Commit
•
30a8e1c
1
Parent(s):
47aa759
init
Browse files- all_results.json +9 -0
- config.json +115 -0
- preprocessor_config.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1135 -0
- training_args.bin +3 -0
- vocab.json +1 -0
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 2.7374813261347353e+19,
|
| 4 |
+
"train_loss": 0.0985529641951284,
|
| 5 |
+
"train_runtime": 13747.1961,
|
| 6 |
+
"train_samples": 3965,
|
| 7 |
+
"train_samples_per_second": 17.305,
|
| 8 |
+
"train_steps_per_second": 0.541
|
| 9 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/wav2vec2-large-xlsr-53",
|
| 3 |
+
"activation_dropout": 0.055,
|
| 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": 0.094,
|
| 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": true,
|
| 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": 0.04,
|
| 53 |
+
"feat_quantizer_dropout": 0.0,
|
| 54 |
+
"final_dropout": 0.0,
|
| 55 |
+
"hidden_act": "gelu",
|
| 56 |
+
"hidden_dropout": 0.047,
|
| 57 |
+
"hidden_size": 1024,
|
| 58 |
+
"initializer_range": 0.02,
|
| 59 |
+
"intermediate_size": 4096,
|
| 60 |
+
"layer_norm_eps": 1e-05,
|
| 61 |
+
"layerdrop": 0.041,
|
| 62 |
+
"mask_channel_length": 10,
|
| 63 |
+
"mask_channel_min_space": 1,
|
| 64 |
+
"mask_channel_other": 0.0,
|
| 65 |
+
"mask_channel_prob": 0.0,
|
| 66 |
+
"mask_channel_selection": "static",
|
| 67 |
+
"mask_feature_length": 10,
|
| 68 |
+
"mask_feature_min_masks": 0,
|
| 69 |
+
"mask_feature_prob": 0.0,
|
| 70 |
+
"mask_time_length": 10,
|
| 71 |
+
"mask_time_min_masks": 2,
|
| 72 |
+
"mask_time_min_space": 1,
|
| 73 |
+
"mask_time_other": 0.0,
|
| 74 |
+
"mask_time_prob": 0.4,
|
| 75 |
+
"mask_time_selection": "static",
|
| 76 |
+
"model_type": "wav2vec2",
|
| 77 |
+
"num_adapter_layers": 3,
|
| 78 |
+
"num_attention_heads": 16,
|
| 79 |
+
"num_codevector_groups": 2,
|
| 80 |
+
"num_codevectors_per_group": 320,
|
| 81 |
+
"num_conv_pos_embedding_groups": 16,
|
| 82 |
+
"num_conv_pos_embeddings": 128,
|
| 83 |
+
"num_feat_extract_layers": 7,
|
| 84 |
+
"num_hidden_layers": 24,
|
| 85 |
+
"num_negatives": 100,
|
| 86 |
+
"output_hidden_size": 1024,
|
| 87 |
+
"pad_token_id": 28,
|
| 88 |
+
"proj_codevector_dim": 768,
|
| 89 |
+
"tdnn_dilation": [
|
| 90 |
+
1,
|
| 91 |
+
2,
|
| 92 |
+
3,
|
| 93 |
+
1,
|
| 94 |
+
1
|
| 95 |
+
],
|
| 96 |
+
"tdnn_dim": [
|
| 97 |
+
512,
|
| 98 |
+
512,
|
| 99 |
+
512,
|
| 100 |
+
512,
|
| 101 |
+
1500
|
| 102 |
+
],
|
| 103 |
+
"tdnn_kernel": [
|
| 104 |
+
5,
|
| 105 |
+
3,
|
| 106 |
+
3,
|
| 107 |
+
1,
|
| 108 |
+
1
|
| 109 |
+
],
|
| 110 |
+
"torch_dtype": "float32",
|
| 111 |
+
"transformers_version": "4.18.0",
|
| 112 |
+
"use_weighted_layer_sum": false,
|
| 113 |
+
"vocab_size": 29,
|
| 114 |
+
"xvector_output_dim": 512
|
| 115 |
+
}
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
| 4 |
+
"feature_size": 1,
|
| 5 |
+
"padding_side": "right",
|
| 6 |
+
"padding_value": 0.0,
|
| 7 |
+
"return_attention_mask": true,
|
| 8 |
+
"sampling_rate": 16000
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"total_flos": 2.7374813261347353e+19,
|
| 4 |
+
"train_loss": 0.0985529641951284,
|
| 5 |
+
"train_runtime": 13747.1961,
|
| 6 |
+
"train_samples": 3965,
|
| 7 |
+
"train_samples_per_second": 17.305,
|
| 8 |
+
"train_steps_per_second": 0.541
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1135 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"global_step": 7440,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.81,
|
| 12 |
+
"learning_rate": 1.2903225806451613e-05,
|
| 13 |
+
"loss": 13.4856,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.81,
|
| 18 |
+
"eval_loss": 12.348929405212402,
|
| 19 |
+
"eval_runtime": 128.1609,
|
| 20 |
+
"eval_samples_per_second": 14.388,
|
| 21 |
+
"eval_steps_per_second": 1.802,
|
| 22 |
+
"eval_wer": 1.0,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.61,
|
| 27 |
+
"learning_rate": 2.6344086021505376e-05,
|
| 28 |
+
"loss": 4.9759,
|
| 29 |
+
"step": 200
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 1.61,
|
| 33 |
+
"eval_loss": 3.1773388385772705,
|
| 34 |
+
"eval_runtime": 128.7202,
|
| 35 |
+
"eval_samples_per_second": 14.326,
|
| 36 |
+
"eval_steps_per_second": 1.795,
|
| 37 |
+
"eval_wer": 1.0,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.42,
|
| 42 |
+
"learning_rate": 3.978494623655914e-05,
|
| 43 |
+
"loss": 2.9905,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.42,
|
| 48 |
+
"eval_loss": 2.9219164848327637,
|
| 49 |
+
"eval_runtime": 129.5213,
|
| 50 |
+
"eval_samples_per_second": 14.237,
|
| 51 |
+
"eval_steps_per_second": 1.783,
|
| 52 |
+
"eval_wer": 1.0,
|
| 53 |
+
"step": 300
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 3.23,
|
| 57 |
+
"learning_rate": 5.32258064516129e-05,
|
| 58 |
+
"loss": 2.8774,
|
| 59 |
+
"step": 400
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 3.23,
|
| 63 |
+
"eval_loss": 2.8899402618408203,
|
| 64 |
+
"eval_runtime": 129.0058,
|
| 65 |
+
"eval_samples_per_second": 14.294,
|
| 66 |
+
"eval_steps_per_second": 1.791,
|
| 67 |
+
"eval_wer": 1.0,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 4.03,
|
| 72 |
+
"learning_rate": 6.666666666666667e-05,
|
| 73 |
+
"loss": 2.832,
|
| 74 |
+
"step": 500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 4.03,
|
| 78 |
+
"eval_loss": 2.787729501724243,
|
| 79 |
+
"eval_runtime": 129.1257,
|
| 80 |
+
"eval_samples_per_second": 14.281,
|
| 81 |
+
"eval_steps_per_second": 1.789,
|
| 82 |
+
"eval_wer": 1.0,
|
| 83 |
+
"step": 500
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 4.84,
|
| 87 |
+
"learning_rate": 8.010752688172043e-05,
|
| 88 |
+
"loss": 2.6921,
|
| 89 |
+
"step": 600
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 4.84,
|
| 93 |
+
"eval_loss": 2.3222157955169678,
|
| 94 |
+
"eval_runtime": 129.5689,
|
| 95 |
+
"eval_samples_per_second": 14.232,
|
| 96 |
+
"eval_steps_per_second": 1.783,
|
| 97 |
+
"eval_wer": 1.020324445273168,
|
| 98 |
+
"step": 600
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 5.65,
|
| 102 |
+
"learning_rate": 9.35483870967742e-05,
|
| 103 |
+
"loss": 1.7609,
|
| 104 |
+
"step": 700
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 5.65,
|
| 108 |
+
"eval_loss": 0.7123318910598755,
|
| 109 |
+
"eval_runtime": 129.8973,
|
| 110 |
+
"eval_samples_per_second": 14.196,
|
| 111 |
+
"eval_steps_per_second": 1.778,
|
| 112 |
+
"eval_wer": 0.7778295730001865,
|
| 113 |
+
"step": 700
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 6.45,
|
| 117 |
+
"learning_rate": 0.0001,
|
| 118 |
+
"loss": 1.0829,
|
| 119 |
+
"step": 800
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 6.45,
|
| 123 |
+
"eval_loss": 0.5219887495040894,
|
| 124 |
+
"eval_runtime": 130.0287,
|
| 125 |
+
"eval_samples_per_second": 14.181,
|
| 126 |
+
"eval_steps_per_second": 1.777,
|
| 127 |
+
"eval_wer": 0.6609173969793026,
|
| 128 |
+
"step": 800
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 7.26,
|
| 132 |
+
"learning_rate": 0.0001,
|
| 133 |
+
"loss": 0.8991,
|
| 134 |
+
"step": 900
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 7.26,
|
| 138 |
+
"eval_loss": 0.4353589415550232,
|
| 139 |
+
"eval_runtime": 129.1246,
|
| 140 |
+
"eval_samples_per_second": 14.281,
|
| 141 |
+
"eval_steps_per_second": 1.789,
|
| 142 |
+
"eval_wer": 0.5778482192802535,
|
| 143 |
+
"step": 900
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 8.06,
|
| 147 |
+
"learning_rate": 0.0001,
|
| 148 |
+
"loss": 0.7943,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 8.06,
|
| 153 |
+
"eval_loss": 0.409180223941803,
|
| 154 |
+
"eval_runtime": 129.6166,
|
| 155 |
+
"eval_samples_per_second": 14.227,
|
| 156 |
+
"eval_steps_per_second": 1.782,
|
| 157 |
+
"eval_wer": 0.5485735595748649,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 8.87,
|
| 162 |
+
"learning_rate": 0.0001,
|
| 163 |
+
"loss": 0.7319,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 8.87,
|
| 168 |
+
"eval_loss": 0.37231260538101196,
|
| 169 |
+
"eval_runtime": 129.5968,
|
| 170 |
+
"eval_samples_per_second": 14.229,
|
| 171 |
+
"eval_steps_per_second": 1.782,
|
| 172 |
+
"eval_wer": 0.5166884206600784,
|
| 173 |
+
"step": 1100
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 9.68,
|
| 177 |
+
"learning_rate": 0.0001,
|
| 178 |
+
"loss": 0.6875,
|
| 179 |
+
"step": 1200
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 9.68,
|
| 183 |
+
"eval_loss": 0.3546667993068695,
|
| 184 |
+
"eval_runtime": 129.8844,
|
| 185 |
+
"eval_samples_per_second": 14.197,
|
| 186 |
+
"eval_steps_per_second": 1.779,
|
| 187 |
+
"eval_wer": 0.5021443222077195,
|
| 188 |
+
"step": 1200
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 10.48,
|
| 192 |
+
"learning_rate": 0.0001,
|
| 193 |
+
"loss": 0.6538,
|
| 194 |
+
"step": 1300
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 10.48,
|
| 198 |
+
"eval_loss": 0.3382996916770935,
|
| 199 |
+
"eval_runtime": 130.4091,
|
| 200 |
+
"eval_samples_per_second": 14.14,
|
| 201 |
+
"eval_steps_per_second": 1.771,
|
| 202 |
+
"eval_wer": 0.5028901734104047,
|
| 203 |
+
"step": 1300
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 11.29,
|
| 207 |
+
"learning_rate": 0.0001,
|
| 208 |
+
"loss": 0.6385,
|
| 209 |
+
"step": 1400
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 11.29,
|
| 213 |
+
"eval_loss": 0.33781182765960693,
|
| 214 |
+
"eval_runtime": 129.3767,
|
| 215 |
+
"eval_samples_per_second": 14.253,
|
| 216 |
+
"eval_steps_per_second": 1.785,
|
| 217 |
+
"eval_wer": 0.4781838523214619,
|
| 218 |
+
"step": 1400
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 12.1,
|
| 222 |
+
"learning_rate": 0.0001,
|
| 223 |
+
"loss": 0.6031,
|
| 224 |
+
"step": 1500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 12.1,
|
| 228 |
+
"eval_loss": 0.33670100569725037,
|
| 229 |
+
"eval_runtime": 130.1665,
|
| 230 |
+
"eval_samples_per_second": 14.166,
|
| 231 |
+
"eval_steps_per_second": 1.775,
|
| 232 |
+
"eval_wer": 0.47445459630803655,
|
| 233 |
+
"step": 1500
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 12.9,
|
| 237 |
+
"learning_rate": 0.0001,
|
| 238 |
+
"loss": 0.5778,
|
| 239 |
+
"step": 1600
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 12.9,
|
| 243 |
+
"eval_loss": 0.3219880163669586,
|
| 244 |
+
"eval_runtime": 130.1154,
|
| 245 |
+
"eval_samples_per_second": 14.172,
|
| 246 |
+
"eval_steps_per_second": 1.775,
|
| 247 |
+
"eval_wer": 0.4559015476412456,
|
| 248 |
+
"step": 1600
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 13.71,
|
| 252 |
+
"learning_rate": 0.0001,
|
| 253 |
+
"loss": 0.5531,
|
| 254 |
+
"step": 1700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 13.71,
|
| 258 |
+
"eval_loss": 0.3436262309551239,
|
| 259 |
+
"eval_runtime": 130.5395,
|
| 260 |
+
"eval_samples_per_second": 14.126,
|
| 261 |
+
"eval_steps_per_second": 1.77,
|
| 262 |
+
"eval_wer": 0.4655043818758158,
|
| 263 |
+
"step": 1700
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 14.52,
|
| 267 |
+
"learning_rate": 0.0001,
|
| 268 |
+
"loss": 0.5506,
|
| 269 |
+
"step": 1800
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 14.52,
|
| 273 |
+
"eval_loss": 0.32741010189056396,
|
| 274 |
+
"eval_runtime": 130.5027,
|
| 275 |
+
"eval_samples_per_second": 14.13,
|
| 276 |
+
"eval_steps_per_second": 1.77,
|
| 277 |
+
"eval_wer": 0.4504008950214432,
|
| 278 |
+
"step": 1800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 15.32,
|
| 282 |
+
"learning_rate": 0.0001,
|
| 283 |
+
"loss": 0.5186,
|
| 284 |
+
"step": 1900
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 15.32,
|
| 288 |
+
"eval_loss": 0.3106687664985657,
|
| 289 |
+
"eval_runtime": 131.9389,
|
| 290 |
+
"eval_samples_per_second": 13.976,
|
| 291 |
+
"eval_steps_per_second": 1.751,
|
| 292 |
+
"eval_wer": 0.44368823419727765,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 16.13,
|
| 297 |
+
"learning_rate": 0.0001,
|
| 298 |
+
"loss": 0.5065,
|
| 299 |
+
"step": 2000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 16.13,
|
| 303 |
+
"eval_loss": 0.3165331482887268,
|
| 304 |
+
"eval_runtime": 130.835,
|
| 305 |
+
"eval_samples_per_second": 14.094,
|
| 306 |
+
"eval_steps_per_second": 1.766,
|
| 307 |
+
"eval_wer": 0.44005220958418795,
|
| 308 |
+
"step": 2000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 16.94,
|
| 312 |
+
"learning_rate": 0.0001,
|
| 313 |
+
"loss": 0.5016,
|
| 314 |
+
"step": 2100
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 16.94,
|
| 318 |
+
"eval_loss": 0.3141399919986725,
|
| 319 |
+
"eval_runtime": 131.1616,
|
| 320 |
+
"eval_samples_per_second": 14.059,
|
| 321 |
+
"eval_steps_per_second": 1.761,
|
| 322 |
+
"eval_wer": 0.440798060786873,
|
| 323 |
+
"step": 2100
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 17.74,
|
| 327 |
+
"learning_rate": 0.0001,
|
| 328 |
+
"loss": 0.4878,
|
| 329 |
+
"step": 2200
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 17.74,
|
| 333 |
+
"eval_loss": 0.323138564825058,
|
| 334 |
+
"eval_runtime": 129.6671,
|
| 335 |
+
"eval_samples_per_second": 14.221,
|
| 336 |
+
"eval_steps_per_second": 1.781,
|
| 337 |
+
"eval_wer": 0.4408912921872086,
|
| 338 |
+
"step": 2200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 18.55,
|
| 342 |
+
"learning_rate": 0.0001,
|
| 343 |
+
"loss": 0.4701,
|
| 344 |
+
"step": 2300
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 18.55,
|
| 348 |
+
"eval_loss": 0.314998984336853,
|
| 349 |
+
"eval_runtime": 130.2103,
|
| 350 |
+
"eval_samples_per_second": 14.162,
|
| 351 |
+
"eval_steps_per_second": 1.774,
|
| 352 |
+
"eval_wer": 0.441823606190565,
|
| 353 |
+
"step": 2300
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 19.35,
|
| 357 |
+
"learning_rate": 0.0001,
|
| 358 |
+
"loss": 0.4643,
|
| 359 |
+
"step": 2400
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 19.35,
|
| 363 |
+
"eval_loss": 0.32002875208854675,
|
| 364 |
+
"eval_runtime": 129.7219,
|
| 365 |
+
"eval_samples_per_second": 14.215,
|
| 366 |
+
"eval_steps_per_second": 1.781,
|
| 367 |
+
"eval_wer": 0.43641618497109824,
|
| 368 |
+
"step": 2400
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 20.16,
|
| 372 |
+
"learning_rate": 0.0001,
|
| 373 |
+
"loss": 0.4623,
|
| 374 |
+
"step": 2500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 20.16,
|
| 378 |
+
"eval_loss": 0.3018810749053955,
|
| 379 |
+
"eval_runtime": 127.3741,
|
| 380 |
+
"eval_samples_per_second": 14.477,
|
| 381 |
+
"eval_steps_per_second": 1.814,
|
| 382 |
+
"eval_wer": 0.42187208651873953,
|
| 383 |
+
"step": 2500
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 20.97,
|
| 387 |
+
"learning_rate": 0.0001,
|
| 388 |
+
"loss": 0.4497,
|
| 389 |
+
"step": 2600
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 20.97,
|
| 393 |
+
"eval_loss": 0.3108297288417816,
|
| 394 |
+
"eval_runtime": 127.9529,
|
| 395 |
+
"eval_samples_per_second": 14.412,
|
| 396 |
+
"eval_steps_per_second": 1.805,
|
| 397 |
+
"eval_wer": 0.43054260674995337,
|
| 398 |
+
"step": 2600
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 21.77,
|
| 402 |
+
"learning_rate": 0.0001,
|
| 403 |
+
"loss": 0.4439,
|
| 404 |
+
"step": 2700
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 21.77,
|
| 408 |
+
"eval_loss": 0.31016433238983154,
|
| 409 |
+
"eval_runtime": 127.0701,
|
| 410 |
+
"eval_samples_per_second": 14.512,
|
| 411 |
+
"eval_steps_per_second": 1.818,
|
| 412 |
+
"eval_wer": 0.42429610292746595,
|
| 413 |
+
"step": 2700
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 22.58,
|
| 417 |
+
"learning_rate": 0.0001,
|
| 418 |
+
"loss": 0.4289,
|
| 419 |
+
"step": 2800
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 22.58,
|
| 423 |
+
"eval_loss": 0.29791948199272156,
|
| 424 |
+
"eval_runtime": 128.4647,
|
| 425 |
+
"eval_samples_per_second": 14.354,
|
| 426 |
+
"eval_steps_per_second": 1.798,
|
| 427 |
+
"eval_wer": 0.41935483870967744,
|
| 428 |
+
"step": 2800
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 23.39,
|
| 432 |
+
"learning_rate": 0.0001,
|
| 433 |
+
"loss": 0.423,
|
| 434 |
+
"step": 2900
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 23.39,
|
| 438 |
+
"eval_loss": 0.3108172118663788,
|
| 439 |
+
"eval_runtime": 128.2048,
|
| 440 |
+
"eval_samples_per_second": 14.383,
|
| 441 |
+
"eval_steps_per_second": 1.802,
|
| 442 |
+
"eval_wer": 0.41832929330598545,
|
| 443 |
+
"step": 2900
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 24.19,
|
| 447 |
+
"learning_rate": 0.0001,
|
| 448 |
+
"loss": 0.4214,
|
| 449 |
+
"step": 3000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 24.19,
|
| 453 |
+
"eval_loss": 0.3556348383426666,
|
| 454 |
+
"eval_runtime": 128.3694,
|
| 455 |
+
"eval_samples_per_second": 14.365,
|
| 456 |
+
"eval_steps_per_second": 1.799,
|
| 457 |
+
"eval_wer": 0.42000745851202687,
|
| 458 |
+
"step": 3000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 25.0,
|
| 462 |
+
"learning_rate": 0.0001,
|
| 463 |
+
"loss": 0.4139,
|
| 464 |
+
"step": 3100
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 25.0,
|
| 468 |
+
"eval_loss": 0.3197577893733978,
|
| 469 |
+
"eval_runtime": 128.1255,
|
| 470 |
+
"eval_samples_per_second": 14.392,
|
| 471 |
+
"eval_steps_per_second": 1.803,
|
| 472 |
+
"eval_wer": 0.42243147492075334,
|
| 473 |
+
"step": 3100
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 25.81,
|
| 477 |
+
"learning_rate": 0.0001,
|
| 478 |
+
"loss": 0.4015,
|
| 479 |
+
"step": 3200
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 25.81,
|
| 483 |
+
"eval_loss": 0.29826977849006653,
|
| 484 |
+
"eval_runtime": 128.9516,
|
| 485 |
+
"eval_samples_per_second": 14.3,
|
| 486 |
+
"eval_steps_per_second": 1.791,
|
| 487 |
+
"eval_wer": 0.4151594256945739,
|
| 488 |
+
"step": 3200
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 26.61,
|
| 492 |
+
"learning_rate": 0.0001,
|
| 493 |
+
"loss": 0.4044,
|
| 494 |
+
"step": 3300
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 26.61,
|
| 498 |
+
"eval_loss": 0.28930172324180603,
|
| 499 |
+
"eval_runtime": 128.281,
|
| 500 |
+
"eval_samples_per_second": 14.375,
|
| 501 |
+
"eval_steps_per_second": 1.801,
|
| 502 |
+
"eval_wer": 0.4065821368636957,
|
| 503 |
+
"step": 3300
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 27.42,
|
| 507 |
+
"learning_rate": 0.0001,
|
| 508 |
+
"loss": 0.4013,
|
| 509 |
+
"step": 3400
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 27.42,
|
| 513 |
+
"eval_loss": 0.32470086216926575,
|
| 514 |
+
"eval_runtime": 128.6984,
|
| 515 |
+
"eval_samples_per_second": 14.328,
|
| 516 |
+
"eval_steps_per_second": 1.795,
|
| 517 |
+
"eval_wer": 0.4160917396979303,
|
| 518 |
+
"step": 3400
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 28.23,
|
| 522 |
+
"learning_rate": 0.0001,
|
| 523 |
+
"loss": 0.3959,
|
| 524 |
+
"step": 3500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 28.23,
|
| 528 |
+
"eval_loss": 0.32055872678756714,
|
| 529 |
+
"eval_runtime": 128.9252,
|
| 530 |
+
"eval_samples_per_second": 14.303,
|
| 531 |
+
"eval_steps_per_second": 1.792,
|
| 532 |
+
"eval_wer": 0.4156255826962521,
|
| 533 |
+
"step": 3500
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 29.03,
|
| 537 |
+
"learning_rate": 0.0001,
|
| 538 |
+
"loss": 0.3923,
|
| 539 |
+
"step": 3600
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 29.03,
|
| 543 |
+
"eval_loss": 0.32669639587402344,
|
| 544 |
+
"eval_runtime": 129.4219,
|
| 545 |
+
"eval_samples_per_second": 14.248,
|
| 546 |
+
"eval_steps_per_second": 1.785,
|
| 547 |
+
"eval_wer": 0.41432034309155324,
|
| 548 |
+
"step": 3600
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 29.84,
|
| 552 |
+
"learning_rate": 0.0001,
|
| 553 |
+
"loss": 0.3789,
|
| 554 |
+
"step": 3700
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 29.84,
|
| 558 |
+
"eval_loss": 0.31051748991012573,
|
| 559 |
+
"eval_runtime": 129.0113,
|
| 560 |
+
"eval_samples_per_second": 14.293,
|
| 561 |
+
"eval_steps_per_second": 1.791,
|
| 562 |
+
"eval_wer": 0.4084467648704084,
|
| 563 |
+
"step": 3700
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 30.65,
|
| 567 |
+
"learning_rate": 9.795698924731184e-05,
|
| 568 |
+
"loss": 0.371,
|
| 569 |
+
"step": 3800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 30.65,
|
| 573 |
+
"eval_loss": 0.31601375341415405,
|
| 574 |
+
"eval_runtime": 130.3347,
|
| 575 |
+
"eval_samples_per_second": 14.148,
|
| 576 |
+
"eval_steps_per_second": 1.772,
|
| 577 |
+
"eval_wer": 0.41031139287712104,
|
| 578 |
+
"step": 3800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 31.45,
|
| 582 |
+
"learning_rate": 9.526881720430108e-05,
|
| 583 |
+
"loss": 0.3813,
|
| 584 |
+
"step": 3900
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 31.45,
|
| 588 |
+
"eval_loss": 0.32537344098091125,
|
| 589 |
+
"eval_runtime": 129.4643,
|
| 590 |
+
"eval_samples_per_second": 14.243,
|
| 591 |
+
"eval_steps_per_second": 1.784,
|
| 592 |
+
"eval_wer": 0.4071415252657095,
|
| 593 |
+
"step": 3900
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 32.26,
|
| 597 |
+
"learning_rate": 9.258064516129033e-05,
|
| 598 |
+
"loss": 0.3575,
|
| 599 |
+
"step": 4000
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 32.26,
|
| 603 |
+
"eval_loss": 0.35630208253860474,
|
| 604 |
+
"eval_runtime": 129.9641,
|
| 605 |
+
"eval_samples_per_second": 14.189,
|
| 606 |
+
"eval_steps_per_second": 1.777,
|
| 607 |
+
"eval_wer": 0.40928584747342905,
|
| 608 |
+
"step": 4000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 33.06,
|
| 612 |
+
"learning_rate": 8.989247311827958e-05,
|
| 613 |
+
"loss": 0.3642,
|
| 614 |
+
"step": 4100
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 33.06,
|
| 618 |
+
"eval_loss": 0.29261836409568787,
|
| 619 |
+
"eval_runtime": 129.3284,
|
| 620 |
+
"eval_samples_per_second": 14.258,
|
| 621 |
+
"eval_steps_per_second": 1.786,
|
| 622 |
+
"eval_wer": 0.4012679470445646,
|
| 623 |
+
"step": 4100
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 33.87,
|
| 627 |
+
"learning_rate": 8.720430107526883e-05,
|
| 628 |
+
"loss": 0.3494,
|
| 629 |
+
"step": 4200
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 33.87,
|
| 633 |
+
"eval_loss": 0.3040522634983063,
|
| 634 |
+
"eval_runtime": 129.9364,
|
| 635 |
+
"eval_samples_per_second": 14.192,
|
| 636 |
+
"eval_steps_per_second": 1.778,
|
| 637 |
+
"eval_wer": 0.39912362483684505,
|
| 638 |
+
"step": 4200
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 34.68,
|
| 642 |
+
"learning_rate": 8.451612903225808e-05,
|
| 643 |
+
"loss": 0.3446,
|
| 644 |
+
"step": 4300
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 34.68,
|
| 648 |
+
"eval_loss": 0.30367347598075867,
|
| 649 |
+
"eval_runtime": 129.7515,
|
| 650 |
+
"eval_samples_per_second": 14.212,
|
| 651 |
+
"eval_steps_per_second": 1.78,
|
| 652 |
+
"eval_wer": 0.3964199142271117,
|
| 653 |
+
"step": 4300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 35.48,
|
| 657 |
+
"learning_rate": 8.182795698924732e-05,
|
| 658 |
+
"loss": 0.3458,
|
| 659 |
+
"step": 4400
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 35.48,
|
| 663 |
+
"eval_loss": 0.3123793303966522,
|
| 664 |
+
"eval_runtime": 129.7166,
|
| 665 |
+
"eval_samples_per_second": 14.216,
|
| 666 |
+
"eval_steps_per_second": 1.781,
|
| 667 |
+
"eval_wer": 0.3993100876375163,
|
| 668 |
+
"step": 4400
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 36.29,
|
| 672 |
+
"learning_rate": 7.913978494623657e-05,
|
| 673 |
+
"loss": 0.3418,
|
| 674 |
+
"step": 4500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 36.29,
|
| 678 |
+
"eval_loss": 0.33982059359550476,
|
| 679 |
+
"eval_runtime": 130.2572,
|
| 680 |
+
"eval_samples_per_second": 14.157,
|
| 681 |
+
"eval_steps_per_second": 1.773,
|
| 682 |
+
"eval_wer": 0.40052209584187953,
|
| 683 |
+
"step": 4500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 37.1,
|
| 687 |
+
"learning_rate": 7.647849462365591e-05,
|
| 688 |
+
"loss": 0.3424,
|
| 689 |
+
"step": 4600
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 37.1,
|
| 693 |
+
"eval_loss": 0.3201097548007965,
|
| 694 |
+
"eval_runtime": 130.7802,
|
| 695 |
+
"eval_samples_per_second": 14.1,
|
| 696 |
+
"eval_steps_per_second": 1.766,
|
| 697 |
+
"eval_wer": 0.4044378146559761,
|
| 698 |
+
"step": 4600
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 37.9,
|
| 702 |
+
"learning_rate": 7.379032258064516e-05,
|
| 703 |
+
"loss": 0.3256,
|
| 704 |
+
"step": 4700
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 37.9,
|
| 708 |
+
"eval_loss": 0.31234973669052124,
|
| 709 |
+
"eval_runtime": 129.793,
|
| 710 |
+
"eval_samples_per_second": 14.207,
|
| 711 |
+
"eval_steps_per_second": 1.78,
|
| 712 |
+
"eval_wer": 0.39912362483684505,
|
| 713 |
+
"step": 4700
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 38.71,
|
| 717 |
+
"learning_rate": 7.110215053763441e-05,
|
| 718 |
+
"loss": 0.3183,
|
| 719 |
+
"step": 4800
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 38.71,
|
| 723 |
+
"eval_loss": 0.3197832405567169,
|
| 724 |
+
"eval_runtime": 131.4284,
|
| 725 |
+
"eval_samples_per_second": 14.03,
|
| 726 |
+
"eval_steps_per_second": 1.758,
|
| 727 |
+
"eval_wer": 0.39707253402946113,
|
| 728 |
+
"step": 4800
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 39.52,
|
| 732 |
+
"learning_rate": 6.841397849462365e-05,
|
| 733 |
+
"loss": 0.3189,
|
| 734 |
+
"step": 4900
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 39.52,
|
| 738 |
+
"eval_loss": 0.32954466342926025,
|
| 739 |
+
"eval_runtime": 135.8464,
|
| 740 |
+
"eval_samples_per_second": 13.574,
|
| 741 |
+
"eval_steps_per_second": 1.7,
|
| 742 |
+
"eval_wer": 0.4004288644415439,
|
| 743 |
+
"step": 4900
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 40.32,
|
| 747 |
+
"learning_rate": 6.57258064516129e-05,
|
| 748 |
+
"loss": 0.3227,
|
| 749 |
+
"step": 5000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 40.32,
|
| 753 |
+
"eval_loss": 0.32881081104278564,
|
| 754 |
+
"eval_runtime": 135.8344,
|
| 755 |
+
"eval_samples_per_second": 13.575,
|
| 756 |
+
"eval_steps_per_second": 1.701,
|
| 757 |
+
"eval_wer": 0.39604698862576915,
|
| 758 |
+
"step": 5000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 41.13,
|
| 762 |
+
"learning_rate": 6.303763440860215e-05,
|
| 763 |
+
"loss": 0.3007,
|
| 764 |
+
"step": 5100
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 41.13,
|
| 768 |
+
"eval_loss": 0.3239193856716156,
|
| 769 |
+
"eval_runtime": 131.4702,
|
| 770 |
+
"eval_samples_per_second": 14.026,
|
| 771 |
+
"eval_steps_per_second": 1.757,
|
| 772 |
+
"eval_wer": 0.39558083162409097,
|
| 773 |
+
"step": 5100
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 41.94,
|
| 777 |
+
"learning_rate": 6.0349462365591405e-05,
|
| 778 |
+
"loss": 0.3013,
|
| 779 |
+
"step": 5200
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 41.94,
|
| 783 |
+
"eval_loss": 0.32908666133880615,
|
| 784 |
+
"eval_runtime": 131.5819,
|
| 785 |
+
"eval_samples_per_second": 14.014,
|
| 786 |
+
"eval_steps_per_second": 1.756,
|
| 787 |
+
"eval_wer": 0.39464851762073466,
|
| 788 |
+
"step": 5200
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 42.74,
|
| 792 |
+
"learning_rate": 5.7661290322580655e-05,
|
| 793 |
+
"loss": 0.3067,
|
| 794 |
+
"step": 5300
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 42.74,
|
| 798 |
+
"eval_loss": 0.3361692726612091,
|
| 799 |
+
"eval_runtime": 131.6013,
|
| 800 |
+
"eval_samples_per_second": 14.012,
|
| 801 |
+
"eval_steps_per_second": 1.755,
|
| 802 |
+
"eval_wer": 0.3944620548200634,
|
| 803 |
+
"step": 5300
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 43.55,
|
| 807 |
+
"learning_rate": 5.497311827956989e-05,
|
| 808 |
+
"loss": 0.3047,
|
| 809 |
+
"step": 5400
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 43.55,
|
| 813 |
+
"eval_loss": 0.32479631900787354,
|
| 814 |
+
"eval_runtime": 131.6497,
|
| 815 |
+
"eval_samples_per_second": 14.007,
|
| 816 |
+
"eval_steps_per_second": 1.755,
|
| 817 |
+
"eval_wer": 0.39297035241469325,
|
| 818 |
+
"step": 5400
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 44.35,
|
| 822 |
+
"learning_rate": 5.228494623655914e-05,
|
| 823 |
+
"loss": 0.2939,
|
| 824 |
+
"step": 5500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 44.35,
|
| 828 |
+
"eval_loss": 0.3141416013240814,
|
| 829 |
+
"eval_runtime": 131.4484,
|
| 830 |
+
"eval_samples_per_second": 14.028,
|
| 831 |
+
"eval_steps_per_second": 1.757,
|
| 832 |
+
"eval_wer": 0.392783889614022,
|
| 833 |
+
"step": 5500
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 45.16,
|
| 837 |
+
"learning_rate": 4.959677419354839e-05,
|
| 838 |
+
"loss": 0.2795,
|
| 839 |
+
"step": 5600
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 45.16,
|
| 843 |
+
"eval_loss": 0.3212459981441498,
|
| 844 |
+
"eval_runtime": 132.0759,
|
| 845 |
+
"eval_samples_per_second": 13.962,
|
| 846 |
+
"eval_steps_per_second": 1.749,
|
| 847 |
+
"eval_wer": 0.392783889614022,
|
| 848 |
+
"step": 5600
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 45.97,
|
| 852 |
+
"learning_rate": 4.690860215053764e-05,
|
| 853 |
+
"loss": 0.295,
|
| 854 |
+
"step": 5700
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 45.97,
|
| 858 |
+
"eval_loss": 0.33260008692741394,
|
| 859 |
+
"eval_runtime": 131.9753,
|
| 860 |
+
"eval_samples_per_second": 13.972,
|
| 861 |
+
"eval_steps_per_second": 1.75,
|
| 862 |
+
"eval_wer": 0.39091926160730933,
|
| 863 |
+
"step": 5700
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 46.77,
|
| 867 |
+
"learning_rate": 4.4220430107526885e-05,
|
| 868 |
+
"loss": 0.2819,
|
| 869 |
+
"step": 5800
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 46.77,
|
| 873 |
+
"eval_loss": 0.3328990936279297,
|
| 874 |
+
"eval_runtime": 132.6165,
|
| 875 |
+
"eval_samples_per_second": 13.905,
|
| 876 |
+
"eval_steps_per_second": 1.742,
|
| 877 |
+
"eval_wer": 0.39408912921872086,
|
| 878 |
+
"step": 5800
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 47.58,
|
| 882 |
+
"learning_rate": 4.1532258064516135e-05,
|
| 883 |
+
"loss": 0.2881,
|
| 884 |
+
"step": 5900
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 47.58,
|
| 888 |
+
"eval_loss": 0.32989582419395447,
|
| 889 |
+
"eval_runtime": 132.6745,
|
| 890 |
+
"eval_samples_per_second": 13.899,
|
| 891 |
+
"eval_steps_per_second": 1.741,
|
| 892 |
+
"eval_wer": 0.392783889614022,
|
| 893 |
+
"step": 5900
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 48.39,
|
| 897 |
+
"learning_rate": 3.884408602150538e-05,
|
| 898 |
+
"loss": 0.2826,
|
| 899 |
+
"step": 6000
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 48.39,
|
| 903 |
+
"eval_loss": 0.33995482325553894,
|
| 904 |
+
"eval_runtime": 132.1443,
|
| 905 |
+
"eval_samples_per_second": 13.954,
|
| 906 |
+
"eval_steps_per_second": 1.748,
|
| 907 |
+
"eval_wer": 0.3911057244079806,
|
| 908 |
+
"step": 6000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 49.19,
|
| 912 |
+
"learning_rate": 3.615591397849463e-05,
|
| 913 |
+
"loss": 0.2741,
|
| 914 |
+
"step": 6100
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 49.19,
|
| 918 |
+
"eval_loss": 0.3233118951320648,
|
| 919 |
+
"eval_runtime": 132.6042,
|
| 920 |
+
"eval_samples_per_second": 13.906,
|
| 921 |
+
"eval_steps_per_second": 1.742,
|
| 922 |
+
"eval_wer": 0.38784262539623343,
|
| 923 |
+
"step": 6100
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 50.0,
|
| 927 |
+
"learning_rate": 3.346774193548387e-05,
|
| 928 |
+
"loss": 0.2621,
|
| 929 |
+
"step": 6200
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 50.0,
|
| 933 |
+
"eval_loss": 0.31878435611724854,
|
| 934 |
+
"eval_runtime": 133.6242,
|
| 935 |
+
"eval_samples_per_second": 13.8,
|
| 936 |
+
"eval_steps_per_second": 1.729,
|
| 937 |
+
"eval_wer": 0.38905463360059667,
|
| 938 |
+
"step": 6200
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 50.81,
|
| 942 |
+
"learning_rate": 3.077956989247312e-05,
|
| 943 |
+
"loss": 0.2743,
|
| 944 |
+
"step": 6300
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 50.81,
|
| 948 |
+
"eval_loss": 0.32761555910110474,
|
| 949 |
+
"eval_runtime": 133.7291,
|
| 950 |
+
"eval_samples_per_second": 13.789,
|
| 951 |
+
"eval_steps_per_second": 1.727,
|
| 952 |
+
"eval_wer": 0.3868170799925415,
|
| 953 |
+
"step": 6300
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 51.61,
|
| 957 |
+
"learning_rate": 2.8091397849462365e-05,
|
| 958 |
+
"loss": 0.272,
|
| 959 |
+
"step": 6400
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 51.61,
|
| 963 |
+
"eval_loss": 0.3271888196468353,
|
| 964 |
+
"eval_runtime": 133.5583,
|
| 965 |
+
"eval_samples_per_second": 13.807,
|
| 966 |
+
"eval_steps_per_second": 1.73,
|
| 967 |
+
"eval_wer": 0.38551184038784264,
|
| 968 |
+
"step": 6400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 52.42,
|
| 972 |
+
"learning_rate": 2.5403225806451615e-05,
|
| 973 |
+
"loss": 0.2717,
|
| 974 |
+
"step": 6500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 52.42,
|
| 978 |
+
"eval_loss": 0.32686877250671387,
|
| 979 |
+
"eval_runtime": 133.4336,
|
| 980 |
+
"eval_samples_per_second": 13.82,
|
| 981 |
+
"eval_steps_per_second": 1.731,
|
| 982 |
+
"eval_wer": 0.38299459257878055,
|
| 983 |
+
"step": 6500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 53.23,
|
| 987 |
+
"learning_rate": 2.271505376344086e-05,
|
| 988 |
+
"loss": 0.261,
|
| 989 |
+
"step": 6600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 53.23,
|
| 993 |
+
"eval_loss": 0.3300979435443878,
|
| 994 |
+
"eval_runtime": 133.7969,
|
| 995 |
+
"eval_samples_per_second": 13.782,
|
| 996 |
+
"eval_steps_per_second": 1.726,
|
| 997 |
+
"eval_wer": 0.3838336751818012,
|
| 998 |
+
"step": 6600
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 54.03,
|
| 1002 |
+
"learning_rate": 2.002688172043011e-05,
|
| 1003 |
+
"loss": 0.2499,
|
| 1004 |
+
"step": 6700
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 54.03,
|
| 1008 |
+
"eval_loss": 0.3309214413166046,
|
| 1009 |
+
"eval_runtime": 133.9684,
|
| 1010 |
+
"eval_samples_per_second": 13.764,
|
| 1011 |
+
"eval_steps_per_second": 1.724,
|
| 1012 |
+
"eval_wer": 0.38560507178817827,
|
| 1013 |
+
"step": 6700
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 54.84,
|
| 1017 |
+
"learning_rate": 1.733870967741936e-05,
|
| 1018 |
+
"loss": 0.2617,
|
| 1019 |
+
"step": 6800
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 54.84,
|
| 1023 |
+
"eval_loss": 0.32907187938690186,
|
| 1024 |
+
"eval_runtime": 135.3114,
|
| 1025 |
+
"eval_samples_per_second": 13.628,
|
| 1026 |
+
"eval_steps_per_second": 1.707,
|
| 1027 |
+
"eval_wer": 0.3820622785754242,
|
| 1028 |
+
"step": 6800
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 55.65,
|
| 1032 |
+
"learning_rate": 1.4650537634408603e-05,
|
| 1033 |
+
"loss": 0.2628,
|
| 1034 |
+
"step": 6900
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 55.65,
|
| 1038 |
+
"eval_loss": 0.32681843638420105,
|
| 1039 |
+
"eval_runtime": 135.4461,
|
| 1040 |
+
"eval_samples_per_second": 13.614,
|
| 1041 |
+
"eval_steps_per_second": 1.705,
|
| 1042 |
+
"eval_wer": 0.38085027037106095,
|
| 1043 |
+
"step": 6900
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 56.45,
|
| 1047 |
+
"learning_rate": 1.196236559139785e-05,
|
| 1048 |
+
"loss": 0.2528,
|
| 1049 |
+
"step": 7000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 56.45,
|
| 1053 |
+
"eval_loss": 0.32666248083114624,
|
| 1054 |
+
"eval_runtime": 134.5769,
|
| 1055 |
+
"eval_samples_per_second": 13.702,
|
| 1056 |
+
"eval_steps_per_second": 1.716,
|
| 1057 |
+
"eval_wer": 0.3807570389707253,
|
| 1058 |
+
"step": 7000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 57.26,
|
| 1062 |
+
"learning_rate": 9.274193548387097e-06,
|
| 1063 |
+
"loss": 0.2475,
|
| 1064 |
+
"step": 7100
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 57.26,
|
| 1068 |
+
"eval_loss": 0.32469746470451355,
|
| 1069 |
+
"eval_runtime": 135.31,
|
| 1070 |
+
"eval_samples_per_second": 13.628,
|
| 1071 |
+
"eval_steps_per_second": 1.707,
|
| 1072 |
+
"eval_wer": 0.37963826216669777,
|
| 1073 |
+
"step": 7100
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 58.06,
|
| 1077 |
+
"learning_rate": 6.586021505376344e-06,
|
| 1078 |
+
"loss": 0.2545,
|
| 1079 |
+
"step": 7200
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 58.06,
|
| 1083 |
+
"eval_loss": 0.32185879349708557,
|
| 1084 |
+
"eval_runtime": 135.4299,
|
| 1085 |
+
"eval_samples_per_second": 13.616,
|
| 1086 |
+
"eval_steps_per_second": 1.706,
|
| 1087 |
+
"eval_wer": 0.3789856423643483,
|
| 1088 |
+
"step": 7200
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 58.87,
|
| 1092 |
+
"learning_rate": 3.8978494623655915e-06,
|
| 1093 |
+
"loss": 0.2571,
|
| 1094 |
+
"step": 7300
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 58.87,
|
| 1098 |
+
"eval_loss": 0.32077959179878235,
|
| 1099 |
+
"eval_runtime": 135.1873,
|
| 1100 |
+
"eval_samples_per_second": 13.64,
|
| 1101 |
+
"eval_steps_per_second": 1.709,
|
| 1102 |
+
"eval_wer": 0.37889241096401266,
|
| 1103 |
+
"step": 7300
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 59.68,
|
| 1107 |
+
"learning_rate": 1.2096774193548388e-06,
|
| 1108 |
+
"loss": 0.2546,
|
| 1109 |
+
"step": 7400
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 59.68,
|
| 1113 |
+
"eval_loss": 0.3218778669834137,
|
| 1114 |
+
"eval_runtime": 134.9556,
|
| 1115 |
+
"eval_samples_per_second": 13.664,
|
| 1116 |
+
"eval_steps_per_second": 1.712,
|
| 1117 |
+
"eval_wer": 0.37945179936602647,
|
| 1118 |
+
"step": 7400
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 60.0,
|
| 1122 |
+
"step": 7440,
|
| 1123 |
+
"total_flos": 2.7374813261347353e+19,
|
| 1124 |
+
"train_loss": 0.0985529641951284,
|
| 1125 |
+
"train_runtime": 13747.1961,
|
| 1126 |
+
"train_samples_per_second": 17.305,
|
| 1127 |
+
"train_steps_per_second": 0.541
|
| 1128 |
+
}
|
| 1129 |
+
],
|
| 1130 |
+
"max_steps": 7440,
|
| 1131 |
+
"num_train_epochs": 60,
|
| 1132 |
+
"total_flos": 2.7374813261347353e+19,
|
| 1133 |
+
"trial_name": null,
|
| 1134 |
+
"trial_params": null
|
| 1135 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72b28d478aa55f881af82235351b3768624d58ffaaa46aa93e2d12c1abca3dd8
|
| 3 |
+
size 3183
|
vocab.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
| 1 |
+
{"a": 0, "b": 1, "c": 2, "d": 3, "e": 4, "f": 5, "g": 6, "h": 7, "i": 8, "j": 9, "k": 10, "l": 11, "m": 12, "n": 13, "o": 14, "p": 15, "q": 16, "r": 17, "s": 18, "t": 19, "u": 20, "v": 21, "w": 22, "x": 23, "y": 24, "z": 25, "|": 26, "[UNK]": 27, "[PAD]": 28}
|