add checkpoint 135k
Browse files- config.json +85 -0
- preprocessor_config.json +9 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2041 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/wav2vec2-large-xlsr-53",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"apply_spec_augment": true,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"Wav2Vec2ForCTC"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 1,
|
| 10 |
+
"codevector_dim": 768,
|
| 11 |
+
"contrastive_logits_temperature": 0.1,
|
| 12 |
+
"conv_bias": true,
|
| 13 |
+
"conv_dim": [
|
| 14 |
+
512,
|
| 15 |
+
512,
|
| 16 |
+
512,
|
| 17 |
+
512,
|
| 18 |
+
512,
|
| 19 |
+
512,
|
| 20 |
+
512
|
| 21 |
+
],
|
| 22 |
+
"conv_kernel": [
|
| 23 |
+
10,
|
| 24 |
+
3,
|
| 25 |
+
3,
|
| 26 |
+
3,
|
| 27 |
+
3,
|
| 28 |
+
2,
|
| 29 |
+
2
|
| 30 |
+
],
|
| 31 |
+
"conv_stride": [
|
| 32 |
+
5,
|
| 33 |
+
2,
|
| 34 |
+
2,
|
| 35 |
+
2,
|
| 36 |
+
2,
|
| 37 |
+
2,
|
| 38 |
+
2
|
| 39 |
+
],
|
| 40 |
+
"ctc_loss_reduction": "mean",
|
| 41 |
+
"ctc_zero_infinity": false,
|
| 42 |
+
"diversity_loss_weight": 0.1,
|
| 43 |
+
"do_stable_layer_norm": true,
|
| 44 |
+
"eos_token_id": 2,
|
| 45 |
+
"feat_extract_activation": "gelu",
|
| 46 |
+
"feat_extract_dropout": 0.0,
|
| 47 |
+
"feat_extract_norm": "layer",
|
| 48 |
+
"feat_proj_dropout": 0.0,
|
| 49 |
+
"feat_quantizer_dropout": 0.0,
|
| 50 |
+
"final_dropout": 0.0,
|
| 51 |
+
"gradient_checkpointing": true,
|
| 52 |
+
"hidden_act": "gelu",
|
| 53 |
+
"hidden_dropout": 0.1,
|
| 54 |
+
"hidden_size": 1024,
|
| 55 |
+
"initializer_range": 0.02,
|
| 56 |
+
"intermediate_size": 4096,
|
| 57 |
+
"layer_norm_eps": 1e-05,
|
| 58 |
+
"layerdrop": 0.1,
|
| 59 |
+
"mask_channel_length": 10,
|
| 60 |
+
"mask_channel_min_space": 1,
|
| 61 |
+
"mask_channel_other": 0.0,
|
| 62 |
+
"mask_channel_prob": 0.0,
|
| 63 |
+
"mask_channel_selection": "static",
|
| 64 |
+
"mask_feature_length": 10,
|
| 65 |
+
"mask_feature_prob": 0.0,
|
| 66 |
+
"mask_time_length": 10,
|
| 67 |
+
"mask_time_min_space": 1,
|
| 68 |
+
"mask_time_other": 0.0,
|
| 69 |
+
"mask_time_prob": 0.05,
|
| 70 |
+
"mask_time_selection": "static",
|
| 71 |
+
"model_type": "wav2vec2",
|
| 72 |
+
"num_attention_heads": 16,
|
| 73 |
+
"num_codevector_groups": 2,
|
| 74 |
+
"num_codevectors_per_group": 320,
|
| 75 |
+
"num_conv_pos_embedding_groups": 16,
|
| 76 |
+
"num_conv_pos_embeddings": 128,
|
| 77 |
+
"num_feat_extract_layers": 7,
|
| 78 |
+
"num_hidden_layers": 24,
|
| 79 |
+
"num_negatives": 100,
|
| 80 |
+
"pad_token_id": 69,
|
| 81 |
+
"proj_codevector_dim": 768,
|
| 82 |
+
"torch_dtype": "float32",
|
| 83 |
+
"transformers_version": "4.9.1",
|
| 84 |
+
"vocab_size": 70
|
| 85 |
+
}
|
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": false,
|
| 8 |
+
"sampling_rate": 16000
|
| 9 |
+
}
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:040892c8367ed2663b43590614fcaee20e683cc064af55f65e3baf94db0473d8
|
| 3 |
+
size 15587
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eaad0a550aad796c00c2715fb3dd07c74e5ca1520507a15a505e4faf0123239f
|
| 3 |
+
size 559
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb118fee7b3cd82205b90d8e4ffc01eee7dfb1f3f32f5210cf6b36e69150bc49
|
| 3 |
+
size 623
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2041 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0,
|
| 3 |
+
"best_model_checkpoint": "../data/wav2vec2-large-xlsr-53-thai/checkpoint-1000",
|
| 4 |
+
"epoch": 99.77827050997783,
|
| 5 |
+
"global_step": 135000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.74,
|
| 12 |
+
"learning_rate": 9.970000000000001e-05,
|
| 13 |
+
"loss": 7.1725,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.74,
|
| 18 |
+
"eval_loss": 3.039940595626831,
|
| 19 |
+
"eval_runtime": 185.7874,
|
| 20 |
+
"eval_samples_per_second": 16.293,
|
| 21 |
+
"eval_steps_per_second": 0.511,
|
| 22 |
+
"eval_wer": 1.0,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.48,
|
| 27 |
+
"learning_rate": 9.925763216679077e-05,
|
| 28 |
+
"loss": 1.0788,
|
| 29 |
+
"step": 2000
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 1.48,
|
| 33 |
+
"eval_loss": 0.6130112409591675,
|
| 34 |
+
"eval_runtime": 176.9194,
|
| 35 |
+
"eval_samples_per_second": 17.109,
|
| 36 |
+
"eval_steps_per_second": 0.537,
|
| 37 |
+
"eval_wer": 0.6820256330103157,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.22,
|
| 42 |
+
"learning_rate": 9.851303052866717e-05,
|
| 43 |
+
"loss": 0.6339,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.22,
|
| 48 |
+
"eval_loss": 0.49624550342559814,
|
| 49 |
+
"eval_runtime": 175.6715,
|
| 50 |
+
"eval_samples_per_second": 17.231,
|
| 51 |
+
"eval_steps_per_second": 0.541,
|
| 52 |
+
"eval_wer": 0.6168802750859643,
|
| 53 |
+
"step": 3000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.96,
|
| 57 |
+
"learning_rate": 9.776842889054357e-05,
|
| 58 |
+
"loss": 0.5295,
|
| 59 |
+
"step": 4000
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 2.96,
|
| 63 |
+
"eval_loss": 0.4389719069004059,
|
| 64 |
+
"eval_runtime": 176.6072,
|
| 65 |
+
"eval_samples_per_second": 17.14,
|
| 66 |
+
"eval_steps_per_second": 0.538,
|
| 67 |
+
"eval_wer": 0.577117849327915,
|
| 68 |
+
"step": 4000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 3.7,
|
| 72 |
+
"learning_rate": 9.702382725241996e-05,
|
| 73 |
+
"loss": 0.4616,
|
| 74 |
+
"step": 5000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 3.7,
|
| 78 |
+
"eval_loss": 0.4208119809627533,
|
| 79 |
+
"eval_runtime": 196.7481,
|
| 80 |
+
"eval_samples_per_second": 15.385,
|
| 81 |
+
"eval_steps_per_second": 0.483,
|
| 82 |
+
"eval_wer": 0.5618005626758362,
|
| 83 |
+
"step": 5000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 4.43,
|
| 87 |
+
"learning_rate": 9.627922561429635e-05,
|
| 88 |
+
"loss": 0.4206,
|
| 89 |
+
"step": 6000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 4.43,
|
| 93 |
+
"eval_loss": 0.4009915888309479,
|
| 94 |
+
"eval_runtime": 180.4803,
|
| 95 |
+
"eval_samples_per_second": 16.772,
|
| 96 |
+
"eval_steps_per_second": 0.526,
|
| 97 |
+
"eval_wer": 0.5411691153485464,
|
| 98 |
+
"step": 6000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 5.17,
|
| 102 |
+
"learning_rate": 9.553462397617276e-05,
|
| 103 |
+
"loss": 0.3908,
|
| 104 |
+
"step": 7000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 5.17,
|
| 108 |
+
"eval_loss": 0.399889200925827,
|
| 109 |
+
"eval_runtime": 189.7255,
|
| 110 |
+
"eval_samples_per_second": 15.955,
|
| 111 |
+
"eval_steps_per_second": 0.501,
|
| 112 |
+
"eval_wer": 0.5384807752422632,
|
| 113 |
+
"step": 7000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 5.91,
|
| 117 |
+
"learning_rate": 9.479002233804914e-05,
|
| 118 |
+
"loss": 0.3659,
|
| 119 |
+
"step": 8000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 5.91,
|
| 123 |
+
"eval_loss": 0.3719017505645752,
|
| 124 |
+
"eval_runtime": 184.5994,
|
| 125 |
+
"eval_samples_per_second": 16.398,
|
| 126 |
+
"eval_steps_per_second": 0.515,
|
| 127 |
+
"eval_wer": 0.5141606752110034,
|
| 128 |
+
"step": 8000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 6.65,
|
| 132 |
+
"learning_rate": 9.404542069992555e-05,
|
| 133 |
+
"loss": 0.338,
|
| 134 |
+
"step": 9000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 6.65,
|
| 138 |
+
"eval_loss": 0.35533592104911804,
|
| 139 |
+
"eval_runtime": 188.5325,
|
| 140 |
+
"eval_samples_per_second": 16.056,
|
| 141 |
+
"eval_steps_per_second": 0.504,
|
| 142 |
+
"eval_wer": 0.5101594248202563,
|
| 143 |
+
"step": 9000
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 7.39,
|
| 147 |
+
"learning_rate": 9.330156366344007e-05,
|
| 148 |
+
"loss": 0.3222,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 7.39,
|
| 153 |
+
"eval_loss": 0.40722423791885376,
|
| 154 |
+
"eval_runtime": 195.2723,
|
| 155 |
+
"eval_samples_per_second": 15.501,
|
| 156 |
+
"eval_steps_per_second": 0.487,
|
| 157 |
+
"eval_wer": 0.49309159112222567,
|
| 158 |
+
"step": 10000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 8.13,
|
| 162 |
+
"learning_rate": 9.255696202531646e-05,
|
| 163 |
+
"loss": 0.31,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 8.13,
|
| 168 |
+
"eval_loss": 0.37148234248161316,
|
| 169 |
+
"eval_runtime": 182.6314,
|
| 170 |
+
"eval_samples_per_second": 16.574,
|
| 171 |
+
"eval_steps_per_second": 0.52,
|
| 172 |
+
"eval_wer": 0.48715223507346045,
|
| 173 |
+
"step": 11000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 8.87,
|
| 177 |
+
"learning_rate": 9.181310498883098e-05,
|
| 178 |
+
"loss": 0.2901,
|
| 179 |
+
"step": 12000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 8.87,
|
| 183 |
+
"eval_loss": 0.365583598613739,
|
| 184 |
+
"eval_runtime": 175.3968,
|
| 185 |
+
"eval_samples_per_second": 17.258,
|
| 186 |
+
"eval_steps_per_second": 0.542,
|
| 187 |
+
"eval_wer": 0.4882775867458581,
|
| 188 |
+
"step": 12000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 9.61,
|
| 192 |
+
"learning_rate": 9.106850335070738e-05,
|
| 193 |
+
"loss": 0.276,
|
| 194 |
+
"step": 13000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 9.61,
|
| 198 |
+
"eval_loss": 0.37835174798965454,
|
| 199 |
+
"eval_runtime": 175.9962,
|
| 200 |
+
"eval_samples_per_second": 17.199,
|
| 201 |
+
"eval_steps_per_second": 0.54,
|
| 202 |
+
"eval_wer": 0.4897155361050328,
|
| 203 |
+
"step": 13000
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 10.35,
|
| 207 |
+
"learning_rate": 9.032539091586003e-05,
|
| 208 |
+
"loss": 0.2628,
|
| 209 |
+
"step": 14000
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 10.35,
|
| 213 |
+
"eval_loss": 0.4127887189388275,
|
| 214 |
+
"eval_runtime": 180.0512,
|
| 215 |
+
"eval_samples_per_second": 16.812,
|
| 216 |
+
"eval_steps_per_second": 0.528,
|
| 217 |
+
"eval_wer": 0.48871522350734603,
|
| 218 |
+
"step": 14000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 11.09,
|
| 222 |
+
"learning_rate": 8.958078927773641e-05,
|
| 223 |
+
"loss": 0.256,
|
| 224 |
+
"step": 15000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 11.09,
|
| 228 |
+
"eval_loss": 0.3795180022716522,
|
| 229 |
+
"eval_runtime": 174.1363,
|
| 230 |
+
"eval_samples_per_second": 17.383,
|
| 231 |
+
"eval_steps_per_second": 0.546,
|
| 232 |
+
"eval_wer": 0.46939668646452015,
|
| 233 |
+
"step": 15000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 11.83,
|
| 237 |
+
"learning_rate": 8.88361876396128e-05,
|
| 238 |
+
"loss": 0.2455,
|
| 239 |
+
"step": 16000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 11.83,
|
| 243 |
+
"eval_loss": 0.3748489320278168,
|
| 244 |
+
"eval_runtime": 174.8629,
|
| 245 |
+
"eval_samples_per_second": 17.311,
|
| 246 |
+
"eval_steps_per_second": 0.543,
|
| 247 |
+
"eval_wer": 0.46970928415129726,
|
| 248 |
+
"step": 16000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 12.56,
|
| 252 |
+
"learning_rate": 8.809233060312732e-05,
|
| 253 |
+
"loss": 0.2377,
|
| 254 |
+
"step": 17000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 12.56,
|
| 258 |
+
"eval_loss": 0.4371410012245178,
|
| 259 |
+
"eval_runtime": 173.1913,
|
| 260 |
+
"eval_samples_per_second": 17.478,
|
| 261 |
+
"eval_steps_per_second": 0.549,
|
| 262 |
+
"eval_wer": 0.4650203188496405,
|
| 263 |
+
"step": 17000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 13.3,
|
| 267 |
+
"learning_rate": 8.734772896500373e-05,
|
| 268 |
+
"loss": 0.2279,
|
| 269 |
+
"step": 18000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 13.3,
|
| 273 |
+
"eval_loss": 0.38019847869873047,
|
| 274 |
+
"eval_runtime": 177.0444,
|
| 275 |
+
"eval_samples_per_second": 17.097,
|
| 276 |
+
"eval_steps_per_second": 0.537,
|
| 277 |
+
"eval_wer": 0.45339168490153176,
|
| 278 |
+
"step": 18000
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 14.04,
|
| 282 |
+
"learning_rate": 8.660387192851825e-05,
|
| 283 |
+
"loss": 0.2247,
|
| 284 |
+
"step": 19000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 14.04,
|
| 288 |
+
"eval_loss": 0.3687661588191986,
|
| 289 |
+
"eval_runtime": 197.8618,
|
| 290 |
+
"eval_samples_per_second": 15.299,
|
| 291 |
+
"eval_steps_per_second": 0.48,
|
| 292 |
+
"eval_wer": 0.4567677399187246,
|
| 293 |
+
"step": 19000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 14.78,
|
| 297 |
+
"learning_rate": 8.585927029039465e-05,
|
| 298 |
+
"loss": 0.2139,
|
| 299 |
+
"step": 20000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 14.78,
|
| 303 |
+
"eval_loss": 0.3616667091846466,
|
| 304 |
+
"eval_runtime": 177.8968,
|
| 305 |
+
"eval_samples_per_second": 17.015,
|
| 306 |
+
"eval_steps_per_second": 0.534,
|
| 307 |
+
"eval_wer": 0.446639574867146,
|
| 308 |
+
"step": 20000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 15.52,
|
| 312 |
+
"learning_rate": 8.511541325390916e-05,
|
| 313 |
+
"loss": 0.2072,
|
| 314 |
+
"step": 21000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 15.52,
|
| 318 |
+
"eval_loss": 0.4171736538410187,
|
| 319 |
+
"eval_runtime": 171.8647,
|
| 320 |
+
"eval_samples_per_second": 17.613,
|
| 321 |
+
"eval_steps_per_second": 0.553,
|
| 322 |
+
"eval_wer": 0.45883088465145355,
|
| 323 |
+
"step": 21000
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 16.26,
|
| 327 |
+
"learning_rate": 8.437081161578556e-05,
|
| 328 |
+
"loss": 0.1999,
|
| 329 |
+
"step": 22000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 16.26,
|
| 333 |
+
"eval_loss": 0.7755644917488098,
|
| 334 |
+
"eval_runtime": 171.1132,
|
| 335 |
+
"eval_samples_per_second": 17.69,
|
| 336 |
+
"eval_steps_per_second": 0.555,
|
| 337 |
+
"eval_wer": 0.45126602063144733,
|
| 338 |
+
"step": 22000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 17.0,
|
| 342 |
+
"learning_rate": 8.362695457930007e-05,
|
| 343 |
+
"loss": 0.1958,
|
| 344 |
+
"step": 23000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 17.0,
|
| 348 |
+
"eval_loss": 0.41745948791503906,
|
| 349 |
+
"eval_runtime": 172.9131,
|
| 350 |
+
"eval_samples_per_second": 17.506,
|
| 351 |
+
"eval_steps_per_second": 0.549,
|
| 352 |
+
"eval_wer": 0.44382619568615195,
|
| 353 |
+
"step": 23000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 17.74,
|
| 357 |
+
"learning_rate": 8.288235294117647e-05,
|
| 358 |
+
"loss": 0.1866,
|
| 359 |
+
"step": 24000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 17.74,
|
| 363 |
+
"eval_loss": 0.4865190386772156,
|
| 364 |
+
"eval_runtime": 173.9742,
|
| 365 |
+
"eval_samples_per_second": 17.399,
|
| 366 |
+
"eval_steps_per_second": 0.546,
|
| 367 |
+
"eval_wer": 0.45401688027508597,
|
| 368 |
+
"step": 24000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 18.48,
|
| 372 |
+
"learning_rate": 8.213924050632912e-05,
|
| 373 |
+
"loss": 0.1854,
|
| 374 |
+
"step": 25000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 18.48,
|
| 378 |
+
"eval_loss": 0.5420244336128235,
|
| 379 |
+
"eval_runtime": 175.6968,
|
| 380 |
+
"eval_samples_per_second": 17.229,
|
| 381 |
+
"eval_steps_per_second": 0.541,
|
| 382 |
+
"eval_wer": 0.4475148483901219,
|
| 383 |
+
"step": 25000
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 19.22,
|
| 387 |
+
"learning_rate": 8.139463886820552e-05,
|
| 388 |
+
"loss": 0.1777,
|
| 389 |
+
"step": 26000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 19.22,
|
| 393 |
+
"eval_loss": 0.4428083300590515,
|
| 394 |
+
"eval_runtime": 174.8946,
|
| 395 |
+
"eval_samples_per_second": 17.308,
|
| 396 |
+
"eval_steps_per_second": 0.543,
|
| 397 |
+
"eval_wer": 0.4501406689590497,
|
| 398 |
+
"step": 26000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 19.96,
|
| 402 |
+
"learning_rate": 8.065003723008191e-05,
|
| 403 |
+
"loss": 0.1764,
|
| 404 |
+
"step": 27000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 19.96,
|
| 408 |
+
"eval_loss": 0.44021618366241455,
|
| 409 |
+
"eval_runtime": 174.1095,
|
| 410 |
+
"eval_samples_per_second": 17.386,
|
| 411 |
+
"eval_steps_per_second": 0.546,
|
| 412 |
+
"eval_wer": 0.44307596123788684,
|
| 413 |
+
"step": 27000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 20.69,
|
| 417 |
+
"learning_rate": 7.990543559195831e-05,
|
| 418 |
+
"loss": 0.1698,
|
| 419 |
+
"step": 28000
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 20.69,
|
| 423 |
+
"eval_loss": 0.48001521825790405,
|
| 424 |
+
"eval_runtime": 173.9598,
|
| 425 |
+
"eval_samples_per_second": 17.401,
|
| 426 |
+
"eval_steps_per_second": 0.546,
|
| 427 |
+
"eval_wer": 0.4456392622694592,
|
| 428 |
+
"step": 28000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 21.43,
|
| 432 |
+
"learning_rate": 7.91608339538347e-05,
|
| 433 |
+
"loss": 0.1625,
|
| 434 |
+
"step": 29000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 21.43,
|
| 438 |
+
"eval_loss": 0.4736146628856659,
|
| 439 |
+
"eval_runtime": 173.7886,
|
| 440 |
+
"eval_samples_per_second": 17.418,
|
| 441 |
+
"eval_steps_per_second": 0.547,
|
| 442 |
+
"eval_wer": 0.44676461394185685,
|
| 443 |
+
"step": 29000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 22.17,
|
| 447 |
+
"learning_rate": 7.841697691734922e-05,
|
| 448 |
+
"loss": 0.1611,
|
| 449 |
+
"step": 30000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 22.17,
|
| 453 |
+
"eval_loss": 0.4669973850250244,
|
| 454 |
+
"eval_runtime": 174.8827,
|
| 455 |
+
"eval_samples_per_second": 17.309,
|
| 456 |
+
"eval_steps_per_second": 0.543,
|
| 457 |
+
"eval_wer": 0.43813691778680836,
|
| 458 |
+
"step": 30000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 22.91,
|
| 462 |
+
"learning_rate": 7.767237527922562e-05,
|
| 463 |
+
"loss": 0.1557,
|
| 464 |
+
"step": 31000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 22.91,
|
| 468 |
+
"eval_loss": 0.49789130687713623,
|
| 469 |
+
"eval_runtime": 176.2993,
|
| 470 |
+
"eval_samples_per_second": 17.17,
|
| 471 |
+
"eval_steps_per_second": 0.539,
|
| 472 |
+
"eval_wer": 0.4390121913097843,
|
| 473 |
+
"step": 31000
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 23.65,
|
| 477 |
+
"learning_rate": 7.692851824274014e-05,
|
| 478 |
+
"loss": 0.1531,
|
| 479 |
+
"step": 32000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 23.65,
|
| 483 |
+
"eval_loss": 0.49486255645751953,
|
| 484 |
+
"eval_runtime": 178.461,
|
| 485 |
+
"eval_samples_per_second": 16.962,
|
| 486 |
+
"eval_steps_per_second": 0.532,
|
| 487 |
+
"eval_wer": 0.4311347296030009,
|
| 488 |
+
"step": 32000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 24.39,
|
| 492 |
+
"learning_rate": 7.618391660461653e-05,
|
| 493 |
+
"loss": 0.1485,
|
| 494 |
+
"step": 33000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 24.39,
|
| 498 |
+
"eval_loss": 0.5052071809768677,
|
| 499 |
+
"eval_runtime": 197.1526,
|
| 500 |
+
"eval_samples_per_second": 15.354,
|
| 501 |
+
"eval_steps_per_second": 0.482,
|
| 502 |
+
"eval_wer": 0.43432322600812756,
|
| 503 |
+
"step": 33000
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 25.13,
|
| 507 |
+
"learning_rate": 7.543931496649293e-05,
|
| 508 |
+
"loss": 0.149,
|
| 509 |
+
"step": 34000
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 25.13,
|
| 513 |
+
"eval_loss": 0.4628446400165558,
|
| 514 |
+
"eval_runtime": 191.152,
|
| 515 |
+
"eval_samples_per_second": 15.836,
|
| 516 |
+
"eval_steps_per_second": 0.497,
|
| 517 |
+
"eval_wer": 0.42638324476398876,
|
| 518 |
+
"step": 34000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 25.87,
|
| 522 |
+
"learning_rate": 7.469545793000745e-05,
|
| 523 |
+
"loss": 0.1439,
|
| 524 |
+
"step": 35000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 25.87,
|
| 528 |
+
"eval_loss": 0.4972988963127136,
|
| 529 |
+
"eval_runtime": 199.2887,
|
| 530 |
+
"eval_samples_per_second": 15.189,
|
| 531 |
+
"eval_steps_per_second": 0.477,
|
| 532 |
+
"eval_wer": 0.4331353547983745,
|
| 533 |
+
"step": 35000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 26.61,
|
| 537 |
+
"learning_rate": 7.395160089352196e-05,
|
| 538 |
+
"loss": 0.14,
|
| 539 |
+
"step": 36000
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 26.61,
|
| 543 |
+
"eval_loss": 0.5113205313682556,
|
| 544 |
+
"eval_runtime": 177.8533,
|
| 545 |
+
"eval_samples_per_second": 17.02,
|
| 546 |
+
"eval_steps_per_second": 0.534,
|
| 547 |
+
"eval_wer": 0.4366364488902782,
|
| 548 |
+
"step": 36000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 27.35,
|
| 552 |
+
"learning_rate": 7.320699925539837e-05,
|
| 553 |
+
"loss": 0.1373,
|
| 554 |
+
"step": 37000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 27.35,
|
| 558 |
+
"eval_loss": 0.5227766036987305,
|
| 559 |
+
"eval_runtime": 197.1721,
|
| 560 |
+
"eval_samples_per_second": 15.352,
|
| 561 |
+
"eval_steps_per_second": 0.482,
|
| 562 |
+
"eval_wer": 0.4340731478587058,
|
| 563 |
+
"step": 37000
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 28.09,
|
| 567 |
+
"learning_rate": 7.246239761727476e-05,
|
| 568 |
+
"loss": 0.134,
|
| 569 |
+
"step": 38000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 28.09,
|
| 573 |
+
"eval_loss": 0.5289518237113953,
|
| 574 |
+
"eval_runtime": 186.9942,
|
| 575 |
+
"eval_samples_per_second": 16.188,
|
| 576 |
+
"eval_steps_per_second": 0.508,
|
| 577 |
+
"eval_wer": 0.42744607689903097,
|
| 578 |
+
"step": 38000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 28.82,
|
| 582 |
+
"learning_rate": 7.171779597915115e-05,
|
| 583 |
+
"loss": 0.1309,
|
| 584 |
+
"step": 39000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 28.82,
|
| 588 |
+
"eval_loss": 0.5079456567764282,
|
| 589 |
+
"eval_runtime": 177.9839,
|
| 590 |
+
"eval_samples_per_second": 17.007,
|
| 591 |
+
"eval_steps_per_second": 0.534,
|
| 592 |
+
"eval_wer": 0.4363238512035011,
|
| 593 |
+
"step": 39000
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 29.56,
|
| 597 |
+
"learning_rate": 7.097319434102755e-05,
|
| 598 |
+
"loss": 0.129,
|
| 599 |
+
"step": 40000
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 29.56,
|
| 603 |
+
"eval_loss": 0.5366144776344299,
|
| 604 |
+
"eval_runtime": 174.4769,
|
| 605 |
+
"eval_samples_per_second": 17.349,
|
| 606 |
+
"eval_steps_per_second": 0.544,
|
| 607 |
+
"eval_wer": 0.4248202563301032,
|
| 608 |
+
"step": 40000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 30.3,
|
| 612 |
+
"learning_rate": 7.02300819061802e-05,
|
| 613 |
+
"loss": 0.126,
|
| 614 |
+
"step": 41000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 30.3,
|
| 618 |
+
"eval_loss": 0.5051801204681396,
|
| 619 |
+
"eval_runtime": 174.1717,
|
| 620 |
+
"eval_samples_per_second": 17.379,
|
| 621 |
+
"eval_steps_per_second": 0.545,
|
| 622 |
+
"eval_wer": 0.42275711159737417,
|
| 623 |
+
"step": 41000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 31.04,
|
| 627 |
+
"learning_rate": 6.94854802680566e-05,
|
| 628 |
+
"loss": 0.1255,
|
| 629 |
+
"step": 42000
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 31.04,
|
| 633 |
+
"eval_loss": 0.5653140544891357,
|
| 634 |
+
"eval_runtime": 175.3338,
|
| 635 |
+
"eval_samples_per_second": 17.264,
|
| 636 |
+
"eval_steps_per_second": 0.542,
|
| 637 |
+
"eval_wer": 0.43063457330415755,
|
| 638 |
+
"step": 42000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 31.78,
|
| 642 |
+
"learning_rate": 6.874087862993299e-05,
|
| 643 |
+
"loss": 0.122,
|
| 644 |
+
"step": 43000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 31.78,
|
| 648 |
+
"eval_loss": 0.5453576445579529,
|
| 649 |
+
"eval_runtime": 176.4408,
|
| 650 |
+
"eval_samples_per_second": 17.156,
|
| 651 |
+
"eval_steps_per_second": 0.538,
|
| 652 |
+
"eval_wer": 0.42800875273522976,
|
| 653 |
+
"step": 43000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 32.52,
|
| 657 |
+
"learning_rate": 6.799627699180938e-05,
|
| 658 |
+
"loss": 0.1191,
|
| 659 |
+
"step": 44000
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 32.52,
|
| 663 |
+
"eval_loss": 0.6267871260643005,
|
| 664 |
+
"eval_runtime": 172.551,
|
| 665 |
+
"eval_samples_per_second": 17.543,
|
| 666 |
+
"eval_steps_per_second": 0.551,
|
| 667 |
+
"eval_wer": 0.42400750234448265,
|
| 668 |
+
"step": 44000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 33.26,
|
| 672 |
+
"learning_rate": 6.725167535368579e-05,
|
| 673 |
+
"loss": 0.1151,
|
| 674 |
+
"step": 45000
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 33.26,
|
| 678 |
+
"eval_loss": 0.5381397008895874,
|
| 679 |
+
"eval_runtime": 174.0505,
|
| 680 |
+
"eval_samples_per_second": 17.392,
|
| 681 |
+
"eval_steps_per_second": 0.546,
|
| 682 |
+
"eval_wer": 0.4221944357611754,
|
| 683 |
+
"step": 45000
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 34.0,
|
| 687 |
+
"learning_rate": 6.650707371556217e-05,
|
| 688 |
+
"loss": 0.115,
|
| 689 |
+
"step": 46000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 34.0,
|
| 693 |
+
"eval_loss": 0.5648242235183716,
|
| 694 |
+
"eval_runtime": 174.2088,
|
| 695 |
+
"eval_samples_per_second": 17.376,
|
| 696 |
+
"eval_steps_per_second": 0.545,
|
| 697 |
+
"eval_wer": 0.4198812128790247,
|
| 698 |
+
"step": 46000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 34.74,
|
| 702 |
+
"learning_rate": 6.576247207743858e-05,
|
| 703 |
+
"loss": 0.1124,
|
| 704 |
+
"step": 47000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 34.74,
|
| 708 |
+
"eval_loss": 0.613673985004425,
|
| 709 |
+
"eval_runtime": 193.9522,
|
| 710 |
+
"eval_samples_per_second": 15.607,
|
| 711 |
+
"eval_steps_per_second": 0.49,
|
| 712 |
+
"eval_wer": 0.4295092216317599,
|
| 713 |
+
"step": 47000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 35.48,
|
| 717 |
+
"learning_rate": 6.50186150409531e-05,
|
| 718 |
+
"loss": 0.1101,
|
| 719 |
+
"step": 48000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 35.48,
|
| 723 |
+
"eval_loss": 0.5470528602600098,
|
| 724 |
+
"eval_runtime": 193.2438,
|
| 725 |
+
"eval_samples_per_second": 15.664,
|
| 726 |
+
"eval_steps_per_second": 0.492,
|
| 727 |
+
"eval_wer": 0.42206939668646454,
|
| 728 |
+
"step": 48000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 36.22,
|
| 732 |
+
"learning_rate": 6.427475800446761e-05,
|
| 733 |
+
"loss": 0.1061,
|
| 734 |
+
"step": 49000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 36.22,
|
| 738 |
+
"eval_loss": 0.5942440032958984,
|
| 739 |
+
"eval_runtime": 200.9573,
|
| 740 |
+
"eval_samples_per_second": 15.063,
|
| 741 |
+
"eval_steps_per_second": 0.473,
|
| 742 |
+
"eval_wer": 0.4258205689277899,
|
| 743 |
+
"step": 49000
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 36.95,
|
| 747 |
+
"learning_rate": 6.353015636634401e-05,
|
| 748 |
+
"loss": 0.1065,
|
| 749 |
+
"step": 50000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 36.95,
|
| 753 |
+
"eval_loss": 0.5583751201629639,
|
| 754 |
+
"eval_runtime": 178.0909,
|
| 755 |
+
"eval_samples_per_second": 16.997,
|
| 756 |
+
"eval_steps_per_second": 0.533,
|
| 757 |
+
"eval_wer": 0.4255704907783682,
|
| 758 |
+
"step": 50000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 37.69,
|
| 762 |
+
"learning_rate": 6.27855547282204e-05,
|
| 763 |
+
"loss": 0.1043,
|
| 764 |
+
"step": 51000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 37.69,
|
| 768 |
+
"eval_loss": 0.7426488399505615,
|
| 769 |
+
"eval_runtime": 188.9691,
|
| 770 |
+
"eval_samples_per_second": 16.018,
|
| 771 |
+
"eval_steps_per_second": 0.503,
|
| 772 |
+
"eval_wer": 0.4210065645514223,
|
| 773 |
+
"step": 51000
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 38.43,
|
| 777 |
+
"learning_rate": 6.204169769173492e-05,
|
| 778 |
+
"loss": 0.103,
|
| 779 |
+
"step": 52000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 38.43,
|
| 783 |
+
"eval_loss": 0.7028630375862122,
|
| 784 |
+
"eval_runtime": 182.3453,
|
| 785 |
+
"eval_samples_per_second": 16.6,
|
| 786 |
+
"eval_steps_per_second": 0.521,
|
| 787 |
+
"eval_wer": 0.42338230697092843,
|
| 788 |
+
"step": 52000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 39.17,
|
| 792 |
+
"learning_rate": 6.129709605361132e-05,
|
| 793 |
+
"loss": 0.1016,
|
| 794 |
+
"step": 53000
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 39.17,
|
| 798 |
+
"eval_loss": 0.644480288028717,
|
| 799 |
+
"eval_runtime": 172.8589,
|
| 800 |
+
"eval_samples_per_second": 17.511,
|
| 801 |
+
"eval_steps_per_second": 0.55,
|
| 802 |
+
"eval_wer": 0.42319474835886217,
|
| 803 |
+
"step": 53000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 39.91,
|
| 807 |
+
"learning_rate": 6.055323901712584e-05,
|
| 808 |
+
"loss": 0.0977,
|
| 809 |
+
"step": 54000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 39.91,
|
| 813 |
+
"eval_loss": 0.6561093926429749,
|
| 814 |
+
"eval_runtime": 174.7926,
|
| 815 |
+
"eval_samples_per_second": 17.318,
|
| 816 |
+
"eval_steps_per_second": 0.544,
|
| 817 |
+
"eval_wer": 0.41394185683025947,
|
| 818 |
+
"step": 54000
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 40.65,
|
| 822 |
+
"learning_rate": 5.980863737900224e-05,
|
| 823 |
+
"loss": 0.097,
|
| 824 |
+
"step": 55000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 40.65,
|
| 828 |
+
"eval_loss": 0.6229739785194397,
|
| 829 |
+
"eval_runtime": 176.8753,
|
| 830 |
+
"eval_samples_per_second": 17.114,
|
| 831 |
+
"eval_steps_per_second": 0.537,
|
| 832 |
+
"eval_wer": 0.41450453266645826,
|
| 833 |
+
"step": 55000
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 41.39,
|
| 837 |
+
"learning_rate": 5.9064780342516755e-05,
|
| 838 |
+
"loss": 0.0963,
|
| 839 |
+
"step": 56000
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 41.39,
|
| 843 |
+
"eval_loss": 0.5873845219612122,
|
| 844 |
+
"eval_runtime": 175.9014,
|
| 845 |
+
"eval_samples_per_second": 17.209,
|
| 846 |
+
"eval_steps_per_second": 0.54,
|
| 847 |
+
"eval_wer": 0.41713035323538605,
|
| 848 |
+
"step": 56000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 42.13,
|
| 852 |
+
"learning_rate": 5.832017870439316e-05,
|
| 853 |
+
"loss": 0.0943,
|
| 854 |
+
"step": 57000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 42.13,
|
| 858 |
+
"eval_loss": 0.5930323600769043,
|
| 859 |
+
"eval_runtime": 174.9181,
|
| 860 |
+
"eval_samples_per_second": 17.305,
|
| 861 |
+
"eval_steps_per_second": 0.543,
|
| 862 |
+
"eval_wer": 0.418005626758362,
|
| 863 |
+
"step": 57000
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 42.87,
|
| 867 |
+
"learning_rate": 5.757557706626955e-05,
|
| 868 |
+
"loss": 0.093,
|
| 869 |
+
"step": 58000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 42.87,
|
| 873 |
+
"eval_loss": 0.6292551755905151,
|
| 874 |
+
"eval_runtime": 175.0843,
|
| 875 |
+
"eval_samples_per_second": 17.289,
|
| 876 |
+
"eval_steps_per_second": 0.543,
|
| 877 |
+
"eval_wer": 0.4175054704595186,
|
| 878 |
+
"step": 58000
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 43.61,
|
| 882 |
+
"learning_rate": 5.683097542814595e-05,
|
| 883 |
+
"loss": 0.0896,
|
| 884 |
+
"step": 59000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 43.61,
|
| 888 |
+
"eval_loss": 0.6171667575836182,
|
| 889 |
+
"eval_runtime": 175.3688,
|
| 890 |
+
"eval_samples_per_second": 17.261,
|
| 891 |
+
"eval_steps_per_second": 0.542,
|
| 892 |
+
"eval_wer": 0.41319162238199436,
|
| 893 |
+
"step": 59000
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 44.35,
|
| 897 |
+
"learning_rate": 5.608637379002234e-05,
|
| 898 |
+
"loss": 0.0911,
|
| 899 |
+
"step": 60000
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 44.35,
|
| 903 |
+
"eval_loss": 0.6143432259559631,
|
| 904 |
+
"eval_runtime": 178.3103,
|
| 905 |
+
"eval_samples_per_second": 16.976,
|
| 906 |
+
"eval_steps_per_second": 0.533,
|
| 907 |
+
"eval_wer": 0.4172553923100969,
|
| 908 |
+
"step": 60000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 45.08,
|
| 912 |
+
"learning_rate": 5.534326135517498e-05,
|
| 913 |
+
"loss": 0.0885,
|
| 914 |
+
"step": 61000
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 45.08,
|
| 918 |
+
"eval_loss": 0.6145235300064087,
|
| 919 |
+
"eval_runtime": 191.9563,
|
| 920 |
+
"eval_samples_per_second": 15.769,
|
| 921 |
+
"eval_steps_per_second": 0.495,
|
| 922 |
+
"eval_wer": 0.4133166614567052,
|
| 923 |
+
"step": 61000
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 45.82,
|
| 927 |
+
"learning_rate": 5.459865971705138e-05,
|
| 928 |
+
"loss": 0.0872,
|
| 929 |
+
"step": 62000
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 45.82,
|
| 933 |
+
"eval_loss": 0.8499487638473511,
|
| 934 |
+
"eval_runtime": 177.455,
|
| 935 |
+
"eval_samples_per_second": 17.058,
|
| 936 |
+
"eval_steps_per_second": 0.535,
|
| 937 |
+
"eval_wer": 0.41187871209753046,
|
| 938 |
+
"step": 62000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 46.56,
|
| 942 |
+
"learning_rate": 5.385405807892777e-05,
|
| 943 |
+
"loss": 0.0883,
|
| 944 |
+
"step": 63000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 46.56,
|
| 948 |
+
"eval_loss": 0.7073235511779785,
|
| 949 |
+
"eval_runtime": 199.205,
|
| 950 |
+
"eval_samples_per_second": 15.195,
|
| 951 |
+
"eval_steps_per_second": 0.477,
|
| 952 |
+
"eval_wer": 0.4190684588934042,
|
| 953 |
+
"step": 63000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 47.3,
|
| 957 |
+
"learning_rate": 5.3109456440804175e-05,
|
| 958 |
+
"loss": 0.0844,
|
| 959 |
+
"step": 64000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 47.3,
|
| 963 |
+
"eval_loss": 0.6707965135574341,
|
| 964 |
+
"eval_runtime": 203.0283,
|
| 965 |
+
"eval_samples_per_second": 14.909,
|
| 966 |
+
"eval_steps_per_second": 0.468,
|
| 967 |
+
"eval_wer": 0.4145670522038137,
|
| 968 |
+
"step": 64000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 48.04,
|
| 972 |
+
"learning_rate": 5.2365599404318685e-05,
|
| 973 |
+
"loss": 0.0851,
|
| 974 |
+
"step": 65000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 48.04,
|
| 978 |
+
"eval_loss": 0.6518837213516235,
|
| 979 |
+
"eval_runtime": 184.8621,
|
| 980 |
+
"eval_samples_per_second": 16.374,
|
| 981 |
+
"eval_steps_per_second": 0.514,
|
| 982 |
+
"eval_wer": 0.4101281650515786,
|
| 983 |
+
"step": 65000
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 48.78,
|
| 987 |
+
"learning_rate": 5.162099776619509e-05,
|
| 988 |
+
"loss": 0.0832,
|
| 989 |
+
"step": 66000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 48.78,
|
| 993 |
+
"eval_loss": 0.6594052910804749,
|
| 994 |
+
"eval_runtime": 196.4902,
|
| 995 |
+
"eval_samples_per_second": 15.405,
|
| 996 |
+
"eval_steps_per_second": 0.483,
|
| 997 |
+
"eval_wer": 0.41387933729290405,
|
| 998 |
+
"step": 66000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 49.52,
|
| 1002 |
+
"learning_rate": 5.0877140729709605e-05,
|
| 1003 |
+
"loss": 0.0822,
|
| 1004 |
+
"step": 67000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 49.52,
|
| 1008 |
+
"eval_loss": 0.653843104839325,
|
| 1009 |
+
"eval_runtime": 186.7472,
|
| 1010 |
+
"eval_samples_per_second": 16.209,
|
| 1011 |
+
"eval_steps_per_second": 0.509,
|
| 1012 |
+
"eval_wer": 0.415942482025633,
|
| 1013 |
+
"step": 67000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 50.26,
|
| 1017 |
+
"learning_rate": 5.013253909158601e-05,
|
| 1018 |
+
"loss": 0.0833,
|
| 1019 |
+
"step": 68000
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 50.26,
|
| 1023 |
+
"eval_loss": 0.6541770100593567,
|
| 1024 |
+
"eval_runtime": 175.2691,
|
| 1025 |
+
"eval_samples_per_second": 17.271,
|
| 1026 |
+
"eval_steps_per_second": 0.542,
|
| 1027 |
+
"eval_wer": 0.41844326351984995,
|
| 1028 |
+
"step": 68000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 51.0,
|
| 1032 |
+
"learning_rate": 4.9387937453462405e-05,
|
| 1033 |
+
"loss": 0.079,
|
| 1034 |
+
"step": 69000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 51.0,
|
| 1038 |
+
"eval_loss": 0.6602481007575989,
|
| 1039 |
+
"eval_runtime": 176.6295,
|
| 1040 |
+
"eval_samples_per_second": 17.138,
|
| 1041 |
+
"eval_steps_per_second": 0.538,
|
| 1042 |
+
"eval_wer": 0.41212879024695215,
|
| 1043 |
+
"step": 69000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 51.74,
|
| 1047 |
+
"learning_rate": 4.8644080416976914e-05,
|
| 1048 |
+
"loss": 0.0768,
|
| 1049 |
+
"step": 70000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 51.74,
|
| 1053 |
+
"eval_loss": 0.6583812832832336,
|
| 1054 |
+
"eval_runtime": 175.3489,
|
| 1055 |
+
"eval_samples_per_second": 17.263,
|
| 1056 |
+
"eval_steps_per_second": 0.542,
|
| 1057 |
+
"eval_wer": 0.41556736480150047,
|
| 1058 |
+
"step": 70000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 52.48,
|
| 1062 |
+
"learning_rate": 4.790022338049144e-05,
|
| 1063 |
+
"loss": 0.0775,
|
| 1064 |
+
"step": 71000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 52.48,
|
| 1068 |
+
"eval_loss": 0.6354698538780212,
|
| 1069 |
+
"eval_runtime": 173.883,
|
| 1070 |
+
"eval_samples_per_second": 17.408,
|
| 1071 |
+
"eval_steps_per_second": 0.546,
|
| 1072 |
+
"eval_wer": 0.4156924038762113,
|
| 1073 |
+
"step": 71000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 53.22,
|
| 1077 |
+
"learning_rate": 4.7155621742367834e-05,
|
| 1078 |
+
"loss": 0.0762,
|
| 1079 |
+
"step": 72000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 53.22,
|
| 1083 |
+
"eval_loss": 0.6745539307594299,
|
| 1084 |
+
"eval_runtime": 174.1736,
|
| 1085 |
+
"eval_samples_per_second": 17.379,
|
| 1086 |
+
"eval_steps_per_second": 0.545,
|
| 1087 |
+
"eval_wer": 0.40675211003438577,
|
| 1088 |
+
"step": 72000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 53.95,
|
| 1092 |
+
"learning_rate": 4.641176470588236e-05,
|
| 1093 |
+
"loss": 0.0761,
|
| 1094 |
+
"step": 73000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 53.95,
|
| 1098 |
+
"eval_loss": 0.6509293913841248,
|
| 1099 |
+
"eval_runtime": 176.7298,
|
| 1100 |
+
"eval_samples_per_second": 17.128,
|
| 1101 |
+
"eval_steps_per_second": 0.538,
|
| 1102 |
+
"eval_wer": 0.4095029696780244,
|
| 1103 |
+
"step": 73000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 54.69,
|
| 1107 |
+
"learning_rate": 4.566716306775875e-05,
|
| 1108 |
+
"loss": 0.0742,
|
| 1109 |
+
"step": 74000
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 54.69,
|
| 1113 |
+
"eval_loss": 0.6646940112113953,
|
| 1114 |
+
"eval_runtime": 177.5457,
|
| 1115 |
+
"eval_samples_per_second": 17.049,
|
| 1116 |
+
"eval_steps_per_second": 0.535,
|
| 1117 |
+
"eval_wer": 0.41312910284463894,
|
| 1118 |
+
"step": 74000
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 55.43,
|
| 1122 |
+
"learning_rate": 4.492330603127327e-05,
|
| 1123 |
+
"loss": 0.0738,
|
| 1124 |
+
"step": 75000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 55.43,
|
| 1128 |
+
"eval_loss": 0.6746073365211487,
|
| 1129 |
+
"eval_runtime": 182.277,
|
| 1130 |
+
"eval_samples_per_second": 16.607,
|
| 1131 |
+
"eval_steps_per_second": 0.521,
|
| 1132 |
+
"eval_wer": 0.4097530478274461,
|
| 1133 |
+
"step": 75000
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 56.17,
|
| 1137 |
+
"learning_rate": 4.417870439314967e-05,
|
| 1138 |
+
"loss": 0.072,
|
| 1139 |
+
"step": 76000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 56.17,
|
| 1143 |
+
"eval_loss": 0.6681593060493469,
|
| 1144 |
+
"eval_runtime": 189.5798,
|
| 1145 |
+
"eval_samples_per_second": 15.967,
|
| 1146 |
+
"eval_steps_per_second": 0.501,
|
| 1147 |
+
"eval_wer": 0.41081587996248825,
|
| 1148 |
+
"step": 76000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 56.91,
|
| 1152 |
+
"learning_rate": 4.3434102755026064e-05,
|
| 1153 |
+
"loss": 0.0715,
|
| 1154 |
+
"step": 77000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 56.91,
|
| 1158 |
+
"eval_loss": 0.6641326546669006,
|
| 1159 |
+
"eval_runtime": 199.1459,
|
| 1160 |
+
"eval_samples_per_second": 15.2,
|
| 1161 |
+
"eval_steps_per_second": 0.477,
|
| 1162 |
+
"eval_wer": 0.4114410753360425,
|
| 1163 |
+
"step": 77000
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 57.65,
|
| 1167 |
+
"learning_rate": 4.2690245718540587e-05,
|
| 1168 |
+
"loss": 0.0717,
|
| 1169 |
+
"step": 78000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 57.65,
|
| 1173 |
+
"eval_loss": 0.6959729790687561,
|
| 1174 |
+
"eval_runtime": 198.8176,
|
| 1175 |
+
"eval_samples_per_second": 15.225,
|
| 1176 |
+
"eval_steps_per_second": 0.478,
|
| 1177 |
+
"eval_wer": 0.41106595811191,
|
| 1178 |
+
"step": 78000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 58.39,
|
| 1182 |
+
"learning_rate": 4.194564408041698e-05,
|
| 1183 |
+
"loss": 0.0711,
|
| 1184 |
+
"step": 79000
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 58.39,
|
| 1188 |
+
"eval_loss": 0.663927435874939,
|
| 1189 |
+
"eval_runtime": 201.2001,
|
| 1190 |
+
"eval_samples_per_second": 15.045,
|
| 1191 |
+
"eval_steps_per_second": 0.472,
|
| 1192 |
+
"eval_wer": 0.4081900593935605,
|
| 1193 |
+
"step": 79000
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 59.13,
|
| 1197 |
+
"learning_rate": 4.1201042442293374e-05,
|
| 1198 |
+
"loss": 0.0706,
|
| 1199 |
+
"step": 80000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 59.13,
|
| 1203 |
+
"eval_loss": 0.6455733180046082,
|
| 1204 |
+
"eval_runtime": 180.6704,
|
| 1205 |
+
"eval_samples_per_second": 16.754,
|
| 1206 |
+
"eval_steps_per_second": 0.526,
|
| 1207 |
+
"eval_wer": 0.41156611441075336,
|
| 1208 |
+
"step": 80000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 59.87,
|
| 1212 |
+
"learning_rate": 4.045644080416977e-05,
|
| 1213 |
+
"loss": 0.0685,
|
| 1214 |
+
"step": 81000
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 59.87,
|
| 1218 |
+
"eval_loss": 0.6929445862770081,
|
| 1219 |
+
"eval_runtime": 176.1838,
|
| 1220 |
+
"eval_samples_per_second": 17.181,
|
| 1221 |
+
"eval_steps_per_second": 0.539,
|
| 1222 |
+
"eval_wer": 0.4169427946233198,
|
| 1223 |
+
"step": 81000
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 60.61,
|
| 1227 |
+
"learning_rate": 3.971258376768429e-05,
|
| 1228 |
+
"loss": 0.0666,
|
| 1229 |
+
"step": 82000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 60.61,
|
| 1233 |
+
"eval_loss": 0.6526830792427063,
|
| 1234 |
+
"eval_runtime": 175.1162,
|
| 1235 |
+
"eval_samples_per_second": 17.286,
|
| 1236 |
+
"eval_steps_per_second": 0.542,
|
| 1237 |
+
"eval_wer": 0.40987808690215694,
|
| 1238 |
+
"step": 82000
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 61.35,
|
| 1242 |
+
"learning_rate": 3.896798212956069e-05,
|
| 1243 |
+
"loss": 0.0677,
|
| 1244 |
+
"step": 83000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 61.35,
|
| 1248 |
+
"eval_loss": 0.708766758441925,
|
| 1249 |
+
"eval_runtime": 174.527,
|
| 1250 |
+
"eval_samples_per_second": 17.344,
|
| 1251 |
+
"eval_steps_per_second": 0.544,
|
| 1252 |
+
"eval_wer": 0.4113785557986871,
|
| 1253 |
+
"step": 83000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 62.08,
|
| 1257 |
+
"learning_rate": 3.822338049143708e-05,
|
| 1258 |
+
"loss": 0.0667,
|
| 1259 |
+
"step": 84000
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 62.08,
|
| 1263 |
+
"eval_loss": 0.6928330063819885,
|
| 1264 |
+
"eval_runtime": 178.7497,
|
| 1265 |
+
"eval_samples_per_second": 16.934,
|
| 1266 |
+
"eval_steps_per_second": 0.531,
|
| 1267 |
+
"eval_wer": 0.4085026570803376,
|
| 1268 |
+
"step": 84000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 62.82,
|
| 1272 |
+
"learning_rate": 3.74795234549516e-05,
|
| 1273 |
+
"loss": 0.065,
|
| 1274 |
+
"step": 85000
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 62.82,
|
| 1278 |
+
"eval_loss": 0.7296735048294067,
|
| 1279 |
+
"eval_runtime": 174.5663,
|
| 1280 |
+
"eval_samples_per_second": 17.34,
|
| 1281 |
+
"eval_steps_per_second": 0.544,
|
| 1282 |
+
"eval_wer": 0.40962800875273525,
|
| 1283 |
+
"step": 85000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 63.56,
|
| 1287 |
+
"learning_rate": 3.6734921816828e-05,
|
| 1288 |
+
"loss": 0.0655,
|
| 1289 |
+
"step": 86000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 63.56,
|
| 1293 |
+
"eval_loss": 0.7482656836509705,
|
| 1294 |
+
"eval_runtime": 173.9987,
|
| 1295 |
+
"eval_samples_per_second": 17.397,
|
| 1296 |
+
"eval_steps_per_second": 0.546,
|
| 1297 |
+
"eval_wer": 0.4106908408877774,
|
| 1298 |
+
"step": 86000
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 64.3,
|
| 1302 |
+
"learning_rate": 3.59903201787044e-05,
|
| 1303 |
+
"loss": 0.0636,
|
| 1304 |
+
"step": 87000
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 64.3,
|
| 1308 |
+
"eval_loss": 0.7502027750015259,
|
| 1309 |
+
"eval_runtime": 172.4262,
|
| 1310 |
+
"eval_samples_per_second": 17.555,
|
| 1311 |
+
"eval_steps_per_second": 0.551,
|
| 1312 |
+
"eval_wer": 0.41875586120662706,
|
| 1313 |
+
"step": 87000
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 65.04,
|
| 1317 |
+
"learning_rate": 3.5245718540580794e-05,
|
| 1318 |
+
"loss": 0.0645,
|
| 1319 |
+
"step": 88000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 65.04,
|
| 1323 |
+
"eval_loss": 0.7395268678665161,
|
| 1324 |
+
"eval_runtime": 172.5532,
|
| 1325 |
+
"eval_samples_per_second": 17.542,
|
| 1326 |
+
"eval_steps_per_second": 0.551,
|
| 1327 |
+
"eval_wer": 0.40793998124413877,
|
| 1328 |
+
"step": 88000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 65.78,
|
| 1332 |
+
"learning_rate": 3.450186150409531e-05,
|
| 1333 |
+
"loss": 0.0628,
|
| 1334 |
+
"step": 89000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 65.78,
|
| 1338 |
+
"eval_loss": 0.6981734037399292,
|
| 1339 |
+
"eval_runtime": 186.2785,
|
| 1340 |
+
"eval_samples_per_second": 16.25,
|
| 1341 |
+
"eval_steps_per_second": 0.51,
|
| 1342 |
+
"eval_wer": 0.4075023444826508,
|
| 1343 |
+
"step": 89000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 66.52,
|
| 1347 |
+
"learning_rate": 3.375725986597171e-05,
|
| 1348 |
+
"loss": 0.0618,
|
| 1349 |
+
"step": 90000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 66.52,
|
| 1353 |
+
"eval_loss": 0.7291818261146545,
|
| 1354 |
+
"eval_runtime": 195.0277,
|
| 1355 |
+
"eval_samples_per_second": 15.521,
|
| 1356 |
+
"eval_steps_per_second": 0.487,
|
| 1357 |
+
"eval_wer": 0.4113160362613317,
|
| 1358 |
+
"step": 90000
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 67.26,
|
| 1362 |
+
"learning_rate": 3.301340282948623e-05,
|
| 1363 |
+
"loss": 0.0612,
|
| 1364 |
+
"step": 91000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 67.26,
|
| 1368 |
+
"eval_loss": 0.7428854703903198,
|
| 1369 |
+
"eval_runtime": 201.9566,
|
| 1370 |
+
"eval_samples_per_second": 14.988,
|
| 1371 |
+
"eval_steps_per_second": 0.47,
|
| 1372 |
+
"eval_wer": 0.40994060643951236,
|
| 1373 |
+
"step": 91000
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 68.0,
|
| 1377 |
+
"learning_rate": 3.2269545793000746e-05,
|
| 1378 |
+
"loss": 0.0613,
|
| 1379 |
+
"step": 92000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 68.0,
|
| 1383 |
+
"eval_loss": 0.7314553260803223,
|
| 1384 |
+
"eval_runtime": 181.5248,
|
| 1385 |
+
"eval_samples_per_second": 16.675,
|
| 1386 |
+
"eval_steps_per_second": 0.523,
|
| 1387 |
+
"eval_wer": 0.41150359487339794,
|
| 1388 |
+
"step": 92000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 68.74,
|
| 1392 |
+
"learning_rate": 3.152494415487714e-05,
|
| 1393 |
+
"loss": 0.0607,
|
| 1394 |
+
"step": 93000
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 68.74,
|
| 1398 |
+
"eval_loss": 0.7307026386260986,
|
| 1399 |
+
"eval_runtime": 263.2133,
|
| 1400 |
+
"eval_samples_per_second": 11.5,
|
| 1401 |
+
"eval_steps_per_second": 0.361,
|
| 1402 |
+
"eval_wer": 0.40725226633322914,
|
| 1403 |
+
"step": 93000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 69.48,
|
| 1407 |
+
"learning_rate": 3.078034251675354e-05,
|
| 1408 |
+
"loss": 0.0597,
|
| 1409 |
+
"step": 94000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 69.48,
|
| 1413 |
+
"eval_loss": 0.6700393557548523,
|
| 1414 |
+
"eval_runtime": 261.1836,
|
| 1415 |
+
"eval_samples_per_second": 11.59,
|
| 1416 |
+
"eval_steps_per_second": 0.364,
|
| 1417 |
+
"eval_wer": 0.40206314473272897,
|
| 1418 |
+
"step": 94000
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 70.21,
|
| 1422 |
+
"learning_rate": 3.0035740878629936e-05,
|
| 1423 |
+
"loss": 0.0586,
|
| 1424 |
+
"step": 95000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 70.21,
|
| 1428 |
+
"eval_loss": 0.7129278779029846,
|
| 1429 |
+
"eval_runtime": 174.8869,
|
| 1430 |
+
"eval_samples_per_second": 17.308,
|
| 1431 |
+
"eval_steps_per_second": 0.543,
|
| 1432 |
+
"eval_wer": 0.40737730540794,
|
| 1433 |
+
"step": 95000
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 70.95,
|
| 1437 |
+
"learning_rate": 2.9291883842144452e-05,
|
| 1438 |
+
"loss": 0.0585,
|
| 1439 |
+
"step": 96000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 70.95,
|
| 1443 |
+
"eval_loss": 0.7538037896156311,
|
| 1444 |
+
"eval_runtime": 175.7698,
|
| 1445 |
+
"eval_samples_per_second": 17.221,
|
| 1446 |
+
"eval_steps_per_second": 0.54,
|
| 1447 |
+
"eval_wer": 0.4142544545170366,
|
| 1448 |
+
"step": 96000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 71.69,
|
| 1452 |
+
"learning_rate": 2.8547282204020846e-05,
|
| 1453 |
+
"loss": 0.0573,
|
| 1454 |
+
"step": 97000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 71.69,
|
| 1458 |
+
"eval_loss": 0.703994631767273,
|
| 1459 |
+
"eval_runtime": 176.9961,
|
| 1460 |
+
"eval_samples_per_second": 17.102,
|
| 1461 |
+
"eval_steps_per_second": 0.537,
|
| 1462 |
+
"eval_wer": 0.40487652391372303,
|
| 1463 |
+
"step": 97000
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 72.43,
|
| 1467 |
+
"learning_rate": 2.780342516753537e-05,
|
| 1468 |
+
"loss": 0.0575,
|
| 1469 |
+
"step": 98000
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 72.43,
|
| 1473 |
+
"eval_loss": 0.7418561577796936,
|
| 1474 |
+
"eval_runtime": 177.6888,
|
| 1475 |
+
"eval_samples_per_second": 17.035,
|
| 1476 |
+
"eval_steps_per_second": 0.535,
|
| 1477 |
+
"eval_wer": 0.4070647077211629,
|
| 1478 |
+
"step": 98000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 73.17,
|
| 1482 |
+
"learning_rate": 2.7058823529411766e-05,
|
| 1483 |
+
"loss": 0.0567,
|
| 1484 |
+
"step": 99000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 73.17,
|
| 1488 |
+
"eval_loss": 0.7433472871780396,
|
| 1489 |
+
"eval_runtime": 175.8142,
|
| 1490 |
+
"eval_samples_per_second": 17.217,
|
| 1491 |
+
"eval_steps_per_second": 0.54,
|
| 1492 |
+
"eval_wer": 0.4093154110659581,
|
| 1493 |
+
"step": 99000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 73.91,
|
| 1497 |
+
"learning_rate": 2.631422189128816e-05,
|
| 1498 |
+
"loss": 0.054,
|
| 1499 |
+
"step": 100000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 73.91,
|
| 1503 |
+
"eval_loss": 0.6921048760414124,
|
| 1504 |
+
"eval_runtime": 180.3464,
|
| 1505 |
+
"eval_samples_per_second": 16.784,
|
| 1506 |
+
"eval_steps_per_second": 0.527,
|
| 1507 |
+
"eval_wer": 0.40256330103157234,
|
| 1508 |
+
"step": 100000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 74.65,
|
| 1512 |
+
"learning_rate": 2.5570364854802682e-05,
|
| 1513 |
+
"loss": 0.0571,
|
| 1514 |
+
"step": 101000
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 74.65,
|
| 1518 |
+
"eval_loss": 0.7526727318763733,
|
| 1519 |
+
"eval_runtime": 176.5656,
|
| 1520 |
+
"eval_samples_per_second": 17.144,
|
| 1521 |
+
"eval_steps_per_second": 0.538,
|
| 1522 |
+
"eval_wer": 0.4104407627383557,
|
| 1523 |
+
"step": 101000
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 75.39,
|
| 1527 |
+
"learning_rate": 2.48265078183172e-05,
|
| 1528 |
+
"loss": 0.0532,
|
| 1529 |
+
"step": 102000
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 75.39,
|
| 1533 |
+
"eval_loss": 0.717917263507843,
|
| 1534 |
+
"eval_runtime": 180.7508,
|
| 1535 |
+
"eval_samples_per_second": 16.747,
|
| 1536 |
+
"eval_steps_per_second": 0.526,
|
| 1537 |
+
"eval_wer": 0.40693966864645204,
|
| 1538 |
+
"step": 102000
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 76.13,
|
| 1542 |
+
"learning_rate": 2.40819061801936e-05,
|
| 1543 |
+
"loss": 0.0549,
|
| 1544 |
+
"step": 103000
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 76.13,
|
| 1548 |
+
"eval_loss": 0.7699146866798401,
|
| 1549 |
+
"eval_runtime": 199.8758,
|
| 1550 |
+
"eval_samples_per_second": 15.144,
|
| 1551 |
+
"eval_steps_per_second": 0.475,
|
| 1552 |
+
"eval_wer": 0.4080650203188496,
|
| 1553 |
+
"step": 103000
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 76.87,
|
| 1557 |
+
"learning_rate": 2.3338049143708118e-05,
|
| 1558 |
+
"loss": 0.0533,
|
| 1559 |
+
"step": 104000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 76.87,
|
| 1563 |
+
"eval_loss": 0.7196021676063538,
|
| 1564 |
+
"eval_runtime": 186.8201,
|
| 1565 |
+
"eval_samples_per_second": 16.203,
|
| 1566 |
+
"eval_steps_per_second": 0.509,
|
| 1567 |
+
"eval_wer": 0.4058143169740544,
|
| 1568 |
+
"step": 104000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 77.61,
|
| 1572 |
+
"learning_rate": 2.2593447505584515e-05,
|
| 1573 |
+
"loss": 0.0538,
|
| 1574 |
+
"step": 105000
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 77.61,
|
| 1578 |
+
"eval_loss": 0.7532729506492615,
|
| 1579 |
+
"eval_runtime": 186.7527,
|
| 1580 |
+
"eval_samples_per_second": 16.209,
|
| 1581 |
+
"eval_steps_per_second": 0.509,
|
| 1582 |
+
"eval_wer": 0.4068146295717412,
|
| 1583 |
+
"step": 105000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 78.34,
|
| 1587 |
+
"learning_rate": 2.1848845867460908e-05,
|
| 1588 |
+
"loss": 0.0533,
|
| 1589 |
+
"step": 106000
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 78.34,
|
| 1593 |
+
"eval_loss": 0.7514676451683044,
|
| 1594 |
+
"eval_runtime": 203.9001,
|
| 1595 |
+
"eval_samples_per_second": 14.846,
|
| 1596 |
+
"eval_steps_per_second": 0.466,
|
| 1597 |
+
"eval_wer": 0.40762738355736167,
|
| 1598 |
+
"step": 106000
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 79.08,
|
| 1602 |
+
"learning_rate": 2.1104244229337305e-05,
|
| 1603 |
+
"loss": 0.0517,
|
| 1604 |
+
"step": 107000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 79.08,
|
| 1608 |
+
"eval_loss": 0.766864001750946,
|
| 1609 |
+
"eval_runtime": 183.9434,
|
| 1610 |
+
"eval_samples_per_second": 16.456,
|
| 1611 |
+
"eval_steps_per_second": 0.516,
|
| 1612 |
+
"eval_wer": 0.4061894341981869,
|
| 1613 |
+
"step": 107000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 79.82,
|
| 1617 |
+
"learning_rate": 2.0359642591213702e-05,
|
| 1618 |
+
"loss": 0.0521,
|
| 1619 |
+
"step": 108000
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 79.82,
|
| 1623 |
+
"eval_loss": 0.7399023771286011,
|
| 1624 |
+
"eval_runtime": 175.691,
|
| 1625 |
+
"eval_samples_per_second": 17.229,
|
| 1626 |
+
"eval_steps_per_second": 0.541,
|
| 1627 |
+
"eval_wer": 0.4015004688965302,
|
| 1628 |
+
"step": 108000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 80.56,
|
| 1632 |
+
"learning_rate": 1.961578555472822e-05,
|
| 1633 |
+
"loss": 0.0525,
|
| 1634 |
+
"step": 109000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 80.56,
|
| 1638 |
+
"eval_loss": 0.7517969608306885,
|
| 1639 |
+
"eval_runtime": 185.511,
|
| 1640 |
+
"eval_samples_per_second": 16.317,
|
| 1641 |
+
"eval_steps_per_second": 0.512,
|
| 1642 |
+
"eval_wer": 0.4044388871522351,
|
| 1643 |
+
"step": 109000
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 81.3,
|
| 1647 |
+
"learning_rate": 1.887118391660462e-05,
|
| 1648 |
+
"loss": 0.0505,
|
| 1649 |
+
"step": 110000
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 81.3,
|
| 1653 |
+
"eval_loss": 0.7839346528053284,
|
| 1654 |
+
"eval_runtime": 181.6126,
|
| 1655 |
+
"eval_samples_per_second": 16.667,
|
| 1656 |
+
"eval_steps_per_second": 0.523,
|
| 1657 |
+
"eval_wer": 0.4077524226320725,
|
| 1658 |
+
"step": 110000
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 82.04,
|
| 1662 |
+
"learning_rate": 1.8126582278481012e-05,
|
| 1663 |
+
"loss": 0.0508,
|
| 1664 |
+
"step": 111000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 82.04,
|
| 1668 |
+
"eval_loss": 0.7692478895187378,
|
| 1669 |
+
"eval_runtime": 182.2532,
|
| 1670 |
+
"eval_samples_per_second": 16.609,
|
| 1671 |
+
"eval_steps_per_second": 0.521,
|
| 1672 |
+
"eval_wer": 0.4070647077211629,
|
| 1673 |
+
"step": 111000
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 82.78,
|
| 1677 |
+
"learning_rate": 1.7383469843633657e-05,
|
| 1678 |
+
"loss": 0.0496,
|
| 1679 |
+
"step": 112000
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 82.78,
|
| 1683 |
+
"eval_loss": 0.7821455597877502,
|
| 1684 |
+
"eval_runtime": 227.3787,
|
| 1685 |
+
"eval_samples_per_second": 13.313,
|
| 1686 |
+
"eval_steps_per_second": 0.418,
|
| 1687 |
+
"eval_wer": 0.40987808690215694,
|
| 1688 |
+
"step": 112000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 83.52,
|
| 1692 |
+
"learning_rate": 1.663886820551005e-05,
|
| 1693 |
+
"loss": 0.0493,
|
| 1694 |
+
"step": 113000
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 83.52,
|
| 1698 |
+
"eval_loss": 0.7528874278068542,
|
| 1699 |
+
"eval_runtime": 217.8709,
|
| 1700 |
+
"eval_samples_per_second": 13.894,
|
| 1701 |
+
"eval_steps_per_second": 0.436,
|
| 1702 |
+
"eval_wer": 0.4055642388246327,
|
| 1703 |
+
"step": 113000
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 84.26,
|
| 1707 |
+
"learning_rate": 1.5894266567386448e-05,
|
| 1708 |
+
"loss": 0.0498,
|
| 1709 |
+
"step": 114000
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 84.26,
|
| 1713 |
+
"eval_loss": 0.7717328071594238,
|
| 1714 |
+
"eval_runtime": 216.4941,
|
| 1715 |
+
"eval_samples_per_second": 13.982,
|
| 1716 |
+
"eval_steps_per_second": 0.439,
|
| 1717 |
+
"eval_wer": 0.40493904345107845,
|
| 1718 |
+
"step": 114000
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 85.0,
|
| 1722 |
+
"learning_rate": 1.5149664929262845e-05,
|
| 1723 |
+
"loss": 0.0499,
|
| 1724 |
+
"step": 115000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 85.0,
|
| 1728 |
+
"eval_loss": 0.757793128490448,
|
| 1729 |
+
"eval_runtime": 213.4299,
|
| 1730 |
+
"eval_samples_per_second": 14.183,
|
| 1731 |
+
"eval_steps_per_second": 0.445,
|
| 1732 |
+
"eval_wer": 0.40406376992810256,
|
| 1733 |
+
"step": 115000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 85.74,
|
| 1737 |
+
"learning_rate": 1.4405807892777364e-05,
|
| 1738 |
+
"loss": 0.0485,
|
| 1739 |
+
"step": 116000
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 85.74,
|
| 1743 |
+
"eval_loss": 0.7553907632827759,
|
| 1744 |
+
"eval_runtime": 181.6761,
|
| 1745 |
+
"eval_samples_per_second": 16.662,
|
| 1746 |
+
"eval_steps_per_second": 0.523,
|
| 1747 |
+
"eval_wer": 0.4044388871522351,
|
| 1748 |
+
"step": 116000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 86.47,
|
| 1752 |
+
"learning_rate": 1.3661206254653761e-05,
|
| 1753 |
+
"loss": 0.0479,
|
| 1754 |
+
"step": 117000
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 86.47,
|
| 1758 |
+
"eval_loss": 0.8072551488876343,
|
| 1759 |
+
"eval_runtime": 188.7645,
|
| 1760 |
+
"eval_samples_per_second": 16.036,
|
| 1761 |
+
"eval_steps_per_second": 0.503,
|
| 1762 |
+
"eval_wer": 0.40625195373554235,
|
| 1763 |
+
"step": 117000
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 87.21,
|
| 1767 |
+
"learning_rate": 1.2917349218168282e-05,
|
| 1768 |
+
"loss": 0.0462,
|
| 1769 |
+
"step": 118000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 87.21,
|
| 1773 |
+
"eval_loss": 0.7592765092849731,
|
| 1774 |
+
"eval_runtime": 187.3254,
|
| 1775 |
+
"eval_samples_per_second": 16.159,
|
| 1776 |
+
"eval_steps_per_second": 0.507,
|
| 1777 |
+
"eval_wer": 0.4025007814942169,
|
| 1778 |
+
"step": 118000
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 87.95,
|
| 1782 |
+
"learning_rate": 1.2172747580044677e-05,
|
| 1783 |
+
"loss": 0.0475,
|
| 1784 |
+
"step": 119000
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 87.95,
|
| 1788 |
+
"eval_loss": 0.7812408208847046,
|
| 1789 |
+
"eval_runtime": 227.3028,
|
| 1790 |
+
"eval_samples_per_second": 13.317,
|
| 1791 |
+
"eval_steps_per_second": 0.418,
|
| 1792 |
+
"eval_wer": 0.40493904345107845,
|
| 1793 |
+
"step": 119000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 88.69,
|
| 1797 |
+
"learning_rate": 1.1428890543559197e-05,
|
| 1798 |
+
"loss": 0.0455,
|
| 1799 |
+
"step": 120000
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 88.69,
|
| 1803 |
+
"eval_loss": 0.779983401298523,
|
| 1804 |
+
"eval_runtime": 223.0101,
|
| 1805 |
+
"eval_samples_per_second": 13.573,
|
| 1806 |
+
"eval_steps_per_second": 0.426,
|
| 1807 |
+
"eval_wer": 0.40487652391372303,
|
| 1808 |
+
"step": 120000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 89.43,
|
| 1812 |
+
"learning_rate": 1.0684288905435592e-05,
|
| 1813 |
+
"loss": 0.0451,
|
| 1814 |
+
"step": 121000
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 89.43,
|
| 1818 |
+
"eval_loss": 0.8092254400253296,
|
| 1819 |
+
"eval_runtime": 251.7762,
|
| 1820 |
+
"eval_samples_per_second": 12.023,
|
| 1821 |
+
"eval_steps_per_second": 0.377,
|
| 1822 |
+
"eval_wer": 0.40737730540794,
|
| 1823 |
+
"step": 121000
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 90.17,
|
| 1827 |
+
"learning_rate": 9.939687267311989e-06,
|
| 1828 |
+
"loss": 0.0455,
|
| 1829 |
+
"step": 122000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 90.17,
|
| 1833 |
+
"eval_loss": 0.7754979133605957,
|
| 1834 |
+
"eval_runtime": 271.2522,
|
| 1835 |
+
"eval_samples_per_second": 11.159,
|
| 1836 |
+
"eval_steps_per_second": 0.35,
|
| 1837 |
+
"eval_wer": 0.4074398249452954,
|
| 1838 |
+
"step": 122000
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 90.91,
|
| 1842 |
+
"learning_rate": 9.195085629188386e-06,
|
| 1843 |
+
"loss": 0.0476,
|
| 1844 |
+
"step": 123000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 90.91,
|
| 1848 |
+
"eval_loss": 0.7846540212631226,
|
| 1849 |
+
"eval_runtime": 244.3416,
|
| 1850 |
+
"eval_samples_per_second": 12.388,
|
| 1851 |
+
"eval_steps_per_second": 0.389,
|
| 1852 |
+
"eval_wer": 0.40593935604876524,
|
| 1853 |
+
"step": 123000
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 91.65,
|
| 1857 |
+
"learning_rate": 8.451228592702905e-06,
|
| 1858 |
+
"loss": 0.0461,
|
| 1859 |
+
"step": 124000
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 91.65,
|
| 1863 |
+
"eval_loss": 0.7784237861633301,
|
| 1864 |
+
"eval_runtime": 267.8746,
|
| 1865 |
+
"eval_samples_per_second": 11.3,
|
| 1866 |
+
"eval_steps_per_second": 0.355,
|
| 1867 |
+
"eval_wer": 0.4034385745545483,
|
| 1868 |
+
"step": 124000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 92.39,
|
| 1872 |
+
"learning_rate": 7.7066269545793e-06,
|
| 1873 |
+
"loss": 0.0454,
|
| 1874 |
+
"step": 125000
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 92.39,
|
| 1878 |
+
"eval_loss": 0.7823372483253479,
|
| 1879 |
+
"eval_runtime": 207.7458,
|
| 1880 |
+
"eval_samples_per_second": 14.571,
|
| 1881 |
+
"eval_steps_per_second": 0.457,
|
| 1882 |
+
"eval_wer": 0.403251015942482,
|
| 1883 |
+
"step": 125000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 93.13,
|
| 1887 |
+
"learning_rate": 6.9620253164556965e-06,
|
| 1888 |
+
"loss": 0.0453,
|
| 1889 |
+
"step": 126000
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 93.13,
|
| 1893 |
+
"eval_loss": 0.7685350179672241,
|
| 1894 |
+
"eval_runtime": 180.1025,
|
| 1895 |
+
"eval_samples_per_second": 16.807,
|
| 1896 |
+
"eval_steps_per_second": 0.527,
|
| 1897 |
+
"eval_wer": 0.4015004688965302,
|
| 1898 |
+
"step": 126000
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 93.87,
|
| 1902 |
+
"learning_rate": 6.2174236783320925e-06,
|
| 1903 |
+
"loss": 0.0451,
|
| 1904 |
+
"step": 127000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 93.87,
|
| 1908 |
+
"eval_loss": 0.7680614590644836,
|
| 1909 |
+
"eval_runtime": 177.9917,
|
| 1910 |
+
"eval_samples_per_second": 17.006,
|
| 1911 |
+
"eval_steps_per_second": 0.534,
|
| 1912 |
+
"eval_wer": 0.4,
|
| 1913 |
+
"step": 127000
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 94.6,
|
| 1917 |
+
"learning_rate": 5.473566641846612e-06,
|
| 1918 |
+
"loss": 0.0428,
|
| 1919 |
+
"step": 128000
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 94.6,
|
| 1923 |
+
"eval_loss": 0.7936143279075623,
|
| 1924 |
+
"eval_runtime": 180.9951,
|
| 1925 |
+
"eval_samples_per_second": 16.724,
|
| 1926 |
+
"eval_steps_per_second": 0.525,
|
| 1927 |
+
"eval_wer": 0.4034385745545483,
|
| 1928 |
+
"step": 128000
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 95.34,
|
| 1932 |
+
"learning_rate": 4.728965003723009e-06,
|
| 1933 |
+
"loss": 0.0445,
|
| 1934 |
+
"step": 129000
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 95.34,
|
| 1938 |
+
"eval_loss": 0.7666231989860535,
|
| 1939 |
+
"eval_runtime": 203.0139,
|
| 1940 |
+
"eval_samples_per_second": 14.91,
|
| 1941 |
+
"eval_steps_per_second": 0.468,
|
| 1942 |
+
"eval_wer": 0.39968740231322286,
|
| 1943 |
+
"step": 129000
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 96.08,
|
| 1947 |
+
"learning_rate": 3.984363365599404e-06,
|
| 1948 |
+
"loss": 0.0441,
|
| 1949 |
+
"step": 130000
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 96.08,
|
| 1953 |
+
"eval_loss": 0.7681093811988831,
|
| 1954 |
+
"eval_runtime": 206.6264,
|
| 1955 |
+
"eval_samples_per_second": 14.65,
|
| 1956 |
+
"eval_steps_per_second": 0.46,
|
| 1957 |
+
"eval_wer": 0.4008752735229759,
|
| 1958 |
+
"step": 130000
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 96.82,
|
| 1962 |
+
"learning_rate": 3.2412509307520474e-06,
|
| 1963 |
+
"loss": 0.043,
|
| 1964 |
+
"step": 131000
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 96.82,
|
| 1968 |
+
"eval_loss": 0.7704082131385803,
|
| 1969 |
+
"eval_runtime": 196.2406,
|
| 1970 |
+
"eval_samples_per_second": 15.425,
|
| 1971 |
+
"eval_steps_per_second": 0.484,
|
| 1972 |
+
"eval_wer": 0.4008752735229759,
|
| 1973 |
+
"step": 131000
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 97.56,
|
| 1977 |
+
"learning_rate": 2.496649292628444e-06,
|
| 1978 |
+
"loss": 0.043,
|
| 1979 |
+
"step": 132000
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 97.56,
|
| 1983 |
+
"eval_loss": 0.7688604593276978,
|
| 1984 |
+
"eval_runtime": 208.2948,
|
| 1985 |
+
"eval_samples_per_second": 14.532,
|
| 1986 |
+
"eval_steps_per_second": 0.456,
|
| 1987 |
+
"eval_wer": 0.3981244138793373,
|
| 1988 |
+
"step": 132000
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 98.3,
|
| 1992 |
+
"learning_rate": 1.7520476545048399e-06,
|
| 1993 |
+
"loss": 0.0429,
|
| 1994 |
+
"step": 133000
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 98.3,
|
| 1998 |
+
"eval_loss": 0.7736847996711731,
|
| 1999 |
+
"eval_runtime": 207.1539,
|
| 2000 |
+
"eval_samples_per_second": 14.612,
|
| 2001 |
+
"eval_steps_per_second": 0.459,
|
| 2002 |
+
"eval_wer": 0.3992497655517349,
|
| 2003 |
+
"step": 133000
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 99.04,
|
| 2007 |
+
"learning_rate": 1.0081906180193596e-06,
|
| 2008 |
+
"loss": 0.0434,
|
| 2009 |
+
"step": 134000
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 99.04,
|
| 2013 |
+
"eval_loss": 0.7795138955116272,
|
| 2014 |
+
"eval_runtime": 196.4909,
|
| 2015 |
+
"eval_samples_per_second": 15.405,
|
| 2016 |
+
"eval_steps_per_second": 0.483,
|
| 2017 |
+
"eval_wer": 0.4012503907471085,
|
| 2018 |
+
"step": 134000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 99.78,
|
| 2022 |
+
"learning_rate": 2.635889798957558e-07,
|
| 2023 |
+
"loss": 0.0433,
|
| 2024 |
+
"step": 135000
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 99.78,
|
| 2028 |
+
"eval_loss": 0.7773757576942444,
|
| 2029 |
+
"eval_runtime": 198.9794,
|
| 2030 |
+
"eval_samples_per_second": 15.213,
|
| 2031 |
+
"eval_steps_per_second": 0.477,
|
| 2032 |
+
"eval_wer": 0.4008127539856205,
|
| 2033 |
+
"step": 135000
|
| 2034 |
+
}
|
| 2035 |
+
],
|
| 2036 |
+
"max_steps": 135300,
|
| 2037 |
+
"num_train_epochs": 100,
|
| 2038 |
+
"total_flos": 1.184566523613751e+21,
|
| 2039 |
+
"trial_name": null,
|
| 2040 |
+
"trial_params": null
|
| 2041 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b84fdc2b5205eb905d41c1c2102380221b9f615acfe849482ef528e3e37c723
|
| 3 |
+
size 2671
|