
|
| 395 |
+
"step": 5700
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 6.82,
|
| 399 |
+
"learning_rate": 3.074188894812617e-05,
|
| 400 |
+
"loss": 0.9491,
|
| 401 |
+
"step": 5800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 6.94,
|
| 405 |
+
"learning_rate": 2.7021311283113737e-05,
|
| 406 |
+
"loss": 0.9332,
|
| 407 |
+
"step": 5900
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 7.06,
|
| 411 |
+
"learning_rate": 2.3455651659094773e-05,
|
| 412 |
+
"loss": 0.9,
|
| 413 |
+
"step": 6000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 7.06,
|
| 417 |
+
"eval_loss": 0.42447754740715027,
|
| 418 |
+
"eval_runtime": 108.0825,
|
| 419 |
+
"eval_samples_per_second": 38.979,
|
| 420 |
+
"eval_steps_per_second": 0.611,
|
| 421 |
+
"eval_wer": 0.4364881192106323,
|
| 422 |
+
"step": 6000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 7.18,
|
| 426 |
+
"learning_rate": 2.0068949089374516e-05,
|
| 427 |
+
"loss": 0.9037,
|
| 428 |
+
"step": 6100
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 7.29,
|
| 432 |
+
"learning_rate": 1.688403609211766e-05,
|
| 433 |
+
"loss": 0.863,
|
| 434 |
+
"step": 6200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 7.41,
|
| 438 |
+
"learning_rate": 1.3922384757761936e-05,
|
| 439 |
+
"loss": 0.8822,
|
| 440 |
+
"step": 6300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 7.53,
|
| 444 |
+
"learning_rate": 1.1203961988182233e-05,
|
| 445 |
+
"loss": 0.8838,
|
| 446 |
+
"step": 6400
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 7.65,
|
| 450 |
+
"learning_rate": 8.74709488355579e-06,
|
| 451 |
+
"loss": 0.8905,
|
| 452 |
+
"step": 6500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 7.76,
|
| 456 |
+
"learning_rate": 6.568347184464973e-06,
|
| 457 |
+
"loss": 0.8846,
|
| 458 |
+
"step": 6600
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 7.88,
|
| 462 |
+
"learning_rate": 4.682407602242322e-06,
|
| 463 |
+
"loss": 0.8865,
|
| 464 |
+
"step": 6700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 8.0,
|
| 468 |
+
"learning_rate": 3.1019907904135227e-06,
|
| 469 |
+
"loss": 0.8802,
|
| 470 |
+
"step": 6800
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 8.12,
|
| 474 |
+
"learning_rate": 1.8377516248706705e-06,
|
| 475 |
+
"loss": 0.8641,
|
| 476 |
+
"step": 6900
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 8.24,
|
| 480 |
+
"learning_rate": 8.982133706828155e-07,
|
| 481 |
+
"loss": 0.8817,
|
| 482 |
+
"step": 7000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 8.24,
|
| 486 |
+
"eval_loss": 0.4170154631137848,
|
| 487 |
+
"eval_runtime": 105.9513,
|
| 488 |
+
"eval_samples_per_second": 39.764,
|
| 489 |
+
"eval_steps_per_second": 0.623,
|
| 490 |
+
"eval_wer": 0.4307289569069674,
|
| 491 |
+
"step": 7000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 8.35,
|
| 495 |
+
"learning_rate": 2.8971021982976964e-07,
|
| 496 |
+
"loss": 0.8626,
|
| 497 |
+
"step": 7100
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 8.47,
|
| 501 |
+
"learning_rate": 1.6344587259148336e-08,
|
| 502 |
+
"loss": 0.8632,
|
| 503 |
+
"step": 7200
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 8.59,
|
| 507 |
+
"learning_rate": 9.992004054683143e-05,
|
| 508 |
+
"loss": 0.8675,
|
| 509 |
+
"step": 7300
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 8.71,
|
| 513 |
+
"learning_rate": 9.951987406202364e-05,
|
| 514 |
+
"loss": 0.9324,
|
| 515 |
+
"step": 7400
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 8.82,
|
| 519 |
+
"learning_rate": 9.878585380625223e-05,
|
| 520 |
+
"loss": 0.9278,
|
| 521 |
+
"step": 7500
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 8.94,
|
| 525 |
+
"learning_rate": 9.772292840741648e-05,
|
| 526 |
+
"loss": 0.9294,
|
| 527 |
+
"step": 7600
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 9.06,
|
| 531 |
+
"learning_rate": 9.633826391065235e-05,
|
| 532 |
+
"loss": 0.9066,
|
| 533 |
+
"step": 7700
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 9.18,
|
| 537 |
+
"learning_rate": 9.46411954661903e-05,
|
| 538 |
+
"loss": 0.908,
|
| 539 |
+
"step": 7800
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 9.29,
|
| 543 |
+
"learning_rate": 9.264316439351065e-05,
|
| 544 |
+
"loss": 0.901,
|
| 545 |
+
"step": 7900
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 9.41,
|
| 549 |
+
"learning_rate": 9.035764104609774e-05,
|
| 550 |
+
"loss": 0.8884,
|
| 551 |
+
"step": 8000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 9.41,
|
| 555 |
+
"eval_loss": 0.4221749007701874,
|
| 556 |
+
"eval_runtime": 105.5743,
|
| 557 |
+
"eval_samples_per_second": 39.906,
|
| 558 |
+
"eval_steps_per_second": 0.625,
|
| 559 |
+
"eval_wer": 0.4426500201369311,
|
| 560 |
+
"step": 8000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 9.53,
|
| 564 |
+
"learning_rate": 8.780003399682484e-05,
|
| 565 |
+
"loss": 0.9003,
|
| 566 |
+
"step": 8100
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 9.65,
|
| 570 |
+
"learning_rate": 8.498758615622466e-05,
|
| 571 |
+
"loss": 0.8945,
|
| 572 |
+
"step": 8200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 9.76,
|
| 576 |
+
"learning_rate": 8.193925852399696e-05,
|
| 577 |
+
"loss": 0.9117,
|
| 578 |
+
"step": 8300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 9.88,
|
| 582 |
+
"learning_rate": 7.867560235747892e-05,
|
| 583 |
+
"loss": 0.8941,
|
| 584 |
+
"step": 8400
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 10.0,
|
| 588 |
+
"learning_rate": 7.521862061889445e-05,
|
| 589 |
+
"loss": 0.8736,
|
| 590 |
+
"step": 8500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 10.12,
|
| 594 |
+
"learning_rate": 7.159161963547964e-05,
|
| 595 |
+
"loss": 0.8359,
|
| 596 |
+
"step": 8600
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 10.24,
|
| 600 |
+
"learning_rate": 6.781905197256384e-05,
|
| 601 |
+
"loss": 0.8247,
|
| 602 |
+
"step": 8700
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 10.35,
|
| 606 |
+
"learning_rate": 6.392635157892686e-05,
|
| 607 |
+
"loss": 0.8058,
|
| 608 |
+
"step": 8800
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 10.47,
|
| 612 |
+
"learning_rate": 5.993976231585202e-05,
|
| 613 |
+
"loss": 0.808,
|
| 614 |
+
"step": 8900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 10.59,
|
| 618 |
+
"learning_rate": 5.588616102589939e-05,
|
| 619 |
+
"loss": 0.8199,
|
| 620 |
+
"step": 9000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 10.59,
|
| 624 |
+
"eval_loss": 0.4070005416870117,
|
| 625 |
+
"eval_runtime": 104.9334,
|
| 626 |
+
"eval_samples_per_second": 40.149,
|
| 627 |
+
"eval_steps_per_second": 0.629,
|
| 628 |
+
"eval_wer": 0.41913008457511075,
|
| 629 |
+
"step": 9000
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 10.71,
|
| 633 |
+
"learning_rate": 5.17928763342375e-05,
|
| 634 |
+
"loss": 0.8051,
|
| 635 |
+
"step": 9100
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 10.82,
|
| 639 |
+
"learning_rate": 4.768750440414064e-05,
|
| 640 |
+
"loss": 0.8283,
|
| 641 |
+
"step": 9200
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 10.94,
|
| 645 |
+
"learning_rate": 4.3597722888794156e-05,
|
| 646 |
+
"loss": 0.82,
|
| 647 |
+
"step": 9300
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 11.06,
|
| 651 |
+
"learning_rate": 3.955110433370994e-05,
|
| 652 |
+
"loss": 0.7931,
|
| 653 |
+
"step": 9400
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 11.18,
|
| 657 |
+
"learning_rate": 3.5574930287758105e-05,
|
| 658 |
+
"loss": 0.7729,
|
| 659 |
+
"step": 9500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 11.29,
|
| 663 |
+
"learning_rate": 3.1696007376044074e-05,
|
| 664 |
+
"loss": 0.7711,
|
| 665 |
+
"step": 9600
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 11.41,
|
| 669 |
+
"learning_rate": 2.794048657463315e-05,
|
| 670 |
+
"loss": 0.7513,
|
| 671 |
+
"step": 9700
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 11.53,
|
| 675 |
+
"learning_rate": 2.4333686905538853e-05,
|
| 676 |
+
"loss": 0.7546,
|
| 677 |
+
"step": 9800
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 11.65,
|
| 681 |
+
"learning_rate": 2.089992474059062e-05,
|
| 682 |
+
"loss": 0.7501,
|
| 683 |
+
"step": 9900
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 11.76,
|
| 687 |
+
"learning_rate": 1.7662349864982235e-05,
|
| 688 |
+
"loss": 0.7504,
|
| 689 |
+
"step": 10000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 11.76,
|
| 693 |
+
"eval_loss": 0.3886597156524658,
|
| 694 |
+
"eval_runtime": 105.3872,
|
| 695 |
+
"eval_samples_per_second": 39.976,
|
| 696 |
+
"eval_steps_per_second": 0.626,
|
| 697 |
+
"eval_wer": 0.404430124848973,
|
| 698 |
+
"step": 10000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 11.88,
|
| 702 |
+
"learning_rate": 1.4642789405730323e-05,
|
| 703 |
+
"loss": 0.7703,
|
| 704 |
+
"step": 10100
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 12.0,
|
| 708 |
+
"learning_rate": 1.1861600677247902e-05,
|
| 709 |
+
"loss": 0.7348,
|
| 710 |
+
"step": 10200
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 12.12,
|
| 714 |
+
"learning_rate": 9.361444478984221e-06,
|
| 715 |
+
"loss": 0.7326,
|
| 716 |
+
"step": 10300
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 12.24,
|
| 720 |
+
"learning_rate": 7.108696871859649e-06,
|
| 721 |
+
"loss": 0.7105,
|
| 722 |
+
"step": 10400
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 12.35,
|
| 726 |
+
"learning_rate": 5.1451144446169495e-06,
|
| 727 |
+
"loss": 0.7248,
|
| 728 |
+
"step": 10500
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 12.47,
|
| 732 |
+
"learning_rate": 3.483935304386882e-06,
|
| 733 |
+
"loss": 0.7322,
|
| 734 |
+
"step": 10600
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 12.59,
|
| 738 |
+
"learning_rate": 2.1363588116427025e-06,
|
| 739 |
+
"loss": 0.7395,
|
| 740 |
+
"step": 10700
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 12.71,
|
| 744 |
+
"learning_rate": 1.120097142214388e-06,
|
| 745 |
+
"loss": 0.7192,
|
| 746 |
+
"step": 10800
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 12.82,
|
| 750 |
+
"learning_rate": 4.214828618490041e-07,
|
| 751 |
+
"loss": 0.7141,
|
| 752 |
+
"step": 10900
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 12.94,
|
| 756 |
+
"learning_rate": 5.711771314791392e-08,
|
| 757 |
+
"loss": 0.7213,
|
| 758 |
+
"step": 11000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 12.94,
|
| 762 |
+
"eval_loss": 0.3896392285823822,
|
| 763 |
+
"eval_runtime": 105.3159,
|
| 764 |
+
"eval_samples_per_second": 40.003,
|
| 765 |
+
"eval_steps_per_second": 0.627,
|
| 766 |
+
"eval_wer": 0.40378574305275877,
|
| 767 |
+
"step": 11000
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 13.06,
|
| 771 |
+
"learning_rate": 9.997054182189645e-05,
|
| 772 |
+
"loss": 0.7318,
|
| 773 |
+
"step": 11100
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 13.18,
|
| 777 |
+
"learning_rate": 9.966130926784864e-05,
|
| 778 |
+
"loss": 0.7748,
|
| 779 |
+
"step": 11200
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 13.29,
|
| 783 |
+
"learning_rate": 9.901726941301295e-05,
|
| 784 |
+
"loss": 0.773,
|
| 785 |
+
"step": 11300
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 13.41,
|
| 789 |
+
"learning_rate": 9.804276425417875e-05,
|
| 790 |
+
"loss": 0.8016,
|
| 791 |
+
"step": 11400
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 13.53,
|
| 795 |
+
"learning_rate": 9.67443637236866e-05,
|
| 796 |
+
"loss": 0.7831,
|
| 797 |
+
"step": 11500
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 13.65,
|
| 801 |
+
"learning_rate": 9.51308213961676e-05,
|
| 802 |
+
"loss": 0.8006,
|
| 803 |
+
"step": 11600
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 13.76,
|
| 807 |
+
"learning_rate": 9.321301547357316e-05,
|
| 808 |
+
"loss": 0.7881,
|
| 809 |
+
"step": 11700
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 13.88,
|
| 813 |
+
"learning_rate": 9.100387544636259e-05,
|
| 814 |
+
"loss": 0.7888,
|
| 815 |
+
"step": 11800
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 14.0,
|
| 819 |
+
"learning_rate": 8.851829492528598e-05,
|
| 820 |
+
"loss": 0.7791,
|
| 821 |
+
"step": 11900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 14.12,
|
| 825 |
+
"learning_rate": 8.57730312314333e-05,
|
| 826 |
+
"loss": 0.7637,
|
| 827 |
+
"step": 12000
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 14.12,
|
| 831 |
+
"eval_loss": 0.4040660262107849,
|
| 832 |
+
"eval_runtime": 105.4427,
|
| 833 |
+
"eval_samples_per_second": 39.955,
|
| 834 |
+
"eval_steps_per_second": 0.626,
|
| 835 |
+
"eval_wer": 0.408135320177205,
|
| 836 |
+
"step": 12000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 14.24,
|
| 840 |
+
"learning_rate": 8.278659242149581e-05,
|
| 841 |
+
"loss": 0.7497,
|
| 842 |
+
"step": 12100
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 14.35,
|
| 846 |
+
"learning_rate": 7.957911250989374e-05,
|
| 847 |
+
"loss": 0.7491,
|
| 848 |
+
"step": 12200
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 14.47,
|
| 852 |
+
"learning_rate": 7.617221572899955e-05,
|
| 853 |
+
"loss": 0.7514,
|
| 854 |
+
"step": 12300
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 14.59,
|
| 858 |
+
"learning_rate": 7.258887074258988e-05,
|
| 859 |
+
"loss": 0.7589,
|
| 860 |
+
"step": 12400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 14.71,
|
| 864 |
+
"learning_rate": 6.88532357953903e-05,
|
| 865 |
+
"loss": 0.7158,
|
| 866 |
+
"step": 12500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 14.82,
|
| 870 |
+
"learning_rate": 6.49904958426878e-05,
|
| 871 |
+
"loss": 0.7142,
|
| 872 |
+
"step": 12600
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 14.94,
|
| 876 |
+
"learning_rate": 6.1026692758051155e-05,
|
| 877 |
+
"loss": 0.7198,
|
| 878 |
+
"step": 12700
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 15.06,
|
| 882 |
+
"learning_rate": 5.698854976386778e-05,
|
| 883 |
+
"loss": 0.7107,
|
| 884 |
+
"step": 12800
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 15.18,
|
| 888 |
+
"learning_rate": 5.2903291268354404e-05,
|
| 889 |
+
"loss": 0.6709,
|
| 890 |
+
"step": 12900
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 15.29,
|
| 894 |
+
"learning_rate": 4.8798459323665145e-05,
|
| 895 |
+
"loss": 0.6743,
|
| 896 |
+
"step": 13000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 15.29,
|
| 900 |
+
"eval_loss": 0.4003716707229614,
|
| 901 |
+
"eval_runtime": 104.8505,
|
| 902 |
+
"eval_samples_per_second": 40.181,
|
| 903 |
+
"eval_steps_per_second": 0.629,
|
| 904 |
+
"eval_wer": 0.39714055577929924,
|
| 905 |
+
"step": 13000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 15.41,
|
| 909 |
+
"learning_rate": 4.470172794250435e-05,
|
| 910 |
+
"loss": 0.7087,
|
| 911 |
+
"step": 13100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 15.53,
|
| 915 |
+
"learning_rate": 4.064071652508525e-05,
|
| 916 |
+
"loss": 0.6737,
|
| 917 |
+
"step": 13200
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 15.65,
|
| 921 |
+
"learning_rate": 3.6642803654275966e-05,
|
| 922 |
+
"loss": 0.6752,
|
| 923 |
+
"step": 13300
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 15.76,
|
| 927 |
+
"learning_rate": 3.2734942514291747e-05,
|
| 928 |
+
"loss": 0.6622,
|
| 929 |
+
"step": 13400
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 15.88,
|
| 933 |
+
"learning_rate": 2.8943479177344868e-05,
|
| 934 |
+
"loss": 0.6889,
|
| 935 |
+
"step": 13500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 16.0,
|
| 939 |
+
"learning_rate": 2.5293974983331425e-05,
|
| 940 |
+
"loss": 0.6654,
|
| 941 |
+
"step": 13600
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 16.12,
|
| 945 |
+
"learning_rate": 2.181103421003689e-05,
|
| 946 |
+
"loss": 0.6492,
|
| 947 |
+
"step": 13700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 16.24,
|
| 951 |
+
"learning_rate": 1.851813819567597e-05,
|
| 952 |
+
"loss": 0.6363,
|
| 953 |
+
"step": 13800
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 16.35,
|
| 957 |
+
"learning_rate": 1.543748703208282e-05,
|
| 958 |
+
"loss": 0.6505,
|
| 959 |
+
"step": 13900
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 16.47,
|
| 963 |
+
"learning_rate": 1.2589849895824684e-05,
|
| 964 |
+
"loss": 0.6261,
|
| 965 |
+
"step": 14000
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 16.47,
|
| 969 |
+
"eval_loss": 0.38466739654541016,
|
| 970 |
+
"eval_runtime": 105.1698,
|
| 971 |
+
"eval_samples_per_second": 40.059,
|
| 972 |
+
"eval_steps_per_second": 0.628,
|
| 973 |
+
"eval_wer": 0.38771647200966575,
|
| 974 |
+
"step": 14000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 16.59,
|
| 978 |
+
"learning_rate": 9.994425026280951e-06,
|
| 979 |
+
"loss": 0.6213,
|
| 980 |
+
"step": 14100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 16.71,
|
| 984 |
+
"learning_rate": 7.668710294687454e-06,
|
| 985 |
+
"loss": 0.6348,
|
| 986 |
+
"step": 14200
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 16.82,
|
| 990 |
+
"learning_rate": 5.6283852367463915e-06,
|
| 991 |
+
"loss": 0.6538,
|
| 992 |
+
"step": 14300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 16.94,
|
| 996 |
+
"learning_rate": 3.8872053441162186e-06,
|
| 997 |
+
"loss": 0.6341,
|
| 998 |
+
"step": 14400
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 17.06,
|
| 1002 |
+
"learning_rate": 2.456909327447748e-06,
|
| 1003 |
+
"loss": 0.6269,
|
| 1004 |
+
"step": 14500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 17.18,
|
| 1008 |
+
"learning_rate": 1.3471399761826819e-06,
|
| 1009 |
+
"loss": 0.6414,
|
| 1010 |
+
"step": 14600
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 17.29,
|
| 1014 |
+
"learning_rate": 5.65379148662526e-07,
|
| 1015 |
+
"loss": 0.6082,
|
| 1016 |
+
"step": 14700
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 17.41,
|
| 1020 |
+
"learning_rate": 1.168973308325827e-07,
|
| 1021 |
+
"loss": 0.6305,
|
| 1022 |
+
"step": 14800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 17.53,
|
| 1026 |
+
"learning_rate": 9.999528189639287e-05,
|
| 1027 |
+
"loss": 0.6196,
|
| 1028 |
+
"step": 14900
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 17.65,
|
| 1032 |
+
"learning_rate": 9.977040224155017e-05,
|
| 1033 |
+
"loss": 0.6818,
|
| 1034 |
+
"step": 15000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 17.65,
|
| 1038 |
+
"eval_loss": 0.4097123444080353,
|
| 1039 |
+
"eval_runtime": 105.0656,
|
| 1040 |
+
"eval_samples_per_second": 40.099,
|
| 1041 |
+
"eval_steps_per_second": 0.628,
|
| 1042 |
+
"eval_wer": 0.3959323399113975,
|
| 1043 |
+
"step": 15000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 17.76,
|
| 1047 |
+
"learning_rate": 9.920997980140259e-05,
|
| 1048 |
+
"loss": 0.7001,
|
| 1049 |
+
"step": 15100
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 17.88,
|
| 1053 |
+
"learning_rate": 9.831779283969819e-05,
|
| 1054 |
+
"loss": 0.7143,
|
| 1055 |
+
"step": 15200
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 18.0,
|
| 1059 |
+
"learning_rate": 9.709985631481902e-05,
|
| 1060 |
+
"loss": 0.6883,
|
| 1061 |
+
"step": 15300
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 18.12,
|
| 1065 |
+
"learning_rate": 9.556438132805179e-05,
|
| 1066 |
+
"loss": 0.6889,
|
| 1067 |
+
"step": 15400
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 18.24,
|
| 1071 |
+
"learning_rate": 9.372171976587154e-05,
|
| 1072 |
+
"loss": 0.6823,
|
| 1073 |
+
"step": 15500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 18.35,
|
| 1077 |
+
"learning_rate": 9.158429450945069e-05,
|
| 1078 |
+
"loss": 0.6682,
|
| 1079 |
+
"step": 15600
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 18.47,
|
| 1083 |
+
"learning_rate": 8.916651568190705e-05,
|
| 1084 |
+
"loss": 0.6735,
|
| 1085 |
+
"step": 15700
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 18.59,
|
| 1089 |
+
"learning_rate": 8.648468349793653e-05,
|
| 1090 |
+
"loss": 0.6835,
|
| 1091 |
+
"step": 15800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 18.71,
|
| 1095 |
+
"learning_rate": 8.355687837080098e-05,
|
| 1096 |
+
"loss": 0.684,
|
| 1097 |
+
"step": 15900
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 18.82,
|
| 1101 |
+
"learning_rate": 8.040283901754784e-05,
|
| 1102 |
+
"loss": 0.7017,
|
| 1103 |
+
"step": 16000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 18.82,
|
| 1107 |
+
"eval_loss": 0.39238452911376953,
|
| 1108 |
+
"eval_runtime": 107.5856,
|
| 1109 |
+
"eval_samples_per_second": 39.16,
|
| 1110 |
+
"eval_steps_per_second": 0.613,
|
| 1111 |
+
"eval_wer": 0.3910189287152638,
|
| 1112 |
+
"step": 16000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 18.94,
|
| 1116 |
+
"learning_rate": 7.704382938425531e-05,
|
| 1117 |
+
"loss": 0.6519,
|
| 1118 |
+
"step": 16100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 19.06,
|
| 1122 |
+
"learning_rate": 7.350249528846678e-05,
|
| 1123 |
+
"loss": 0.6436,
|
| 1124 |
+
"step": 16200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 19.18,
|
| 1128 |
+
"learning_rate": 6.987810048655369e-05,
|
| 1129 |
+
"loss": 0.64,
|
| 1130 |
+
"step": 16300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 19.29,
|
| 1134 |
+
"learning_rate": 6.604723032247485e-05,
|
| 1135 |
+
"loss": 0.6514,
|
| 1136 |
+
"step": 16400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 19.41,
|
| 1140 |
+
"learning_rate": 6.210817271934942e-05,
|
| 1141 |
+
"loss": 0.6311,
|
| 1142 |
+
"step": 16500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 19.53,
|
| 1146 |
+
"learning_rate": 5.8087484070135967e-05,
|
| 1147 |
+
"loss": 0.6429,
|
| 1148 |
+
"step": 16600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 19.65,
|
| 1152 |
+
"learning_rate": 5.4012271109106916e-05,
|
| 1153 |
+
"loss": 0.6398,
|
| 1154 |
+
"step": 16700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 19.76,
|
| 1158 |
+
"learning_rate": 4.991000816329353e-05,
|
| 1159 |
+
"loss": 0.6158,
|
| 1160 |
+
"step": 16800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 19.88,
|
| 1164 |
+
"learning_rate": 4.580835192568782e-05,
|
| 1165 |
+
"loss": 0.6142,
|
| 1166 |
+
"step": 16900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 20.0,
|
| 1170 |
+
"learning_rate": 4.173495499896826e-05,
|
| 1171 |
+
"loss": 0.6221,
|
| 1172 |
+
"step": 17000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 20.0,
|
| 1176 |
+
"eval_loss": 0.3881571292877197,
|
| 1177 |
+
"eval_runtime": 104.9541,
|
| 1178 |
+
"eval_samples_per_second": 40.141,
|
| 1179 |
+
"eval_steps_per_second": 0.629,
|
| 1180 |
+
"eval_wer": 0.381876761981474,
|
| 1181 |
+
"step": 17000
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 20.12,
|
| 1185 |
+
"learning_rate": 3.7717279466800834e-05,
|
| 1186 |
+
"loss": 0.6,
|
| 1187 |
+
"step": 17100
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 20.24,
|
| 1191 |
+
"learning_rate": 3.3782411749583854e-05,
|
| 1192 |
+
"loss": 0.5918,
|
| 1193 |
+
"step": 17200
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 20.35,
|
| 1197 |
+
"learning_rate": 2.9956879992845265e-05,
|
| 1198 |
+
"loss": 0.569,
|
| 1199 |
+
"step": 17300
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 20.47,
|
| 1203 |
+
"learning_rate": 2.6266475219424137e-05,
|
| 1204 |
+
"loss": 0.5599,
|
| 1205 |
+
"step": 17400
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 20.59,
|
| 1209 |
+
"learning_rate": 2.2736077451197112e-05,
|
| 1210 |
+
"loss": 0.5745,
|
| 1211 |
+
"step": 17500
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 20.71,
|
| 1215 |
+
"learning_rate": 1.9389487972603084e-05,
|
| 1216 |
+
"loss": 0.5829,
|
| 1217 |
+
"step": 17600
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 20.82,
|
| 1221 |
+
"learning_rate": 1.6249268866814737e-05,
|
| 1222 |
+
"loss": 0.5823,
|
| 1223 |
+
"step": 17700
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 20.94,
|
| 1227 |
+
"learning_rate": 1.3336590906375096e-05,
|
| 1228 |
+
"loss": 0.5708,
|
| 1229 |
+
"step": 17800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 21.06,
|
| 1233 |
+
"learning_rate": 1.067109082379022e-05,
|
| 1234 |
+
"loss": 0.5556,
|
| 1235 |
+
"step": 17900
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 21.18,
|
| 1239 |
+
"learning_rate": 8.270738924335186e-06,
|
| 1240 |
+
"loss": 0.5568,
|
| 1241 |
+
"step": 18000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 21.18,
|
| 1245 |
+
"eval_loss": 0.3854876756668091,
|
| 1246 |
+
"eval_runtime": 105.4508,
|
| 1247 |
+
"eval_samples_per_second": 39.952,
|
| 1248 |
+
"eval_steps_per_second": 0.626,
|
| 1249 |
+
"eval_wer": 0.3769230769230769,
|
| 1250 |
+
"step": 18000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 21.29,
|
| 1254 |
+
"learning_rate": 6.151717933601536e-06,
|
| 1255 |
+
"loss": 0.5469,
|
| 1256 |
+
"step": 18100
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 21.41,
|
| 1260 |
+
"learning_rate": 4.328313896574343e-06,
|
| 1261 |
+
"loss": 0.5706,
|
| 1262 |
+
"step": 18200
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 21.53,
|
| 1266 |
+
"learning_rate": 2.812819863776944e-06,
|
| 1267 |
+
"loss": 0.553,
|
| 1268 |
+
"step": 18300
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 21.65,
|
| 1272 |
+
"learning_rate": 1.6154530138129676e-06,
|
| 1273 |
+
"loss": 0.5454,
|
| 1274 |
+
"step": 18400
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 21.76,
|
| 1278 |
+
"learning_rate": 7.442857710505957e-07,
|
| 1279 |
+
"loss": 0.5453,
|
| 1280 |
+
"step": 18500
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 21.88,
|
| 1284 |
+
"learning_rate": 2.051913828404417e-07,
|
| 1285 |
+
"loss": 0.5537,
|
| 1286 |
+
"step": 18600
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 22.0,
|
| 1290 |
+
"learning_rate": 1.8043231753583911e-09,
|
| 1291 |
+
"loss": 0.554,
|
| 1292 |
+
"step": 18700
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 22.12,
|
| 1296 |
+
"learning_rate": 9.986450421025644e-05,
|
| 1297 |
+
"loss": 0.5931,
|
| 1298 |
+
"step": 18800
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 22.24,
|
| 1302 |
+
"learning_rate": 9.93946355404309e-05,
|
| 1303 |
+
"loss": 0.6287,
|
| 1304 |
+
"step": 18900
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 22.35,
|
| 1308 |
+
"learning_rate": 9.859175743443652e-05,
|
| 1309 |
+
"loss": 0.6087,
|
| 1310 |
+
"step": 19000
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 22.35,
|
| 1314 |
+
"eval_loss": 0.394783079624176,
|
| 1315 |
+
"eval_runtime": 106.0156,
|
| 1316 |
+
"eval_samples_per_second": 39.739,
|
| 1317 |
+
"eval_steps_per_second": 0.623,
|
| 1318 |
+
"eval_wer": 0.38787756745871926,
|
| 1319 |
+
"step": 19000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 22.47,
|
| 1323 |
+
"learning_rate": 9.746128274697833e-05,
|
| 1324 |
+
"loss": 0.6281,
|
| 1325 |
+
"step": 19100
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 22.59,
|
| 1329 |
+
"learning_rate": 9.601083292791942e-05,
|
| 1330 |
+
"loss": 0.6269,
|
| 1331 |
+
"step": 19200
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 22.71,
|
| 1335 |
+
"learning_rate": 9.425018663988473e-05,
|
| 1336 |
+
"loss": 0.6171,
|
| 1337 |
+
"step": 19300
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 22.82,
|
| 1341 |
+
"learning_rate": 9.219121383233855e-05,
|
| 1342 |
+
"loss": 0.6135,
|
| 1343 |
+
"step": 19400
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 22.94,
|
| 1347 |
+
"learning_rate": 8.984779571659518e-05,
|
| 1348 |
+
"loss": 0.6233,
|
| 1349 |
+
"step": 19500
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 23.06,
|
| 1353 |
+
"learning_rate": 8.723573118127667e-05,
|
| 1354 |
+
"loss": 0.6153,
|
| 1355 |
+
"step": 19600
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 23.18,
|
| 1359 |
+
"learning_rate": 8.43726302791475e-05,
|
| 1360 |
+
"loss": 0.6014,
|
| 1361 |
+
"step": 19700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 23.29,
|
| 1365 |
+
"learning_rate": 8.127779550341711e-05,
|
| 1366 |
+
"loss": 0.5969,
|
| 1367 |
+
"step": 19800
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 23.41,
|
| 1371 |
+
"learning_rate": 7.797209165392523e-05,
|
| 1372 |
+
"loss": 0.579,
|
| 1373 |
+
"step": 19900
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 23.53,
|
| 1377 |
+
"learning_rate": 7.451360514395531e-05,
|
| 1378 |
+
"loss": 0.5808,
|
| 1379 |
+
"step": 20000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 23.53,
|
| 1383 |
+
"eval_loss": 0.4057287871837616,
|
| 1384 |
+
"eval_runtime": 104.5213,
|
| 1385 |
+
"eval_samples_per_second": 40.308,
|
| 1386 |
+
"eval_steps_per_second": 0.631,
|
| 1387 |
+
"eval_wer": 0.3873942811115586,
|
| 1388 |
+
"step": 20000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 23.65,
|
| 1392 |
+
"learning_rate": 7.0855823703674e-05,
|
| 1393 |
+
"loss": 0.5769,
|
| 1394 |
+
"step": 20100
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 23.76,
|
| 1398 |
+
"learning_rate": 6.70574361830682e-05,
|
| 1399 |
+
"loss": 0.5719,
|
| 1400 |
+
"step": 20200
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 23.88,
|
| 1404 |
+
"learning_rate": 6.314405060359115e-05,
|
| 1405 |
+
"loss": 0.5757,
|
| 1406 |
+
"step": 20300
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 24.0,
|
| 1410 |
+
"learning_rate": 5.91420502821974e-05,
|
| 1411 |
+
"loss": 0.5741,
|
| 1412 |
+
"step": 20400
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 24.12,
|
| 1416 |
+
"learning_rate": 5.507841595993307e-05,
|
| 1417 |
+
"loss": 0.5505,
|
| 1418 |
+
"step": 20500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 24.24,
|
| 1422 |
+
"learning_rate": 5.098054390280248e-05,
|
| 1423 |
+
"loss": 0.5231,
|
| 1424 |
+
"step": 20600
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 24.35,
|
| 1428 |
+
"learning_rate": 4.691704783544732e-05,
|
| 1429 |
+
"loss": 0.5522,
|
| 1430 |
+
"step": 20700
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 24.47,
|
| 1434 |
+
"learning_rate": 4.28332788746043e-05,
|
| 1435 |
+
"loss": 0.5458,
|
| 1436 |
+
"step": 20800
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 24.59,
|
| 1440 |
+
"learning_rate": 3.879782661323741e-05,
|
| 1441 |
+
"loss": 0.5427,
|
| 1442 |
+
"step": 20900
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 24.71,
|
| 1446 |
+
"learning_rate": 3.483789731914372e-05,
|
| 1447 |
+
"loss": 0.5413,
|
| 1448 |
+
"step": 21000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 24.71,
|
| 1452 |
+
"eval_loss": 0.3984732925891876,
|
| 1453 |
+
"eval_runtime": 107.6668,
|
| 1454 |
+
"eval_samples_per_second": 39.13,
|
| 1455 |
+
"eval_steps_per_second": 0.613,
|
| 1456 |
+
"eval_wer": 0.3810712847362062,
|
| 1457 |
+
"step": 21000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 24.82,
|
| 1461 |
+
"learning_rate": 3.098018809833621e-05,
|
| 1462 |
+
"loss": 0.5312,
|
| 1463 |
+
"step": 21100
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 24.94,
|
| 1467 |
+
"learning_rate": 2.7250706908125788e-05,
|
| 1468 |
+
"loss": 0.5168,
|
| 1469 |
+
"step": 21200
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 25.06,
|
| 1473 |
+
"learning_rate": 2.3674597216315607e-05,
|
| 1474 |
+
"loss": 0.4865,
|
| 1475 |
+
"step": 21300
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 25.18,
|
| 1479 |
+
"learning_rate": 2.0275968488622116e-05,
|
| 1480 |
+
"loss": 0.4999,
|
| 1481 |
+
"step": 21400
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 25.29,
|
| 1485 |
+
"learning_rate": 1.707773364714661e-05,
|
| 1486 |
+
"loss": 0.509,
|
| 1487 |
+
"step": 21500
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 25.41,
|
| 1491 |
+
"learning_rate": 1.4101454595720243e-05,
|
| 1492 |
+
"loss": 0.4925,
|
| 1493 |
+
"step": 21600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 25.53,
|
| 1497 |
+
"learning_rate": 1.1367196853562878e-05,
|
| 1498 |
+
"loss": 0.4885,
|
| 1499 |
+
"step": 21700
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 25.65,
|
| 1503 |
+
"learning_rate": 8.89339427728868e-06,
|
| 1504 |
+
"loss": 0.4897,
|
| 1505 |
+
"step": 21800
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 25.76,
|
| 1509 |
+
"learning_rate": 6.696724783276548e-06,
|
| 1510 |
+
"loss": 0.5034,
|
| 1511 |
+
"step": 21900
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 25.88,
|
| 1515 |
+
"learning_rate": 4.791997908263651e-06,
|
| 1516 |
+
"loss": 0.4731,
|
| 1517 |
+
"step": 22000
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 25.88,
|
| 1521 |
+
"eval_loss": 0.3930116593837738,
|
| 1522 |
+
"eval_runtime": 106.2837,
|
| 1523 |
+
"eval_samples_per_second": 39.639,
|
| 1524 |
+
"eval_steps_per_second": 0.621,
|
| 1525 |
+
"eval_wer": 0.36939186467982277,
|
| 1526 |
+
"step": 22000
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 26.0,
|
| 1530 |
+
"learning_rate": 3.1920549662057043e-06,
|
| 1531 |
+
"loss": 0.4943,
|
| 1532 |
+
"step": 22100
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 26.12,
|
| 1536 |
+
"learning_rate": 1.9076824745293208e-06,
|
| 1537 |
+
"loss": 0.4817,
|
| 1538 |
+
"step": 22200
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 26.24,
|
| 1542 |
+
"learning_rate": 9.475394334397813e-07,
|
| 1543 |
+
"loss": 0.4809,
|
| 1544 |
+
"step": 22300
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 26.35,
|
| 1548 |
+
"learning_rate": 3.180989485537966e-07,
|
| 1549 |
+
"loss": 0.478,
|
| 1550 |
+
"step": 22400
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 26.47,
|
| 1554 |
+
"learning_rate": 2.3604590426795636e-08,
|
| 1555 |
+
"loss": 0.4911,
|
| 1556 |
+
"step": 22500
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 26.59,
|
| 1560 |
+
"learning_rate": 9.993395821480979e-05,
|
| 1561 |
+
"loss": 0.5161,
|
| 1562 |
+
"step": 22600
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 26.71,
|
| 1566 |
+
"learning_rate": 9.95548755708218e-05,
|
| 1567 |
+
"loss": 0.5251,
|
| 1568 |
+
"step": 22700
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 26.82,
|
| 1572 |
+
"learning_rate": 9.884170318221471e-05,
|
| 1573 |
+
"loss": 0.5496,
|
| 1574 |
+
"step": 22800
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 26.94,
|
| 1578 |
+
"learning_rate": 9.779924912444897e-05,
|
| 1579 |
+
"loss": 0.5606,
|
| 1580 |
+
"step": 22900
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 27.06,
|
| 1584 |
+
"learning_rate": 9.643454142869053e-05,
|
| 1585 |
+
"loss": 0.5363,
|
| 1586 |
+
"step": 23000
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 27.06,
|
| 1590 |
+
"eval_loss": 0.4152192175388336,
|
| 1591 |
+
"eval_runtime": 104.8412,
|
| 1592 |
+
"eval_samples_per_second": 40.185,
|
| 1593 |
+
"eval_steps_per_second": 0.63,
|
| 1594 |
+
"eval_wer": 0.3826419653644785,
|
| 1595 |
+
"step": 23000
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 27.18,
|
| 1599 |
+
"learning_rate": 9.47567807001332e-05,
|
| 1600 |
+
"loss": 0.5215,
|
| 1601 |
+
"step": 23100
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 27.29,
|
| 1605 |
+
"learning_rate": 9.277727808923149e-05,
|
| 1606 |
+
"loss": 0.5371,
|
| 1607 |
+
"step": 23200
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 27.41,
|
| 1611 |
+
"learning_rate": 9.05093790340303e-05,
|
| 1612 |
+
"loss": 0.5601,
|
| 1613 |
+
"step": 23300
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 27.53,
|
| 1617 |
+
"learning_rate": 8.796837328770604e-05,
|
| 1618 |
+
"loss": 0.5388,
|
| 1619 |
+
"step": 23400
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 27.65,
|
| 1623 |
+
"learning_rate": 8.5171391837898e-05,
|
| 1624 |
+
"loss": 0.5428,
|
| 1625 |
+
"step": 23500
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 27.76,
|
| 1629 |
+
"learning_rate": 8.213729141277887e-05,
|
| 1630 |
+
"loss": 0.5581,
|
| 1631 |
+
"step": 23600
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 27.88,
|
| 1635 |
+
"learning_rate": 7.888652735250435e-05,
|
| 1636 |
+
"loss": 0.5678,
|
| 1637 |
+
"step": 23700
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 28.0,
|
| 1641 |
+
"learning_rate": 7.544101570311861e-05,
|
| 1642 |
+
"loss": 0.5225,
|
| 1643 |
+
"step": 23800
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 28.12,
|
| 1647 |
+
"learning_rate": 7.182398546265166e-05,
|
| 1648 |
+
"loss": 0.5327,
|
| 1649 |
+
"step": 23900
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 28.24,
|
| 1653 |
+
"learning_rate": 6.805982197554043e-05,
|
| 1654 |
+
"loss": 0.5235,
|
| 1655 |
+
"step": 24000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 28.24,
|
| 1659 |
+
"eval_loss": 0.40879756212234497,
|
| 1660 |
+
"eval_runtime": 106.7199,
|
| 1661 |
+
"eval_samples_per_second": 39.477,
|
| 1662 |
+
"eval_steps_per_second": 0.618,
|
| 1663 |
+
"eval_wer": 0.3765606121627064,
|
| 1664 |
+
"step": 24000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 28.35,
|
| 1668 |
+
"learning_rate": 6.417390253117627e-05,
|
| 1669 |
+
"loss": 0.5206,
|
| 1670 |
+
"step": 24100
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 28.47,
|
| 1674 |
+
"learning_rate": 6.0192425274944955e-05,
|
| 1675 |
+
"loss": 0.5239,
|
| 1676 |
+
"step": 24200
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 28.59,
|
| 1680 |
+
"learning_rate": 5.614223258520837e-05,
|
| 1681 |
+
"loss": 0.4954,
|
| 1682 |
+
"step": 24300
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 28.71,
|
| 1686 |
+
"learning_rate": 5.2050630106986734e-05,
|
| 1687 |
+
"loss": 0.4954,
|
| 1688 |
+
"step": 24400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 28.82,
|
| 1692 |
+
"learning_rate": 4.79452026623847e-05,
|
| 1693 |
+
"loss": 0.4967,
|
| 1694 |
+
"step": 24500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 28.94,
|
| 1698 |
+
"learning_rate": 4.385362827885607e-05,
|
| 1699 |
+
"loss": 0.4762,
|
| 1700 |
+
"step": 24600
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 29.06,
|
| 1704 |
+
"learning_rate": 3.980349158909619e-05,
|
| 1705 |
+
"loss": 0.476,
|
| 1706 |
+
"step": 24700
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 29.18,
|
| 1710 |
+
"learning_rate": 3.582209786058425e-05,
|
| 1711 |
+
"loss": 0.4622,
|
| 1712 |
+
"step": 24800
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 29.29,
|
| 1716 |
+
"learning_rate": 3.193628890855341e-05,
|
| 1717 |
+
"loss": 0.4724,
|
| 1718 |
+
"step": 24900
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 29.41,
|
| 1722 |
+
"learning_rate": 2.8172262133470838e-05,
|
| 1723 |
+
"loss": 0.4951,
|
| 1724 |
+
"step": 25000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 29.41,
|
| 1728 |
+
"eval_loss": 0.41040393710136414,
|
| 1729 |
+
"eval_runtime": 105.7062,
|
| 1730 |
+
"eval_samples_per_second": 39.856,
|
| 1731 |
+
"eval_steps_per_second": 0.624,
|
| 1732 |
+
"eval_wer": 0.37140555779299234,
|
| 1733 |
+
"step": 25000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 29.53,
|
| 1737 |
+
"learning_rate": 2.455539390304083e-05,
|
| 1738 |
+
"loss": 0.448,
|
| 1739 |
+
"step": 25100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 29.65,
|
| 1743 |
+
"learning_rate": 2.1110068469458716e-05,
|
| 1744 |
+
"loss": 0.4546,
|
| 1745 |
+
"step": 25200
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 29.76,
|
| 1749 |
+
"learning_rate": 1.785951357532256e-05,
|
| 1750 |
+
"loss": 0.4388,
|
| 1751 |
+
"step": 25300
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 29.88,
|
| 1755 |
+
"learning_rate": 1.4825643856517025e-05,
|
| 1756 |
+
"loss": 0.4521,
|
| 1757 |
+
"step": 25400
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 30.0,
|
| 1761 |
+
"learning_rate": 1.2028913097818905e-05,
|
| 1762 |
+
"loss": 0.4474,
|
| 1763 |
+
"step": 25500
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 30.12,
|
| 1767 |
+
"learning_rate": 9.488176337288151e-06,
|
| 1768 |
+
"loss": 0.431,
|
| 1769 |
+
"step": 25600
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 30.24,
|
| 1773 |
+
"learning_rate": 7.2205627491117756e-06,
|
| 1774 |
+
"loss": 0.4444,
|
| 1775 |
+
"step": 25700
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 30.35,
|
| 1779 |
+
"learning_rate": 5.2596790812937e-06,
|
| 1780 |
+
"loss": 0.4279,
|
| 1781 |
+
"step": 25800
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 30.47,
|
| 1785 |
+
"learning_rate": 3.579153928305451e-06,
|
| 1786 |
+
"loss": 0.4397,
|
| 1787 |
+
"step": 25900
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 30.59,
|
| 1791 |
+
"learning_rate": 2.2115894765690538e-06,
|
| 1792 |
+
"loss": 0.4422,
|
| 1793 |
+
"step": 26000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 30.59,
|
| 1797 |
+
"eval_loss": 0.4197729825973511,
|
| 1798 |
+
"eval_runtime": 105.5173,
|
| 1799 |
+
"eval_samples_per_second": 39.927,
|
| 1800 |
+
"eval_steps_per_second": 0.625,
|
| 1801 |
+
"eval_wer": 0.3667740636327024,
|
| 1802 |
+
"step": 26000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 30.71,
|
| 1806 |
+
"learning_rate": 1.1662055909953062e-06,
|
| 1807 |
+
"loss": 0.4333,
|
| 1808 |
+
"step": 26100
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 30.82,
|
| 1812 |
+
"learning_rate": 4.500500551037179e-07,
|
| 1813 |
+
"loss": 0.4262,
|
| 1814 |
+
"step": 26200
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 30.94,
|
| 1818 |
+
"learning_rate": 6.795105617835584e-08,
|
| 1819 |
+
"loss": 0.423,
|
| 1820 |
+
"step": 26300
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 31.06,
|
| 1824 |
+
"learning_rate": 9.997751536547195e-05,
|
| 1825 |
+
"loss": 0.4465,
|
| 1826 |
+
"step": 26400
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 31.18,
|
| 1830 |
+
"learning_rate": 9.96860426836977e-05,
|
| 1831 |
+
"loss": 0.4738,
|
| 1832 |
+
"step": 26500
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 31.29,
|
| 1836 |
+
"learning_rate": 9.905959595305095e-05,
|
| 1837 |
+
"loss": 0.5112,
|
| 1838 |
+
"step": 26600
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 31.41,
|
| 1842 |
+
"learning_rate": 9.81023985607534e-05,
|
| 1843 |
+
"loss": 0.4893,
|
| 1844 |
+
"step": 26700
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 31.53,
|
| 1848 |
+
"learning_rate": 9.682090375340612e-05,
|
| 1849 |
+
"loss": 0.5042,
|
| 1850 |
+
"step": 26800
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 31.65,
|
| 1854 |
+
"learning_rate": 9.522375113040246e-05,
|
| 1855 |
+
"loss": 0.5001,
|
| 1856 |
+
"step": 26900
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 31.76,
|
| 1860 |
+
"learning_rate": 9.332170839735737e-05,
|
| 1861 |
+
"loss": 0.4968,
|
| 1862 |
+
"step": 27000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 31.76,
|
| 1866 |
+
"eval_loss": 0.4139433801174164,
|
| 1867 |
+
"eval_runtime": 105.1015,
|
| 1868 |
+
"eval_samples_per_second": 40.085,
|
| 1869 |
+
"eval_steps_per_second": 0.628,
|
| 1870 |
+
"eval_wer": 0.3725734997986307,
|
| 1871 |
+
"step": 27000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 31.88,
|
| 1875 |
+
"learning_rate": 9.112759877224125e-05,
|
| 1876 |
+
"loss": 0.5126,
|
| 1877 |
+
"step": 27100
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 32.0,
|
| 1881 |
+
"learning_rate": 8.865621453363166e-05,
|
| 1882 |
+
"loss": 0.527,
|
| 1883 |
+
"step": 27200
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 32.12,
|
| 1887 |
+
"learning_rate": 8.592421729392543e-05,
|
| 1888 |
+
"loss": 0.4913,
|
| 1889 |
+
"step": 27300
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 32.24,
|
| 1893 |
+
"learning_rate": 8.295002566985007e-05,
|
| 1894 |
+
"loss": 0.4655,
|
| 1895 |
+
"step": 27400
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 32.35,
|
| 1899 |
+
"learning_rate": 7.975369110757874e-05,
|
| 1900 |
+
"loss": 0.465,
|
| 1901 |
+
"step": 27500
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 32.47,
|
| 1905 |
+
"learning_rate": 7.635676269961402e-05,
|
| 1906 |
+
"loss": 0.4849,
|
| 1907 |
+
"step": 27600
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 32.59,
|
| 1911 |
+
"learning_rate": 7.278214190481817e-05,
|
| 1912 |
+
"loss": 0.4657,
|
| 1913 |
+
"step": 27700
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 32.71,
|
| 1917 |
+
"learning_rate": 6.905392815104131e-05,
|
| 1918 |
+
"loss": 0.4617,
|
| 1919 |
+
"step": 27800
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 32.82,
|
| 1923 |
+
"learning_rate": 6.51972563612662e-05,
|
| 1924 |
+
"loss": 0.4816,
|
| 1925 |
+
"step": 27900
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 32.94,
|
| 1929 |
+
"learning_rate": 6.123812749863665e-05,
|
| 1930 |
+
"loss": 0.4697,
|
| 1931 |
+
"step": 28000
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 32.94,
|
| 1935 |
+
"eval_loss": 0.427847295999527,
|
| 1936 |
+
"eval_runtime": 105.1717,
|
| 1937 |
+
"eval_samples_per_second": 40.058,
|
| 1938 |
+
"eval_steps_per_second": 0.628,
|
| 1939 |
+
"eval_wer": 0.3773660894079742,
|
| 1940 |
+
"step": 28000
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 33.06,
|
| 1944 |
+
"learning_rate": 5.7203233272805804e-05,
|
| 1945 |
+
"loss": 0.4574,
|
| 1946 |
+
"step": 28100
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 33.18,
|
| 1950 |
+
"learning_rate": 5.311977618939895e-05,
|
| 1951 |
+
"loss": 0.4532,
|
| 1952 |
+
"step": 28200
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 33.29,
|
| 1956 |
+
"learning_rate": 4.9015286155785297e-05,
|
| 1957 |
+
"loss": 0.4403,
|
| 1958 |
+
"step": 28300
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 33.41,
|
| 1962 |
+
"learning_rate": 4.4917434879568456e-05,
|
| 1963 |
+
"loss": 0.435,
|
| 1964 |
+
"step": 28400
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 33.53,
|
| 1968 |
+
"learning_rate": 4.085384931108528e-05,
|
| 1969 |
+
"loss": 0.4277,
|
| 1970 |
+
"step": 28500
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 33.65,
|
| 1974 |
+
"learning_rate": 3.6851925387651344e-05,
|
| 1975 |
+
"loss": 0.4153,
|
| 1976 |
+
"step": 28600
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 33.76,
|
| 1980 |
+
"learning_rate": 3.293864333525176e-05,
|
| 1981 |
+
"loss": 0.4235,
|
| 1982 |
+
"step": 28700
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 33.88,
|
| 1986 |
+
"learning_rate": 2.914038577288113e-05,
|
| 1987 |
+
"loss": 0.4215,
|
| 1988 |
+
"step": 28800
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 34.0,
|
| 1992 |
+
"learning_rate": 2.5482759845838445e-05,
|
| 1993 |
+
"loss": 0.4057,
|
| 1994 |
+
"step": 28900
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 34.12,
|
| 1998 |
+
"learning_rate": 2.1990424587120746e-05,
|
| 1999 |
+
"loss": 0.399,
|
| 2000 |
+
"step": 29000
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 34.12,
|
| 2004 |
+
"eval_loss": 0.4379839301109314,
|
| 2005 |
+
"eval_runtime": 105.199,
|
| 2006 |
+
"eval_samples_per_second": 40.048,
|
| 2007 |
+
"eval_steps_per_second": 0.627,
|
| 2008 |
+
"eval_wer": 0.3649617398308498,
|
| 2009 |
+
"step": 29000
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 34.24,
|
| 2013 |
+
"learning_rate": 1.868692467081464e-05,
|
| 2014 |
+
"loss": 0.3893,
|
| 2015 |
+
"step": 29100
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 34.35,
|
| 2019 |
+
"learning_rate": 1.559453167828669e-05,
|
| 2020 |
+
"loss": 0.404,
|
| 2021 |
+
"step": 29200
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 34.47,
|
| 2025 |
+
"learning_rate": 1.2734093947328829e-05,
|
| 2026 |
+
"loss": 0.3977,
|
| 2027 |
+
"step": 29300
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 34.59,
|
| 2031 |
+
"learning_rate": 1.0124896016546576e-05,
|
| 2032 |
+
"loss": 0.4139,
|
| 2033 |
+
"step": 29400
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 34.71,
|
| 2037 |
+
"learning_rate": 7.784528612590087e-06,
|
| 2038 |
+
"loss": 0.3903,
|
| 2039 |
+
"step": 29500
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 34.82,
|
| 2043 |
+
"learning_rate": 5.728770056750632e-06,
|
| 2044 |
+
"loss": 0.3923,
|
| 2045 |
+
"step": 29600
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 34.94,
|
| 2049 |
+
"learning_rate": 3.971479890455959e-06,
|
| 2050 |
+
"loss": 0.3927,
|
| 2051 |
+
"step": 29700
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 35.06,
|
| 2055 |
+
"learning_rate": 2.5245054368224164e-06,
|
| 2056 |
+
"loss": 0.3755,
|
| 2057 |
+
"step": 29800
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 35.18,
|
| 2061 |
+
"learning_rate": 1.4072598492477495e-06,
|
| 2062 |
+
"loss": 0.385,
|
| 2063 |
+
"step": 29900
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 35.29,
|
| 2067 |
+
"learning_rate": 6.047175874878519e-07,
|
| 2068 |
+
"loss": 0.3929,
|
| 2069 |
+
"step": 30000
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 35.29,
|
| 2073 |
+
"eval_loss": 0.441159188747406,
|
| 2074 |
+
"eval_runtime": 104.6164,
|
| 2075 |
+
"eval_samples_per_second": 40.271,
|
| 2076 |
+
"eval_steps_per_second": 0.631,
|
| 2077 |
+
"eval_wer": 0.3600483286347161,
|
| 2078 |
+
"step": 30000
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 35.41,
|
| 2082 |
+
"learning_rate": 1.3518912298790543e-07,
|
| 2083 |
+
"loss": 0.3965,
|
| 2084 |
+
"step": 30100
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 35.53,
|
| 2088 |
+
"learning_rate": 9.999816007076456e-05,
|
| 2089 |
+
"loss": 0.3789,
|
| 2090 |
+
"step": 30200
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 35.65,
|
| 2094 |
+
"learning_rate": 9.979443097833314e-05,
|
| 2095 |
+
"loss": 0.4195,
|
| 2096 |
+
"step": 30300
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 35.76,
|
| 2100 |
+
"learning_rate": 9.92549971033273e-05,
|
| 2101 |
+
"loss": 0.4553,
|
| 2102 |
+
"step": 30400
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 35.88,
|
| 2106 |
+
"learning_rate": 9.83834952084967e-05,
|
| 2107 |
+
"loss": 0.4391,
|
| 2108 |
+
"step": 30500
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 36.0,
|
| 2112 |
+
"learning_rate": 9.718580079735384e-05,
|
| 2113 |
+
"loss": 0.4444,
|
| 2114 |
+
"step": 30600
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 36.12,
|
| 2118 |
+
"learning_rate": 9.566998850262337e-05,
|
| 2119 |
+
"loss": 0.4345,
|
| 2120 |
+
"step": 30700
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 36.24,
|
| 2124 |
+
"learning_rate": 9.384627764857165e-05,
|
| 2125 |
+
"loss": 0.4462,
|
| 2126 |
+
"step": 30800
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 36.35,
|
| 2130 |
+
"learning_rate": 9.172696335422526e-05,
|
| 2131 |
+
"loss": 0.4393,
|
| 2132 |
+
"step": 30900
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 36.47,
|
| 2136 |
+
"learning_rate": 8.932633364196772e-05,
|
| 2137 |
+
"loss": 0.4403,
|
| 2138 |
+
"step": 31000
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 36.47,
|
| 2142 |
+
"eval_loss": 0.4456220269203186,
|
| 2143 |
+
"eval_runtime": 104.0661,
|
| 2144 |
+
"eval_samples_per_second": 40.484,
|
| 2145 |
+
"eval_steps_per_second": 0.634,
|
| 2146 |
+
"eval_wer": 0.3653644784534837,
|
| 2147 |
+
"step": 31000
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 36.59,
|
| 2151 |
+
"learning_rate": 8.66605731103526e-05,
|
| 2152 |
+
"loss": 0.4542,
|
| 2153 |
+
"step": 31100
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 36.71,
|
| 2157 |
+
"learning_rate": 8.374765382055385e-05,
|
| 2158 |
+
"loss": 0.4522,
|
| 2159 |
+
"step": 31200
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 36.82,
|
| 2163 |
+
"learning_rate": 8.060721413207527e-05,
|
| 2164 |
+
"loss": 0.4581,
|
| 2165 |
+
"step": 31300
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 36.94,
|
| 2169 |
+
"learning_rate": 7.726042630458715e-05,
|
| 2170 |
+
"loss": 0.439,
|
| 2171 |
+
"step": 31400
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 37.06,
|
| 2175 |
+
"learning_rate": 7.372985375849379e-05,
|
| 2176 |
+
"loss": 0.4288,
|
| 2177 |
+
"step": 31500
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 37.18,
|
| 2181 |
+
"learning_rate": 7.003929895655356e-05,
|
| 2182 |
+
"loss": 0.4188,
|
| 2183 |
+
"step": 31600
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 37.29,
|
| 2187 |
+
"learning_rate": 6.621364293210756e-05,
|
| 2188 |
+
"loss": 0.4072,
|
| 2189 |
+
"step": 31700
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 37.41,
|
| 2193 |
+
"learning_rate": 6.227867754578468e-05,
|
| 2194 |
+
"loss": 0.4275,
|
| 2195 |
+
"step": 31800
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 37.53,
|
| 2199 |
+
"learning_rate": 5.83014302576376e-05,
|
| 2200 |
+
"loss": 0.4064,
|
| 2201 |
+
"step": 31900
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 37.65,
|
| 2205 |
+
"learning_rate": 5.42284121214101e-05,
|
| 2206 |
+
"loss": 0.3923,
|
| 2207 |
+
"step": 32000
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 37.65,
|
| 2211 |
+
"eval_loss": 0.4511985182762146,
|
| 2212 |
+
"eval_runtime": 104.7283,
|
| 2213 |
+
"eval_samples_per_second": 40.228,
|
| 2214 |
+
"eval_steps_per_second": 0.63,
|
| 2215 |
+
"eval_wer": 0.36568666935159083,
|
| 2216 |
+
"step": 32000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 37.76,
|
| 2220 |
+
"learning_rate": 5.012688681793577e-05,
|
| 2221 |
+
"loss": 0.3951,
|
| 2222 |
+
"step": 32100
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 37.88,
|
| 2226 |
+
"learning_rate": 4.6024506067159605e-05,
|
| 2227 |
+
"loss": 0.4101,
|
| 2228 |
+
"step": 32200
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 38.0,
|
| 2232 |
+
"learning_rate": 4.19489273562929e-05,
|
| 2233 |
+
"loss": 0.4005,
|
| 2234 |
+
"step": 32300
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 38.12,
|
| 2238 |
+
"learning_rate": 3.792762747818408e-05,
|
| 2239 |
+
"loss": 0.3798,
|
| 2240 |
+
"step": 32400
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 38.24,
|
| 2244 |
+
"learning_rate": 3.398771728789724e-05,
|
| 2245 |
+
"loss": 0.3547,
|
| 2246 |
+
"step": 32500
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 38.35,
|
| 2250 |
+
"learning_rate": 3.0155758926374532e-05,
|
| 2251 |
+
"loss": 0.3815,
|
| 2252 |
+
"step": 32600
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 38.47,
|
| 2256 |
+
"learning_rate": 2.6457586743428208e-05,
|
| 2257 |
+
"loss": 0.3657,
|
| 2258 |
+
"step": 32700
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 38.59,
|
| 2262 |
+
"learning_rate": 2.2918133127363645e-05,
|
| 2263 |
+
"loss": 0.3665,
|
| 2264 |
+
"step": 32800
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 38.71,
|
| 2268 |
+
"learning_rate": 1.9561260415459726e-05,
|
| 2269 |
+
"loss": 0.3522,
|
| 2270 |
+
"step": 32900
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 38.82,
|
| 2274 |
+
"learning_rate": 1.640960001853372e-05,
|
| 2275 |
+
"loss": 0.3571,
|
| 2276 |
+
"step": 33000
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 38.82,
|
| 2280 |
+
"eval_loss": 0.45314452052116394,
|
| 2281 |
+
"eval_runtime": 105.5791,
|
| 2282 |
+
"eval_samples_per_second": 39.904,
|
| 2283 |
+
"eval_steps_per_second": 0.625,
|
| 2284 |
+
"eval_wer": 0.3557792992347966,
|
| 2285 |
+
"step": 33000
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 38.94,
|
| 2289 |
+
"learning_rate": 1.3484399844182983e-05,
|
| 2290 |
+
"loss": 0.3616,
|
| 2291 |
+
"step": 33100
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 39.06,
|
| 2295 |
+
"learning_rate": 1.0805381047348467e-05,
|
| 2296 |
+
"loss": 0.3522,
|
| 2297 |
+
"step": 33200
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 39.18,
|
| 2301 |
+
"learning_rate": 8.390605073959684e-06,
|
| 2302 |
+
"loss": 0.3527,
|
| 2303 |
+
"step": 33300
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 39.29,
|
| 2307 |
+
"learning_rate": 6.256351894029011e-06,
|
| 2308 |
+
"loss": 0.3405,
|
| 2309 |
+
"step": 33400
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 39.41,
|
| 2313 |
+
"learning_rate": 4.4170102451251805e-06,
|
| 2314 |
+
"loss": 0.3488,
|
| 2315 |
+
"step": 33500
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 39.53,
|
| 2319 |
+
"learning_rate": 2.884980626183248e-06,
|
| 2320 |
+
"loss": 0.3519,
|
| 2321 |
+
"step": 33600
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 39.65,
|
| 2325 |
+
"learning_rate": 1.67059169565002e-06,
|
| 2326 |
+
"loss": 0.3395,
|
| 2327 |
+
"step": 33700
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 39.76,
|
| 2331 |
+
"learning_rate": 7.820306375915687e-07,
|
| 2332 |
+
"loss": 0.3404,
|
| 2333 |
+
"step": 33800
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 39.88,
|
| 2337 |
+
"learning_rate": 2.2528796522271467e-07,
|
| 2338 |
+
"loss": 0.3447,
|
| 2339 |
+
"step": 33900
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 40.0,
|
| 2343 |
+
"learning_rate": 4.117133982245447e-09,
|
| 2344 |
+
"loss": 0.3482,
|
| 2345 |
+
"step": 34000
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 40.0,
|
| 2349 |
+
"eval_loss": 0.4571866989135742,
|
| 2350 |
+
"eval_runtime": 103.9248,
|
| 2351 |
+
"eval_samples_per_second": 40.539,
|
| 2352 |
+
"eval_steps_per_second": 0.635,
|
| 2353 |
+
"eval_wer": 0.3579943616592831,
|
| 2354 |
+
"step": 34000
|
| 2355 |
+
}
|
| 2356 |
+
],
|
| 2357 |
+
"max_steps": 34000,
|
| 2358 |
+
"num_train_epochs": 40,
|
| 2359 |
+
"total_flos": 3.007490366928798e+20,
|
| 2360 |
+
"trial_name": null,
|
| 2361 |
+
"trial_params": null
|
| 2362 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd3f14de6cf1bf80b9b0b7e0038a72df02945b2e8e66e4118507b85fb159854a
|
| 3 |
+
size 3951
|