Hugging Face's logo

yesj1234
/
enko_xlsr_100p_sup

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
./train_dataset_sup.py
Generated from Trainer
Inference Endpoints
Model card Files Community
yesj1234 commited on
Commit
d923e5b
•
1 Parent(s): 3c937f2

Upload folder using huggingface_hub

Browse files
Files changed (42) hide show
  1. README.md +60 -0
  2. added_tokens.json +4 -0
  3. all_results.json +8 -0
  4. checkpoint-1231340/added_tokens.json +4 -0
  5. checkpoint-1231340/config.json +117 -0
  6. checkpoint-1231340/model.safetensors +3 -0
  7. checkpoint-1231340/optimizer.pt +3 -0
  8. checkpoint-1231340/preprocessor_config.json +10 -0
  9. checkpoint-1231340/rng_state_0.pth +3 -0
  10. checkpoint-1231340/rng_state_1.pth +3 -0
  11. checkpoint-1231340/rng_state_2.pth +3 -0
  12. checkpoint-1231340/rng_state_3.pth +3 -0
  13. checkpoint-1231340/scheduler.pt +3 -0
  14. checkpoint-1231340/special_tokens_map.json +30 -0
  15. checkpoint-1231340/tokenizer_config.json +51 -0
  16. checkpoint-1231340/trainer_state.json +3711 -0
  17. checkpoint-1231340/training_args.bin +3 -0
  18. checkpoint-1231340/vocab.json +98 -0
  19. checkpoint-1273800/added_tokens.json +4 -0
  20. checkpoint-1273800/config.json +117 -0
  21. checkpoint-1273800/model.safetensors +3 -0
  22. checkpoint-1273800/optimizer.pt +3 -0
  23. checkpoint-1273800/preprocessor_config.json +10 -0
  24. checkpoint-1273800/rng_state_0.pth +3 -0
  25. checkpoint-1273800/rng_state_1.pth +3 -0
  26. checkpoint-1273800/rng_state_2.pth +3 -0
  27. checkpoint-1273800/rng_state_3.pth +3 -0
  28. checkpoint-1273800/scheduler.pt +3 -0
  29. checkpoint-1273800/special_tokens_map.json +30 -0
  30. checkpoint-1273800/tokenizer_config.json +51 -0
  31. checkpoint-1273800/trainer_state.json +3837 -0
  32. checkpoint-1273800/training_args.bin +3 -0
  33. checkpoint-1273800/vocab.json +98 -0
  34. config.json +117 -0
  35. model.safetensors +3 -0
  36. preprocessor_config.json +10 -0
  37. special_tokens_map.json +30 -0
  38. tokenizer_config.json +51 -0
  39. train_results.json +8 -0
  40. trainer_state.json +3846 -0
  41. training_args.bin +3 -0
  42. vocab.json +98 -0
README.md ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: facebook/wav2vec2-large-xlsr-53
4
+ tags:
5
+ - automatic-speech-recognition
6
+ - ./train_dataset_sup.py
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: finetuned_model
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # finetuned_model
17
+
18
+ This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the ./TRAIN_DATASET_SUP.PY - NA dataset.
19
+
20
+ ## Model description
21
+
22
+ More information needed
23
+
24
+ ## Intended uses & limitations
25
+
26
+ More information needed
27
+
28
+ ## Training and evaluation data
29
+
30
+ More information needed
31
+
32
+ ## Training procedure
33
+
34
+ ### Training hyperparameters
35
+
36
+ The following hyperparameters were used during training:
37
+ - learning_rate: 0.0003
38
+ - train_batch_size: 2
39
+ - eval_batch_size: 8
40
+ - seed: 42
41
+ - distributed_type: multi-GPU
42
+ - num_devices: 4
43
+ - gradient_accumulation_steps: 2
44
+ - total_train_batch_size: 16
45
+ - total_eval_batch_size: 32
46
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
47
+ - lr_scheduler_type: linear
48
+ - lr_scheduler_warmup_steps: 2000
49
+ - num_epochs: 30
50
+
51
+ ### Training results
52
+
53
+
54
+
55
+ ### Framework versions
56
+
57
+ - Transformers 4.36.2
58
+ - Pytorch 2.1.2+cu121
59
+ - Datasets 2.16.1
60
+ - Tokenizers 0.15.0
added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "</s>": 97,
3
+ "<s>": 96
4
+ }
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 30.0,
3
+ "train_loss": 0.6109381775774412,
4
+ "train_runtime": 724079.1012,
5
+ "train_samples": 679360,
6
+ "train_samples_per_second": 28.147,
7
+ "train_steps_per_second": 1.759
8
+ }
checkpoint-1231340/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "</s>": 97,
3
+ "<s>": 96
4
+ }
checkpoint-1231340/config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
3
+ "activation_dropout": 0.0,
4
+ "adapter_attn_dim": null,
5
+ "adapter_kernel_size": 3,
6
+ "adapter_stride": 2,
7
+ "add_adapter": false,
8
+ "apply_spec_augment": true,
9
+ "architectures": [
10
+ "Wav2Vec2ForCTC"
11
+ ],
12
+ "attention_dropout": 0.05,
13
+ "bos_token_id": 1,
14
+ "classifier_proj_size": 256,
15
+ "codevector_dim": 768,
16
+ "contrastive_logits_temperature": 0.1,
17
+ "conv_bias": true,
18
+ "conv_dim": [
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512,
25
+ 512
26
+ ],
27
+ "conv_kernel": [
28
+ 10,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 3,
33
+ 2,
34
+ 2
35
+ ],
36
+ "conv_stride": [
37
+ 5,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2,
43
+ 2
44
+ ],
45
+ "ctc_loss_reduction": "mean",
46
+ "ctc_zero_infinity": true,
47
+ "diversity_loss_weight": 0.1,
48
+ "do_stable_layer_norm": true,
49
+ "eos_token_id": 2,
50
+ "feat_extract_activation": "gelu",
51
+ "feat_extract_dropout": 0.0,
52
+ "feat_extract_norm": "layer",
53
+ "feat_proj_dropout": 0.05,
54
+ "feat_quantizer_dropout": 0.0,
55
+ "final_dropout": 0.0,
56
+ "gradient_checkpointing": false,
57
+ "hidden_act": "gelu",
58
+ "hidden_dropout": 0.05,
59
+ "hidden_size": 1024,
60
+ "initializer_range": 0.02,
61
+ "intermediate_size": 4096,
62
+ "layer_norm_eps": 1e-05,
63
+ "layerdrop": 0.05,
64
+ "mask_channel_length": 10,
65
+ "mask_channel_min_space": 1,
66
+ "mask_channel_other": 0.0,
67
+ "mask_channel_prob": 0.0,
68
+ "mask_channel_selection": "static",
69
+ "mask_feature_length": 10,
70
+ "mask_feature_min_masks": 0,
71
+ "mask_feature_prob": 0.0,
72
+ "mask_time_length": 10,
73
+ "mask_time_min_masks": 2,
74
+ "mask_time_min_space": 1,
75
+ "mask_time_other": 0.0,
76
+ "mask_time_prob": 0.05,
77
+ "mask_time_selection": "static",
78
+ "model_type": "wav2vec2",
79
+ "num_adapter_layers": 3,
80
+ "num_attention_heads": 16,
81
+ "num_codevector_groups": 2,
82
+ "num_codevectors_per_group": 320,
83
+ "num_conv_pos_embedding_groups": 16,
84
+ "num_conv_pos_embeddings": 128,
85
+ "num_feat_extract_layers": 7,
86
+ "num_hidden_layers": 24,
87
+ "num_negatives": 100,
88
+ "output_hidden_size": 1024,
89
+ "pad_token_id": 95,
90
+ "proj_codevector_dim": 768,
91
+ "tdnn_dilation": [
92
+ 1,
93
+ 2,
94
+ 3,
95
+ 1,
96
+ 1
97
+ ],
98
+ "tdnn_dim": [
99
+ 512,
100
+ 512,
101
+ 512,
102
+ 512,
103
+ 1500
104
+ ],
105
+ "tdnn_kernel": [
106
+ 5,
107
+ 3,
108
+ 3,
109
+ 1,
110
+ 1
111
+ ],
112
+ "torch_dtype": "float32",
113
+ "transformers_version": "4.36.2",
114
+ "use_weighted_layer_sum": false,
115
+ "vocab_size": 98,
116
+ "xvector_output_dim": 512
117
+ }
checkpoint-1231340/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:615a1b12a3db19ff0ff805f58d8d7454e1dc707d80805e5b4f605035c3bec7a9
3
+ size 1262209280
checkpoint-1231340/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63d5d290ccc89c157302d05433c6ed41dec2bd92c88a6662338634579d7b2c5c
3
+ size 2490963382
checkpoint-1231340/preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
checkpoint-1231340/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:879682baf33cb19919209ea5c8335ba8e418be760f748ec86358eb4669894dac
3
+ size 15088
checkpoint-1231340/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f923e5b9a0705318dbc3ebf1b9e56986b395019607d311bc7fbdcc9b39ab9c3
3
+ size 15088
checkpoint-1231340/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6def3dcc91d318ceda8ecb1c32718e8a2148a7156470ed35160c51d182583e6
3
+ size 15024
checkpoint-1231340/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4440e42f183b7eb198241e0c003a3398a58a790fbbcb8bac1e32092ebf43b2bc
3
+ size 15024
checkpoint-1231340/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:763bd062100ffb9900b0cf3a5b0cf6dbb68122a6209f77d24f8302beed6f87e5
3
+ size 1064
checkpoint-1231340/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": true,
19
+ "normalized": false,
20
+ "rstrip": true,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "[UNK]",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": true,
28
+ "single_word": false
29
+ }
30
+ }
checkpoint-1231340/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "94": {
4
+ "content": "[UNK]",
5
+ "lstrip": true,
6
+ "normalized": false,
7
+ "rstrip": true,
8
+ "single_word": false,
9
+ "special": false
10
+ },
11
+ "95": {
12
+ "content": "[PAD]",
13
+ "lstrip": true,
14
+ "normalized": false,
15
+ "rstrip": true,
16
+ "single_word": false,
17
+ "special": false
18
+ },
19
+ "96": {
20
+ "content": "<s>",
21
+ "lstrip": false,
22
+ "normalized": true,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "97": {
28
+ "content": "</s>",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<s>",
37
+ "clean_up_tokenization_spaces": true,
38
+ "config": null,
39
+ "do_lower_case": false,
40
+ "eos_token": "</s>",
41
+ "model_max_length": 1000000000000000019884624838656,
42
+ "pad_token": "[PAD]",
43
+ "processor_class": "Wav2Vec2Processor",
44
+ "replace_word_delimiter_char": " ",
45
+ "target_lang": null,
46
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
47
+ "tokenizer_type": "wav2vec2",
48
+ "trust_remote_code": false,
49
+ "unk_token": "[UNK]",
50
+ "word_delimiter_token": "|"
51
+ }
checkpoint-1231340/trainer_state.json ADDED
@@ -0,0 +1,3711 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 29.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1231340,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.05,
13
+ "learning_rate": 0.0003,
14
+ "loss": 3.0982,
15
+ "step": 2000
16
+ },
17
+ {
18
+ "epoch": 0.09,
19
+ "learning_rate": 0.0002995282277087592,
20
+ "loss": 1.1999,
21
+ "step": 4000
22
+ },
23
+ {
24
+ "epoch": 0.14,
25
+ "learning_rate": 0.00029905645541751844,
26
+ "loss": 1.0995,
27
+ "step": 6000
28
+ },
29
+ {
30
+ "epoch": 0.19,
31
+ "learning_rate": 0.0002985846831262777,
32
+ "loss": 1.0716,
33
+ "step": 8000
34
+ },
35
+ {
36
+ "epoch": 0.24,
37
+ "learning_rate": 0.0002981129108350369,
38
+ "loss": 1.0664,
39
+ "step": 10000
40
+ },
41
+ {
42
+ "epoch": 0.28,
43
+ "learning_rate": 0.0002976411385437962,
44
+ "loss": 1.0223,
45
+ "step": 12000
46
+ },
47
+ {
48
+ "epoch": 0.33,
49
+ "learning_rate": 0.00029716936625255537,
50
+ "loss": 0.9955,
51
+ "step": 14000
52
+ },
53
+ {
54
+ "epoch": 0.38,
55
+ "learning_rate": 0.00029669759396131466,
56
+ "loss": 0.9925,
57
+ "step": 16000
58
+ },
59
+ {
60
+ "epoch": 0.42,
61
+ "learning_rate": 0.0002962258216700739,
62
+ "loss": 0.9951,
63
+ "step": 18000
64
+ },
65
+ {
66
+ "epoch": 0.47,
67
+ "learning_rate": 0.0002957540493788331,
68
+ "loss": 0.9811,
69
+ "step": 20000
70
+ },
71
+ {
72
+ "epoch": 0.52,
73
+ "learning_rate": 0.00029528227708759236,
74
+ "loss": 0.9727,
75
+ "step": 22000
76
+ },
77
+ {
78
+ "epoch": 0.57,
79
+ "learning_rate": 0.0002948105047963516,
80
+ "loss": 0.9655,
81
+ "step": 24000
82
+ },
83
+ {
84
+ "epoch": 0.61,
85
+ "learning_rate": 0.0002943387325051109,
86
+ "loss": 0.959,
87
+ "step": 26000
88
+ },
89
+ {
90
+ "epoch": 0.66,
91
+ "learning_rate": 0.00029386696021387005,
92
+ "loss": 0.9591,
93
+ "step": 28000
94
+ },
95
+ {
96
+ "epoch": 0.71,
97
+ "learning_rate": 0.00029339518792262934,
98
+ "loss": 0.9455,
99
+ "step": 30000
100
+ },
101
+ {
102
+ "epoch": 0.75,
103
+ "learning_rate": 0.0002929234156313886,
104
+ "loss": 0.9425,
105
+ "step": 32000
106
+ },
107
+ {
108
+ "epoch": 0.8,
109
+ "learning_rate": 0.0002924516433401478,
110
+ "loss": 0.9373,
111
+ "step": 34000
112
+ },
113
+ {
114
+ "epoch": 0.85,
115
+ "learning_rate": 0.00029197987104890704,
116
+ "loss": 0.9341,
117
+ "step": 36000
118
+ },
119
+ {
120
+ "epoch": 0.89,
121
+ "learning_rate": 0.00029150809875766627,
122
+ "loss": 0.93,
123
+ "step": 38000
124
+ },
125
+ {
126
+ "epoch": 0.94,
127
+ "learning_rate": 0.0002910363264664255,
128
+ "loss": 0.9184,
129
+ "step": 40000
130
+ },
131
+ {
132
+ "epoch": 0.99,
133
+ "learning_rate": 0.00029056455417518474,
134
+ "loss": 0.9222,
135
+ "step": 42000
136
+ },
137
+ {
138
+ "epoch": 1.04,
139
+ "learning_rate": 0.000290092781883944,
140
+ "loss": 0.8941,
141
+ "step": 44000
142
+ },
143
+ {
144
+ "epoch": 1.08,
145
+ "learning_rate": 0.0002896210095927032,
146
+ "loss": 0.8868,
147
+ "step": 46000
148
+ },
149
+ {
150
+ "epoch": 1.13,
151
+ "learning_rate": 0.0002891492373014625,
152
+ "loss": 0.8894,
153
+ "step": 48000
154
+ },
155
+ {
156
+ "epoch": 1.18,
157
+ "learning_rate": 0.0002886774650102217,
158
+ "loss": 0.8799,
159
+ "step": 50000
160
+ },
161
+ {
162
+ "epoch": 1.22,
163
+ "learning_rate": 0.00028820569271898095,
164
+ "loss": 0.8781,
165
+ "step": 52000
166
+ },
167
+ {
168
+ "epoch": 1.27,
169
+ "learning_rate": 0.0002877339204277402,
170
+ "loss": 0.8853,
171
+ "step": 54000
172
+ },
173
+ {
174
+ "epoch": 1.32,
175
+ "learning_rate": 0.0002872621481364994,
176
+ "loss": 0.8829,
177
+ "step": 56000
178
+ },
179
+ {
180
+ "epoch": 1.37,
181
+ "learning_rate": 0.00028679037584525865,
182
+ "loss": 0.8747,
183
+ "step": 58000
184
+ },
185
+ {
186
+ "epoch": 1.41,
187
+ "learning_rate": 0.0002863186035540179,
188
+ "loss": 0.8806,
189
+ "step": 60000
190
+ },
191
+ {
192
+ "epoch": 1.46,
193
+ "learning_rate": 0.00028584683126277717,
194
+ "loss": 0.8868,
195
+ "step": 62000
196
+ },
197
+ {
198
+ "epoch": 1.51,
199
+ "learning_rate": 0.00028537505897153635,
200
+ "loss": 0.8806,
201
+ "step": 64000
202
+ },
203
+ {
204
+ "epoch": 1.55,
205
+ "learning_rate": 0.00028490328668029564,
206
+ "loss": 0.8771,
207
+ "step": 66000
208
+ },
209
+ {
210
+ "epoch": 1.6,
211
+ "learning_rate": 0.00028443151438905487,
212
+ "loss": 0.8648,
213
+ "step": 68000
214
+ },
215
+ {
216
+ "epoch": 1.65,
217
+ "learning_rate": 0.0002839597420978141,
218
+ "loss": 0.8747,
219
+ "step": 70000
220
+ },
221
+ {
222
+ "epoch": 1.7,
223
+ "learning_rate": 0.00028348796980657334,
224
+ "loss": 0.8719,
225
+ "step": 72000
226
+ },
227
+ {
228
+ "epoch": 1.74,
229
+ "learning_rate": 0.00028301619751533257,
230
+ "loss": 0.8641,
231
+ "step": 74000
232
+ },
233
+ {
234
+ "epoch": 1.79,
235
+ "learning_rate": 0.0002825444252240918,
236
+ "loss": 0.8555,
237
+ "step": 76000
238
+ },
239
+ {
240
+ "epoch": 1.84,
241
+ "learning_rate": 0.00028207265293285103,
242
+ "loss": 0.8583,
243
+ "step": 78000
244
+ },
245
+ {
246
+ "epoch": 1.88,
247
+ "learning_rate": 0.0002816008806416103,
248
+ "loss": 0.8624,
249
+ "step": 80000
250
+ },
251
+ {
252
+ "epoch": 1.93,
253
+ "learning_rate": 0.0002811291083503695,
254
+ "loss": 0.8578,
255
+ "step": 82000
256
+ },
257
+ {
258
+ "epoch": 1.98,
259
+ "learning_rate": 0.0002806573360591288,
260
+ "loss": 0.8474,
261
+ "step": 84000
262
+ },
263
+ {
264
+ "epoch": 2.03,
265
+ "learning_rate": 0.000280185563767888,
266
+ "loss": 0.8366,
267
+ "step": 86000
268
+ },
269
+ {
270
+ "epoch": 2.07,
271
+ "learning_rate": 0.00027971379147664725,
272
+ "loss": 0.8226,
273
+ "step": 88000
274
+ },
275
+ {
276
+ "epoch": 2.12,
277
+ "learning_rate": 0.0002792420191854065,
278
+ "loss": 0.8267,
279
+ "step": 90000
280
+ },
281
+ {
282
+ "epoch": 2.17,
283
+ "learning_rate": 0.0002787702468941657,
284
+ "loss": 0.8231,
285
+ "step": 92000
286
+ },
287
+ {
288
+ "epoch": 2.21,
289
+ "learning_rate": 0.00027829847460292495,
290
+ "loss": 0.8284,
291
+ "step": 94000
292
+ },
293
+ {
294
+ "epoch": 2.26,
295
+ "learning_rate": 0.0002778267023116842,
296
+ "loss": 0.8153,
297
+ "step": 96000
298
+ },
299
+ {
300
+ "epoch": 2.31,
301
+ "learning_rate": 0.00027735493002044347,
302
+ "loss": 0.8241,
303
+ "step": 98000
304
+ },
305
+ {
306
+ "epoch": 2.36,
307
+ "learning_rate": 0.00027688315772920265,
308
+ "loss": 0.8246,
309
+ "step": 100000
310
+ },
311
+ {
312
+ "epoch": 2.4,
313
+ "learning_rate": 0.00027641138543796194,
314
+ "loss": 0.8219,
315
+ "step": 102000
316
+ },
317
+ {
318
+ "epoch": 2.45,
319
+ "learning_rate": 0.00027593961314672117,
320
+ "loss": 0.8244,
321
+ "step": 104000
322
+ },
323
+ {
324
+ "epoch": 2.5,
325
+ "learning_rate": 0.0002754678408554804,
326
+ "loss": 0.8201,
327
+ "step": 106000
328
+ },
329
+ {
330
+ "epoch": 2.54,
331
+ "learning_rate": 0.00027499606856423963,
332
+ "loss": 0.8193,
333
+ "step": 108000
334
+ },
335
+ {
336
+ "epoch": 2.59,
337
+ "learning_rate": 0.00027452429627299887,
338
+ "loss": 0.8225,
339
+ "step": 110000
340
+ },
341
+ {
342
+ "epoch": 2.64,
343
+ "learning_rate": 0.0002740525239817581,
344
+ "loss": 0.8105,
345
+ "step": 112000
346
+ },
347
+ {
348
+ "epoch": 2.68,
349
+ "learning_rate": 0.00027358075169051733,
350
+ "loss": 0.82,
351
+ "step": 114000
352
+ },
353
+ {
354
+ "epoch": 2.73,
355
+ "learning_rate": 0.0002731089793992766,
356
+ "loss": 0.8191,
357
+ "step": 116000
358
+ },
359
+ {
360
+ "epoch": 2.78,
361
+ "learning_rate": 0.00027263720710803585,
362
+ "loss": 0.8102,
363
+ "step": 118000
364
+ },
365
+ {
366
+ "epoch": 2.83,
367
+ "learning_rate": 0.0002721654348167951,
368
+ "loss": 0.8189,
369
+ "step": 120000
370
+ },
371
+ {
372
+ "epoch": 2.87,
373
+ "learning_rate": 0.0002716936625255543,
374
+ "loss": 0.8128,
375
+ "step": 122000
376
+ },
377
+ {
378
+ "epoch": 2.92,
379
+ "learning_rate": 0.00027122189023431355,
380
+ "loss": 0.815,
381
+ "step": 124000
382
+ },
383
+ {
384
+ "epoch": 2.97,
385
+ "learning_rate": 0.0002707501179430728,
386
+ "loss": 0.8123,
387
+ "step": 126000
388
+ },
389
+ {
390
+ "epoch": 3.01,
391
+ "learning_rate": 0.000270278345651832,
392
+ "loss": 0.7986,
393
+ "step": 128000
394
+ },
395
+ {
396
+ "epoch": 3.06,
397
+ "learning_rate": 0.00026980657336059125,
398
+ "loss": 0.7701,
399
+ "step": 130000
400
+ },
401
+ {
402
+ "epoch": 3.11,
403
+ "learning_rate": 0.0002693348010693505,
404
+ "loss": 0.7819,
405
+ "step": 132000
406
+ },
407
+ {
408
+ "epoch": 3.16,
409
+ "learning_rate": 0.00026886302877810977,
410
+ "loss": 0.7826,
411
+ "step": 134000
412
+ },
413
+ {
414
+ "epoch": 3.2,
415
+ "learning_rate": 0.000268391256486869,
416
+ "loss": 0.7868,
417
+ "step": 136000
418
+ },
419
+ {
420
+ "epoch": 3.25,
421
+ "learning_rate": 0.00026791948419562823,
422
+ "loss": 0.7843,
423
+ "step": 138000
424
+ },
425
+ {
426
+ "epoch": 3.3,
427
+ "learning_rate": 0.00026744771190438747,
428
+ "loss": 0.7841,
429
+ "step": 140000
430
+ },
431
+ {
432
+ "epoch": 3.34,
433
+ "learning_rate": 0.0002669759396131467,
434
+ "loss": 0.7845,
435
+ "step": 142000
436
+ },
437
+ {
438
+ "epoch": 3.39,
439
+ "learning_rate": 0.00026650416732190593,
440
+ "loss": 0.7859,
441
+ "step": 144000
442
+ },
443
+ {
444
+ "epoch": 3.44,
445
+ "learning_rate": 0.00026603239503066516,
446
+ "loss": 0.7746,
447
+ "step": 146000
448
+ },
449
+ {
450
+ "epoch": 3.49,
451
+ "learning_rate": 0.0002655606227394244,
452
+ "loss": 0.7749,
453
+ "step": 148000
454
+ },
455
+ {
456
+ "epoch": 3.53,
457
+ "learning_rate": 0.00026508885044818363,
458
+ "loss": 0.7747,
459
+ "step": 150000
460
+ },
461
+ {
462
+ "epoch": 3.58,
463
+ "learning_rate": 0.0002646170781569429,
464
+ "loss": 0.7816,
465
+ "step": 152000
466
+ },
467
+ {
468
+ "epoch": 3.63,
469
+ "learning_rate": 0.00026414530586570215,
470
+ "loss": 0.7807,
471
+ "step": 154000
472
+ },
473
+ {
474
+ "epoch": 3.67,
475
+ "learning_rate": 0.0002636735335744614,
476
+ "loss": 0.7817,
477
+ "step": 156000
478
+ },
479
+ {
480
+ "epoch": 3.72,
481
+ "learning_rate": 0.0002632017612832206,
482
+ "loss": 0.787,
483
+ "step": 158000
484
+ },
485
+ {
486
+ "epoch": 3.77,
487
+ "learning_rate": 0.00026272998899197985,
488
+ "loss": 0.7796,
489
+ "step": 160000
490
+ },
491
+ {
492
+ "epoch": 3.82,
493
+ "learning_rate": 0.0002622582167007391,
494
+ "loss": 0.7662,
495
+ "step": 162000
496
+ },
497
+ {
498
+ "epoch": 3.86,
499
+ "learning_rate": 0.0002617864444094983,
500
+ "loss": 0.7898,
501
+ "step": 164000
502
+ },
503
+ {
504
+ "epoch": 3.91,
505
+ "learning_rate": 0.00026131467211825755,
506
+ "loss": 0.7705,
507
+ "step": 166000
508
+ },
509
+ {
510
+ "epoch": 3.96,
511
+ "learning_rate": 0.0002608428998270168,
512
+ "loss": 0.7876,
513
+ "step": 168000
514
+ },
515
+ {
516
+ "epoch": 4.0,
517
+ "learning_rate": 0.00026037112753577607,
518
+ "loss": 0.7747,
519
+ "step": 170000
520
+ },
521
+ {
522
+ "epoch": 4.05,
523
+ "learning_rate": 0.0002598993552445353,
524
+ "loss": 0.7517,
525
+ "step": 172000
526
+ },
527
+ {
528
+ "epoch": 4.1,
529
+ "learning_rate": 0.00025942758295329453,
530
+ "loss": 0.7465,
531
+ "step": 174000
532
+ },
533
+ {
534
+ "epoch": 4.15,
535
+ "learning_rate": 0.00025895581066205376,
536
+ "loss": 0.7485,
537
+ "step": 176000
538
+ },
539
+ {
540
+ "epoch": 4.19,
541
+ "learning_rate": 0.000258484038370813,
542
+ "loss": 0.7531,
543
+ "step": 178000
544
+ },
545
+ {
546
+ "epoch": 4.24,
547
+ "learning_rate": 0.00025801226607957223,
548
+ "loss": 0.7521,
549
+ "step": 180000
550
+ },
551
+ {
552
+ "epoch": 4.29,
553
+ "learning_rate": 0.00025754049378833146,
554
+ "loss": 0.7477,
555
+ "step": 182000
556
+ },
557
+ {
558
+ "epoch": 4.33,
559
+ "learning_rate": 0.0002570687214970907,
560
+ "loss": 0.7621,
561
+ "step": 184000
562
+ },
563
+ {
564
+ "epoch": 4.38,
565
+ "learning_rate": 0.00025659694920585,
566
+ "loss": 0.7552,
567
+ "step": 186000
568
+ },
569
+ {
570
+ "epoch": 4.43,
571
+ "learning_rate": 0.0002561251769146092,
572
+ "loss": 0.7592,
573
+ "step": 188000
574
+ },
575
+ {
576
+ "epoch": 4.47,
577
+ "learning_rate": 0.00025565340462336845,
578
+ "loss": 0.7508,
579
+ "step": 190000
580
+ },
581
+ {
582
+ "epoch": 4.52,
583
+ "learning_rate": 0.0002551816323321277,
584
+ "loss": 0.7547,
585
+ "step": 192000
586
+ },
587
+ {
588
+ "epoch": 4.57,
589
+ "learning_rate": 0.0002547098600408869,
590
+ "loss": 0.7439,
591
+ "step": 194000
592
+ },
593
+ {
594
+ "epoch": 4.62,
595
+ "learning_rate": 0.00025423808774964615,
596
+ "loss": 0.762,
597
+ "step": 196000
598
+ },
599
+ {
600
+ "epoch": 4.66,
601
+ "learning_rate": 0.0002537663154584054,
602
+ "loss": 0.7507,
603
+ "step": 198000
604
+ },
605
+ {
606
+ "epoch": 4.71,
607
+ "learning_rate": 0.0002532945431671646,
608
+ "loss": 0.7553,
609
+ "step": 200000
610
+ },
611
+ {
612
+ "epoch": 4.76,
613
+ "learning_rate": 0.00025282277087592384,
614
+ "loss": 0.7498,
615
+ "step": 202000
616
+ },
617
+ {
618
+ "epoch": 4.8,
619
+ "learning_rate": 0.00025235099858468313,
620
+ "loss": 0.7466,
621
+ "step": 204000
622
+ },
623
+ {
624
+ "epoch": 4.85,
625
+ "learning_rate": 0.0002518792262934423,
626
+ "loss": 0.7496,
627
+ "step": 206000
628
+ },
629
+ {
630
+ "epoch": 4.9,
631
+ "learning_rate": 0.0002514074540022016,
632
+ "loss": 0.7406,
633
+ "step": 208000
634
+ },
635
+ {
636
+ "epoch": 4.95,
637
+ "learning_rate": 0.00025093568171096083,
638
+ "loss": 0.7447,
639
+ "step": 210000
640
+ },
641
+ {
642
+ "epoch": 4.99,
643
+ "learning_rate": 0.00025046390941972006,
644
+ "loss": 0.7532,
645
+ "step": 212000
646
+ },
647
+ {
648
+ "epoch": 5.04,
649
+ "learning_rate": 0.0002499921371284793,
650
+ "loss": 0.7339,
651
+ "step": 214000
652
+ },
653
+ {
654
+ "epoch": 5.09,
655
+ "learning_rate": 0.00024952036483723853,
656
+ "loss": 0.7214,
657
+ "step": 216000
658
+ },
659
+ {
660
+ "epoch": 5.13,
661
+ "learning_rate": 0.00024904859254599776,
662
+ "loss": 0.7227,
663
+ "step": 218000
664
+ },
665
+ {
666
+ "epoch": 5.18,
667
+ "learning_rate": 0.000248576820254757,
668
+ "loss": 0.7344,
669
+ "step": 220000
670
+ },
671
+ {
672
+ "epoch": 5.23,
673
+ "learning_rate": 0.0002481050479635163,
674
+ "loss": 0.7272,
675
+ "step": 222000
676
+ },
677
+ {
678
+ "epoch": 5.28,
679
+ "learning_rate": 0.00024763327567227546,
680
+ "loss": 0.7271,
681
+ "step": 224000
682
+ },
683
+ {
684
+ "epoch": 5.32,
685
+ "learning_rate": 0.00024716150338103475,
686
+ "loss": 0.7302,
687
+ "step": 226000
688
+ },
689
+ {
690
+ "epoch": 5.37,
691
+ "learning_rate": 0.000246689731089794,
692
+ "loss": 0.7218,
693
+ "step": 228000
694
+ },
695
+ {
696
+ "epoch": 5.42,
697
+ "learning_rate": 0.0002462179587985532,
698
+ "loss": 0.7242,
699
+ "step": 230000
700
+ },
701
+ {
702
+ "epoch": 5.46,
703
+ "learning_rate": 0.00024574618650731244,
704
+ "loss": 0.7306,
705
+ "step": 232000
706
+ },
707
+ {
708
+ "epoch": 5.51,
709
+ "learning_rate": 0.0002452744142160717,
710
+ "loss": 0.7351,
711
+ "step": 234000
712
+ },
713
+ {
714
+ "epoch": 5.56,
715
+ "learning_rate": 0.0002448026419248309,
716
+ "loss": 0.7225,
717
+ "step": 236000
718
+ },
719
+ {
720
+ "epoch": 5.61,
721
+ "learning_rate": 0.00024433086963359014,
722
+ "loss": 0.7256,
723
+ "step": 238000
724
+ },
725
+ {
726
+ "epoch": 5.65,
727
+ "learning_rate": 0.0002438590973423494,
728
+ "loss": 0.7256,
729
+ "step": 240000
730
+ },
731
+ {
732
+ "epoch": 5.7,
733
+ "learning_rate": 0.00024338732505110863,
734
+ "loss": 0.7293,
735
+ "step": 242000
736
+ },
737
+ {
738
+ "epoch": 5.75,
739
+ "learning_rate": 0.0002429155527598679,
740
+ "loss": 0.7272,
741
+ "step": 244000
742
+ },
743
+ {
744
+ "epoch": 5.79,
745
+ "learning_rate": 0.00024244378046862713,
746
+ "loss": 0.7154,
747
+ "step": 246000
748
+ },
749
+ {
750
+ "epoch": 5.84,
751
+ "learning_rate": 0.00024197200817738636,
752
+ "loss": 0.7274,
753
+ "step": 248000
754
+ },
755
+ {
756
+ "epoch": 5.89,
757
+ "learning_rate": 0.00024150023588614562,
758
+ "loss": 0.7184,
759
+ "step": 250000
760
+ },
761
+ {
762
+ "epoch": 5.93,
763
+ "learning_rate": 0.00024102846359490483,
764
+ "loss": 0.7218,
765
+ "step": 252000
766
+ },
767
+ {
768
+ "epoch": 5.98,
769
+ "learning_rate": 0.00024055669130366409,
770
+ "loss": 0.7246,
771
+ "step": 254000
772
+ },
773
+ {
774
+ "epoch": 6.03,
775
+ "learning_rate": 0.00024008491901242332,
776
+ "loss": 0.7036,
777
+ "step": 256000
778
+ },
779
+ {
780
+ "epoch": 6.08,
781
+ "learning_rate": 0.00023961314672118255,
782
+ "loss": 0.6992,
783
+ "step": 258000
784
+ },
785
+ {
786
+ "epoch": 6.12,
787
+ "learning_rate": 0.00023914137442994178,
788
+ "loss": 0.7029,
789
+ "step": 260000
790
+ },
791
+ {
792
+ "epoch": 6.17,
793
+ "learning_rate": 0.00023866960213870104,
794
+ "loss": 0.7097,
795
+ "step": 262000
796
+ },
797
+ {
798
+ "epoch": 6.22,
799
+ "learning_rate": 0.00023819782984746028,
800
+ "loss": 0.7042,
801
+ "step": 264000
802
+ },
803
+ {
804
+ "epoch": 6.26,
805
+ "learning_rate": 0.0002377260575562195,
806
+ "loss": 0.7076,
807
+ "step": 266000
808
+ },
809
+ {
810
+ "epoch": 6.31,
811
+ "learning_rate": 0.00023725428526497877,
812
+ "loss": 0.6972,
813
+ "step": 268000
814
+ },
815
+ {
816
+ "epoch": 6.36,
817
+ "learning_rate": 0.00023678251297373797,
818
+ "loss": 0.7097,
819
+ "step": 270000
820
+ },
821
+ {
822
+ "epoch": 6.41,
823
+ "learning_rate": 0.00023631074068249723,
824
+ "loss": 0.7025,
825
+ "step": 272000
826
+ },
827
+ {
828
+ "epoch": 6.45,
829
+ "learning_rate": 0.00023583896839125647,
830
+ "loss": 0.7041,
831
+ "step": 274000
832
+ },
833
+ {
834
+ "epoch": 6.5,
835
+ "learning_rate": 0.0002353671961000157,
836
+ "loss": 0.7098,
837
+ "step": 276000
838
+ },
839
+ {
840
+ "epoch": 6.55,
841
+ "learning_rate": 0.00023489542380877493,
842
+ "loss": 0.704,
843
+ "step": 278000
844
+ },
845
+ {
846
+ "epoch": 6.59,
847
+ "learning_rate": 0.0002344236515175342,
848
+ "loss": 0.7046,
849
+ "step": 280000
850
+ },
851
+ {
852
+ "epoch": 6.64,
853
+ "learning_rate": 0.00023395187922629343,
854
+ "loss": 0.701,
855
+ "step": 282000
856
+ },
857
+ {
858
+ "epoch": 6.69,
859
+ "learning_rate": 0.00023348010693505266,
860
+ "loss": 0.7107,
861
+ "step": 284000
862
+ },
863
+ {
864
+ "epoch": 6.74,
865
+ "learning_rate": 0.00023300833464381192,
866
+ "loss": 0.7107,
867
+ "step": 286000
868
+ },
869
+ {
870
+ "epoch": 6.78,
871
+ "learning_rate": 0.00023253656235257112,
872
+ "loss": 0.7051,
873
+ "step": 288000
874
+ },
875
+ {
876
+ "epoch": 6.83,
877
+ "learning_rate": 0.00023206479006133038,
878
+ "loss": 0.7055,
879
+ "step": 290000
880
+ },
881
+ {
882
+ "epoch": 6.88,
883
+ "learning_rate": 0.00023159301777008962,
884
+ "loss": 0.7004,
885
+ "step": 292000
886
+ },
887
+ {
888
+ "epoch": 6.92,
889
+ "learning_rate": 0.00023112124547884885,
890
+ "loss": 0.7051,
891
+ "step": 294000
892
+ },
893
+ {
894
+ "epoch": 6.97,
895
+ "learning_rate": 0.00023064947318760808,
896
+ "loss": 0.7025,
897
+ "step": 296000
898
+ },
899
+ {
900
+ "epoch": 7.02,
901
+ "learning_rate": 0.00023017770089636734,
902
+ "loss": 0.6976,
903
+ "step": 298000
904
+ },
905
+ {
906
+ "epoch": 7.07,
907
+ "learning_rate": 0.0002297059286051266,
908
+ "loss": 0.6785,
909
+ "step": 300000
910
+ },
911
+ {
912
+ "epoch": 7.11,
913
+ "learning_rate": 0.0002292341563138858,
914
+ "loss": 0.6773,
915
+ "step": 302000
916
+ },
917
+ {
918
+ "epoch": 7.16,
919
+ "learning_rate": 0.00022876238402264507,
920
+ "loss": 0.6763,
921
+ "step": 304000
922
+ },
923
+ {
924
+ "epoch": 7.21,
925
+ "learning_rate": 0.00022829061173140427,
926
+ "loss": 0.6774,
927
+ "step": 306000
928
+ },
929
+ {
930
+ "epoch": 7.25,
931
+ "learning_rate": 0.00022781883944016353,
932
+ "loss": 0.6786,
933
+ "step": 308000
934
+ },
935
+ {
936
+ "epoch": 7.3,
937
+ "learning_rate": 0.00022734706714892276,
938
+ "loss": 0.6885,
939
+ "step": 310000
940
+ },
941
+ {
942
+ "epoch": 7.35,
943
+ "learning_rate": 0.000226875294857682,
944
+ "loss": 0.6852,
945
+ "step": 312000
946
+ },
947
+ {
948
+ "epoch": 7.4,
949
+ "learning_rate": 0.00022640352256644123,
950
+ "loss": 0.6842,
951
+ "step": 314000
952
+ },
953
+ {
954
+ "epoch": 7.44,
955
+ "learning_rate": 0.0002259317502752005,
956
+ "loss": 0.6868,
957
+ "step": 316000
958
+ },
959
+ {
960
+ "epoch": 7.49,
961
+ "learning_rate": 0.00022545997798395975,
962
+ "loss": 0.6918,
963
+ "step": 318000
964
+ },
965
+ {
966
+ "epoch": 7.54,
967
+ "learning_rate": 0.00022498820569271896,
968
+ "loss": 0.6846,
969
+ "step": 320000
970
+ },
971
+ {
972
+ "epoch": 7.58,
973
+ "learning_rate": 0.00022451643340147822,
974
+ "loss": 0.6881,
975
+ "step": 322000
976
+ },
977
+ {
978
+ "epoch": 7.63,
979
+ "learning_rate": 0.00022404466111023742,
980
+ "loss": 0.69,
981
+ "step": 324000
982
+ },
983
+ {
984
+ "epoch": 7.68,
985
+ "learning_rate": 0.00022357288881899668,
986
+ "loss": 0.6823,
987
+ "step": 326000
988
+ },
989
+ {
990
+ "epoch": 7.72,
991
+ "learning_rate": 0.00022310111652775591,
992
+ "loss": 0.6896,
993
+ "step": 328000
994
+ },
995
+ {
996
+ "epoch": 7.77,
997
+ "learning_rate": 0.00022262934423651517,
998
+ "loss": 0.6862,
999
+ "step": 330000
1000
+ },
1001
+ {
1002
+ "epoch": 7.82,
1003
+ "learning_rate": 0.00022215757194527438,
1004
+ "loss": 0.6858,
1005
+ "step": 332000
1006
+ },
1007
+ {
1008
+ "epoch": 7.87,
1009
+ "learning_rate": 0.00022168579965403364,
1010
+ "loss": 0.691,
1011
+ "step": 334000
1012
+ },
1013
+ {
1014
+ "epoch": 7.91,
1015
+ "learning_rate": 0.00022121402736279284,
1016
+ "loss": 0.6895,
1017
+ "step": 336000
1018
+ },
1019
+ {
1020
+ "epoch": 7.96,
1021
+ "learning_rate": 0.0002207422550715521,
1022
+ "loss": 0.6924,
1023
+ "step": 338000
1024
+ },
1025
+ {
1026
+ "epoch": 8.01,
1027
+ "learning_rate": 0.00022027048278031136,
1028
+ "loss": 0.6903,
1029
+ "step": 340000
1030
+ },
1031
+ {
1032
+ "epoch": 8.05,
1033
+ "learning_rate": 0.0002197987104890706,
1034
+ "loss": 0.6637,
1035
+ "step": 342000
1036
+ },
1037
+ {
1038
+ "epoch": 8.1,
1039
+ "learning_rate": 0.00021932693819782983,
1040
+ "loss": 0.6612,
1041
+ "step": 344000
1042
+ },
1043
+ {
1044
+ "epoch": 8.15,
1045
+ "learning_rate": 0.00021885516590658906,
1046
+ "loss": 0.6665,
1047
+ "step": 346000
1048
+ },
1049
+ {
1050
+ "epoch": 8.2,
1051
+ "learning_rate": 0.00021838339361534832,
1052
+ "loss": 0.6702,
1053
+ "step": 348000
1054
+ },
1055
+ {
1056
+ "epoch": 8.24,
1057
+ "learning_rate": 0.00021791162132410753,
1058
+ "loss": 0.6667,
1059
+ "step": 350000
1060
+ },
1061
+ {
1062
+ "epoch": 8.29,
1063
+ "learning_rate": 0.0002174398490328668,
1064
+ "loss": 0.6674,
1065
+ "step": 352000
1066
+ },
1067
+ {
1068
+ "epoch": 8.34,
1069
+ "learning_rate": 0.000216968076741626,
1070
+ "loss": 0.6719,
1071
+ "step": 354000
1072
+ },
1073
+ {
1074
+ "epoch": 8.38,
1075
+ "learning_rate": 0.00021649630445038525,
1076
+ "loss": 0.6671,
1077
+ "step": 356000
1078
+ },
1079
+ {
1080
+ "epoch": 8.43,
1081
+ "learning_rate": 0.0002160245321591445,
1082
+ "loss": 0.6647,
1083
+ "step": 358000
1084
+ },
1085
+ {
1086
+ "epoch": 8.48,
1087
+ "learning_rate": 0.00021555275986790375,
1088
+ "loss": 0.6671,
1089
+ "step": 360000
1090
+ },
1091
+ {
1092
+ "epoch": 8.53,
1093
+ "learning_rate": 0.00021508098757666298,
1094
+ "loss": 0.6681,
1095
+ "step": 362000
1096
+ },
1097
+ {
1098
+ "epoch": 8.57,
1099
+ "learning_rate": 0.0002146092152854222,
1100
+ "loss": 0.6727,
1101
+ "step": 364000
1102
+ },
1103
+ {
1104
+ "epoch": 8.62,
1105
+ "learning_rate": 0.00021413744299418147,
1106
+ "loss": 0.6767,
1107
+ "step": 366000
1108
+ },
1109
+ {
1110
+ "epoch": 8.67,
1111
+ "learning_rate": 0.00021366567070294068,
1112
+ "loss": 0.6749,
1113
+ "step": 368000
1114
+ },
1115
+ {
1116
+ "epoch": 8.71,
1117
+ "learning_rate": 0.00021319389841169994,
1118
+ "loss": 0.6704,
1119
+ "step": 370000
1120
+ },
1121
+ {
1122
+ "epoch": 8.76,
1123
+ "learning_rate": 0.00021272212612045917,
1124
+ "loss": 0.6729,
1125
+ "step": 372000
1126
+ },
1127
+ {
1128
+ "epoch": 8.81,
1129
+ "learning_rate": 0.0002122503538292184,
1130
+ "loss": 0.6641,
1131
+ "step": 374000
1132
+ },
1133
+ {
1134
+ "epoch": 8.86,
1135
+ "learning_rate": 0.00021177858153797766,
1136
+ "loss": 0.6678,
1137
+ "step": 376000
1138
+ },
1139
+ {
1140
+ "epoch": 8.9,
1141
+ "learning_rate": 0.0002113068092467369,
1142
+ "loss": 0.6677,
1143
+ "step": 378000
1144
+ },
1145
+ {
1146
+ "epoch": 8.95,
1147
+ "learning_rate": 0.00021083503695549613,
1148
+ "loss": 0.6683,
1149
+ "step": 380000
1150
+ },
1151
+ {
1152
+ "epoch": 9.0,
1153
+ "learning_rate": 0.00021036326466425536,
1154
+ "loss": 0.6746,
1155
+ "step": 382000
1156
+ },
1157
+ {
1158
+ "epoch": 9.04,
1159
+ "learning_rate": 0.00020989149237301462,
1160
+ "loss": 0.6469,
1161
+ "step": 384000
1162
+ },
1163
+ {
1164
+ "epoch": 9.09,
1165
+ "learning_rate": 0.00020941972008177383,
1166
+ "loss": 0.6496,
1167
+ "step": 386000
1168
+ },
1169
+ {
1170
+ "epoch": 9.14,
1171
+ "learning_rate": 0.00020894794779053309,
1172
+ "loss": 0.6513,
1173
+ "step": 388000
1174
+ },
1175
+ {
1176
+ "epoch": 9.19,
1177
+ "learning_rate": 0.00020847617549929232,
1178
+ "loss": 0.6516,
1179
+ "step": 390000
1180
+ },
1181
+ {
1182
+ "epoch": 9.23,
1183
+ "learning_rate": 0.00020800440320805155,
1184
+ "loss": 0.6556,
1185
+ "step": 392000
1186
+ },
1187
+ {
1188
+ "epoch": 9.28,
1189
+ "learning_rate": 0.0002075326309168108,
1190
+ "loss": 0.6477,
1191
+ "step": 394000
1192
+ },
1193
+ {
1194
+ "epoch": 9.33,
1195
+ "learning_rate": 0.00020706085862557004,
1196
+ "loss": 0.6571,
1197
+ "step": 396000
1198
+ },
1199
+ {
1200
+ "epoch": 9.37,
1201
+ "learning_rate": 0.00020658908633432928,
1202
+ "loss": 0.6479,
1203
+ "step": 398000
1204
+ },
1205
+ {
1206
+ "epoch": 9.42,
1207
+ "learning_rate": 0.0002061173140430885,
1208
+ "loss": 0.65,
1209
+ "step": 400000
1210
+ },
1211
+ {
1212
+ "epoch": 9.47,
1213
+ "learning_rate": 0.00020564554175184777,
1214
+ "loss": 0.6557,
1215
+ "step": 402000
1216
+ },
1217
+ {
1218
+ "epoch": 9.51,
1219
+ "learning_rate": 0.00020517376946060697,
1220
+ "loss": 0.6644,
1221
+ "step": 404000
1222
+ },
1223
+ {
1224
+ "epoch": 9.56,
1225
+ "learning_rate": 0.00020470199716936623,
1226
+ "loss": 0.6558,
1227
+ "step": 406000
1228
+ },
1229
+ {
1230
+ "epoch": 9.61,
1231
+ "learning_rate": 0.00020423022487812547,
1232
+ "loss": 0.6503,
1233
+ "step": 408000
1234
+ },
1235
+ {
1236
+ "epoch": 9.66,
1237
+ "learning_rate": 0.0002037584525868847,
1238
+ "loss": 0.6569,
1239
+ "step": 410000
1240
+ },
1241
+ {
1242
+ "epoch": 9.7,
1243
+ "learning_rate": 0.00020328668029564396,
1244
+ "loss": 0.66,
1245
+ "step": 412000
1246
+ },
1247
+ {
1248
+ "epoch": 9.75,
1249
+ "learning_rate": 0.0002028149080044032,
1250
+ "loss": 0.6528,
1251
+ "step": 414000
1252
+ },
1253
+ {
1254
+ "epoch": 9.8,
1255
+ "learning_rate": 0.00020234313571316245,
1256
+ "loss": 0.6493,
1257
+ "step": 416000
1258
+ },
1259
+ {
1260
+ "epoch": 9.84,
1261
+ "learning_rate": 0.00020187136342192166,
1262
+ "loss": 0.6599,
1263
+ "step": 418000
1264
+ },
1265
+ {
1266
+ "epoch": 9.89,
1267
+ "learning_rate": 0.00020139959113068092,
1268
+ "loss": 0.6534,
1269
+ "step": 420000
1270
+ },
1271
+ {
1272
+ "epoch": 9.94,
1273
+ "learning_rate": 0.00020092781883944012,
1274
+ "loss": 0.6546,
1275
+ "step": 422000
1276
+ },
1277
+ {
1278
+ "epoch": 9.99,
1279
+ "learning_rate": 0.00020045604654819938,
1280
+ "loss": 0.6488,
1281
+ "step": 424000
1282
+ },
1283
+ {
1284
+ "epoch": 10.03,
1285
+ "learning_rate": 0.00019998427425695862,
1286
+ "loss": 0.643,
1287
+ "step": 426000
1288
+ },
1289
+ {
1290
+ "epoch": 10.08,
1291
+ "learning_rate": 0.00019951250196571788,
1292
+ "loss": 0.6465,
1293
+ "step": 428000
1294
+ },
1295
+ {
1296
+ "epoch": 10.13,
1297
+ "learning_rate": 0.0001990407296744771,
1298
+ "loss": 0.6324,
1299
+ "step": 430000
1300
+ },
1301
+ {
1302
+ "epoch": 10.17,
1303
+ "learning_rate": 0.00019856895738323634,
1304
+ "loss": 0.6446,
1305
+ "step": 432000
1306
+ },
1307
+ {
1308
+ "epoch": 10.22,
1309
+ "learning_rate": 0.0001980971850919956,
1310
+ "loss": 0.6283,
1311
+ "step": 434000
1312
+ },
1313
+ {
1314
+ "epoch": 10.27,
1315
+ "learning_rate": 0.0001976254128007548,
1316
+ "loss": 0.635,
1317
+ "step": 436000
1318
+ },
1319
+ {
1320
+ "epoch": 10.32,
1321
+ "learning_rate": 0.00019715364050951407,
1322
+ "loss": 0.6333,
1323
+ "step": 438000
1324
+ },
1325
+ {
1326
+ "epoch": 10.36,
1327
+ "learning_rate": 0.0001966818682182733,
1328
+ "loss": 0.6378,
1329
+ "step": 440000
1330
+ },
1331
+ {
1332
+ "epoch": 10.41,
1333
+ "learning_rate": 0.00019621009592703253,
1334
+ "loss": 0.6369,
1335
+ "step": 442000
1336
+ },
1337
+ {
1338
+ "epoch": 10.46,
1339
+ "learning_rate": 0.00019573832363579177,
1340
+ "loss": 0.6348,
1341
+ "step": 444000
1342
+ },
1343
+ {
1344
+ "epoch": 10.5,
1345
+ "learning_rate": 0.00019526655134455103,
1346
+ "loss": 0.6439,
1347
+ "step": 446000
1348
+ },
1349
+ {
1350
+ "epoch": 10.55,
1351
+ "learning_rate": 0.00019479477905331026,
1352
+ "loss": 0.6342,
1353
+ "step": 448000
1354
+ },
1355
+ {
1356
+ "epoch": 10.6,
1357
+ "learning_rate": 0.0001943230067620695,
1358
+ "loss": 0.6424,
1359
+ "step": 450000
1360
+ },
1361
+ {
1362
+ "epoch": 10.65,
1363
+ "learning_rate": 0.00019385123447082875,
1364
+ "loss": 0.6408,
1365
+ "step": 452000
1366
+ },
1367
+ {
1368
+ "epoch": 10.69,
1369
+ "learning_rate": 0.00019337946217958796,
1370
+ "loss": 0.6413,
1371
+ "step": 454000
1372
+ },
1373
+ {
1374
+ "epoch": 10.74,
1375
+ "learning_rate": 0.00019290768988834722,
1376
+ "loss": 0.643,
1377
+ "step": 456000
1378
+ },
1379
+ {
1380
+ "epoch": 10.79,
1381
+ "learning_rate": 0.00019243591759710645,
1382
+ "loss": 0.6421,
1383
+ "step": 458000
1384
+ },
1385
+ {
1386
+ "epoch": 10.83,
1387
+ "learning_rate": 0.00019196414530586568,
1388
+ "loss": 0.6393,
1389
+ "step": 460000
1390
+ },
1391
+ {
1392
+ "epoch": 10.88,
1393
+ "learning_rate": 0.00019149237301462491,
1394
+ "loss": 0.6385,
1395
+ "step": 462000
1396
+ },
1397
+ {
1398
+ "epoch": 10.93,
1399
+ "learning_rate": 0.00019102060072338417,
1400
+ "loss": 0.6387,
1401
+ "step": 464000
1402
+ },
1403
+ {
1404
+ "epoch": 10.98,
1405
+ "learning_rate": 0.00019054882843214338,
1406
+ "loss": 0.6442,
1407
+ "step": 466000
1408
+ },
1409
+ {
1410
+ "epoch": 11.02,
1411
+ "learning_rate": 0.00019007705614090264,
1412
+ "loss": 0.6294,
1413
+ "step": 468000
1414
+ },
1415
+ {
1416
+ "epoch": 11.07,
1417
+ "learning_rate": 0.0001896052838496619,
1418
+ "loss": 0.6191,
1419
+ "step": 470000
1420
+ },
1421
+ {
1422
+ "epoch": 11.12,
1423
+ "learning_rate": 0.0001891335115584211,
1424
+ "loss": 0.6226,
1425
+ "step": 472000
1426
+ },
1427
+ {
1428
+ "epoch": 11.16,
1429
+ "learning_rate": 0.00018866173926718036,
1430
+ "loss": 0.6207,
1431
+ "step": 474000
1432
+ },
1433
+ {
1434
+ "epoch": 11.21,
1435
+ "learning_rate": 0.0001881899669759396,
1436
+ "loss": 0.6282,
1437
+ "step": 476000
1438
+ },
1439
+ {
1440
+ "epoch": 11.26,
1441
+ "learning_rate": 0.00018771819468469883,
1442
+ "loss": 0.615,
1443
+ "step": 478000
1444
+ },
1445
+ {
1446
+ "epoch": 11.3,
1447
+ "learning_rate": 0.00018724642239345806,
1448
+ "loss": 0.6201,
1449
+ "step": 480000
1450
+ },
1451
+ {
1452
+ "epoch": 11.35,
1453
+ "learning_rate": 0.00018677465010221732,
1454
+ "loss": 0.6199,
1455
+ "step": 482000
1456
+ },
1457
+ {
1458
+ "epoch": 11.4,
1459
+ "learning_rate": 0.00018630287781097653,
1460
+ "loss": 0.6211,
1461
+ "step": 484000
1462
+ },
1463
+ {
1464
+ "epoch": 11.45,
1465
+ "learning_rate": 0.0001858311055197358,
1466
+ "loss": 0.6372,
1467
+ "step": 486000
1468
+ },
1469
+ {
1470
+ "epoch": 11.49,
1471
+ "learning_rate": 0.00018535933322849505,
1472
+ "loss": 0.6268,
1473
+ "step": 488000
1474
+ },
1475
+ {
1476
+ "epoch": 11.54,
1477
+ "learning_rate": 0.00018488756093725425,
1478
+ "loss": 0.6283,
1479
+ "step": 490000
1480
+ },
1481
+ {
1482
+ "epoch": 11.59,
1483
+ "learning_rate": 0.00018441578864601351,
1484
+ "loss": 0.6398,
1485
+ "step": 492000
1486
+ },
1487
+ {
1488
+ "epoch": 11.63,
1489
+ "learning_rate": 0.00018394401635477275,
1490
+ "loss": 0.6282,
1491
+ "step": 494000
1492
+ },
1493
+ {
1494
+ "epoch": 11.68,
1495
+ "learning_rate": 0.00018347224406353198,
1496
+ "loss": 0.6327,
1497
+ "step": 496000
1498
+ },
1499
+ {
1500
+ "epoch": 11.73,
1501
+ "learning_rate": 0.0001830004717722912,
1502
+ "loss": 0.6226,
1503
+ "step": 498000
1504
+ },
1505
+ {
1506
+ "epoch": 11.78,
1507
+ "learning_rate": 0.00018252869948105047,
1508
+ "loss": 0.6321,
1509
+ "step": 500000
1510
+ },
1511
+ {
1512
+ "epoch": 11.82,
1513
+ "learning_rate": 0.00018205692718980968,
1514
+ "loss": 0.6244,
1515
+ "step": 502000
1516
+ },
1517
+ {
1518
+ "epoch": 11.87,
1519
+ "learning_rate": 0.00018158515489856894,
1520
+ "loss": 0.6257,
1521
+ "step": 504000
1522
+ },
1523
+ {
1524
+ "epoch": 11.92,
1525
+ "learning_rate": 0.0001811133826073282,
1526
+ "loss": 0.6291,
1527
+ "step": 506000
1528
+ },
1529
+ {
1530
+ "epoch": 11.96,
1531
+ "learning_rate": 0.0001806416103160874,
1532
+ "loss": 0.6283,
1533
+ "step": 508000
1534
+ },
1535
+ {
1536
+ "epoch": 12.01,
1537
+ "learning_rate": 0.00018016983802484666,
1538
+ "loss": 0.6246,
1539
+ "step": 510000
1540
+ },
1541
+ {
1542
+ "epoch": 12.06,
1543
+ "learning_rate": 0.0001796980657336059,
1544
+ "loss": 0.6133,
1545
+ "step": 512000
1546
+ },
1547
+ {
1548
+ "epoch": 12.11,
1549
+ "learning_rate": 0.00017922629344236516,
1550
+ "loss": 0.6069,
1551
+ "step": 514000
1552
+ },
1553
+ {
1554
+ "epoch": 12.15,
1555
+ "learning_rate": 0.00017875452115112436,
1556
+ "loss": 0.6163,
1557
+ "step": 516000
1558
+ },
1559
+ {
1560
+ "epoch": 12.2,
1561
+ "learning_rate": 0.00017828274885988362,
1562
+ "loss": 0.6131,
1563
+ "step": 518000
1564
+ },
1565
+ {
1566
+ "epoch": 12.25,
1567
+ "learning_rate": 0.00017781097656864283,
1568
+ "loss": 0.6096,
1569
+ "step": 520000
1570
+ },
1571
+ {
1572
+ "epoch": 12.29,
1573
+ "learning_rate": 0.00017733920427740209,
1574
+ "loss": 0.6073,
1575
+ "step": 522000
1576
+ },
1577
+ {
1578
+ "epoch": 12.34,
1579
+ "learning_rate": 0.00017686743198616135,
1580
+ "loss": 0.6068,
1581
+ "step": 524000
1582
+ },
1583
+ {
1584
+ "epoch": 12.39,
1585
+ "learning_rate": 0.00017639565969492058,
1586
+ "loss": 0.6114,
1587
+ "step": 526000
1588
+ },
1589
+ {
1590
+ "epoch": 12.44,
1591
+ "learning_rate": 0.0001759238874036798,
1592
+ "loss": 0.6189,
1593
+ "step": 528000
1594
+ },
1595
+ {
1596
+ "epoch": 12.48,
1597
+ "learning_rate": 0.00017545211511243904,
1598
+ "loss": 0.6111,
1599
+ "step": 530000
1600
+ },
1601
+ {
1602
+ "epoch": 12.53,
1603
+ "learning_rate": 0.0001749803428211983,
1604
+ "loss": 0.6085,
1605
+ "step": 532000
1606
+ },
1607
+ {
1608
+ "epoch": 12.58,
1609
+ "learning_rate": 0.0001745085705299575,
1610
+ "loss": 0.6164,
1611
+ "step": 534000
1612
+ },
1613
+ {
1614
+ "epoch": 12.62,
1615
+ "learning_rate": 0.00017403679823871677,
1616
+ "loss": 0.6169,
1617
+ "step": 536000
1618
+ },
1619
+ {
1620
+ "epoch": 12.67,
1621
+ "learning_rate": 0.00017356502594747598,
1622
+ "loss": 0.615,
1623
+ "step": 538000
1624
+ },
1625
+ {
1626
+ "epoch": 12.72,
1627
+ "learning_rate": 0.00017309325365623524,
1628
+ "loss": 0.6145,
1629
+ "step": 540000
1630
+ },
1631
+ {
1632
+ "epoch": 12.76,
1633
+ "learning_rate": 0.0001726214813649945,
1634
+ "loss": 0.6149,
1635
+ "step": 542000
1636
+ },
1637
+ {
1638
+ "epoch": 12.81,
1639
+ "learning_rate": 0.00017214970907375373,
1640
+ "loss": 0.6128,
1641
+ "step": 544000
1642
+ },
1643
+ {
1644
+ "epoch": 12.86,
1645
+ "learning_rate": 0.00017167793678251296,
1646
+ "loss": 0.6123,
1647
+ "step": 546000
1648
+ },
1649
+ {
1650
+ "epoch": 12.91,
1651
+ "learning_rate": 0.0001712061644912722,
1652
+ "loss": 0.6132,
1653
+ "step": 548000
1654
+ },
1655
+ {
1656
+ "epoch": 12.95,
1657
+ "learning_rate": 0.00017073439220003145,
1658
+ "loss": 0.6154,
1659
+ "step": 550000
1660
+ },
1661
+ {
1662
+ "epoch": 13.0,
1663
+ "learning_rate": 0.00017026261990879066,
1664
+ "loss": 0.6168,
1665
+ "step": 552000
1666
+ },
1667
+ {
1668
+ "epoch": 13.05,
1669
+ "learning_rate": 0.00016979084761754992,
1670
+ "loss": 0.5926,
1671
+ "step": 554000
1672
+ },
1673
+ {
1674
+ "epoch": 13.09,
1675
+ "learning_rate": 0.00016931907532630915,
1676
+ "loss": 0.5924,
1677
+ "step": 556000
1678
+ },
1679
+ {
1680
+ "epoch": 13.14,
1681
+ "learning_rate": 0.00016884730303506838,
1682
+ "loss": 0.5949,
1683
+ "step": 558000
1684
+ },
1685
+ {
1686
+ "epoch": 13.19,
1687
+ "learning_rate": 0.00016837553074382764,
1688
+ "loss": 0.5947,
1689
+ "step": 560000
1690
+ },
1691
+ {
1692
+ "epoch": 13.24,
1693
+ "learning_rate": 0.00016790375845258688,
1694
+ "loss": 0.5971,
1695
+ "step": 562000
1696
+ },
1697
+ {
1698
+ "epoch": 13.28,
1699
+ "learning_rate": 0.0001674319861613461,
1700
+ "loss": 0.5976,
1701
+ "step": 564000
1702
+ },
1703
+ {
1704
+ "epoch": 13.33,
1705
+ "learning_rate": 0.00016696021387010534,
1706
+ "loss": 0.5984,
1707
+ "step": 566000
1708
+ },
1709
+ {
1710
+ "epoch": 13.38,
1711
+ "learning_rate": 0.0001664884415788646,
1712
+ "loss": 0.606,
1713
+ "step": 568000
1714
+ },
1715
+ {
1716
+ "epoch": 13.42,
1717
+ "learning_rate": 0.0001660166692876238,
1718
+ "loss": 0.6091,
1719
+ "step": 570000
1720
+ },
1721
+ {
1722
+ "epoch": 13.47,
1723
+ "learning_rate": 0.00016554489699638307,
1724
+ "loss": 0.5927,
1725
+ "step": 572000
1726
+ },
1727
+ {
1728
+ "epoch": 13.52,
1729
+ "learning_rate": 0.0001650731247051423,
1730
+ "loss": 0.6067,
1731
+ "step": 574000
1732
+ },
1733
+ {
1734
+ "epoch": 13.57,
1735
+ "learning_rate": 0.00016460135241390153,
1736
+ "loss": 0.6024,
1737
+ "step": 576000
1738
+ },
1739
+ {
1740
+ "epoch": 13.61,
1741
+ "learning_rate": 0.0001641295801226608,
1742
+ "loss": 0.6055,
1743
+ "step": 578000
1744
+ },
1745
+ {
1746
+ "epoch": 13.66,
1747
+ "learning_rate": 0.00016365780783142003,
1748
+ "loss": 0.6003,
1749
+ "step": 580000
1750
+ },
1751
+ {
1752
+ "epoch": 13.71,
1753
+ "learning_rate": 0.00016318603554017929,
1754
+ "loss": 0.6076,
1755
+ "step": 582000
1756
+ },
1757
+ {
1758
+ "epoch": 13.75,
1759
+ "learning_rate": 0.0001627142632489385,
1760
+ "loss": 0.606,
1761
+ "step": 584000
1762
+ },
1763
+ {
1764
+ "epoch": 13.8,
1765
+ "learning_rate": 0.00016224249095769775,
1766
+ "loss": 0.5901,
1767
+ "step": 586000
1768
+ },
1769
+ {
1770
+ "epoch": 13.85,
1771
+ "learning_rate": 0.00016177071866645696,
1772
+ "loss": 0.6084,
1773
+ "step": 588000
1774
+ },
1775
+ {
1776
+ "epoch": 13.9,
1777
+ "learning_rate": 0.00016129894637521622,
1778
+ "loss": 0.6075,
1779
+ "step": 590000
1780
+ },
1781
+ {
1782
+ "epoch": 13.94,
1783
+ "learning_rate": 0.00016082717408397545,
1784
+ "loss": 0.6085,
1785
+ "step": 592000
1786
+ },
1787
+ {
1788
+ "epoch": 13.99,
1789
+ "learning_rate": 0.00016035540179273468,
1790
+ "loss": 0.6009,
1791
+ "step": 594000
1792
+ },
1793
+ {
1794
+ "epoch": 14.04,
1795
+ "learning_rate": 0.00015988362950149391,
1796
+ "loss": 0.5894,
1797
+ "step": 596000
1798
+ },
1799
+ {
1800
+ "epoch": 14.08,
1801
+ "learning_rate": 0.00015941185721025317,
1802
+ "loss": 0.5835,
1803
+ "step": 598000
1804
+ },
1805
+ {
1806
+ "epoch": 14.13,
1807
+ "learning_rate": 0.00015894008491901243,
1808
+ "loss": 0.5859,
1809
+ "step": 600000
1810
+ },
1811
+ {
1812
+ "epoch": 14.18,
1813
+ "learning_rate": 0.00015846831262777164,
1814
+ "loss": 0.5929,
1815
+ "step": 602000
1816
+ },
1817
+ {
1818
+ "epoch": 14.23,
1819
+ "learning_rate": 0.0001579965403365309,
1820
+ "loss": 0.59,
1821
+ "step": 604000
1822
+ },
1823
+ {
1824
+ "epoch": 14.27,
1825
+ "learning_rate": 0.0001575247680452901,
1826
+ "loss": 0.582,
1827
+ "step": 606000
1828
+ },
1829
+ {
1830
+ "epoch": 14.32,
1831
+ "learning_rate": 0.00015705299575404937,
1832
+ "loss": 0.588,
1833
+ "step": 608000
1834
+ },
1835
+ {
1836
+ "epoch": 14.37,
1837
+ "learning_rate": 0.0001565812234628086,
1838
+ "loss": 0.5923,
1839
+ "step": 610000
1840
+ },
1841
+ {
1842
+ "epoch": 14.41,
1843
+ "learning_rate": 0.00015610945117156786,
1844
+ "loss": 0.5837,
1845
+ "step": 612000
1846
+ },
1847
+ {
1848
+ "epoch": 14.46,
1849
+ "learning_rate": 0.00015563767888032706,
1850
+ "loss": 0.5898,
1851
+ "step": 614000
1852
+ },
1853
+ {
1854
+ "epoch": 14.51,
1855
+ "learning_rate": 0.00015516590658908632,
1856
+ "loss": 0.5904,
1857
+ "step": 616000
1858
+ },
1859
+ {
1860
+ "epoch": 14.55,
1861
+ "learning_rate": 0.00015469413429784558,
1862
+ "loss": 0.5892,
1863
+ "step": 618000
1864
+ },
1865
+ {
1866
+ "epoch": 14.6,
1867
+ "learning_rate": 0.0001542223620066048,
1868
+ "loss": 0.5887,
1869
+ "step": 620000
1870
+ },
1871
+ {
1872
+ "epoch": 14.65,
1873
+ "learning_rate": 0.00015375058971536405,
1874
+ "loss": 0.5843,
1875
+ "step": 622000
1876
+ },
1877
+ {
1878
+ "epoch": 14.7,
1879
+ "learning_rate": 0.00015327881742412328,
1880
+ "loss": 0.5853,
1881
+ "step": 624000
1882
+ },
1883
+ {
1884
+ "epoch": 14.74,
1885
+ "learning_rate": 0.00015280704513288251,
1886
+ "loss": 0.5919,
1887
+ "step": 626000
1888
+ },
1889
+ {
1890
+ "epoch": 14.79,
1891
+ "learning_rate": 0.00015233527284164175,
1892
+ "loss": 0.5915,
1893
+ "step": 628000
1894
+ },
1895
+ {
1896
+ "epoch": 14.84,
1897
+ "learning_rate": 0.000151863500550401,
1898
+ "loss": 0.5918,
1899
+ "step": 630000
1900
+ },
1901
+ {
1902
+ "epoch": 14.88,
1903
+ "learning_rate": 0.0001513917282591602,
1904
+ "loss": 0.5932,
1905
+ "step": 632000
1906
+ },
1907
+ {
1908
+ "epoch": 14.93,
1909
+ "learning_rate": 0.00015091995596791947,
1910
+ "loss": 0.5891,
1911
+ "step": 634000
1912
+ },
1913
+ {
1914
+ "epoch": 14.98,
1915
+ "learning_rate": 0.00015044818367667873,
1916
+ "loss": 0.5888,
1917
+ "step": 636000
1918
+ },
1919
+ {
1920
+ "epoch": 15.03,
1921
+ "learning_rate": 0.00014997641138543794,
1922
+ "loss": 0.5817,
1923
+ "step": 638000
1924
+ },
1925
+ {
1926
+ "epoch": 15.07,
1927
+ "learning_rate": 0.00014950463909419717,
1928
+ "loss": 0.574,
1929
+ "step": 640000
1930
+ },
1931
+ {
1932
+ "epoch": 15.12,
1933
+ "learning_rate": 0.00014903286680295643,
1934
+ "loss": 0.5723,
1935
+ "step": 642000
1936
+ },
1937
+ {
1938
+ "epoch": 15.17,
1939
+ "learning_rate": 0.00014856109451171566,
1940
+ "loss": 0.573,
1941
+ "step": 644000
1942
+ },
1943
+ {
1944
+ "epoch": 15.21,
1945
+ "learning_rate": 0.00014808932222047492,
1946
+ "loss": 0.5758,
1947
+ "step": 646000
1948
+ },
1949
+ {
1950
+ "epoch": 15.26,
1951
+ "learning_rate": 0.00014761754992923416,
1952
+ "loss": 0.5796,
1953
+ "step": 648000
1954
+ },
1955
+ {
1956
+ "epoch": 15.31,
1957
+ "learning_rate": 0.0001471457776379934,
1958
+ "loss": 0.5722,
1959
+ "step": 650000
1960
+ },
1961
+ {
1962
+ "epoch": 15.36,
1963
+ "learning_rate": 0.00014667400534675262,
1964
+ "loss": 0.5825,
1965
+ "step": 652000
1966
+ },
1967
+ {
1968
+ "epoch": 15.4,
1969
+ "learning_rate": 0.00014620223305551185,
1970
+ "loss": 0.5744,
1971
+ "step": 654000
1972
+ },
1973
+ {
1974
+ "epoch": 15.45,
1975
+ "learning_rate": 0.0001457304607642711,
1976
+ "loss": 0.5786,
1977
+ "step": 656000
1978
+ },
1979
+ {
1980
+ "epoch": 15.5,
1981
+ "learning_rate": 0.00014525868847303032,
1982
+ "loss": 0.5808,
1983
+ "step": 658000
1984
+ },
1985
+ {
1986
+ "epoch": 15.54,
1987
+ "learning_rate": 0.00014478691618178958,
1988
+ "loss": 0.5855,
1989
+ "step": 660000
1990
+ },
1991
+ {
1992
+ "epoch": 15.59,
1993
+ "learning_rate": 0.0001443151438905488,
1994
+ "loss": 0.585,
1995
+ "step": 662000
1996
+ },
1997
+ {
1998
+ "epoch": 15.64,
1999
+ "learning_rate": 0.00014384337159930807,
2000
+ "loss": 0.5692,
2001
+ "step": 664000
2002
+ },
2003
+ {
2004
+ "epoch": 15.69,
2005
+ "learning_rate": 0.0001433715993080673,
2006
+ "loss": 0.5789,
2007
+ "step": 666000
2008
+ },
2009
+ {
2010
+ "epoch": 15.73,
2011
+ "learning_rate": 0.00014289982701682654,
2012
+ "loss": 0.5836,
2013
+ "step": 668000
2014
+ },
2015
+ {
2016
+ "epoch": 15.78,
2017
+ "learning_rate": 0.00014242805472558577,
2018
+ "loss": 0.5796,
2019
+ "step": 670000
2020
+ },
2021
+ {
2022
+ "epoch": 15.83,
2023
+ "learning_rate": 0.000141956282434345,
2024
+ "loss": 0.5756,
2025
+ "step": 672000
2026
+ },
2027
+ {
2028
+ "epoch": 15.87,
2029
+ "learning_rate": 0.00014148451014310424,
2030
+ "loss": 0.5667,
2031
+ "step": 674000
2032
+ },
2033
+ {
2034
+ "epoch": 15.92,
2035
+ "learning_rate": 0.0001410127378518635,
2036
+ "loss": 0.5845,
2037
+ "step": 676000
2038
+ },
2039
+ {
2040
+ "epoch": 15.97,
2041
+ "learning_rate": 0.00014054096556062273,
2042
+ "loss": 0.5783,
2043
+ "step": 678000
2044
+ },
2045
+ {
2046
+ "epoch": 16.02,
2047
+ "learning_rate": 0.00014006919326938196,
2048
+ "loss": 0.5749,
2049
+ "step": 680000
2050
+ },
2051
+ {
2052
+ "epoch": 16.06,
2053
+ "learning_rate": 0.00013959742097814122,
2054
+ "loss": 0.5602,
2055
+ "step": 682000
2056
+ },
2057
+ {
2058
+ "epoch": 16.11,
2059
+ "learning_rate": 0.00013912564868690045,
2060
+ "loss": 0.564,
2061
+ "step": 684000
2062
+ },
2063
+ {
2064
+ "epoch": 16.16,
2065
+ "learning_rate": 0.00013865387639565969,
2066
+ "loss": 0.5659,
2067
+ "step": 686000
2068
+ },
2069
+ {
2070
+ "epoch": 16.2,
2071
+ "learning_rate": 0.00013818210410441892,
2072
+ "loss": 0.5608,
2073
+ "step": 688000
2074
+ },
2075
+ {
2076
+ "epoch": 16.25,
2077
+ "learning_rate": 0.00013771033181317815,
2078
+ "loss": 0.5668,
2079
+ "step": 690000
2080
+ },
2081
+ {
2082
+ "epoch": 16.3,
2083
+ "learning_rate": 0.00013723855952193738,
2084
+ "loss": 0.5646,
2085
+ "step": 692000
2086
+ },
2087
+ {
2088
+ "epoch": 16.34,
2089
+ "learning_rate": 0.00013676678723069664,
2090
+ "loss": 0.5663,
2091
+ "step": 694000
2092
+ },
2093
+ {
2094
+ "epoch": 16.39,
2095
+ "learning_rate": 0.00013629501493945588,
2096
+ "loss": 0.5716,
2097
+ "step": 696000
2098
+ },
2099
+ {
2100
+ "epoch": 16.44,
2101
+ "learning_rate": 0.0001358232426482151,
2102
+ "loss": 0.5716,
2103
+ "step": 698000
2104
+ },
2105
+ {
2106
+ "epoch": 16.49,
2107
+ "learning_rate": 0.00013535147035697437,
2108
+ "loss": 0.568,
2109
+ "step": 700000
2110
+ },
2111
+ {
2112
+ "epoch": 16.53,
2113
+ "learning_rate": 0.0001348796980657336,
2114
+ "loss": 0.566,
2115
+ "step": 702000
2116
+ },
2117
+ {
2118
+ "epoch": 16.58,
2119
+ "learning_rate": 0.00013440792577449284,
2120
+ "loss": 0.5657,
2121
+ "step": 704000
2122
+ },
2123
+ {
2124
+ "epoch": 16.63,
2125
+ "learning_rate": 0.00013393615348325207,
2126
+ "loss": 0.5641,
2127
+ "step": 706000
2128
+ },
2129
+ {
2130
+ "epoch": 16.67,
2131
+ "learning_rate": 0.0001334643811920113,
2132
+ "loss": 0.5679,
2133
+ "step": 708000
2134
+ },
2135
+ {
2136
+ "epoch": 16.72,
2137
+ "learning_rate": 0.00013299260890077056,
2138
+ "loss": 0.5652,
2139
+ "step": 710000
2140
+ },
2141
+ {
2142
+ "epoch": 16.77,
2143
+ "learning_rate": 0.0001325208366095298,
2144
+ "loss": 0.5672,
2145
+ "step": 712000
2146
+ },
2147
+ {
2148
+ "epoch": 16.82,
2149
+ "learning_rate": 0.00013204906431828903,
2150
+ "loss": 0.5695,
2151
+ "step": 714000
2152
+ },
2153
+ {
2154
+ "epoch": 16.86,
2155
+ "learning_rate": 0.00013157729202704826,
2156
+ "loss": 0.5719,
2157
+ "step": 716000
2158
+ },
2159
+ {
2160
+ "epoch": 16.91,
2161
+ "learning_rate": 0.00013110551973580752,
2162
+ "loss": 0.571,
2163
+ "step": 718000
2164
+ },
2165
+ {
2166
+ "epoch": 16.96,
2167
+ "learning_rate": 0.00013063374744456675,
2168
+ "loss": 0.5684,
2169
+ "step": 720000
2170
+ },
2171
+ {
2172
+ "epoch": 17.0,
2173
+ "learning_rate": 0.00013016197515332598,
2174
+ "loss": 0.5672,
2175
+ "step": 722000
2176
+ },
2177
+ {
2178
+ "epoch": 17.05,
2179
+ "learning_rate": 0.00012969020286208522,
2180
+ "loss": 0.5566,
2181
+ "step": 724000
2182
+ },
2183
+ {
2184
+ "epoch": 17.1,
2185
+ "learning_rate": 0.00012921843057084445,
2186
+ "loss": 0.5459,
2187
+ "step": 726000
2188
+ },
2189
+ {
2190
+ "epoch": 17.15,
2191
+ "learning_rate": 0.0001287466582796037,
2192
+ "loss": 0.5522,
2193
+ "step": 728000
2194
+ },
2195
+ {
2196
+ "epoch": 17.19,
2197
+ "learning_rate": 0.00012827488598836294,
2198
+ "loss": 0.5586,
2199
+ "step": 730000
2200
+ },
2201
+ {
2202
+ "epoch": 17.24,
2203
+ "learning_rate": 0.00012780311369712217,
2204
+ "loss": 0.5551,
2205
+ "step": 732000
2206
+ },
2207
+ {
2208
+ "epoch": 17.29,
2209
+ "learning_rate": 0.0001273313414058814,
2210
+ "loss": 0.558,
2211
+ "step": 734000
2212
+ },
2213
+ {
2214
+ "epoch": 17.33,
2215
+ "learning_rate": 0.00012685956911464067,
2216
+ "loss": 0.5501,
2217
+ "step": 736000
2218
+ },
2219
+ {
2220
+ "epoch": 17.38,
2221
+ "learning_rate": 0.0001263877968233999,
2222
+ "loss": 0.5496,
2223
+ "step": 738000
2224
+ },
2225
+ {
2226
+ "epoch": 17.43,
2227
+ "learning_rate": 0.00012591602453215913,
2228
+ "loss": 0.5617,
2229
+ "step": 740000
2230
+ },
2231
+ {
2232
+ "epoch": 17.48,
2233
+ "learning_rate": 0.00012544425224091837,
2234
+ "loss": 0.5481,
2235
+ "step": 742000
2236
+ },
2237
+ {
2238
+ "epoch": 17.52,
2239
+ "learning_rate": 0.00012497247994967763,
2240
+ "loss": 0.5627,
2241
+ "step": 744000
2242
+ },
2243
+ {
2244
+ "epoch": 17.57,
2245
+ "learning_rate": 0.00012450070765843686,
2246
+ "loss": 0.5556,
2247
+ "step": 746000
2248
+ },
2249
+ {
2250
+ "epoch": 17.62,
2251
+ "learning_rate": 0.0001240289353671961,
2252
+ "loss": 0.5565,
2253
+ "step": 748000
2254
+ },
2255
+ {
2256
+ "epoch": 17.66,
2257
+ "learning_rate": 0.00012355716307595532,
2258
+ "loss": 0.5591,
2259
+ "step": 750000
2260
+ },
2261
+ {
2262
+ "epoch": 17.71,
2263
+ "learning_rate": 0.00012308539078471456,
2264
+ "loss": 0.5526,
2265
+ "step": 752000
2266
+ },
2267
+ {
2268
+ "epoch": 17.76,
2269
+ "learning_rate": 0.00012261361849347382,
2270
+ "loss": 0.5564,
2271
+ "step": 754000
2272
+ },
2273
+ {
2274
+ "epoch": 17.8,
2275
+ "learning_rate": 0.00012214184620223305,
2276
+ "loss": 0.5541,
2277
+ "step": 756000
2278
+ },
2279
+ {
2280
+ "epoch": 17.85,
2281
+ "learning_rate": 0.00012167007391099228,
2282
+ "loss": 0.5601,
2283
+ "step": 758000
2284
+ },
2285
+ {
2286
+ "epoch": 17.9,
2287
+ "learning_rate": 0.00012119830161975153,
2288
+ "loss": 0.5555,
2289
+ "step": 760000
2290
+ },
2291
+ {
2292
+ "epoch": 17.95,
2293
+ "learning_rate": 0.00012072652932851076,
2294
+ "loss": 0.5525,
2295
+ "step": 762000
2296
+ },
2297
+ {
2298
+ "epoch": 17.99,
2299
+ "learning_rate": 0.00012025475703727,
2300
+ "loss": 0.5577,
2301
+ "step": 764000
2302
+ },
2303
+ {
2304
+ "epoch": 18.04,
2305
+ "learning_rate": 0.00011978298474602924,
2306
+ "loss": 0.5403,
2307
+ "step": 766000
2308
+ },
2309
+ {
2310
+ "epoch": 18.09,
2311
+ "learning_rate": 0.00011931121245478847,
2312
+ "loss": 0.5466,
2313
+ "step": 768000
2314
+ },
2315
+ {
2316
+ "epoch": 18.13,
2317
+ "learning_rate": 0.0001188394401635477,
2318
+ "loss": 0.546,
2319
+ "step": 770000
2320
+ },
2321
+ {
2322
+ "epoch": 18.18,
2323
+ "learning_rate": 0.00011836766787230695,
2324
+ "loss": 0.5435,
2325
+ "step": 772000
2326
+ },
2327
+ {
2328
+ "epoch": 18.23,
2329
+ "learning_rate": 0.0001178958955810662,
2330
+ "loss": 0.5442,
2331
+ "step": 774000
2332
+ },
2333
+ {
2334
+ "epoch": 18.28,
2335
+ "learning_rate": 0.00011742412328982544,
2336
+ "loss": 0.5381,
2337
+ "step": 776000
2338
+ },
2339
+ {
2340
+ "epoch": 18.32,
2341
+ "learning_rate": 0.00011695235099858468,
2342
+ "loss": 0.542,
2343
+ "step": 778000
2344
+ },
2345
+ {
2346
+ "epoch": 18.37,
2347
+ "learning_rate": 0.00011648057870734391,
2348
+ "loss": 0.5458,
2349
+ "step": 780000
2350
+ },
2351
+ {
2352
+ "epoch": 18.42,
2353
+ "learning_rate": 0.00011600880641610316,
2354
+ "loss": 0.5473,
2355
+ "step": 782000
2356
+ },
2357
+ {
2358
+ "epoch": 18.46,
2359
+ "learning_rate": 0.00011553703412486239,
2360
+ "loss": 0.5511,
2361
+ "step": 784000
2362
+ },
2363
+ {
2364
+ "epoch": 18.51,
2365
+ "learning_rate": 0.00011506526183362162,
2366
+ "loss": 0.5415,
2367
+ "step": 786000
2368
+ },
2369
+ {
2370
+ "epoch": 18.56,
2371
+ "learning_rate": 0.00011459348954238087,
2372
+ "loss": 0.5431,
2373
+ "step": 788000
2374
+ },
2375
+ {
2376
+ "epoch": 18.61,
2377
+ "learning_rate": 0.0001141217172511401,
2378
+ "loss": 0.5403,
2379
+ "step": 790000
2380
+ },
2381
+ {
2382
+ "epoch": 18.65,
2383
+ "learning_rate": 0.00011364994495989935,
2384
+ "loss": 0.5402,
2385
+ "step": 792000
2386
+ },
2387
+ {
2388
+ "epoch": 18.7,
2389
+ "learning_rate": 0.00011317817266865859,
2390
+ "loss": 0.5447,
2391
+ "step": 794000
2392
+ },
2393
+ {
2394
+ "epoch": 18.75,
2395
+ "learning_rate": 0.00011270640037741783,
2396
+ "loss": 0.5485,
2397
+ "step": 796000
2398
+ },
2399
+ {
2400
+ "epoch": 18.79,
2401
+ "learning_rate": 0.00011223462808617706,
2402
+ "loss": 0.5485,
2403
+ "step": 798000
2404
+ },
2405
+ {
2406
+ "epoch": 18.84,
2407
+ "learning_rate": 0.0001117628557949363,
2408
+ "loss": 0.5441,
2409
+ "step": 800000
2410
+ },
2411
+ {
2412
+ "epoch": 18.89,
2413
+ "learning_rate": 0.00011129108350369554,
2414
+ "loss": 0.5457,
2415
+ "step": 802000
2416
+ },
2417
+ {
2418
+ "epoch": 18.94,
2419
+ "learning_rate": 0.00011081931121245477,
2420
+ "loss": 0.5416,
2421
+ "step": 804000
2422
+ },
2423
+ {
2424
+ "epoch": 18.98,
2425
+ "learning_rate": 0.00011034753892121402,
2426
+ "loss": 0.5494,
2427
+ "step": 806000
2428
+ },
2429
+ {
2430
+ "epoch": 19.03,
2431
+ "learning_rate": 0.00010987576662997325,
2432
+ "loss": 0.5347,
2433
+ "step": 808000
2434
+ },
2435
+ {
2436
+ "epoch": 19.08,
2437
+ "learning_rate": 0.00010940399433873251,
2438
+ "loss": 0.5257,
2439
+ "step": 810000
2440
+ },
2441
+ {
2442
+ "epoch": 19.12,
2443
+ "learning_rate": 0.00010893222204749174,
2444
+ "loss": 0.5278,
2445
+ "step": 812000
2446
+ },
2447
+ {
2448
+ "epoch": 19.17,
2449
+ "learning_rate": 0.00010846044975625097,
2450
+ "loss": 0.5355,
2451
+ "step": 814000
2452
+ },
2453
+ {
2454
+ "epoch": 19.22,
2455
+ "learning_rate": 0.00010798867746501022,
2456
+ "loss": 0.5378,
2457
+ "step": 816000
2458
+ },
2459
+ {
2460
+ "epoch": 19.27,
2461
+ "learning_rate": 0.00010751690517376945,
2462
+ "loss": 0.537,
2463
+ "step": 818000
2464
+ },
2465
+ {
2466
+ "epoch": 19.31,
2467
+ "learning_rate": 0.00010704513288252869,
2468
+ "loss": 0.5385,
2469
+ "step": 820000
2470
+ },
2471
+ {
2472
+ "epoch": 19.36,
2473
+ "learning_rate": 0.00010657336059128793,
2474
+ "loss": 0.5302,
2475
+ "step": 822000
2476
+ },
2477
+ {
2478
+ "epoch": 19.41,
2479
+ "learning_rate": 0.00010610158830004717,
2480
+ "loss": 0.5409,
2481
+ "step": 824000
2482
+ },
2483
+ {
2484
+ "epoch": 19.45,
2485
+ "learning_rate": 0.0001056298160088064,
2486
+ "loss": 0.5347,
2487
+ "step": 826000
2488
+ },
2489
+ {
2490
+ "epoch": 19.5,
2491
+ "learning_rate": 0.00010515804371756563,
2492
+ "loss": 0.5288,
2493
+ "step": 828000
2494
+ },
2495
+ {
2496
+ "epoch": 19.55,
2497
+ "learning_rate": 0.00010468627142632489,
2498
+ "loss": 0.5276,
2499
+ "step": 830000
2500
+ },
2501
+ {
2502
+ "epoch": 19.59,
2503
+ "learning_rate": 0.00010421449913508412,
2504
+ "loss": 0.5334,
2505
+ "step": 832000
2506
+ },
2507
+ {
2508
+ "epoch": 19.64,
2509
+ "learning_rate": 0.00010374272684384337,
2510
+ "loss": 0.5407,
2511
+ "step": 834000
2512
+ },
2513
+ {
2514
+ "epoch": 19.69,
2515
+ "learning_rate": 0.0001032709545526026,
2516
+ "loss": 0.5354,
2517
+ "step": 836000
2518
+ },
2519
+ {
2520
+ "epoch": 19.74,
2521
+ "learning_rate": 0.00010279918226136184,
2522
+ "loss": 0.5274,
2523
+ "step": 838000
2524
+ },
2525
+ {
2526
+ "epoch": 19.78,
2527
+ "learning_rate": 0.00010232740997012108,
2528
+ "loss": 0.529,
2529
+ "step": 840000
2530
+ },
2531
+ {
2532
+ "epoch": 19.83,
2533
+ "learning_rate": 0.00010185563767888031,
2534
+ "loss": 0.5327,
2535
+ "step": 842000
2536
+ },
2537
+ {
2538
+ "epoch": 19.88,
2539
+ "learning_rate": 0.00010138386538763955,
2540
+ "loss": 0.5383,
2541
+ "step": 844000
2542
+ },
2543
+ {
2544
+ "epoch": 19.92,
2545
+ "learning_rate": 0.0001009120930963988,
2546
+ "loss": 0.5352,
2547
+ "step": 846000
2548
+ },
2549
+ {
2550
+ "epoch": 19.97,
2551
+ "learning_rate": 0.00010044032080515804,
2552
+ "loss": 0.5355,
2553
+ "step": 848000
2554
+ },
2555
+ {
2556
+ "epoch": 20.02,
2557
+ "learning_rate": 9.996854851391727e-05,
2558
+ "loss": 0.5233,
2559
+ "step": 850000
2560
+ },
2561
+ {
2562
+ "epoch": 20.07,
2563
+ "learning_rate": 9.949677622267652e-05,
2564
+ "loss": 0.5255,
2565
+ "step": 852000
2566
+ },
2567
+ {
2568
+ "epoch": 20.11,
2569
+ "learning_rate": 9.902500393143575e-05,
2570
+ "loss": 0.5168,
2571
+ "step": 854000
2572
+ },
2573
+ {
2574
+ "epoch": 20.16,
2575
+ "learning_rate": 9.855323164019498e-05,
2576
+ "loss": 0.5181,
2577
+ "step": 856000
2578
+ },
2579
+ {
2580
+ "epoch": 20.21,
2581
+ "learning_rate": 9.808145934895423e-05,
2582
+ "loss": 0.5245,
2583
+ "step": 858000
2584
+ },
2585
+ {
2586
+ "epoch": 20.25,
2587
+ "learning_rate": 9.760968705771346e-05,
2588
+ "loss": 0.5191,
2589
+ "step": 860000
2590
+ },
2591
+ {
2592
+ "epoch": 20.3,
2593
+ "learning_rate": 9.71379147664727e-05,
2594
+ "loss": 0.5207,
2595
+ "step": 862000
2596
+ },
2597
+ {
2598
+ "epoch": 20.35,
2599
+ "learning_rate": 9.666614247523194e-05,
2600
+ "loss": 0.5251,
2601
+ "step": 864000
2602
+ },
2603
+ {
2604
+ "epoch": 20.4,
2605
+ "learning_rate": 9.619437018399119e-05,
2606
+ "loss": 0.5158,
2607
+ "step": 866000
2608
+ },
2609
+ {
2610
+ "epoch": 20.44,
2611
+ "learning_rate": 9.572259789275043e-05,
2612
+ "loss": 0.5178,
2613
+ "step": 868000
2614
+ },
2615
+ {
2616
+ "epoch": 20.49,
2617
+ "learning_rate": 9.525082560150967e-05,
2618
+ "loss": 0.5213,
2619
+ "step": 870000
2620
+ },
2621
+ {
2622
+ "epoch": 20.54,
2623
+ "learning_rate": 9.47790533102689e-05,
2624
+ "loss": 0.5271,
2625
+ "step": 872000
2626
+ },
2627
+ {
2628
+ "epoch": 20.58,
2629
+ "learning_rate": 9.430728101902815e-05,
2630
+ "loss": 0.5242,
2631
+ "step": 874000
2632
+ },
2633
+ {
2634
+ "epoch": 20.63,
2635
+ "learning_rate": 9.383550872778738e-05,
2636
+ "loss": 0.5278,
2637
+ "step": 876000
2638
+ },
2639
+ {
2640
+ "epoch": 20.68,
2641
+ "learning_rate": 9.336373643654661e-05,
2642
+ "loss": 0.5222,
2643
+ "step": 878000
2644
+ },
2645
+ {
2646
+ "epoch": 20.73,
2647
+ "learning_rate": 9.289196414530586e-05,
2648
+ "loss": 0.5242,
2649
+ "step": 880000
2650
+ },
2651
+ {
2652
+ "epoch": 20.77,
2653
+ "learning_rate": 9.242019185406509e-05,
2654
+ "loss": 0.5271,
2655
+ "step": 882000
2656
+ },
2657
+ {
2658
+ "epoch": 20.82,
2659
+ "learning_rate": 9.194841956282434e-05,
2660
+ "loss": 0.5267,
2661
+ "step": 884000
2662
+ },
2663
+ {
2664
+ "epoch": 20.87,
2665
+ "learning_rate": 9.147664727158358e-05,
2666
+ "loss": 0.5195,
2667
+ "step": 886000
2668
+ },
2669
+ {
2670
+ "epoch": 20.91,
2671
+ "learning_rate": 9.100487498034282e-05,
2672
+ "loss": 0.5162,
2673
+ "step": 888000
2674
+ },
2675
+ {
2676
+ "epoch": 20.96,
2677
+ "learning_rate": 9.053310268910205e-05,
2678
+ "loss": 0.5166,
2679
+ "step": 890000
2680
+ },
2681
+ {
2682
+ "epoch": 21.01,
2683
+ "learning_rate": 9.00613303978613e-05,
2684
+ "loss": 0.5266,
2685
+ "step": 892000
2686
+ },
2687
+ {
2688
+ "epoch": 21.06,
2689
+ "learning_rate": 8.958955810662053e-05,
2690
+ "loss": 0.5149,
2691
+ "step": 894000
2692
+ },
2693
+ {
2694
+ "epoch": 21.1,
2695
+ "learning_rate": 8.911778581537976e-05,
2696
+ "loss": 0.5023,
2697
+ "step": 896000
2698
+ },
2699
+ {
2700
+ "epoch": 21.15,
2701
+ "learning_rate": 8.864601352413901e-05,
2702
+ "loss": 0.5112,
2703
+ "step": 898000
2704
+ },
2705
+ {
2706
+ "epoch": 21.2,
2707
+ "learning_rate": 8.817424123289824e-05,
2708
+ "loss": 0.5084,
2709
+ "step": 900000
2710
+ },
2711
+ {
2712
+ "epoch": 21.24,
2713
+ "learning_rate": 8.770246894165747e-05,
2714
+ "loss": 0.509,
2715
+ "step": 902000
2716
+ },
2717
+ {
2718
+ "epoch": 21.29,
2719
+ "learning_rate": 8.723069665041673e-05,
2720
+ "loss": 0.5043,
2721
+ "step": 904000
2722
+ },
2723
+ {
2724
+ "epoch": 21.34,
2725
+ "learning_rate": 8.675892435917597e-05,
2726
+ "loss": 0.5188,
2727
+ "step": 906000
2728
+ },
2729
+ {
2730
+ "epoch": 21.38,
2731
+ "learning_rate": 8.628715206793521e-05,
2732
+ "loss": 0.5094,
2733
+ "step": 908000
2734
+ },
2735
+ {
2736
+ "epoch": 21.43,
2737
+ "learning_rate": 8.581537977669444e-05,
2738
+ "loss": 0.519,
2739
+ "step": 910000
2740
+ },
2741
+ {
2742
+ "epoch": 21.48,
2743
+ "learning_rate": 8.534360748545368e-05,
2744
+ "loss": 0.5108,
2745
+ "step": 912000
2746
+ },
2747
+ {
2748
+ "epoch": 21.53,
2749
+ "learning_rate": 8.487183519421292e-05,
2750
+ "loss": 0.5054,
2751
+ "step": 914000
2752
+ },
2753
+ {
2754
+ "epoch": 21.57,
2755
+ "learning_rate": 8.440006290297216e-05,
2756
+ "loss": 0.5092,
2757
+ "step": 916000
2758
+ },
2759
+ {
2760
+ "epoch": 21.62,
2761
+ "learning_rate": 8.392829061173139e-05,
2762
+ "loss": 0.5075,
2763
+ "step": 918000
2764
+ },
2765
+ {
2766
+ "epoch": 21.67,
2767
+ "learning_rate": 8.345651832049062e-05,
2768
+ "loss": 0.5024,
2769
+ "step": 920000
2770
+ },
2771
+ {
2772
+ "epoch": 21.71,
2773
+ "learning_rate": 8.298474602924988e-05,
2774
+ "loss": 0.5161,
2775
+ "step": 922000
2776
+ },
2777
+ {
2778
+ "epoch": 21.76,
2779
+ "learning_rate": 8.251297373800911e-05,
2780
+ "loss": 0.5117,
2781
+ "step": 924000
2782
+ },
2783
+ {
2784
+ "epoch": 21.81,
2785
+ "learning_rate": 8.204120144676836e-05,
2786
+ "loss": 0.5136,
2787
+ "step": 926000
2788
+ },
2789
+ {
2790
+ "epoch": 21.86,
2791
+ "learning_rate": 8.15694291555276e-05,
2792
+ "loss": 0.5062,
2793
+ "step": 928000
2794
+ },
2795
+ {
2796
+ "epoch": 21.9,
2797
+ "learning_rate": 8.109765686428683e-05,
2798
+ "loss": 0.5122,
2799
+ "step": 930000
2800
+ },
2801
+ {
2802
+ "epoch": 21.95,
2803
+ "learning_rate": 8.062588457304607e-05,
2804
+ "loss": 0.5113,
2805
+ "step": 932000
2806
+ },
2807
+ {
2808
+ "epoch": 22.0,
2809
+ "learning_rate": 8.01541122818053e-05,
2810
+ "loss": 0.5073,
2811
+ "step": 934000
2812
+ },
2813
+ {
2814
+ "epoch": 22.04,
2815
+ "learning_rate": 7.968233999056454e-05,
2816
+ "loss": 0.4986,
2817
+ "step": 936000
2818
+ },
2819
+ {
2820
+ "epoch": 22.09,
2821
+ "learning_rate": 7.921056769932378e-05,
2822
+ "loss": 0.5061,
2823
+ "step": 938000
2824
+ },
2825
+ {
2826
+ "epoch": 22.14,
2827
+ "learning_rate": 7.873879540808303e-05,
2828
+ "loss": 0.4963,
2829
+ "step": 940000
2830
+ },
2831
+ {
2832
+ "epoch": 22.19,
2833
+ "learning_rate": 7.826702311684228e-05,
2834
+ "loss": 0.4981,
2835
+ "step": 942000
2836
+ },
2837
+ {
2838
+ "epoch": 22.23,
2839
+ "learning_rate": 7.779525082560151e-05,
2840
+ "loss": 0.498,
2841
+ "step": 944000
2842
+ },
2843
+ {
2844
+ "epoch": 22.28,
2845
+ "learning_rate": 7.732347853436074e-05,
2846
+ "loss": 0.4963,
2847
+ "step": 946000
2848
+ },
2849
+ {
2850
+ "epoch": 22.33,
2851
+ "learning_rate": 7.685170624311998e-05,
2852
+ "loss": 0.4987,
2853
+ "step": 948000
2854
+ },
2855
+ {
2856
+ "epoch": 22.37,
2857
+ "learning_rate": 7.637993395187922e-05,
2858
+ "loss": 0.4999,
2859
+ "step": 950000
2860
+ },
2861
+ {
2862
+ "epoch": 22.42,
2863
+ "learning_rate": 7.590816166063845e-05,
2864
+ "loss": 0.4945,
2865
+ "step": 952000
2866
+ },
2867
+ {
2868
+ "epoch": 22.47,
2869
+ "learning_rate": 7.543638936939769e-05,
2870
+ "loss": 0.4986,
2871
+ "step": 954000
2872
+ },
2873
+ {
2874
+ "epoch": 22.52,
2875
+ "learning_rate": 7.496461707815693e-05,
2876
+ "loss": 0.4956,
2877
+ "step": 956000
2878
+ },
2879
+ {
2880
+ "epoch": 22.56,
2881
+ "learning_rate": 7.449284478691618e-05,
2882
+ "loss": 0.4958,
2883
+ "step": 958000
2884
+ },
2885
+ {
2886
+ "epoch": 22.61,
2887
+ "learning_rate": 7.402107249567541e-05,
2888
+ "loss": 0.5033,
2889
+ "step": 960000
2890
+ },
2891
+ {
2892
+ "epoch": 22.66,
2893
+ "learning_rate": 7.354930020443464e-05,
2894
+ "loss": 0.4976,
2895
+ "step": 962000
2896
+ },
2897
+ {
2898
+ "epoch": 22.7,
2899
+ "learning_rate": 7.307752791319389e-05,
2900
+ "loss": 0.5036,
2901
+ "step": 964000
2902
+ },
2903
+ {
2904
+ "epoch": 22.75,
2905
+ "learning_rate": 7.260575562195314e-05,
2906
+ "loss": 0.4973,
2907
+ "step": 966000
2908
+ },
2909
+ {
2910
+ "epoch": 22.8,
2911
+ "learning_rate": 7.213398333071237e-05,
2912
+ "loss": 0.501,
2913
+ "step": 968000
2914
+ },
2915
+ {
2916
+ "epoch": 22.85,
2917
+ "learning_rate": 7.16622110394716e-05,
2918
+ "loss": 0.5005,
2919
+ "step": 970000
2920
+ },
2921
+ {
2922
+ "epoch": 22.89,
2923
+ "learning_rate": 7.119043874823085e-05,
2924
+ "loss": 0.4991,
2925
+ "step": 972000
2926
+ },
2927
+ {
2928
+ "epoch": 22.94,
2929
+ "learning_rate": 7.07186664569901e-05,
2930
+ "loss": 0.495,
2931
+ "step": 974000
2932
+ },
2933
+ {
2934
+ "epoch": 22.99,
2935
+ "learning_rate": 7.024689416574933e-05,
2936
+ "loss": 0.5016,
2937
+ "step": 976000
2938
+ },
2939
+ {
2940
+ "epoch": 23.03,
2941
+ "learning_rate": 6.977512187450856e-05,
2942
+ "loss": 0.4966,
2943
+ "step": 978000
2944
+ },
2945
+ {
2946
+ "epoch": 23.08,
2947
+ "learning_rate": 6.93033495832678e-05,
2948
+ "loss": 0.4909,
2949
+ "step": 980000
2950
+ },
2951
+ {
2952
+ "epoch": 23.13,
2953
+ "learning_rate": 6.883157729202704e-05,
2954
+ "loss": 0.4846,
2955
+ "step": 982000
2956
+ },
2957
+ {
2958
+ "epoch": 23.17,
2959
+ "learning_rate": 6.835980500078629e-05,
2960
+ "loss": 0.4938,
2961
+ "step": 984000
2962
+ },
2963
+ {
2964
+ "epoch": 23.22,
2965
+ "learning_rate": 6.788803270954552e-05,
2966
+ "loss": 0.4967,
2967
+ "step": 986000
2968
+ },
2969
+ {
2970
+ "epoch": 23.27,
2971
+ "learning_rate": 6.741626041830475e-05,
2972
+ "loss": 0.4894,
2973
+ "step": 988000
2974
+ },
2975
+ {
2976
+ "epoch": 23.32,
2977
+ "learning_rate": 6.6944488127064e-05,
2978
+ "loss": 0.4912,
2979
+ "step": 990000
2980
+ },
2981
+ {
2982
+ "epoch": 23.36,
2983
+ "learning_rate": 6.647271583582324e-05,
2984
+ "loss": 0.4898,
2985
+ "step": 992000
2986
+ },
2987
+ {
2988
+ "epoch": 23.41,
2989
+ "learning_rate": 6.600094354458248e-05,
2990
+ "loss": 0.4854,
2991
+ "step": 994000
2992
+ },
2993
+ {
2994
+ "epoch": 23.46,
2995
+ "learning_rate": 6.552917125334171e-05,
2996
+ "loss": 0.4838,
2997
+ "step": 996000
2998
+ },
2999
+ {
3000
+ "epoch": 23.5,
3001
+ "learning_rate": 6.505739896210096e-05,
3002
+ "loss": 0.4846,
3003
+ "step": 998000
3004
+ },
3005
+ {
3006
+ "epoch": 23.55,
3007
+ "learning_rate": 6.45856266708602e-05,
3008
+ "loss": 0.4829,
3009
+ "step": 1000000
3010
+ },
3011
+ {
3012
+ "epoch": 23.6,
3013
+ "learning_rate": 6.411385437961944e-05,
3014
+ "loss": 0.4792,
3015
+ "step": 1002000
3016
+ },
3017
+ {
3018
+ "epoch": 23.65,
3019
+ "learning_rate": 6.364208208837867e-05,
3020
+ "loss": 0.486,
3021
+ "step": 1004000
3022
+ },
3023
+ {
3024
+ "epoch": 23.69,
3025
+ "learning_rate": 6.317030979713791e-05,
3026
+ "loss": 0.4946,
3027
+ "step": 1006000
3028
+ },
3029
+ {
3030
+ "epoch": 23.74,
3031
+ "learning_rate": 6.269853750589715e-05,
3032
+ "loss": 0.4825,
3033
+ "step": 1008000
3034
+ },
3035
+ {
3036
+ "epoch": 23.79,
3037
+ "learning_rate": 6.22267652146564e-05,
3038
+ "loss": 0.4845,
3039
+ "step": 1010000
3040
+ },
3041
+ {
3042
+ "epoch": 23.83,
3043
+ "learning_rate": 6.175499292341563e-05,
3044
+ "loss": 0.4875,
3045
+ "step": 1012000
3046
+ },
3047
+ {
3048
+ "epoch": 23.88,
3049
+ "learning_rate": 6.128322063217486e-05,
3050
+ "loss": 0.4935,
3051
+ "step": 1014000
3052
+ },
3053
+ {
3054
+ "epoch": 23.93,
3055
+ "learning_rate": 6.0811448340934105e-05,
3056
+ "loss": 0.4834,
3057
+ "step": 1016000
3058
+ },
3059
+ {
3060
+ "epoch": 23.98,
3061
+ "learning_rate": 6.033967604969334e-05,
3062
+ "loss": 0.4853,
3063
+ "step": 1018000
3064
+ },
3065
+ {
3066
+ "epoch": 24.02,
3067
+ "learning_rate": 5.9867903758452584e-05,
3068
+ "loss": 0.485,
3069
+ "step": 1020000
3070
+ },
3071
+ {
3072
+ "epoch": 24.07,
3073
+ "learning_rate": 5.9396131467211824e-05,
3074
+ "loss": 0.4821,
3075
+ "step": 1022000
3076
+ },
3077
+ {
3078
+ "epoch": 24.12,
3079
+ "learning_rate": 5.8924359175971057e-05,
3080
+ "loss": 0.4739,
3081
+ "step": 1024000
3082
+ },
3083
+ {
3084
+ "epoch": 24.16,
3085
+ "learning_rate": 5.8452586884730296e-05,
3086
+ "loss": 0.4824,
3087
+ "step": 1026000
3088
+ },
3089
+ {
3090
+ "epoch": 24.21,
3091
+ "learning_rate": 5.798081459348954e-05,
3092
+ "loss": 0.4823,
3093
+ "step": 1028000
3094
+ },
3095
+ {
3096
+ "epoch": 24.26,
3097
+ "learning_rate": 5.7509042302248775e-05,
3098
+ "loss": 0.4805,
3099
+ "step": 1030000
3100
+ },
3101
+ {
3102
+ "epoch": 24.31,
3103
+ "learning_rate": 5.7037270011008015e-05,
3104
+ "loss": 0.4727,
3105
+ "step": 1032000
3106
+ },
3107
+ {
3108
+ "epoch": 24.35,
3109
+ "learning_rate": 5.6565497719767254e-05,
3110
+ "loss": 0.4726,
3111
+ "step": 1034000
3112
+ },
3113
+ {
3114
+ "epoch": 24.4,
3115
+ "learning_rate": 5.609372542852649e-05,
3116
+ "loss": 0.4734,
3117
+ "step": 1036000
3118
+ },
3119
+ {
3120
+ "epoch": 24.45,
3121
+ "learning_rate": 5.562195313728573e-05,
3122
+ "loss": 0.4775,
3123
+ "step": 1038000
3124
+ },
3125
+ {
3126
+ "epoch": 24.49,
3127
+ "learning_rate": 5.515018084604497e-05,
3128
+ "loss": 0.4761,
3129
+ "step": 1040000
3130
+ },
3131
+ {
3132
+ "epoch": 24.54,
3133
+ "learning_rate": 5.467840855480421e-05,
3134
+ "loss": 0.4692,
3135
+ "step": 1042000
3136
+ },
3137
+ {
3138
+ "epoch": 24.59,
3139
+ "learning_rate": 5.4206636263563445e-05,
3140
+ "loss": 0.4826,
3141
+ "step": 1044000
3142
+ },
3143
+ {
3144
+ "epoch": 24.63,
3145
+ "learning_rate": 5.3734863972322684e-05,
3146
+ "loss": 0.4767,
3147
+ "step": 1046000
3148
+ },
3149
+ {
3150
+ "epoch": 24.68,
3151
+ "learning_rate": 5.326309168108193e-05,
3152
+ "loss": 0.4733,
3153
+ "step": 1048000
3154
+ },
3155
+ {
3156
+ "epoch": 24.73,
3157
+ "learning_rate": 5.2791319389841163e-05,
3158
+ "loss": 0.4772,
3159
+ "step": 1050000
3160
+ },
3161
+ {
3162
+ "epoch": 24.78,
3163
+ "learning_rate": 5.23195470986004e-05,
3164
+ "loss": 0.4758,
3165
+ "step": 1052000
3166
+ },
3167
+ {
3168
+ "epoch": 24.82,
3169
+ "learning_rate": 5.184777480735964e-05,
3170
+ "loss": 0.48,
3171
+ "step": 1054000
3172
+ },
3173
+ {
3174
+ "epoch": 24.87,
3175
+ "learning_rate": 5.137600251611889e-05,
3176
+ "loss": 0.477,
3177
+ "step": 1056000
3178
+ },
3179
+ {
3180
+ "epoch": 24.92,
3181
+ "learning_rate": 5.090423022487812e-05,
3182
+ "loss": 0.4715,
3183
+ "step": 1058000
3184
+ },
3185
+ {
3186
+ "epoch": 24.96,
3187
+ "learning_rate": 5.043245793363736e-05,
3188
+ "loss": 0.4696,
3189
+ "step": 1060000
3190
+ },
3191
+ {
3192
+ "epoch": 25.01,
3193
+ "learning_rate": 4.99606856423966e-05,
3194
+ "loss": 0.4682,
3195
+ "step": 1062000
3196
+ },
3197
+ {
3198
+ "epoch": 25.06,
3199
+ "learning_rate": 4.948891335115583e-05,
3200
+ "loss": 0.4725,
3201
+ "step": 1064000
3202
+ },
3203
+ {
3204
+ "epoch": 25.11,
3205
+ "learning_rate": 4.901714105991508e-05,
3206
+ "loss": 0.4655,
3207
+ "step": 1066000
3208
+ },
3209
+ {
3210
+ "epoch": 25.15,
3211
+ "learning_rate": 4.854536876867432e-05,
3212
+ "loss": 0.4629,
3213
+ "step": 1068000
3214
+ },
3215
+ {
3216
+ "epoch": 25.2,
3217
+ "learning_rate": 4.807359647743355e-05,
3218
+ "loss": 0.4654,
3219
+ "step": 1070000
3220
+ },
3221
+ {
3222
+ "epoch": 25.25,
3223
+ "learning_rate": 4.760182418619279e-05,
3224
+ "loss": 0.4663,
3225
+ "step": 1072000
3226
+ },
3227
+ {
3228
+ "epoch": 25.29,
3229
+ "learning_rate": 4.713005189495203e-05,
3230
+ "loss": 0.4682,
3231
+ "step": 1074000
3232
+ },
3233
+ {
3234
+ "epoch": 25.34,
3235
+ "learning_rate": 4.665827960371127e-05,
3236
+ "loss": 0.4647,
3237
+ "step": 1076000
3238
+ },
3239
+ {
3240
+ "epoch": 25.39,
3241
+ "learning_rate": 4.618650731247051e-05,
3242
+ "loss": 0.4722,
3243
+ "step": 1078000
3244
+ },
3245
+ {
3246
+ "epoch": 25.44,
3247
+ "learning_rate": 4.571473502122975e-05,
3248
+ "loss": 0.4604,
3249
+ "step": 1080000
3250
+ },
3251
+ {
3252
+ "epoch": 25.48,
3253
+ "learning_rate": 4.524296272998898e-05,
3254
+ "loss": 0.4585,
3255
+ "step": 1082000
3256
+ },
3257
+ {
3258
+ "epoch": 25.53,
3259
+ "learning_rate": 4.477119043874823e-05,
3260
+ "loss": 0.4619,
3261
+ "step": 1084000
3262
+ },
3263
+ {
3264
+ "epoch": 25.58,
3265
+ "learning_rate": 4.429941814750747e-05,
3266
+ "loss": 0.4661,
3267
+ "step": 1086000
3268
+ },
3269
+ {
3270
+ "epoch": 25.62,
3271
+ "learning_rate": 4.382764585626671e-05,
3272
+ "loss": 0.4724,
3273
+ "step": 1088000
3274
+ },
3275
+ {
3276
+ "epoch": 25.67,
3277
+ "learning_rate": 4.335587356502594e-05,
3278
+ "loss": 0.4667,
3279
+ "step": 1090000
3280
+ },
3281
+ {
3282
+ "epoch": 25.72,
3283
+ "learning_rate": 4.288410127378518e-05,
3284
+ "loss": 0.4692,
3285
+ "step": 1092000
3286
+ },
3287
+ {
3288
+ "epoch": 25.77,
3289
+ "learning_rate": 4.2412328982544426e-05,
3290
+ "loss": 0.4624,
3291
+ "step": 1094000
3292
+ },
3293
+ {
3294
+ "epoch": 25.81,
3295
+ "learning_rate": 4.194055669130366e-05,
3296
+ "loss": 0.4587,
3297
+ "step": 1096000
3298
+ },
3299
+ {
3300
+ "epoch": 25.86,
3301
+ "learning_rate": 4.14687844000629e-05,
3302
+ "loss": 0.4604,
3303
+ "step": 1098000
3304
+ },
3305
+ {
3306
+ "epoch": 25.91,
3307
+ "learning_rate": 4.099701210882214e-05,
3308
+ "loss": 0.4718,
3309
+ "step": 1100000
3310
+ },
3311
+ {
3312
+ "epoch": 25.95,
3313
+ "learning_rate": 4.052523981758137e-05,
3314
+ "loss": 0.4666,
3315
+ "step": 1102000
3316
+ },
3317
+ {
3318
+ "epoch": 26.0,
3319
+ "learning_rate": 4.005346752634062e-05,
3320
+ "loss": 0.4616,
3321
+ "step": 1104000
3322
+ },
3323
+ {
3324
+ "epoch": 26.05,
3325
+ "learning_rate": 3.9581695235099856e-05,
3326
+ "loss": 0.456,
3327
+ "step": 1106000
3328
+ },
3329
+ {
3330
+ "epoch": 26.1,
3331
+ "learning_rate": 3.9109922943859096e-05,
3332
+ "loss": 0.4573,
3333
+ "step": 1108000
3334
+ },
3335
+ {
3336
+ "epoch": 26.14,
3337
+ "learning_rate": 3.863815065261833e-05,
3338
+ "loss": 0.4522,
3339
+ "step": 1110000
3340
+ },
3341
+ {
3342
+ "epoch": 26.19,
3343
+ "learning_rate": 3.8166378361377575e-05,
3344
+ "loss": 0.4604,
3345
+ "step": 1112000
3346
+ },
3347
+ {
3348
+ "epoch": 26.24,
3349
+ "learning_rate": 3.7694606070136815e-05,
3350
+ "loss": 0.4539,
3351
+ "step": 1114000
3352
+ },
3353
+ {
3354
+ "epoch": 26.28,
3355
+ "learning_rate": 3.722283377889605e-05,
3356
+ "loss": 0.4485,
3357
+ "step": 1116000
3358
+ },
3359
+ {
3360
+ "epoch": 26.33,
3361
+ "learning_rate": 3.675106148765529e-05,
3362
+ "loss": 0.4596,
3363
+ "step": 1118000
3364
+ },
3365
+ {
3366
+ "epoch": 26.38,
3367
+ "learning_rate": 3.6279289196414526e-05,
3368
+ "loss": 0.4487,
3369
+ "step": 1120000
3370
+ },
3371
+ {
3372
+ "epoch": 26.42,
3373
+ "learning_rate": 3.5807516905173766e-05,
3374
+ "loss": 0.459,
3375
+ "step": 1122000
3376
+ },
3377
+ {
3378
+ "epoch": 26.47,
3379
+ "learning_rate": 3.5335744613933005e-05,
3380
+ "loss": 0.4609,
3381
+ "step": 1124000
3382
+ },
3383
+ {
3384
+ "epoch": 26.52,
3385
+ "learning_rate": 3.4863972322692245e-05,
3386
+ "loss": 0.4549,
3387
+ "step": 1126000
3388
+ },
3389
+ {
3390
+ "epoch": 26.57,
3391
+ "learning_rate": 3.4392200031451484e-05,
3392
+ "loss": 0.4553,
3393
+ "step": 1128000
3394
+ },
3395
+ {
3396
+ "epoch": 26.61,
3397
+ "learning_rate": 3.3920427740210724e-05,
3398
+ "loss": 0.4552,
3399
+ "step": 1130000
3400
+ },
3401
+ {
3402
+ "epoch": 26.66,
3403
+ "learning_rate": 3.344865544896996e-05,
3404
+ "loss": 0.4534,
3405
+ "step": 1132000
3406
+ },
3407
+ {
3408
+ "epoch": 26.71,
3409
+ "learning_rate": 3.29768831577292e-05,
3410
+ "loss": 0.4522,
3411
+ "step": 1134000
3412
+ },
3413
+ {
3414
+ "epoch": 26.75,
3415
+ "learning_rate": 3.2505110866488436e-05,
3416
+ "loss": 0.4544,
3417
+ "step": 1136000
3418
+ },
3419
+ {
3420
+ "epoch": 26.8,
3421
+ "learning_rate": 3.203333857524768e-05,
3422
+ "loss": 0.458,
3423
+ "step": 1138000
3424
+ },
3425
+ {
3426
+ "epoch": 26.85,
3427
+ "learning_rate": 3.1561566284006915e-05,
3428
+ "loss": 0.4537,
3429
+ "step": 1140000
3430
+ },
3431
+ {
3432
+ "epoch": 26.9,
3433
+ "learning_rate": 3.1089793992766154e-05,
3434
+ "loss": 0.4581,
3435
+ "step": 1142000
3436
+ },
3437
+ {
3438
+ "epoch": 26.94,
3439
+ "learning_rate": 3.0618021701525394e-05,
3440
+ "loss": 0.4525,
3441
+ "step": 1144000
3442
+ },
3443
+ {
3444
+ "epoch": 26.99,
3445
+ "learning_rate": 3.014624941028463e-05,
3446
+ "loss": 0.4562,
3447
+ "step": 1146000
3448
+ },
3449
+ {
3450
+ "epoch": 27.04,
3451
+ "learning_rate": 2.9674477119043873e-05,
3452
+ "loss": 0.4516,
3453
+ "step": 1148000
3454
+ },
3455
+ {
3456
+ "epoch": 27.08,
3457
+ "learning_rate": 2.920270482780311e-05,
3458
+ "loss": 0.4472,
3459
+ "step": 1150000
3460
+ },
3461
+ {
3462
+ "epoch": 27.13,
3463
+ "learning_rate": 2.8730932536562352e-05,
3464
+ "loss": 0.4507,
3465
+ "step": 1152000
3466
+ },
3467
+ {
3468
+ "epoch": 27.18,
3469
+ "learning_rate": 2.8259160245321588e-05,
3470
+ "loss": 0.4421,
3471
+ "step": 1154000
3472
+ },
3473
+ {
3474
+ "epoch": 27.23,
3475
+ "learning_rate": 2.778738795408083e-05,
3476
+ "loss": 0.4487,
3477
+ "step": 1156000
3478
+ },
3479
+ {
3480
+ "epoch": 27.27,
3481
+ "learning_rate": 2.7315615662840067e-05,
3482
+ "loss": 0.4528,
3483
+ "step": 1158000
3484
+ },
3485
+ {
3486
+ "epoch": 27.32,
3487
+ "learning_rate": 2.6843843371599303e-05,
3488
+ "loss": 0.4483,
3489
+ "step": 1160000
3490
+ },
3491
+ {
3492
+ "epoch": 27.37,
3493
+ "learning_rate": 2.6372071080358546e-05,
3494
+ "loss": 0.45,
3495
+ "step": 1162000
3496
+ },
3497
+ {
3498
+ "epoch": 27.41,
3499
+ "learning_rate": 2.5900298789117782e-05,
3500
+ "loss": 0.4517,
3501
+ "step": 1164000
3502
+ },
3503
+ {
3504
+ "epoch": 27.46,
3505
+ "learning_rate": 2.5428526497877022e-05,
3506
+ "loss": 0.4444,
3507
+ "step": 1166000
3508
+ },
3509
+ {
3510
+ "epoch": 27.51,
3511
+ "learning_rate": 2.495675420663626e-05,
3512
+ "loss": 0.4457,
3513
+ "step": 1168000
3514
+ },
3515
+ {
3516
+ "epoch": 27.56,
3517
+ "learning_rate": 2.44849819153955e-05,
3518
+ "loss": 0.4471,
3519
+ "step": 1170000
3520
+ },
3521
+ {
3522
+ "epoch": 27.6,
3523
+ "learning_rate": 2.401320962415474e-05,
3524
+ "loss": 0.4455,
3525
+ "step": 1172000
3526
+ },
3527
+ {
3528
+ "epoch": 27.65,
3529
+ "learning_rate": 2.3541437332913976e-05,
3530
+ "loss": 0.4466,
3531
+ "step": 1174000
3532
+ },
3533
+ {
3534
+ "epoch": 27.7,
3535
+ "learning_rate": 2.3069665041673216e-05,
3536
+ "loss": 0.4421,
3537
+ "step": 1176000
3538
+ },
3539
+ {
3540
+ "epoch": 27.74,
3541
+ "learning_rate": 2.2597892750432455e-05,
3542
+ "loss": 0.446,
3543
+ "step": 1178000
3544
+ },
3545
+ {
3546
+ "epoch": 27.79,
3547
+ "learning_rate": 2.2126120459191695e-05,
3548
+ "loss": 0.4441,
3549
+ "step": 1180000
3550
+ },
3551
+ {
3552
+ "epoch": 27.84,
3553
+ "learning_rate": 2.1654348167950935e-05,
3554
+ "loss": 0.4452,
3555
+ "step": 1182000
3556
+ },
3557
+ {
3558
+ "epoch": 27.89,
3559
+ "learning_rate": 2.1182575876710174e-05,
3560
+ "loss": 0.4418,
3561
+ "step": 1184000
3562
+ },
3563
+ {
3564
+ "epoch": 27.93,
3565
+ "learning_rate": 2.071080358546941e-05,
3566
+ "loss": 0.4396,
3567
+ "step": 1186000
3568
+ },
3569
+ {
3570
+ "epoch": 27.98,
3571
+ "learning_rate": 2.0239031294228653e-05,
3572
+ "loss": 0.4477,
3573
+ "step": 1188000
3574
+ },
3575
+ {
3576
+ "epoch": 28.03,
3577
+ "learning_rate": 1.976725900298789e-05,
3578
+ "loss": 0.4435,
3579
+ "step": 1190000
3580
+ },
3581
+ {
3582
+ "epoch": 28.07,
3583
+ "learning_rate": 1.9295486711747125e-05,
3584
+ "loss": 0.4379,
3585
+ "step": 1192000
3586
+ },
3587
+ {
3588
+ "epoch": 28.12,
3589
+ "learning_rate": 1.8823714420506368e-05,
3590
+ "loss": 0.4442,
3591
+ "step": 1194000
3592
+ },
3593
+ {
3594
+ "epoch": 28.17,
3595
+ "learning_rate": 1.8351942129265604e-05,
3596
+ "loss": 0.4329,
3597
+ "step": 1196000
3598
+ },
3599
+ {
3600
+ "epoch": 28.21,
3601
+ "learning_rate": 1.7880169838024844e-05,
3602
+ "loss": 0.4368,
3603
+ "step": 1198000
3604
+ },
3605
+ {
3606
+ "epoch": 28.26,
3607
+ "learning_rate": 1.7408397546784083e-05,
3608
+ "loss": 0.4427,
3609
+ "step": 1200000
3610
+ },
3611
+ {
3612
+ "epoch": 28.31,
3613
+ "learning_rate": 1.6936625255543323e-05,
3614
+ "loss": 0.4426,
3615
+ "step": 1202000
3616
+ },
3617
+ {
3618
+ "epoch": 28.36,
3619
+ "learning_rate": 1.6464852964302562e-05,
3620
+ "loss": 0.4374,
3621
+ "step": 1204000
3622
+ },
3623
+ {
3624
+ "epoch": 28.4,
3625
+ "learning_rate": 1.5993080673061802e-05,
3626
+ "loss": 0.4372,
3627
+ "step": 1206000
3628
+ },
3629
+ {
3630
+ "epoch": 28.45,
3631
+ "learning_rate": 1.552130838182104e-05,
3632
+ "loss": 0.4365,
3633
+ "step": 1208000
3634
+ },
3635
+ {
3636
+ "epoch": 28.5,
3637
+ "learning_rate": 1.504953609058028e-05,
3638
+ "loss": 0.4329,
3639
+ "step": 1210000
3640
+ },
3641
+ {
3642
+ "epoch": 28.54,
3643
+ "learning_rate": 1.4577763799339517e-05,
3644
+ "loss": 0.4397,
3645
+ "step": 1212000
3646
+ },
3647
+ {
3648
+ "epoch": 28.59,
3649
+ "learning_rate": 1.4105991508098757e-05,
3650
+ "loss": 0.4355,
3651
+ "step": 1214000
3652
+ },
3653
+ {
3654
+ "epoch": 28.64,
3655
+ "learning_rate": 1.3634219216857994e-05,
3656
+ "loss": 0.4386,
3657
+ "step": 1216000
3658
+ },
3659
+ {
3660
+ "epoch": 28.69,
3661
+ "learning_rate": 1.3162446925617234e-05,
3662
+ "loss": 0.4369,
3663
+ "step": 1218000
3664
+ },
3665
+ {
3666
+ "epoch": 28.73,
3667
+ "learning_rate": 1.2690674634376474e-05,
3668
+ "loss": 0.441,
3669
+ "step": 1220000
3670
+ },
3671
+ {
3672
+ "epoch": 28.78,
3673
+ "learning_rate": 1.2218902343135713e-05,
3674
+ "loss": 0.4383,
3675
+ "step": 1222000
3676
+ },
3677
+ {
3678
+ "epoch": 28.83,
3679
+ "learning_rate": 1.1747130051894953e-05,
3680
+ "loss": 0.4338,
3681
+ "step": 1224000
3682
+ },
3683
+ {
3684
+ "epoch": 28.87,
3685
+ "learning_rate": 1.1275357760654189e-05,
3686
+ "loss": 0.4373,
3687
+ "step": 1226000
3688
+ },
3689
+ {
3690
+ "epoch": 28.92,
3691
+ "learning_rate": 1.0803585469413428e-05,
3692
+ "loss": 0.4392,
3693
+ "step": 1228000
3694
+ },
3695
+ {
3696
+ "epoch": 28.97,
3697
+ "learning_rate": 1.0331813178172668e-05,
3698
+ "loss": 0.4401,
3699
+ "step": 1230000
3700
+ }
3701
+ ],
3702
+ "logging_steps": 2000,
3703
+ "max_steps": 1273800,
3704
+ "num_input_tokens_seen": 0,
3705
+ "num_train_epochs": 30,
3706
+ "save_steps": 500,
3707
+ "total_flos": 2.4919775353735186e+21,
3708
+ "train_batch_size": 2,
3709
+ "trial_name": null,
3710
+ "trial_params": null
3711
+ }
checkpoint-1231340/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cecbe2308480a0db70978cbda8460cbfae2844e623308cee5fa4aa246d04a482
3
+ size 4728
checkpoint-1231340/vocab.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "\"": 1,
3
+ ">": 2,
4
+ "[PAD]": 95,
5
+ "[UNK]": 94,
6
+ "\\": 3,
7
+ "a": 4,
8
+ "b": 5,
9
+ "c": 6,
10
+ "d": 7,
11
+ "e": 8,
12
+ "f": 9,
13
+ "g": 10,
14
+ "h": 11,
15
+ "i": 12,
16
+ "j": 13,
17
+ "k": 14,
18
+ "l": 15,
19
+ "m": 16,
20
+ "n": 17,
21
+ "o": 18,
22
+ "p": 19,
23
+ "q": 20,
24
+ "r": 21,
25
+ "s": 22,
26
+ "t": 23,
27
+ "u": 24,
28
+ "v": 25,
29
+ "w": 26,
30
+ "x": 27,
31
+ "y": 28,
32
+ "z": 29,
33
+ "|": 0,
34
+ "ç": 30,
35
+ "é": 31,
36
+ "õ": 32,
37
+ "ş": 33,
38
+ "ơ": 34,
39
+ "ư": 35,
40
+ "̀": 36,
41
+ "υ": 37,
42
+ "а": 38,
43
+ "в": 39,
44
+ "г": 40,
45
+ "д": 41,
46
+ "е": 42,
47
+ "з": 43,
48
+ "к": 44,
49
+ "л": 45,
50
+ "н": 46,
51
+ "п": 47,
52
+ "с": 48,
53
+ "ф": 49,
54
+ "ч": 50,
55
+ "ả": 51,
56
+ "ậ": 52,
57
+ "ồ": 53,
58
+ "ờ": 54,
59
+ "​": 55,
60
+ "’": 56,
61
+ "⁄": 57,
62
+ "れ": 58,
63
+ "ㅑ": 59,
64
+ "ㅛ": 60,
65
+ "각": 61,
66
+ "국": 62,
67
+ "기": 63,
68
+ "는": 64,
69
+ "데": 65,
70
+ "도": 66,
71
+ "드": 67,
72
+ "들": 68,
73
+ "랜": 69,
74
+ "로": 70,
75
+ "를": 71,
76
+ "만": 72,
77
+ "많": 73,
78
+ "부": 74,
79
+ "브": 75,
80
+ "생": 76,
81
+ "솔": 77,
82
+ "스": 78,
83
+ "앤": 79,
84
+ "에": 80,
85
+ "여": 81,
86
+ "온": 82,
87
+ "원": 83,
88
+ "전": 84,
89
+ "제": 85,
90
+ "지": 86,
91
+ "천": 87,
92
+ "초": 88,
93
+ "터": 89,
94
+ "트": 90,
95
+ "한": 91,
96
+ "행": 92,
97
+ "�": 93
98
+ }
checkpoint-1273800/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "</s>": 97,
3
+ "<s>": 96
4
+ }
checkpoint-1273800/config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
3
+ "activation_dropout": 0.0,
4
+ "adapter_attn_dim": null,
5
+ "adapter_kernel_size": 3,
6
+ "adapter_stride": 2,
7
+ "add_adapter": false,
8
+ "apply_spec_augment": true,
9
+ "architectures": [
10
+ "Wav2Vec2ForCTC"
11
+ ],
12
+ "attention_dropout": 0.05,
13
+ "bos_token_id": 1,
14
+ "classifier_proj_size": 256,
15
+ "codevector_dim": 768,
16
+ "contrastive_logits_temperature": 0.1,
17
+ "conv_bias": true,
18
+ "conv_dim": [
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512,
25
+ 512
26
+ ],
27
+ "conv_kernel": [
28
+ 10,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 3,
33
+ 2,
34
+ 2
35
+ ],
36
+ "conv_stride": [
37
+ 5,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2,
43
+ 2
44
+ ],
45
+ "ctc_loss_reduction": "mean",
46
+ "ctc_zero_infinity": true,
47
+ "diversity_loss_weight": 0.1,
48
+ "do_stable_layer_norm": true,
49
+ "eos_token_id": 2,
50
+ "feat_extract_activation": "gelu",
51
+ "feat_extract_dropout": 0.0,
52
+ "feat_extract_norm": "layer",
53
+ "feat_proj_dropout": 0.05,
54
+ "feat_quantizer_dropout": 0.0,
55
+ "final_dropout": 0.0,
56
+ "gradient_checkpointing": false,
57
+ "hidden_act": "gelu",
58
+ "hidden_dropout": 0.05,
59
+ "hidden_size": 1024,
60
+ "initializer_range": 0.02,
61
+ "intermediate_size": 4096,
62
+ "layer_norm_eps": 1e-05,
63
+ "layerdrop": 0.05,
64
+ "mask_channel_length": 10,
65
+ "mask_channel_min_space": 1,
66
+ "mask_channel_other": 0.0,
67
+ "mask_channel_prob": 0.0,
68
+ "mask_channel_selection": "static",
69
+ "mask_feature_length": 10,
70
+ "mask_feature_min_masks": 0,
71
+ "mask_feature_prob": 0.0,
72
+ "mask_time_length": 10,
73
+ "mask_time_min_masks": 2,
74
+ "mask_time_min_space": 1,
75
+ "mask_time_other": 0.0,
76
+ "mask_time_prob": 0.05,
77
+ "mask_time_selection": "static",
78
+ "model_type": "wav2vec2",
79
+ "num_adapter_layers": 3,
80
+ "num_attention_heads": 16,
81
+ "num_codevector_groups": 2,
82
+ "num_codevectors_per_group": 320,
83
+ "num_conv_pos_embedding_groups": 16,
84
+ "num_conv_pos_embeddings": 128,
85
+ "num_feat_extract_layers": 7,
86
+ "num_hidden_layers": 24,
87
+ "num_negatives": 100,
88
+ "output_hidden_size": 1024,
89
+ "pad_token_id": 95,
90
+ "proj_codevector_dim": 768,
91
+ "tdnn_dilation": [
92
+ 1,
93
+ 2,
94
+ 3,
95
+ 1,
96
+ 1
97
+ ],
98
+ "tdnn_dim": [
99
+ 512,
100
+ 512,
101
+ 512,
102
+ 512,
103
+ 1500
104
+ ],
105
+ "tdnn_kernel": [
106
+ 5,
107
+ 3,
108
+ 3,
109
+ 1,
110
+ 1
111
+ ],
112
+ "torch_dtype": "float32",
113
+ "transformers_version": "4.36.2",
114
+ "use_weighted_layer_sum": false,
115
+ "vocab_size": 98,
116
+ "xvector_output_dim": 512
117
+ }
checkpoint-1273800/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dafdcdb0e309cbeeae1382c1347fbaea4b82a930c87f2336462c88682bb1ef7c
3
+ size 1262209280
checkpoint-1273800/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb5cb786d92df412b236a751e25194c39049ee2ba8f9c72ab5847ee16671f3fb
3
+ size 2490963382
checkpoint-1273800/preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
checkpoint-1273800/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f58ce107374ce8151a9d421ba1602c2a828691bff4b25dfe6de1d4d0b89b3a4b
3
+ size 15024
checkpoint-1273800/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91993878a22dd83d10776cc85aa581217d61db79c11b60ad294eb6ce8556640e
3
+ size 15024
checkpoint-1273800/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56199292207511ee71b90351a1fbc5ec0f0869ea7f2b15efeb039548acd8efd8
3
+ size 15024
checkpoint-1273800/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aba8089adbae41e4f8438dbdd4b8edce147ea6e8d9fd666919cd8f907f1b7ec9
3
+ size 15024
checkpoint-1273800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd13603fc888497fe8d1389c056f64e4dbbfc9af902976a83acc427ca15188b8
3
+ size 1064
checkpoint-1273800/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": true,
19
+ "normalized": false,
20
+ "rstrip": true,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "[UNK]",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": true,
28
+ "single_word": false
29
+ }
30
+ }
checkpoint-1273800/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "94": {
4
+ "content": "[UNK]",
5
+ "lstrip": true,
6
+ "normalized": false,
7
+ "rstrip": true,
8
+ "single_word": false,
9
+ "special": false
10
+ },
11
+ "95": {
12
+ "content": "[PAD]",
13
+ "lstrip": true,
14
+ "normalized": false,
15
+ "rstrip": true,
16
+ "single_word": false,
17
+ "special": false
18
+ },
19
+ "96": {
20
+ "content": "<s>",
21
+ "lstrip": false,
22
+ "normalized": true,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "97": {
28
+ "content": "</s>",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<s>",
37
+ "clean_up_tokenization_spaces": true,
38
+ "config": null,
39
+ "do_lower_case": false,
40
+ "eos_token": "</s>",
41
+ "model_max_length": 1000000000000000019884624838656,
42
+ "pad_token": "[PAD]",
43
+ "processor_class": "Wav2Vec2Processor",
44
+ "replace_word_delimiter_char": " ",
45
+ "target_lang": null,
46
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
47
+ "tokenizer_type": "wav2vec2",
48
+ "trust_remote_code": false,
49
+ "unk_token": "[UNK]",
50
+ "word_delimiter_token": "|"
51
+ }
checkpoint-1273800/trainer_state.json ADDED
@@ -0,0 +1,3837 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 30.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1273800,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.05,
13
+ "learning_rate": 0.0003,
14
+ "loss": 3.0982,
15
+ "step": 2000
16
+ },
17
+ {
18
+ "epoch": 0.09,
19
+ "learning_rate": 0.0002995282277087592,
20
+ "loss": 1.1999,
21
+ "step": 4000
22
+ },
23
+ {
24
+ "epoch": 0.14,
25
+ "learning_rate": 0.00029905645541751844,
26
+ "loss": 1.0995,
27
+ "step": 6000
28
+ },
29
+ {
30
+ "epoch": 0.19,
31
+ "learning_rate": 0.0002985846831262777,
32
+ "loss": 1.0716,
33
+ "step": 8000
34
+ },
35
+ {
36
+ "epoch": 0.24,
37
+ "learning_rate": 0.0002981129108350369,
38
+ "loss": 1.0664,
39
+ "step": 10000
40
+ },
41
+ {
42
+ "epoch": 0.28,
43
+ "learning_rate": 0.0002976411385437962,
44
+ "loss": 1.0223,
45
+ "step": 12000
46
+ },
47
+ {
48
+ "epoch": 0.33,
49
+ "learning_rate": 0.00029716936625255537,
50
+ "loss": 0.9955,
51
+ "step": 14000
52
+ },
53
+ {
54
+ "epoch": 0.38,
55
+ "learning_rate": 0.00029669759396131466,
56
+ "loss": 0.9925,
57
+ "step": 16000
58
+ },
59
+ {
60
+ "epoch": 0.42,
61
+ "learning_rate": 0.0002962258216700739,
62
+ "loss": 0.9951,
63
+ "step": 18000
64
+ },
65
+ {
66
+ "epoch": 0.47,
67
+ "learning_rate": 0.0002957540493788331,
68
+ "loss": 0.9811,
69
+ "step": 20000
70
+ },
71
+ {
72
+ "epoch": 0.52,
73
+ "learning_rate": 0.00029528227708759236,
74
+ "loss": 0.9727,
75
+ "step": 22000
76
+ },
77
+ {
78
+ "epoch": 0.57,
79
+ "learning_rate": 0.0002948105047963516,
80
+ "loss": 0.9655,
81
+ "step": 24000
82
+ },
83
+ {
84
+ "epoch": 0.61,
85
+ "learning_rate": 0.0002943387325051109,
86
+ "loss": 0.959,
87
+ "step": 26000
88
+ },
89
+ {
90
+ "epoch": 0.66,
91
+ "learning_rate": 0.00029386696021387005,
92
+ "loss": 0.9591,
93
+ "step": 28000
94
+ },
95
+ {
96
+ "epoch": 0.71,
97
+ "learning_rate": 0.00029339518792262934,
98
+ "loss": 0.9455,
99
+ "step": 30000
100
+ },
101
+ {
102
+ "epoch": 0.75,
103
+ "learning_rate": 0.0002929234156313886,
104
+ "loss": 0.9425,
105
+ "step": 32000
106
+ },
107
+ {
108
+ "epoch": 0.8,
109
+ "learning_rate": 0.0002924516433401478,
110
+ "loss": 0.9373,
111
+ "step": 34000
112
+ },
113
+ {
114
+ "epoch": 0.85,
115
+ "learning_rate": 0.00029197987104890704,
116
+ "loss": 0.9341,
117
+ "step": 36000
118
+ },
119
+ {
120
+ "epoch": 0.89,
121
+ "learning_rate": 0.00029150809875766627,
122
+ "loss": 0.93,
123
+ "step": 38000
124
+ },
125
+ {
126
+ "epoch": 0.94,
127
+ "learning_rate": 0.0002910363264664255,
128
+ "loss": 0.9184,
129
+ "step": 40000
130
+ },
131
+ {
132
+ "epoch": 0.99,
133
+ "learning_rate": 0.00029056455417518474,
134
+ "loss": 0.9222,
135
+ "step": 42000
136
+ },
137
+ {
138
+ "epoch": 1.04,
139
+ "learning_rate": 0.000290092781883944,
140
+ "loss": 0.8941,
141
+ "step": 44000
142
+ },
143
+ {
144
+ "epoch": 1.08,
145
+ "learning_rate": 0.0002896210095927032,
146
+ "loss": 0.8868,
147
+ "step": 46000
148
+ },
149
+ {
150
+ "epoch": 1.13,
151
+ "learning_rate": 0.0002891492373014625,
152
+ "loss": 0.8894,
153
+ "step": 48000
154
+ },
155
+ {
156
+ "epoch": 1.18,
157
+ "learning_rate": 0.0002886774650102217,
158
+ "loss": 0.8799,
159
+ "step": 50000
160
+ },
161
+ {
162
+ "epoch": 1.22,
163
+ "learning_rate": 0.00028820569271898095,
164
+ "loss": 0.8781,
165
+ "step": 52000
166
+ },
167
+ {
168
+ "epoch": 1.27,
169
+ "learning_rate": 0.0002877339204277402,
170
+ "loss": 0.8853,
171
+ "step": 54000
172
+ },
173
+ {
174
+ "epoch": 1.32,
175
+ "learning_rate": 0.0002872621481364994,
176
+ "loss": 0.8829,
177
+ "step": 56000
178
+ },
179
+ {
180
+ "epoch": 1.37,
181
+ "learning_rate": 0.00028679037584525865,
182
+ "loss": 0.8747,
183
+ "step": 58000
184
+ },
185
+ {
186
+ "epoch": 1.41,
187
+ "learning_rate": 0.0002863186035540179,
188
+ "loss": 0.8806,
189
+ "step": 60000
190
+ },
191
+ {
192
+ "epoch": 1.46,
193
+ "learning_rate": 0.00028584683126277717,
194
+ "loss": 0.8868,
195
+ "step": 62000
196
+ },
197
+ {
198
+ "epoch": 1.51,
199
+ "learning_rate": 0.00028537505897153635,
200
+ "loss": 0.8806,
201
+ "step": 64000
202
+ },
203
+ {
204
+ "epoch": 1.55,
205
+ "learning_rate": 0.00028490328668029564,
206
+ "loss": 0.8771,
207
+ "step": 66000
208
+ },
209
+ {
210
+ "epoch": 1.6,
211
+ "learning_rate": 0.00028443151438905487,
212
+ "loss": 0.8648,
213
+ "step": 68000
214
+ },
215
+ {
216
+ "epoch": 1.65,
217
+ "learning_rate": 0.0002839597420978141,
218
+ "loss": 0.8747,
219
+ "step": 70000
220
+ },
221
+ {
222
+ "epoch": 1.7,
223
+ "learning_rate": 0.00028348796980657334,
224
+ "loss": 0.8719,
225
+ "step": 72000
226
+ },
227
+ {
228
+ "epoch": 1.74,
229
+ "learning_rate": 0.00028301619751533257,
230
+ "loss": 0.8641,
231
+ "step": 74000
232
+ },
233
+ {
234
+ "epoch": 1.79,
235
+ "learning_rate": 0.0002825444252240918,
236
+ "loss": 0.8555,
237
+ "step": 76000
238
+ },
239
+ {
240
+ "epoch": 1.84,
241
+ "learning_rate": 0.00028207265293285103,
242
+ "loss": 0.8583,
243
+ "step": 78000
244
+ },
245
+ {
246
+ "epoch": 1.88,
247
+ "learning_rate": 0.0002816008806416103,
248
+ "loss": 0.8624,
249
+ "step": 80000
250
+ },
251
+ {
252
+ "epoch": 1.93,
253
+ "learning_rate": 0.0002811291083503695,
254
+ "loss": 0.8578,
255
+ "step": 82000
256
+ },
257
+ {
258
+ "epoch": 1.98,
259
+ "learning_rate": 0.0002806573360591288,
260
+ "loss": 0.8474,
261
+ "step": 84000
262
+ },
263
+ {
264
+ "epoch": 2.03,
265
+ "learning_rate": 0.000280185563767888,
266
+ "loss": 0.8366,
267
+ "step": 86000
268
+ },
269
+ {
270
+ "epoch": 2.07,
271
+ "learning_rate": 0.00027971379147664725,
272
+ "loss": 0.8226,
273
+ "step": 88000
274
+ },
275
+ {
276
+ "epoch": 2.12,
277
+ "learning_rate": 0.0002792420191854065,
278
+ "loss": 0.8267,
279
+ "step": 90000
280
+ },
281
+ {
282
+ "epoch": 2.17,
283
+ "learning_rate": 0.0002787702468941657,
284
+ "loss": 0.8231,
285
+ "step": 92000
286
+ },
287
+ {
288
+ "epoch": 2.21,
289
+ "learning_rate": 0.00027829847460292495,
290
+ "loss": 0.8284,
291
+ "step": 94000
292
+ },
293
+ {
294
+ "epoch": 2.26,
295
+ "learning_rate": 0.0002778267023116842,
296
+ "loss": 0.8153,
297
+ "step": 96000
298
+ },
299
+ {
300
+ "epoch": 2.31,
301
+ "learning_rate": 0.00027735493002044347,
302
+ "loss": 0.8241,
303
+ "step": 98000
304
+ },
305
+ {
306
+ "epoch": 2.36,
307
+ "learning_rate": 0.00027688315772920265,
308
+ "loss": 0.8246,
309
+ "step": 100000
310
+ },
311
+ {
312
+ "epoch": 2.4,
313
+ "learning_rate": 0.00027641138543796194,
314
+ "loss": 0.8219,
315
+ "step": 102000
316
+ },
317
+ {
318
+ "epoch": 2.45,
319
+ "learning_rate": 0.00027593961314672117,
320
+ "loss": 0.8244,
321
+ "step": 104000
322
+ },
323
+ {
324
+ "epoch": 2.5,
325
+ "learning_rate": 0.0002754678408554804,
326
+ "loss": 0.8201,
327
+ "step": 106000
328
+ },
329
+ {
330
+ "epoch": 2.54,
331
+ "learning_rate": 0.00027499606856423963,
332
+ "loss": 0.8193,
333
+ "step": 108000
334
+ },
335
+ {
336
+ "epoch": 2.59,
337
+ "learning_rate": 0.00027452429627299887,
338
+ "loss": 0.8225,
339
+ "step": 110000
340
+ },
341
+ {
342
+ "epoch": 2.64,
343
+ "learning_rate": 0.0002740525239817581,
344
+ "loss": 0.8105,
345
+ "step": 112000
346
+ },
347
+ {
348
+ "epoch": 2.68,
349
+ "learning_rate": 0.00027358075169051733,
350
+ "loss": 0.82,
351
+ "step": 114000
352
+ },
353
+ {
354
+ "epoch": 2.73,
355
+ "learning_rate": 0.0002731089793992766,
356
+ "loss": 0.8191,
357
+ "step": 116000
358
+ },
359
+ {
360
+ "epoch": 2.78,
361
+ "learning_rate": 0.00027263720710803585,
362
+ "loss": 0.8102,
363
+ "step": 118000
364
+ },
365
+ {
366
+ "epoch": 2.83,
367
+ "learning_rate": 0.0002721654348167951,
368
+ "loss": 0.8189,
369
+ "step": 120000
370
+ },
371
+ {
372
+ "epoch": 2.87,
373
+ "learning_rate": 0.0002716936625255543,
374
+ "loss": 0.8128,
375
+ "step": 122000
376
+ },
377
+ {
378
+ "epoch": 2.92,
379
+ "learning_rate": 0.00027122189023431355,
380
+ "loss": 0.815,
381
+ "step": 124000
382
+ },
383
+ {
384
+ "epoch": 2.97,
385
+ "learning_rate": 0.0002707501179430728,
386
+ "loss": 0.8123,
387
+ "step": 126000
388
+ },
389
+ {
390
+ "epoch": 3.01,
391
+ "learning_rate": 0.000270278345651832,
392
+ "loss": 0.7986,
393
+ "step": 128000
394
+ },
395
+ {
396
+ "epoch": 3.06,
397
+ "learning_rate": 0.00026980657336059125,
398
+ "loss": 0.7701,
399
+ "step": 130000
400
+ },
401
+ {
402
+ "epoch": 3.11,
403
+ "learning_rate": 0.0002693348010693505,
404
+ "loss": 0.7819,
405
+ "step": 132000
406
+ },
407
+ {
408
+ "epoch": 3.16,
409
+ "learning_rate": 0.00026886302877810977,
410
+ "loss": 0.7826,
411
+ "step": 134000
412
+ },
413
+ {
414
+ "epoch": 3.2,
415
+ "learning_rate": 0.000268391256486869,
416
+ "loss": 0.7868,
417
+ "step": 136000
418
+ },
419
+ {
420
+ "epoch": 3.25,
421
+ "learning_rate": 0.00026791948419562823,
422
+ "loss": 0.7843,
423
+ "step": 138000
424
+ },
425
+ {
426
+ "epoch": 3.3,
427
+ "learning_rate": 0.00026744771190438747,
428
+ "loss": 0.7841,
429
+ "step": 140000
430
+ },
431
+ {
432
+ "epoch": 3.34,
433
+ "learning_rate": 0.0002669759396131467,
434
+ "loss": 0.7845,
435
+ "step": 142000
436
+ },
437
+ {
438
+ "epoch": 3.39,
439
+ "learning_rate": 0.00026650416732190593,
440
+ "loss": 0.7859,
441
+ "step": 144000
442
+ },
443
+ {
444
+ "epoch": 3.44,
445
+ "learning_rate": 0.00026603239503066516,
446
+ "loss": 0.7746,
447
+ "step": 146000
448
+ },
449
+ {
450
+ "epoch": 3.49,
451
+ "learning_rate": 0.0002655606227394244,
452
+ "loss": 0.7749,
453
+ "step": 148000
454
+ },
455
+ {
456
+ "epoch": 3.53,
457
+ "learning_rate": 0.00026508885044818363,
458
+ "loss": 0.7747,
459
+ "step": 150000
460
+ },
461
+ {
462
+ "epoch": 3.58,
463
+ "learning_rate": 0.0002646170781569429,
464
+ "loss": 0.7816,
465
+ "step": 152000
466
+ },
467
+ {
468
+ "epoch": 3.63,
469
+ "learning_rate": 0.00026414530586570215,
470
+ "loss": 0.7807,
471
+ "step": 154000
472
+ },
473
+ {
474
+ "epoch": 3.67,
475
+ "learning_rate": 0.0002636735335744614,
476
+ "loss": 0.7817,
477
+ "step": 156000
478
+ },
479
+ {
480
+ "epoch": 3.72,
481
+ "learning_rate": 0.0002632017612832206,
482
+ "loss": 0.787,
483
+ "step": 158000
484
+ },
485
+ {
486
+ "epoch": 3.77,
487
+ "learning_rate": 0.00026272998899197985,
488
+ "loss": 0.7796,
489
+ "step": 160000
490
+ },
491
+ {
492
+ "epoch": 3.82,
493
+ "learning_rate": 0.0002622582167007391,
494
+ "loss": 0.7662,
495
+ "step": 162000
496
+ },
497
+ {
498
+ "epoch": 3.86,
499
+ "learning_rate": 0.0002617864444094983,
500
+ "loss": 0.7898,
501
+ "step": 164000
502
+ },
503
+ {
504
+ "epoch": 3.91,
505
+ "learning_rate": 0.00026131467211825755,
506
+ "loss": 0.7705,
507
+ "step": 166000
508
+ },
509
+ {
510
+ "epoch": 3.96,
511
+ "learning_rate": 0.0002608428998270168,
512
+ "loss": 0.7876,
513
+ "step": 168000
514
+ },
515
+ {
516
+ "epoch": 4.0,
517
+ "learning_rate": 0.00026037112753577607,
518
+ "loss": 0.7747,
519
+ "step": 170000
520
+ },
521
+ {
522
+ "epoch": 4.05,
523
+ "learning_rate": 0.0002598993552445353,
524
+ "loss": 0.7517,
525
+ "step": 172000
526
+ },
527
+ {
528
+ "epoch": 4.1,
529
+ "learning_rate": 0.00025942758295329453,
530
+ "loss": 0.7465,
531
+ "step": 174000
532
+ },
533
+ {
534
+ "epoch": 4.15,
535
+ "learning_rate": 0.00025895581066205376,
536
+ "loss": 0.7485,
537
+ "step": 176000
538
+ },
539
+ {
540
+ "epoch": 4.19,
541
+ "learning_rate": 0.000258484038370813,
542
+ "loss": 0.7531,
543
+ "step": 178000
544
+ },
545
+ {
546
+ "epoch": 4.24,
547
+ "learning_rate": 0.00025801226607957223,
548
+ "loss": 0.7521,
549
+ "step": 180000
550
+ },
551
+ {
552
+ "epoch": 4.29,
553
+ "learning_rate": 0.00025754049378833146,
554
+ "loss": 0.7477,
555
+ "step": 182000
556
+ },
557
+ {
558
+ "epoch": 4.33,
559
+ "learning_rate": 0.0002570687214970907,
560
+ "loss": 0.7621,
561
+ "step": 184000
562
+ },
563
+ {
564
+ "epoch": 4.38,
565
+ "learning_rate": 0.00025659694920585,
566
+ "loss": 0.7552,
567
+ "step": 186000
568
+ },
569
+ {
570
+ "epoch": 4.43,
571
+ "learning_rate": 0.0002561251769146092,
572
+ "loss": 0.7592,
573
+ "step": 188000
574
+ },
575
+ {
576
+ "epoch": 4.47,
577
+ "learning_rate": 0.00025565340462336845,
578
+ "loss": 0.7508,
579
+ "step": 190000
580
+ },
581
+ {
582
+ "epoch": 4.52,
583
+ "learning_rate": 0.0002551816323321277,
584
+ "loss": 0.7547,
585
+ "step": 192000
586
+ },
587
+ {
588
+ "epoch": 4.57,
589
+ "learning_rate": 0.0002547098600408869,
590
+ "loss": 0.7439,
591
+ "step": 194000
592
+ },
593
+ {
594
+ "epoch": 4.62,
595
+ "learning_rate": 0.00025423808774964615,
596
+ "loss": 0.762,
597
+ "step": 196000
598
+ },
599
+ {
600
+ "epoch": 4.66,
601
+ "learning_rate": 0.0002537663154584054,
602
+ "loss": 0.7507,
603
+ "step": 198000
604
+ },
605
+ {
606
+ "epoch": 4.71,
607
+ "learning_rate": 0.0002532945431671646,
608
+ "loss": 0.7553,
609
+ "step": 200000
610
+ },
611
+ {
612
+ "epoch": 4.76,
613
+ "learning_rate": 0.00025282277087592384,
614
+ "loss": 0.7498,
615
+ "step": 202000
616
+ },
617
+ {
618
+ "epoch": 4.8,
619
+ "learning_rate": 0.00025235099858468313,
620
+ "loss": 0.7466,
621
+ "step": 204000
622
+ },
623
+ {
624
+ "epoch": 4.85,
625
+ "learning_rate": 0.0002518792262934423,
626
+ "loss": 0.7496,
627
+ "step": 206000
628
+ },
629
+ {
630
+ "epoch": 4.9,
631
+ "learning_rate": 0.0002514074540022016,
632
+ "loss": 0.7406,
633
+ "step": 208000
634
+ },
635
+ {
636
+ "epoch": 4.95,
637
+ "learning_rate": 0.00025093568171096083,
638
+ "loss": 0.7447,
639
+ "step": 210000
640
+ },
641
+ {
642
+ "epoch": 4.99,
643
+ "learning_rate": 0.00025046390941972006,
644
+ "loss": 0.7532,
645
+ "step": 212000
646
+ },
647
+ {
648
+ "epoch": 5.04,
649
+ "learning_rate": 0.0002499921371284793,
650
+ "loss": 0.7339,
651
+ "step": 214000
652
+ },
653
+ {
654
+ "epoch": 5.09,
655
+ "learning_rate": 0.00024952036483723853,
656
+ "loss": 0.7214,
657
+ "step": 216000
658
+ },
659
+ {
660
+ "epoch": 5.13,
661
+ "learning_rate": 0.00024904859254599776,
662
+ "loss": 0.7227,
663
+ "step": 218000
664
+ },
665
+ {
666
+ "epoch": 5.18,
667
+ "learning_rate": 0.000248576820254757,
668
+ "loss": 0.7344,
669
+ "step": 220000
670
+ },
671
+ {
672
+ "epoch": 5.23,
673
+ "learning_rate": 0.0002481050479635163,
674
+ "loss": 0.7272,
675
+ "step": 222000
676
+ },
677
+ {
678
+ "epoch": 5.28,
679
+ "learning_rate": 0.00024763327567227546,
680
+ "loss": 0.7271,
681
+ "step": 224000
682
+ },
683
+ {
684
+ "epoch": 5.32,
685
+ "learning_rate": 0.00024716150338103475,
686
+ "loss": 0.7302,
687
+ "step": 226000
688
+ },
689
+ {
690
+ "epoch": 5.37,
691
+ "learning_rate": 0.000246689731089794,
692
+ "loss": 0.7218,
693
+ "step": 228000
694
+ },
695
+ {
696
+ "epoch": 5.42,
697
+ "learning_rate": 0.0002462179587985532,
698
+ "loss": 0.7242,
699
+ "step": 230000
700
+ },
701
+ {
702
+ "epoch": 5.46,
703
+ "learning_rate": 0.00024574618650731244,
704
+ "loss": 0.7306,
705
+ "step": 232000
706
+ },
707
+ {
708
+ "epoch": 5.51,
709
+ "learning_rate": 0.0002452744142160717,
710
+ "loss": 0.7351,
711
+ "step": 234000
712
+ },
713
+ {
714
+ "epoch": 5.56,
715
+ "learning_rate": 0.0002448026419248309,
716
+ "loss": 0.7225,
717
+ "step": 236000
718
+ },
719
+ {
720
+ "epoch": 5.61,
721
+ "learning_rate": 0.00024433086963359014,
722
+ "loss": 0.7256,
723
+ "step": 238000
724
+ },
725
+ {
726
+ "epoch": 5.65,
727
+ "learning_rate": 0.0002438590973423494,
728
+ "loss": 0.7256,
729
+ "step": 240000
730
+ },
731
+ {
732
+ "epoch": 5.7,
733
+ "learning_rate": 0.00024338732505110863,
734
+ "loss": 0.7293,
735
+ "step": 242000
736
+ },
737
+ {
738
+ "epoch": 5.75,
739
+ "learning_rate": 0.0002429155527598679,
740
+ "loss": 0.7272,
741
+ "step": 244000
742
+ },
743
+ {
744
+ "epoch": 5.79,
745
+ "learning_rate": 0.00024244378046862713,
746
+ "loss": 0.7154,
747
+ "step": 246000
748
+ },
749
+ {
750
+ "epoch": 5.84,
751
+ "learning_rate": 0.00024197200817738636,
752
+ "loss": 0.7274,
753
+ "step": 248000
754
+ },
755
+ {
756
+ "epoch": 5.89,
757
+ "learning_rate": 0.00024150023588614562,
758
+ "loss": 0.7184,
759
+ "step": 250000
760
+ },
761
+ {
762
+ "epoch": 5.93,
763
+ "learning_rate": 0.00024102846359490483,
764
+ "loss": 0.7218,
765
+ "step": 252000
766
+ },
767
+ {
768
+ "epoch": 5.98,
769
+ "learning_rate": 0.00024055669130366409,
770
+ "loss": 0.7246,
771
+ "step": 254000
772
+ },
773
+ {
774
+ "epoch": 6.03,
775
+ "learning_rate": 0.00024008491901242332,
776
+ "loss": 0.7036,
777
+ "step": 256000
778
+ },
779
+ {
780
+ "epoch": 6.08,
781
+ "learning_rate": 0.00023961314672118255,
782
+ "loss": 0.6992,
783
+ "step": 258000
784
+ },
785
+ {
786
+ "epoch": 6.12,
787
+ "learning_rate": 0.00023914137442994178,
788
+ "loss": 0.7029,
789
+ "step": 260000
790
+ },
791
+ {
792
+ "epoch": 6.17,
793
+ "learning_rate": 0.00023866960213870104,
794
+ "loss": 0.7097,
795
+ "step": 262000
796
+ },
797
+ {
798
+ "epoch": 6.22,
799
+ "learning_rate": 0.00023819782984746028,
800
+ "loss": 0.7042,
801
+ "step": 264000
802
+ },
803
+ {
804
+ "epoch": 6.26,
805
+ "learning_rate": 0.0002377260575562195,
806
+ "loss": 0.7076,
807
+ "step": 266000
808
+ },
809
+ {
810
+ "epoch": 6.31,
811
+ "learning_rate": 0.00023725428526497877,
812
+ "loss": 0.6972,
813
+ "step": 268000
814
+ },
815
+ {
816
+ "epoch": 6.36,
817
+ "learning_rate": 0.00023678251297373797,
818
+ "loss": 0.7097,
819
+ "step": 270000
820
+ },
821
+ {
822
+ "epoch": 6.41,
823
+ "learning_rate": 0.00023631074068249723,
824
+ "loss": 0.7025,
825
+ "step": 272000
826
+ },
827
+ {
828
+ "epoch": 6.45,
829
+ "learning_rate": 0.00023583896839125647,
830
+ "loss": 0.7041,
831
+ "step": 274000
832
+ },
833
+ {
834
+ "epoch": 6.5,
835
+ "learning_rate": 0.0002353671961000157,
836
+ "loss": 0.7098,
837
+ "step": 276000
838
+ },
839
+ {
840
+ "epoch": 6.55,
841
+ "learning_rate": 0.00023489542380877493,
842
+ "loss": 0.704,
843
+ "step": 278000
844
+ },
845
+ {
846
+ "epoch": 6.59,
847
+ "learning_rate": 0.0002344236515175342,
848
+ "loss": 0.7046,
849
+ "step": 280000
850
+ },
851
+ {
852
+ "epoch": 6.64,
853
+ "learning_rate": 0.00023395187922629343,
854
+ "loss": 0.701,
855
+ "step": 282000
856
+ },
857
+ {
858
+ "epoch": 6.69,
859
+ "learning_rate": 0.00023348010693505266,
860
+ "loss": 0.7107,
861
+ "step": 284000
862
+ },
863
+ {
864
+ "epoch": 6.74,
865
+ "learning_rate": 0.00023300833464381192,
866
+ "loss": 0.7107,
867
+ "step": 286000
868
+ },
869
+ {
870
+ "epoch": 6.78,
871
+ "learning_rate": 0.00023253656235257112,
872
+ "loss": 0.7051,
873
+ "step": 288000
874
+ },
875
+ {
876
+ "epoch": 6.83,
877
+ "learning_rate": 0.00023206479006133038,
878
+ "loss": 0.7055,
879
+ "step": 290000
880
+ },
881
+ {
882
+ "epoch": 6.88,
883
+ "learning_rate": 0.00023159301777008962,
884
+ "loss": 0.7004,
885
+ "step": 292000
886
+ },
887
+ {
888
+ "epoch": 6.92,
889
+ "learning_rate": 0.00023112124547884885,
890
+ "loss": 0.7051,
891
+ "step": 294000
892
+ },
893
+ {
894
+ "epoch": 6.97,
895
+ "learning_rate": 0.00023064947318760808,
896
+ "loss": 0.7025,
897
+ "step": 296000
898
+ },
899
+ {
900
+ "epoch": 7.02,
901
+ "learning_rate": 0.00023017770089636734,
902
+ "loss": 0.6976,
903
+ "step": 298000
904
+ },
905
+ {
906
+ "epoch": 7.07,
907
+ "learning_rate": 0.0002297059286051266,
908
+ "loss": 0.6785,
909
+ "step": 300000
910
+ },
911
+ {
912
+ "epoch": 7.11,
913
+ "learning_rate": 0.0002292341563138858,
914
+ "loss": 0.6773,
915
+ "step": 302000
916
+ },
917
+ {
918
+ "epoch": 7.16,
919
+ "learning_rate": 0.00022876238402264507,
920
+ "loss": 0.6763,
921
+ "step": 304000
922
+ },
923
+ {
924
+ "epoch": 7.21,
925
+ "learning_rate": 0.00022829061173140427,
926
+ "loss": 0.6774,
927
+ "step": 306000
928
+ },
929
+ {
930
+ "epoch": 7.25,
931
+ "learning_rate": 0.00022781883944016353,
932
+ "loss": 0.6786,
933
+ "step": 308000
934
+ },
935
+ {
936
+ "epoch": 7.3,
937
+ "learning_rate": 0.00022734706714892276,
938
+ "loss": 0.6885,
939
+ "step": 310000
940
+ },
941
+ {
942
+ "epoch": 7.35,
943
+ "learning_rate": 0.000226875294857682,
944
+ "loss": 0.6852,
945
+ "step": 312000
946
+ },
947
+ {
948
+ "epoch": 7.4,
949
+ "learning_rate": 0.00022640352256644123,
950
+ "loss": 0.6842,
951
+ "step": 314000
952
+ },
953
+ {
954
+ "epoch": 7.44,
955
+ "learning_rate": 0.0002259317502752005,
956
+ "loss": 0.6868,
957
+ "step": 316000
958
+ },
959
+ {
960
+ "epoch": 7.49,
961
+ "learning_rate": 0.00022545997798395975,
962
+ "loss": 0.6918,
963
+ "step": 318000
964
+ },
965
+ {
966
+ "epoch": 7.54,
967
+ "learning_rate": 0.00022498820569271896,
968
+ "loss": 0.6846,
969
+ "step": 320000
970
+ },
971
+ {
972
+ "epoch": 7.58,
973
+ "learning_rate": 0.00022451643340147822,
974
+ "loss": 0.6881,
975
+ "step": 322000
976
+ },
977
+ {
978
+ "epoch": 7.63,
979
+ "learning_rate": 0.00022404466111023742,
980
+ "loss": 0.69,
981
+ "step": 324000
982
+ },
983
+ {
984
+ "epoch": 7.68,
985
+ "learning_rate": 0.00022357288881899668,
986
+ "loss": 0.6823,
987
+ "step": 326000
988
+ },
989
+ {
990
+ "epoch": 7.72,
991
+ "learning_rate": 0.00022310111652775591,
992
+ "loss": 0.6896,
993
+ "step": 328000
994
+ },
995
+ {
996
+ "epoch": 7.77,
997
+ "learning_rate": 0.00022262934423651517,
998
+ "loss": 0.6862,
999
+ "step": 330000
1000
+ },
1001
+ {
1002
+ "epoch": 7.82,
1003
+ "learning_rate": 0.00022215757194527438,
1004
+ "loss": 0.6858,
1005
+ "step": 332000
1006
+ },
1007
+ {
1008
+ "epoch": 7.87,
1009
+ "learning_rate": 0.00022168579965403364,
1010
+ "loss": 0.691,
1011
+ "step": 334000
1012
+ },
1013
+ {
1014
+ "epoch": 7.91,
1015
+ "learning_rate": 0.00022121402736279284,
1016
+ "loss": 0.6895,
1017
+ "step": 336000
1018
+ },
1019
+ {
1020
+ "epoch": 7.96,
1021
+ "learning_rate": 0.0002207422550715521,
1022
+ "loss": 0.6924,
1023
+ "step": 338000
1024
+ },
1025
+ {
1026
+ "epoch": 8.01,
1027
+ "learning_rate": 0.00022027048278031136,
1028
+ "loss": 0.6903,
1029
+ "step": 340000
1030
+ },
1031
+ {
1032
+ "epoch": 8.05,
1033
+ "learning_rate": 0.0002197987104890706,
1034
+ "loss": 0.6637,
1035
+ "step": 342000
1036
+ },
1037
+ {
1038
+ "epoch": 8.1,
1039
+ "learning_rate": 0.00021932693819782983,
1040
+ "loss": 0.6612,
1041
+ "step": 344000
1042
+ },
1043
+ {
1044
+ "epoch": 8.15,
1045
+ "learning_rate": 0.00021885516590658906,
1046
+ "loss": 0.6665,
1047
+ "step": 346000
1048
+ },
1049
+ {
1050
+ "epoch": 8.2,
1051
+ "learning_rate": 0.00021838339361534832,
1052
+ "loss": 0.6702,
1053
+ "step": 348000
1054
+ },
1055
+ {
1056
+ "epoch": 8.24,
1057
+ "learning_rate": 0.00021791162132410753,
1058
+ "loss": 0.6667,
1059
+ "step": 350000
1060
+ },
1061
+ {
1062
+ "epoch": 8.29,
1063
+ "learning_rate": 0.0002174398490328668,
1064
+ "loss": 0.6674,
1065
+ "step": 352000
1066
+ },
1067
+ {
1068
+ "epoch": 8.34,
1069
+ "learning_rate": 0.000216968076741626,
1070
+ "loss": 0.6719,
1071
+ "step": 354000
1072
+ },
1073
+ {
1074
+ "epoch": 8.38,
1075
+ "learning_rate": 0.00021649630445038525,
1076
+ "loss": 0.6671,
1077
+ "step": 356000
1078
+ },
1079
+ {
1080
+ "epoch": 8.43,
1081
+ "learning_rate": 0.0002160245321591445,
1082
+ "loss": 0.6647,
1083
+ "step": 358000
1084
+ },
1085
+ {
1086
+ "epoch": 8.48,
1087
+ "learning_rate": 0.00021555275986790375,
1088
+ "loss": 0.6671,
1089
+ "step": 360000
1090
+ },
1091
+ {
1092
+ "epoch": 8.53,
1093
+ "learning_rate": 0.00021508098757666298,
1094
+ "loss": 0.6681,
1095
+ "step": 362000
1096
+ },
1097
+ {
1098
+ "epoch": 8.57,
1099
+ "learning_rate": 0.0002146092152854222,
1100
+ "loss": 0.6727,
1101
+ "step": 364000
1102
+ },
1103
+ {
1104
+ "epoch": 8.62,
1105
+ "learning_rate": 0.00021413744299418147,
1106
+ "loss": 0.6767,
1107
+ "step": 366000
1108
+ },
1109
+ {
1110
+ "epoch": 8.67,
1111
+ "learning_rate": 0.00021366567070294068,
1112
+ "loss": 0.6749,
1113
+ "step": 368000
1114
+ },
1115
+ {
1116
+ "epoch": 8.71,
1117
+ "learning_rate": 0.00021319389841169994,
1118
+ "loss": 0.6704,
1119
+ "step": 370000
1120
+ },
1121
+ {
1122
+ "epoch": 8.76,
1123
+ "learning_rate": 0.00021272212612045917,
1124
+ "loss": 0.6729,
1125
+ "step": 372000
1126
+ },
1127
+ {
1128
+ "epoch": 8.81,
1129
+ "learning_rate": 0.0002122503538292184,
1130
+ "loss": 0.6641,
1131
+ "step": 374000
1132
+ },
1133
+ {
1134
+ "epoch": 8.86,
1135
+ "learning_rate": 0.00021177858153797766,
1136
+ "loss": 0.6678,
1137
+ "step": 376000
1138
+ },
1139
+ {
1140
+ "epoch": 8.9,
1141
+ "learning_rate": 0.0002113068092467369,
1142
+ "loss": 0.6677,
1143
+ "step": 378000
1144
+ },
1145
+ {
1146
+ "epoch": 8.95,
1147
+ "learning_rate": 0.00021083503695549613,
1148
+ "loss": 0.6683,
1149
+ "step": 380000
1150
+ },
1151
+ {
1152
+ "epoch": 9.0,
1153
+ "learning_rate": 0.00021036326466425536,
1154
+ "loss": 0.6746,
1155
+ "step": 382000
1156
+ },
1157
+ {
1158
+ "epoch": 9.04,
1159
+ "learning_rate": 0.00020989149237301462,
1160
+ "loss": 0.6469,
1161
+ "step": 384000
1162
+ },
1163
+ {
1164
+ "epoch": 9.09,
1165
+ "learning_rate": 0.00020941972008177383,
1166
+ "loss": 0.6496,
1167
+ "step": 386000
1168
+ },
1169
+ {
1170
+ "epoch": 9.14,
1171
+ "learning_rate": 0.00020894794779053309,
1172
+ "loss": 0.6513,
1173
+ "step": 388000
1174
+ },
1175
+ {
1176
+ "epoch": 9.19,
1177
+ "learning_rate": 0.00020847617549929232,
1178
+ "loss": 0.6516,
1179
+ "step": 390000
1180
+ },
1181
+ {
1182
+ "epoch": 9.23,
1183
+ "learning_rate": 0.00020800440320805155,
1184
+ "loss": 0.6556,
1185
+ "step": 392000
1186
+ },
1187
+ {
1188
+ "epoch": 9.28,
1189
+ "learning_rate": 0.0002075326309168108,
1190
+ "loss": 0.6477,
1191
+ "step": 394000
1192
+ },
1193
+ {
1194
+ "epoch": 9.33,
1195
+ "learning_rate": 0.00020706085862557004,
1196
+ "loss": 0.6571,
1197
+ "step": 396000
1198
+ },
1199
+ {
1200
+ "epoch": 9.37,
1201
+ "learning_rate": 0.00020658908633432928,
1202
+ "loss": 0.6479,
1203
+ "step": 398000
1204
+ },
1205
+ {
1206
+ "epoch": 9.42,
1207
+ "learning_rate": 0.0002061173140430885,
1208
+ "loss": 0.65,
1209
+ "step": 400000
1210
+ },
1211
+ {
1212
+ "epoch": 9.47,
1213
+ "learning_rate": 0.00020564554175184777,
1214
+ "loss": 0.6557,
1215
+ "step": 402000
1216
+ },
1217
+ {
1218
+ "epoch": 9.51,
1219
+ "learning_rate": 0.00020517376946060697,
1220
+ "loss": 0.6644,
1221
+ "step": 404000
1222
+ },
1223
+ {
1224
+ "epoch": 9.56,
1225
+ "learning_rate": 0.00020470199716936623,
1226
+ "loss": 0.6558,
1227
+ "step": 406000
1228
+ },
1229
+ {
1230
+ "epoch": 9.61,
1231
+ "learning_rate": 0.00020423022487812547,
1232
+ "loss": 0.6503,
1233
+ "step": 408000
1234
+ },
1235
+ {
1236
+ "epoch": 9.66,
1237
+ "learning_rate": 0.0002037584525868847,
1238
+ "loss": 0.6569,
1239
+ "step": 410000
1240
+ },
1241
+ {
1242
+ "epoch": 9.7,
1243
+ "learning_rate": 0.00020328668029564396,
1244
+ "loss": 0.66,
1245
+ "step": 412000
1246
+ },
1247
+ {
1248
+ "epoch": 9.75,
1249
+ "learning_rate": 0.0002028149080044032,
1250
+ "loss": 0.6528,
1251
+ "step": 414000
1252
+ },
1253
+ {
1254
+ "epoch": 9.8,
1255
+ "learning_rate": 0.00020234313571316245,
1256
+ "loss": 0.6493,
1257
+ "step": 416000
1258
+ },
1259
+ {
1260
+ "epoch": 9.84,
1261
+ "learning_rate": 0.00020187136342192166,
1262
+ "loss": 0.6599,
1263
+ "step": 418000
1264
+ },
1265
+ {
1266
+ "epoch": 9.89,
1267
+ "learning_rate": 0.00020139959113068092,
1268
+ "loss": 0.6534,
1269
+ "step": 420000
1270
+ },
1271
+ {
1272
+ "epoch": 9.94,
1273
+ "learning_rate": 0.00020092781883944012,
1274
+ "loss": 0.6546,
1275
+ "step": 422000
1276
+ },
1277
+ {
1278
+ "epoch": 9.99,
1279
+ "learning_rate": 0.00020045604654819938,
1280
+ "loss": 0.6488,
1281
+ "step": 424000
1282
+ },
1283
+ {
1284
+ "epoch": 10.03,
1285
+ "learning_rate": 0.00019998427425695862,
1286
+ "loss": 0.643,
1287
+ "step": 426000
1288
+ },
1289
+ {
1290
+ "epoch": 10.08,
1291
+ "learning_rate": 0.00019951250196571788,
1292
+ "loss": 0.6465,
1293
+ "step": 428000
1294
+ },
1295
+ {
1296
+ "epoch": 10.13,
1297
+ "learning_rate": 0.0001990407296744771,
1298
+ "loss": 0.6324,
1299
+ "step": 430000
1300
+ },
1301
+ {
1302
+ "epoch": 10.17,
1303
+ "learning_rate": 0.00019856895738323634,
1304
+ "loss": 0.6446,
1305
+ "step": 432000
1306
+ },
1307
+ {
1308
+ "epoch": 10.22,
1309
+ "learning_rate": 0.0001980971850919956,
1310
+ "loss": 0.6283,
1311
+ "step": 434000
1312
+ },
1313
+ {
1314
+ "epoch": 10.27,
1315
+ "learning_rate": 0.0001976254128007548,
1316
+ "loss": 0.635,
1317
+ "step": 436000
1318
+ },
1319
+ {
1320
+ "epoch": 10.32,
1321
+ "learning_rate": 0.00019715364050951407,
1322
+ "loss": 0.6333,
1323
+ "step": 438000
1324
+ },
1325
+ {
1326
+ "epoch": 10.36,
1327
+ "learning_rate": 0.0001966818682182733,
1328
+ "loss": 0.6378,
1329
+ "step": 440000
1330
+ },
1331
+ {
1332
+ "epoch": 10.41,
1333
+ "learning_rate": 0.00019621009592703253,
1334
+ "loss": 0.6369,
1335
+ "step": 442000
1336
+ },
1337
+ {
1338
+ "epoch": 10.46,
1339
+ "learning_rate": 0.00019573832363579177,
1340
+ "loss": 0.6348,
1341
+ "step": 444000
1342
+ },
1343
+ {
1344
+ "epoch": 10.5,
1345
+ "learning_rate": 0.00019526655134455103,
1346
+ "loss": 0.6439,
1347
+ "step": 446000
1348
+ },
1349
+ {
1350
+ "epoch": 10.55,
1351
+ "learning_rate": 0.00019479477905331026,
1352
+ "loss": 0.6342,
1353
+ "step": 448000
1354
+ },
1355
+ {
1356
+ "epoch": 10.6,
1357
+ "learning_rate": 0.0001943230067620695,
1358
+ "loss": 0.6424,
1359
+ "step": 450000
1360
+ },
1361
+ {
1362
+ "epoch": 10.65,
1363
+ "learning_rate": 0.00019385123447082875,
1364
+ "loss": 0.6408,
1365
+ "step": 452000
1366
+ },
1367
+ {
1368
+ "epoch": 10.69,
1369
+ "learning_rate": 0.00019337946217958796,
1370
+ "loss": 0.6413,
1371
+ "step": 454000
1372
+ },
1373
+ {
1374
+ "epoch": 10.74,
1375
+ "learning_rate": 0.00019290768988834722,
1376
+ "loss": 0.643,
1377
+ "step": 456000
1378
+ },
1379
+ {
1380
+ "epoch": 10.79,
1381
+ "learning_rate": 0.00019243591759710645,
1382
+ "loss": 0.6421,
1383
+ "step": 458000
1384
+ },
1385
+ {
1386
+ "epoch": 10.83,
1387
+ "learning_rate": 0.00019196414530586568,
1388
+ "loss": 0.6393,
1389
+ "step": 460000
1390
+ },
1391
+ {
1392
+ "epoch": 10.88,
1393
+ "learning_rate": 0.00019149237301462491,
1394
+ "loss": 0.6385,
1395
+ "step": 462000
1396
+ },
1397
+ {
1398
+ "epoch": 10.93,
1399
+ "learning_rate": 0.00019102060072338417,
1400
+ "loss": 0.6387,
1401
+ "step": 464000
1402
+ },
1403
+ {
1404
+ "epoch": 10.98,
1405
+ "learning_rate": 0.00019054882843214338,
1406
+ "loss": 0.6442,
1407
+ "step": 466000
1408
+ },
1409
+ {
1410
+ "epoch": 11.02,
1411
+ "learning_rate": 0.00019007705614090264,
1412
+ "loss": 0.6294,
1413
+ "step": 468000
1414
+ },
1415
+ {
1416
+ "epoch": 11.07,
1417
+ "learning_rate": 0.0001896052838496619,
1418
+ "loss": 0.6191,
1419
+ "step": 470000
1420
+ },
1421
+ {
1422
+ "epoch": 11.12,
1423
+ "learning_rate": 0.0001891335115584211,
1424
+ "loss": 0.6226,
1425
+ "step": 472000
1426
+ },
1427
+ {
1428
+ "epoch": 11.16,
1429
+ "learning_rate": 0.00018866173926718036,
1430
+ "loss": 0.6207,
1431
+ "step": 474000
1432
+ },
1433
+ {
1434
+ "epoch": 11.21,
1435
+ "learning_rate": 0.0001881899669759396,
1436
+ "loss": 0.6282,
1437
+ "step": 476000
1438
+ },
1439
+ {
1440
+ "epoch": 11.26,
1441
+ "learning_rate": 0.00018771819468469883,
1442
+ "loss": 0.615,
1443
+ "step": 478000
1444
+ },
1445
+ {
1446
+ "epoch": 11.3,
1447
+ "learning_rate": 0.00018724642239345806,
1448
+ "loss": 0.6201,
1449
+ "step": 480000
1450
+ },
1451
+ {
1452
+ "epoch": 11.35,
1453
+ "learning_rate": 0.00018677465010221732,
1454
+ "loss": 0.6199,
1455
+ "step": 482000
1456
+ },
1457
+ {
1458
+ "epoch": 11.4,
1459
+ "learning_rate": 0.00018630287781097653,
1460
+ "loss": 0.6211,
1461
+ "step": 484000
1462
+ },
1463
+ {
1464
+ "epoch": 11.45,
1465
+ "learning_rate": 0.0001858311055197358,
1466
+ "loss": 0.6372,
1467
+ "step": 486000
1468
+ },
1469
+ {
1470
+ "epoch": 11.49,
1471
+ "learning_rate": 0.00018535933322849505,
1472
+ "loss": 0.6268,
1473
+ "step": 488000
1474
+ },
1475
+ {
1476
+ "epoch": 11.54,
1477
+ "learning_rate": 0.00018488756093725425,
1478
+ "loss": 0.6283,
1479
+ "step": 490000
1480
+ },
1481
+ {
1482
+ "epoch": 11.59,
1483
+ "learning_rate": 0.00018441578864601351,
1484
+ "loss": 0.6398,
1485
+ "step": 492000
1486
+ },
1487
+ {
1488
+ "epoch": 11.63,
1489
+ "learning_rate": 0.00018394401635477275,
1490
+ "loss": 0.6282,
1491
+ "step": 494000
1492
+ },
1493
+ {
1494
+ "epoch": 11.68,
1495
+ "learning_rate": 0.00018347224406353198,
1496
+ "loss": 0.6327,
1497
+ "step": 496000
1498
+ },
1499
+ {
1500
+ "epoch": 11.73,
1501
+ "learning_rate": 0.0001830004717722912,
1502
+ "loss": 0.6226,
1503
+ "step": 498000
1504
+ },
1505
+ {
1506
+ "epoch": 11.78,
1507
+ "learning_rate": 0.00018252869948105047,
1508
+ "loss": 0.6321,
1509
+ "step": 500000
1510
+ },
1511
+ {
1512
+ "epoch": 11.82,
1513
+ "learning_rate": 0.00018205692718980968,
1514
+ "loss": 0.6244,
1515
+ "step": 502000
1516
+ },
1517
+ {
1518
+ "epoch": 11.87,
1519
+ "learning_rate": 0.00018158515489856894,
1520
+ "loss": 0.6257,
1521
+ "step": 504000
1522
+ },
1523
+ {
1524
+ "epoch": 11.92,
1525
+ "learning_rate": 0.0001811133826073282,
1526
+ "loss": 0.6291,
1527
+ "step": 506000
1528
+ },
1529
+ {
1530
+ "epoch": 11.96,
1531
+ "learning_rate": 0.0001806416103160874,
1532
+ "loss": 0.6283,
1533
+ "step": 508000
1534
+ },
1535
+ {
1536
+ "epoch": 12.01,
1537
+ "learning_rate": 0.00018016983802484666,
1538
+ "loss": 0.6246,
1539
+ "step": 510000
1540
+ },
1541
+ {
1542
+ "epoch": 12.06,
1543
+ "learning_rate": 0.0001796980657336059,
1544
+ "loss": 0.6133,
1545
+ "step": 512000
1546
+ },
1547
+ {
1548
+ "epoch": 12.11,
1549
+ "learning_rate": 0.00017922629344236516,
1550
+ "loss": 0.6069,
1551
+ "step": 514000
1552
+ },
1553
+ {
1554
+ "epoch": 12.15,
1555
+ "learning_rate": 0.00017875452115112436,
1556
+ "loss": 0.6163,
1557
+ "step": 516000
1558
+ },
1559
+ {
1560
+ "epoch": 12.2,
1561
+ "learning_rate": 0.00017828274885988362,
1562
+ "loss": 0.6131,
1563
+ "step": 518000
1564
+ },
1565
+ {
1566
+ "epoch": 12.25,
1567
+ "learning_rate": 0.00017781097656864283,
1568
+ "loss": 0.6096,
1569
+ "step": 520000
1570
+ },
1571
+ {
1572
+ "epoch": 12.29,
1573
+ "learning_rate": 0.00017733920427740209,
1574
+ "loss": 0.6073,
1575
+ "step": 522000
1576
+ },
1577
+ {
1578
+ "epoch": 12.34,
1579
+ "learning_rate": 0.00017686743198616135,
1580
+ "loss": 0.6068,
1581
+ "step": 524000
1582
+ },
1583
+ {
1584
+ "epoch": 12.39,
1585
+ "learning_rate": 0.00017639565969492058,
1586
+ "loss": 0.6114,
1587
+ "step": 526000
1588
+ },
1589
+ {
1590
+ "epoch": 12.44,
1591
+ "learning_rate": 0.0001759238874036798,
1592
+ "loss": 0.6189,
1593
+ "step": 528000
1594
+ },
1595
+ {
1596
+ "epoch": 12.48,
1597
+ "learning_rate": 0.00017545211511243904,
1598
+ "loss": 0.6111,
1599
+ "step": 530000
1600
+ },
1601
+ {
1602
+ "epoch": 12.53,
1603
+ "learning_rate": 0.0001749803428211983,
1604
+ "loss": 0.6085,
1605
+ "step": 532000
1606
+ },
1607
+ {
1608
+ "epoch": 12.58,
1609
+ "learning_rate": 0.0001745085705299575,
1610
+ "loss": 0.6164,
1611
+ "step": 534000
1612
+ },
1613
+ {
1614
+ "epoch": 12.62,
1615
+ "learning_rate": 0.00017403679823871677,
1616
+ "loss": 0.6169,
1617
+ "step": 536000
1618
+ },
1619
+ {
1620
+ "epoch": 12.67,
1621
+ "learning_rate": 0.00017356502594747598,
1622
+ "loss": 0.615,
1623
+ "step": 538000
1624
+ },
1625
+ {
1626
+ "epoch": 12.72,
1627
+ "learning_rate": 0.00017309325365623524,
1628
+ "loss": 0.6145,
1629
+ "step": 540000
1630
+ },
1631
+ {
1632
+ "epoch": 12.76,
1633
+ "learning_rate": 0.0001726214813649945,
1634
+ "loss": 0.6149,
1635
+ "step": 542000
1636
+ },
1637
+ {
1638
+ "epoch": 12.81,
1639
+ "learning_rate": 0.00017214970907375373,
1640
+ "loss": 0.6128,
1641
+ "step": 544000
1642
+ },
1643
+ {
1644
+ "epoch": 12.86,
1645
+ "learning_rate": 0.00017167793678251296,
1646
+ "loss": 0.6123,
1647
+ "step": 546000
1648
+ },
1649
+ {
1650
+ "epoch": 12.91,
1651
+ "learning_rate": 0.0001712061644912722,
1652
+ "loss": 0.6132,
1653
+ "step": 548000
1654
+ },
1655
+ {
1656
+ "epoch": 12.95,
1657
+ "learning_rate": 0.00017073439220003145,
1658
+ "loss": 0.6154,
1659
+ "step": 550000
1660
+ },
1661
+ {
1662
+ "epoch": 13.0,
1663
+ "learning_rate": 0.00017026261990879066,
1664
+ "loss": 0.6168,
1665
+ "step": 552000
1666
+ },
1667
+ {
1668
+ "epoch": 13.05,
1669
+ "learning_rate": 0.00016979084761754992,
1670
+ "loss": 0.5926,
1671
+ "step": 554000
1672
+ },
1673
+ {
1674
+ "epoch": 13.09,
1675
+ "learning_rate": 0.00016931907532630915,
1676
+ "loss": 0.5924,
1677
+ "step": 556000
1678
+ },
1679
+ {
1680
+ "epoch": 13.14,
1681
+ "learning_rate": 0.00016884730303506838,
1682
+ "loss": 0.5949,
1683
+ "step": 558000
1684
+ },
1685
+ {
1686
+ "epoch": 13.19,
1687
+ "learning_rate": 0.00016837553074382764,
1688
+ "loss": 0.5947,
1689
+ "step": 560000
1690
+ },
1691
+ {
1692
+ "epoch": 13.24,
1693
+ "learning_rate": 0.00016790375845258688,
1694
+ "loss": 0.5971,
1695
+ "step": 562000
1696
+ },
1697
+ {
1698
+ "epoch": 13.28,
1699
+ "learning_rate": 0.0001674319861613461,
1700
+ "loss": 0.5976,
1701
+ "step": 564000
1702
+ },
1703
+ {
1704
+ "epoch": 13.33,
1705
+ "learning_rate": 0.00016696021387010534,
1706
+ "loss": 0.5984,
1707
+ "step": 566000
1708
+ },
1709
+ {
1710
+ "epoch": 13.38,
1711
+ "learning_rate": 0.0001664884415788646,
1712
+ "loss": 0.606,
1713
+ "step": 568000
1714
+ },
1715
+ {
1716
+ "epoch": 13.42,
1717
+ "learning_rate": 0.0001660166692876238,
1718
+ "loss": 0.6091,
1719
+ "step": 570000
1720
+ },
1721
+ {
1722
+ "epoch": 13.47,
1723
+ "learning_rate": 0.00016554489699638307,
1724
+ "loss": 0.5927,
1725
+ "step": 572000
1726
+ },
1727
+ {
1728
+ "epoch": 13.52,
1729
+ "learning_rate": 0.0001650731247051423,
1730
+ "loss": 0.6067,
1731
+ "step": 574000
1732
+ },
1733
+ {
1734
+ "epoch": 13.57,
1735
+ "learning_rate": 0.00016460135241390153,
1736
+ "loss": 0.6024,
1737
+ "step": 576000
1738
+ },
1739
+ {
1740
+ "epoch": 13.61,
1741
+ "learning_rate": 0.0001641295801226608,
1742
+ "loss": 0.6055,
1743
+ "step": 578000
1744
+ },
1745
+ {
1746
+ "epoch": 13.66,
1747
+ "learning_rate": 0.00016365780783142003,
1748
+ "loss": 0.6003,
1749
+ "step": 580000
1750
+ },
1751
+ {
1752
+ "epoch": 13.71,
1753
+ "learning_rate": 0.00016318603554017929,
1754
+ "loss": 0.6076,
1755
+ "step": 582000
1756
+ },
1757
+ {
1758
+ "epoch": 13.75,
1759
+ "learning_rate": 0.0001627142632489385,
1760
+ "loss": 0.606,
1761
+ "step": 584000
1762
+ },
1763
+ {
1764
+ "epoch": 13.8,
1765
+ "learning_rate": 0.00016224249095769775,
1766
+ "loss": 0.5901,
1767
+ "step": 586000
1768
+ },
1769
+ {
1770
+ "epoch": 13.85,
1771
+ "learning_rate": 0.00016177071866645696,
1772
+ "loss": 0.6084,
1773
+ "step": 588000
1774
+ },
1775
+ {
1776
+ "epoch": 13.9,
1777
+ "learning_rate": 0.00016129894637521622,
1778
+ "loss": 0.6075,
1779
+ "step": 590000
1780
+ },
1781
+ {
1782
+ "epoch": 13.94,
1783
+ "learning_rate": 0.00016082717408397545,
1784
+ "loss": 0.6085,
1785
+ "step": 592000
1786
+ },
1787
+ {
1788
+ "epoch": 13.99,
1789
+ "learning_rate": 0.00016035540179273468,
1790
+ "loss": 0.6009,
1791
+ "step": 594000
1792
+ },
1793
+ {
1794
+ "epoch": 14.04,
1795
+ "learning_rate": 0.00015988362950149391,
1796
+ "loss": 0.5894,
1797
+ "step": 596000
1798
+ },
1799
+ {
1800
+ "epoch": 14.08,
1801
+ "learning_rate": 0.00015941185721025317,
1802
+ "loss": 0.5835,
1803
+ "step": 598000
1804
+ },
1805
+ {
1806
+ "epoch": 14.13,
1807
+ "learning_rate": 0.00015894008491901243,
1808
+ "loss": 0.5859,
1809
+ "step": 600000
1810
+ },
1811
+ {
1812
+ "epoch": 14.18,
1813
+ "learning_rate": 0.00015846831262777164,
1814
+ "loss": 0.5929,
1815
+ "step": 602000
1816
+ },
1817
+ {
1818
+ "epoch": 14.23,
1819
+ "learning_rate": 0.0001579965403365309,
1820
+ "loss": 0.59,
1821
+ "step": 604000
1822
+ },
1823
+ {
1824
+ "epoch": 14.27,
1825
+ "learning_rate": 0.0001575247680452901,
1826
+ "loss": 0.582,
1827
+ "step": 606000
1828
+ },
1829
+ {
1830
+ "epoch": 14.32,
1831
+ "learning_rate": 0.00015705299575404937,
1832
+ "loss": 0.588,
1833
+ "step": 608000
1834
+ },
1835
+ {
1836
+ "epoch": 14.37,
1837
+ "learning_rate": 0.0001565812234628086,
1838
+ "loss": 0.5923,
1839
+ "step": 610000
1840
+ },
1841
+ {
1842
+ "epoch": 14.41,
1843
+ "learning_rate": 0.00015610945117156786,
1844
+ "loss": 0.5837,
1845
+ "step": 612000
1846
+ },
1847
+ {
1848
+ "epoch": 14.46,
1849
+ "learning_rate": 0.00015563767888032706,
1850
+ "loss": 0.5898,
1851
+ "step": 614000
1852
+ },
1853
+ {
1854
+ "epoch": 14.51,
1855
+ "learning_rate": 0.00015516590658908632,
1856
+ "loss": 0.5904,
1857
+ "step": 616000
1858
+ },
1859
+ {
1860
+ "epoch": 14.55,
1861
+ "learning_rate": 0.00015469413429784558,
1862
+ "loss": 0.5892,
1863
+ "step": 618000
1864
+ },
1865
+ {
1866
+ "epoch": 14.6,
1867
+ "learning_rate": 0.0001542223620066048,
1868
+ "loss": 0.5887,
1869
+ "step": 620000
1870
+ },
1871
+ {
1872
+ "epoch": 14.65,
1873
+ "learning_rate": 0.00015375058971536405,
1874
+ "loss": 0.5843,
1875
+ "step": 622000
1876
+ },
1877
+ {
1878
+ "epoch": 14.7,
1879
+ "learning_rate": 0.00015327881742412328,
1880
+ "loss": 0.5853,
1881
+ "step": 624000
1882
+ },
1883
+ {
1884
+ "epoch": 14.74,
1885
+ "learning_rate": 0.00015280704513288251,
1886
+ "loss": 0.5919,
1887
+ "step": 626000
1888
+ },
1889
+ {
1890
+ "epoch": 14.79,
1891
+ "learning_rate": 0.00015233527284164175,
1892
+ "loss": 0.5915,
1893
+ "step": 628000
1894
+ },
1895
+ {
1896
+ "epoch": 14.84,
1897
+ "learning_rate": 0.000151863500550401,
1898
+ "loss": 0.5918,
1899
+ "step": 630000
1900
+ },
1901
+ {
1902
+ "epoch": 14.88,
1903
+ "learning_rate": 0.0001513917282591602,
1904
+ "loss": 0.5932,
1905
+ "step": 632000
1906
+ },
1907
+ {
1908
+ "epoch": 14.93,
1909
+ "learning_rate": 0.00015091995596791947,
1910
+ "loss": 0.5891,
1911
+ "step": 634000
1912
+ },
1913
+ {
1914
+ "epoch": 14.98,
1915
+ "learning_rate": 0.00015044818367667873,
1916
+ "loss": 0.5888,
1917
+ "step": 636000
1918
+ },
1919
+ {
1920
+ "epoch": 15.03,
1921
+ "learning_rate": 0.00014997641138543794,
1922
+ "loss": 0.5817,
1923
+ "step": 638000
1924
+ },
1925
+ {
1926
+ "epoch": 15.07,
1927
+ "learning_rate": 0.00014950463909419717,
1928
+ "loss": 0.574,
1929
+ "step": 640000
1930
+ },
1931
+ {
1932
+ "epoch": 15.12,
1933
+ "learning_rate": 0.00014903286680295643,
1934
+ "loss": 0.5723,
1935
+ "step": 642000
1936
+ },
1937
+ {
1938
+ "epoch": 15.17,
1939
+ "learning_rate": 0.00014856109451171566,
1940
+ "loss": 0.573,
1941
+ "step": 644000
1942
+ },
1943
+ {
1944
+ "epoch": 15.21,
1945
+ "learning_rate": 0.00014808932222047492,
1946
+ "loss": 0.5758,
1947
+ "step": 646000
1948
+ },
1949
+ {
1950
+ "epoch": 15.26,
1951
+ "learning_rate": 0.00014761754992923416,
1952
+ "loss": 0.5796,
1953
+ "step": 648000
1954
+ },
1955
+ {
1956
+ "epoch": 15.31,
1957
+ "learning_rate": 0.0001471457776379934,
1958
+ "loss": 0.5722,
1959
+ "step": 650000
1960
+ },
1961
+ {
1962
+ "epoch": 15.36,
1963
+ "learning_rate": 0.00014667400534675262,
1964
+ "loss": 0.5825,
1965
+ "step": 652000
1966
+ },
1967
+ {
1968
+ "epoch": 15.4,
1969
+ "learning_rate": 0.00014620223305551185,
1970
+ "loss": 0.5744,
1971
+ "step": 654000
1972
+ },
1973
+ {
1974
+ "epoch": 15.45,
1975
+ "learning_rate": 0.0001457304607642711,
1976
+ "loss": 0.5786,
1977
+ "step": 656000
1978
+ },
1979
+ {
1980
+ "epoch": 15.5,
1981
+ "learning_rate": 0.00014525868847303032,
1982
+ "loss": 0.5808,
1983
+ "step": 658000
1984
+ },
1985
+ {
1986
+ "epoch": 15.54,
1987
+ "learning_rate": 0.00014478691618178958,
1988
+ "loss": 0.5855,
1989
+ "step": 660000
1990
+ },
1991
+ {
1992
+ "epoch": 15.59,
1993
+ "learning_rate": 0.0001443151438905488,
1994
+ "loss": 0.585,
1995
+ "step": 662000
1996
+ },
1997
+ {
1998
+ "epoch": 15.64,
1999
+ "learning_rate": 0.00014384337159930807,
2000
+ "loss": 0.5692,
2001
+ "step": 664000
2002
+ },
2003
+ {
2004
+ "epoch": 15.69,
2005
+ "learning_rate": 0.0001433715993080673,
2006
+ "loss": 0.5789,
2007
+ "step": 666000
2008
+ },
2009
+ {
2010
+ "epoch": 15.73,
2011
+ "learning_rate": 0.00014289982701682654,
2012
+ "loss": 0.5836,
2013
+ "step": 668000
2014
+ },
2015
+ {
2016
+ "epoch": 15.78,
2017
+ "learning_rate": 0.00014242805472558577,
2018
+ "loss": 0.5796,
2019
+ "step": 670000
2020
+ },
2021
+ {
2022
+ "epoch": 15.83,
2023
+ "learning_rate": 0.000141956282434345,
2024
+ "loss": 0.5756,
2025
+ "step": 672000
2026
+ },
2027
+ {
2028
+ "epoch": 15.87,
2029
+ "learning_rate": 0.00014148451014310424,
2030
+ "loss": 0.5667,
2031
+ "step": 674000
2032
+ },
2033
+ {
2034
+ "epoch": 15.92,
2035
+ "learning_rate": 0.0001410127378518635,
2036
+ "loss": 0.5845,
2037
+ "step": 676000
2038
+ },
2039
+ {
2040
+ "epoch": 15.97,
2041
+ "learning_rate": 0.00014054096556062273,
2042
+ "loss": 0.5783,
2043
+ "step": 678000
2044
+ },
2045
+ {
2046
+ "epoch": 16.02,
2047
+ "learning_rate": 0.00014006919326938196,
2048
+ "loss": 0.5749,
2049
+ "step": 680000
2050
+ },
2051
+ {
2052
+ "epoch": 16.06,
2053
+ "learning_rate": 0.00013959742097814122,
2054
+ "loss": 0.5602,
2055
+ "step": 682000
2056
+ },
2057
+ {
2058
+ "epoch": 16.11,
2059
+ "learning_rate": 0.00013912564868690045,
2060
+ "loss": 0.564,
2061
+ "step": 684000
2062
+ },
2063
+ {
2064
+ "epoch": 16.16,
2065
+ "learning_rate": 0.00013865387639565969,
2066
+ "loss": 0.5659,
2067
+ "step": 686000
2068
+ },
2069
+ {
2070
+ "epoch": 16.2,
2071
+ "learning_rate": 0.00013818210410441892,
2072
+ "loss": 0.5608,
2073
+ "step": 688000
2074
+ },
2075
+ {
2076
+ "epoch": 16.25,
2077
+ "learning_rate": 0.00013771033181317815,
2078
+ "loss": 0.5668,
2079
+ "step": 690000
2080
+ },
2081
+ {
2082
+ "epoch": 16.3,
2083
+ "learning_rate": 0.00013723855952193738,
2084
+ "loss": 0.5646,
2085
+ "step": 692000
2086
+ },
2087
+ {
2088
+ "epoch": 16.34,
2089
+ "learning_rate": 0.00013676678723069664,
2090
+ "loss": 0.5663,
2091
+ "step": 694000
2092
+ },
2093
+ {
2094
+ "epoch": 16.39,
2095
+ "learning_rate": 0.00013629501493945588,
2096
+ "loss": 0.5716,
2097
+ "step": 696000
2098
+ },
2099
+ {
2100
+ "epoch": 16.44,
2101
+ "learning_rate": 0.0001358232426482151,
2102
+ "loss": 0.5716,
2103
+ "step": 698000
2104
+ },
2105
+ {
2106
+ "epoch": 16.49,
2107
+ "learning_rate": 0.00013535147035697437,
2108
+ "loss": 0.568,
2109
+ "step": 700000
2110
+ },
2111
+ {
2112
+ "epoch": 16.53,
2113
+ "learning_rate": 0.0001348796980657336,
2114
+ "loss": 0.566,
2115
+ "step": 702000
2116
+ },
2117
+ {
2118
+ "epoch": 16.58,
2119
+ "learning_rate": 0.00013440792577449284,
2120
+ "loss": 0.5657,
2121
+ "step": 704000
2122
+ },
2123
+ {
2124
+ "epoch": 16.63,
2125
+ "learning_rate": 0.00013393615348325207,
2126
+ "loss": 0.5641,
2127
+ "step": 706000
2128
+ },
2129
+ {
2130
+ "epoch": 16.67,
2131
+ "learning_rate": 0.0001334643811920113,
2132
+ "loss": 0.5679,
2133
+ "step": 708000
2134
+ },
2135
+ {
2136
+ "epoch": 16.72,
2137
+ "learning_rate": 0.00013299260890077056,
2138
+ "loss": 0.5652,
2139
+ "step": 710000
2140
+ },
2141
+ {
2142
+ "epoch": 16.77,
2143
+ "learning_rate": 0.0001325208366095298,
2144
+ "loss": 0.5672,
2145
+ "step": 712000
2146
+ },
2147
+ {
2148
+ "epoch": 16.82,
2149
+ "learning_rate": 0.00013204906431828903,
2150
+ "loss": 0.5695,
2151
+ "step": 714000
2152
+ },
2153
+ {
2154
+ "epoch": 16.86,
2155
+ "learning_rate": 0.00013157729202704826,
2156
+ "loss": 0.5719,
2157
+ "step": 716000
2158
+ },
2159
+ {
2160
+ "epoch": 16.91,
2161
+ "learning_rate": 0.00013110551973580752,
2162
+ "loss": 0.571,
2163
+ "step": 718000
2164
+ },
2165
+ {
2166
+ "epoch": 16.96,
2167
+ "learning_rate": 0.00013063374744456675,
2168
+ "loss": 0.5684,
2169
+ "step": 720000
2170
+ },
2171
+ {
2172
+ "epoch": 17.0,
2173
+ "learning_rate": 0.00013016197515332598,
2174
+ "loss": 0.5672,
2175
+ "step": 722000
2176
+ },
2177
+ {
2178
+ "epoch": 17.05,
2179
+ "learning_rate": 0.00012969020286208522,
2180
+ "loss": 0.5566,
2181
+ "step": 724000
2182
+ },
2183
+ {
2184
+ "epoch": 17.1,
2185
+ "learning_rate": 0.00012921843057084445,
2186
+ "loss": 0.5459,
2187
+ "step": 726000
2188
+ },
2189
+ {
2190
+ "epoch": 17.15,
2191
+ "learning_rate": 0.0001287466582796037,
2192
+ "loss": 0.5522,
2193
+ "step": 728000
2194
+ },
2195
+ {
2196
+ "epoch": 17.19,
2197
+ "learning_rate": 0.00012827488598836294,
2198
+ "loss": 0.5586,
2199
+ "step": 730000
2200
+ },
2201
+ {
2202
+ "epoch": 17.24,
2203
+ "learning_rate": 0.00012780311369712217,
2204
+ "loss": 0.5551,
2205
+ "step": 732000
2206
+ },
2207
+ {
2208
+ "epoch": 17.29,
2209
+ "learning_rate": 0.0001273313414058814,
2210
+ "loss": 0.558,
2211
+ "step": 734000
2212
+ },
2213
+ {
2214
+ "epoch": 17.33,
2215
+ "learning_rate": 0.00012685956911464067,
2216
+ "loss": 0.5501,
2217
+ "step": 736000
2218
+ },
2219
+ {
2220
+ "epoch": 17.38,
2221
+ "learning_rate": 0.0001263877968233999,
2222
+ "loss": 0.5496,
2223
+ "step": 738000
2224
+ },
2225
+ {
2226
+ "epoch": 17.43,
2227
+ "learning_rate": 0.00012591602453215913,
2228
+ "loss": 0.5617,
2229
+ "step": 740000
2230
+ },
2231
+ {
2232
+ "epoch": 17.48,
2233
+ "learning_rate": 0.00012544425224091837,
2234
+ "loss": 0.5481,
2235
+ "step": 742000
2236
+ },
2237
+ {
2238
+ "epoch": 17.52,
2239
+ "learning_rate": 0.00012497247994967763,
2240
+ "loss": 0.5627,
2241
+ "step": 744000
2242
+ },
2243
+ {
2244
+ "epoch": 17.57,
2245
+ "learning_rate": 0.00012450070765843686,
2246
+ "loss": 0.5556,
2247
+ "step": 746000
2248
+ },
2249
+ {
2250
+ "epoch": 17.62,
2251
+ "learning_rate": 0.0001240289353671961,
2252
+ "loss": 0.5565,
2253
+ "step": 748000
2254
+ },
2255
+ {
2256
+ "epoch": 17.66,
2257
+ "learning_rate": 0.00012355716307595532,
2258
+ "loss": 0.5591,
2259
+ "step": 750000
2260
+ },
2261
+ {
2262
+ "epoch": 17.71,
2263
+ "learning_rate": 0.00012308539078471456,
2264
+ "loss": 0.5526,
2265
+ "step": 752000
2266
+ },
2267
+ {
2268
+ "epoch": 17.76,
2269
+ "learning_rate": 0.00012261361849347382,
2270
+ "loss": 0.5564,
2271
+ "step": 754000
2272
+ },
2273
+ {
2274
+ "epoch": 17.8,
2275
+ "learning_rate": 0.00012214184620223305,
2276
+ "loss": 0.5541,
2277
+ "step": 756000
2278
+ },
2279
+ {
2280
+ "epoch": 17.85,
2281
+ "learning_rate": 0.00012167007391099228,
2282
+ "loss": 0.5601,
2283
+ "step": 758000
2284
+ },
2285
+ {
2286
+ "epoch": 17.9,
2287
+ "learning_rate": 0.00012119830161975153,
2288
+ "loss": 0.5555,
2289
+ "step": 760000
2290
+ },
2291
+ {
2292
+ "epoch": 17.95,
2293
+ "learning_rate": 0.00012072652932851076,
2294
+ "loss": 0.5525,
2295
+ "step": 762000
2296
+ },
2297
+ {
2298
+ "epoch": 17.99,
2299
+ "learning_rate": 0.00012025475703727,
2300
+ "loss": 0.5577,
2301
+ "step": 764000
2302
+ },
2303
+ {
2304
+ "epoch": 18.04,
2305
+ "learning_rate": 0.00011978298474602924,
2306
+ "loss": 0.5403,
2307
+ "step": 766000
2308
+ },
2309
+ {
2310
+ "epoch": 18.09,
2311
+ "learning_rate": 0.00011931121245478847,
2312
+ "loss": 0.5466,
2313
+ "step": 768000
2314
+ },
2315
+ {
2316
+ "epoch": 18.13,
2317
+ "learning_rate": 0.0001188394401635477,
2318
+ "loss": 0.546,
2319
+ "step": 770000
2320
+ },
2321
+ {
2322
+ "epoch": 18.18,
2323
+ "learning_rate": 0.00011836766787230695,
2324
+ "loss": 0.5435,
2325
+ "step": 772000
2326
+ },
2327
+ {
2328
+ "epoch": 18.23,
2329
+ "learning_rate": 0.0001178958955810662,
2330
+ "loss": 0.5442,
2331
+ "step": 774000
2332
+ },
2333
+ {
2334
+ "epoch": 18.28,
2335
+ "learning_rate": 0.00011742412328982544,
2336
+ "loss": 0.5381,
2337
+ "step": 776000
2338
+ },
2339
+ {
2340
+ "epoch": 18.32,
2341
+ "learning_rate": 0.00011695235099858468,
2342
+ "loss": 0.542,
2343
+ "step": 778000
2344
+ },
2345
+ {
2346
+ "epoch": 18.37,
2347
+ "learning_rate": 0.00011648057870734391,
2348
+ "loss": 0.5458,
2349
+ "step": 780000
2350
+ },
2351
+ {
2352
+ "epoch": 18.42,
2353
+ "learning_rate": 0.00011600880641610316,
2354
+ "loss": 0.5473,
2355
+ "step": 782000
2356
+ },
2357
+ {
2358
+ "epoch": 18.46,
2359
+ "learning_rate": 0.00011553703412486239,
2360
+ "loss": 0.5511,
2361
+ "step": 784000
2362
+ },
2363
+ {
2364
+ "epoch": 18.51,
2365
+ "learning_rate": 0.00011506526183362162,
2366
+ "loss": 0.5415,
2367
+ "step": 786000
2368
+ },
2369
+ {
2370
+ "epoch": 18.56,
2371
+ "learning_rate": 0.00011459348954238087,
2372
+ "loss": 0.5431,
2373
+ "step": 788000
2374
+ },
2375
+ {
2376
+ "epoch": 18.61,
2377
+ "learning_rate": 0.0001141217172511401,
2378
+ "loss": 0.5403,
2379
+ "step": 790000
2380
+ },
2381
+ {
2382
+ "epoch": 18.65,
2383
+ "learning_rate": 0.00011364994495989935,
2384
+ "loss": 0.5402,
2385
+ "step": 792000
2386
+ },
2387
+ {
2388
+ "epoch": 18.7,
2389
+ "learning_rate": 0.00011317817266865859,
2390
+ "loss": 0.5447,
2391
+ "step": 794000
2392
+ },
2393
+ {
2394
+ "epoch": 18.75,
2395
+ "learning_rate": 0.00011270640037741783,
2396
+ "loss": 0.5485,
2397
+ "step": 796000
2398
+ },
2399
+ {
2400
+ "epoch": 18.79,
2401
+ "learning_rate": 0.00011223462808617706,
2402
+ "loss": 0.5485,
2403
+ "step": 798000
2404
+ },
2405
+ {
2406
+ "epoch": 18.84,
2407
+ "learning_rate": 0.0001117628557949363,
2408
+ "loss": 0.5441,
2409
+ "step": 800000
2410
+ },
2411
+ {
2412
+ "epoch": 18.89,
2413
+ "learning_rate": 0.00011129108350369554,
2414
+ "loss": 0.5457,
2415
+ "step": 802000
2416
+ },
2417
+ {
2418
+ "epoch": 18.94,
2419
+ "learning_rate": 0.00011081931121245477,
2420
+ "loss": 0.5416,
2421
+ "step": 804000
2422
+ },
2423
+ {
2424
+ "epoch": 18.98,
2425
+ "learning_rate": 0.00011034753892121402,
2426
+ "loss": 0.5494,
2427
+ "step": 806000
2428
+ },
2429
+ {
2430
+ "epoch": 19.03,
2431
+ "learning_rate": 0.00010987576662997325,
2432
+ "loss": 0.5347,
2433
+ "step": 808000
2434
+ },
2435
+ {
2436
+ "epoch": 19.08,
2437
+ "learning_rate": 0.00010940399433873251,
2438
+ "loss": 0.5257,
2439
+ "step": 810000
2440
+ },
2441
+ {
2442
+ "epoch": 19.12,
2443
+ "learning_rate": 0.00010893222204749174,
2444
+ "loss": 0.5278,
2445
+ "step": 812000
2446
+ },
2447
+ {
2448
+ "epoch": 19.17,
2449
+ "learning_rate": 0.00010846044975625097,
2450
+ "loss": 0.5355,
2451
+ "step": 814000
2452
+ },
2453
+ {
2454
+ "epoch": 19.22,
2455
+ "learning_rate": 0.00010798867746501022,
2456
+ "loss": 0.5378,
2457
+ "step": 816000
2458
+ },
2459
+ {
2460
+ "epoch": 19.27,
2461
+ "learning_rate": 0.00010751690517376945,
2462
+ "loss": 0.537,
2463
+ "step": 818000
2464
+ },
2465
+ {
2466
+ "epoch": 19.31,
2467
+ "learning_rate": 0.00010704513288252869,
2468
+ "loss": 0.5385,
2469
+ "step": 820000
2470
+ },
2471
+ {
2472
+ "epoch": 19.36,
2473
+ "learning_rate": 0.00010657336059128793,
2474
+ "loss": 0.5302,
2475
+ "step": 822000
2476
+ },
2477
+ {
2478
+ "epoch": 19.41,
2479
+ "learning_rate": 0.00010610158830004717,
2480
+ "loss": 0.5409,
2481
+ "step": 824000
2482
+ },
2483
+ {
2484
+ "epoch": 19.45,
2485
+ "learning_rate": 0.0001056298160088064,
2486
+ "loss": 0.5347,
2487
+ "step": 826000
2488
+ },
2489
+ {
2490
+ "epoch": 19.5,
2491
+ "learning_rate": 0.00010515804371756563,
2492
+ "loss": 0.5288,
2493
+ "step": 828000
2494
+ },
2495
+ {
2496
+ "epoch": 19.55,
2497
+ "learning_rate": 0.00010468627142632489,
2498
+ "loss": 0.5276,
2499
+ "step": 830000
2500
+ },
2501
+ {
2502
+ "epoch": 19.59,
2503
+ "learning_rate": 0.00010421449913508412,
2504
+ "loss": 0.5334,
2505
+ "step": 832000
2506
+ },
2507
+ {
2508
+ "epoch": 19.64,
2509
+ "learning_rate": 0.00010374272684384337,
2510
+ "loss": 0.5407,
2511
+ "step": 834000
2512
+ },
2513
+ {
2514
+ "epoch": 19.69,
2515
+ "learning_rate": 0.0001032709545526026,
2516
+ "loss": 0.5354,
2517
+ "step": 836000
2518
+ },
2519
+ {
2520
+ "epoch": 19.74,
2521
+ "learning_rate": 0.00010279918226136184,
2522
+ "loss": 0.5274,
2523
+ "step": 838000
2524
+ },
2525
+ {
2526
+ "epoch": 19.78,
2527
+ "learning_rate": 0.00010232740997012108,
2528
+ "loss": 0.529,
2529
+ "step": 840000
2530
+ },
2531
+ {
2532
+ "epoch": 19.83,
2533
+ "learning_rate": 0.00010185563767888031,
2534
+ "loss": 0.5327,
2535
+ "step": 842000
2536
+ },
2537
+ {
2538
+ "epoch": 19.88,
2539
+ "learning_rate": 0.00010138386538763955,
2540
+ "loss": 0.5383,
2541
+ "step": 844000
2542
+ },
2543
+ {
2544
+ "epoch": 19.92,
2545
+ "learning_rate": 0.0001009120930963988,
2546
+ "loss": 0.5352,
2547
+ "step": 846000
2548
+ },
2549
+ {
2550
+ "epoch": 19.97,
2551
+ "learning_rate": 0.00010044032080515804,
2552
+ "loss": 0.5355,
2553
+ "step": 848000
2554
+ },
2555
+ {
2556
+ "epoch": 20.02,
2557
+ "learning_rate": 9.996854851391727e-05,
2558
+ "loss": 0.5233,
2559
+ "step": 850000
2560
+ },
2561
+ {
2562
+ "epoch": 20.07,
2563
+ "learning_rate": 9.949677622267652e-05,
2564
+ "loss": 0.5255,
2565
+ "step": 852000
2566
+ },
2567
+ {
2568
+ "epoch": 20.11,
2569
+ "learning_rate": 9.902500393143575e-05,
2570
+ "loss": 0.5168,
2571
+ "step": 854000
2572
+ },
2573
+ {
2574
+ "epoch": 20.16,
2575
+ "learning_rate": 9.855323164019498e-05,
2576
+ "loss": 0.5181,
2577
+ "step": 856000
2578
+ },
2579
+ {
2580
+ "epoch": 20.21,
2581
+ "learning_rate": 9.808145934895423e-05,
2582
+ "loss": 0.5245,
2583
+ "step": 858000
2584
+ },
2585
+ {
2586
+ "epoch": 20.25,
2587
+ "learning_rate": 9.760968705771346e-05,
2588
+ "loss": 0.5191,
2589
+ "step": 860000
2590
+ },
2591
+ {
2592
+ "epoch": 20.3,
2593
+ "learning_rate": 9.71379147664727e-05,
2594
+ "loss": 0.5207,
2595
+ "step": 862000
2596
+ },
2597
+ {
2598
+ "epoch": 20.35,
2599
+ "learning_rate": 9.666614247523194e-05,
2600
+ "loss": 0.5251,
2601
+ "step": 864000
2602
+ },
2603
+ {
2604
+ "epoch": 20.4,
2605
+ "learning_rate": 9.619437018399119e-05,
2606
+ "loss": 0.5158,
2607
+ "step": 866000
2608
+ },
2609
+ {
2610
+ "epoch": 20.44,
2611
+ "learning_rate": 9.572259789275043e-05,
2612
+ "loss": 0.5178,
2613
+ "step": 868000
2614
+ },
2615
+ {
2616
+ "epoch": 20.49,
2617
+ "learning_rate": 9.525082560150967e-05,
2618
+ "loss": 0.5213,
2619
+ "step": 870000
2620
+ },
2621
+ {
2622
+ "epoch": 20.54,
2623
+ "learning_rate": 9.47790533102689e-05,
2624
+ "loss": 0.5271,
2625
+ "step": 872000
2626
+ },
2627
+ {
2628
+ "epoch": 20.58,
2629
+ "learning_rate": 9.430728101902815e-05,
2630
+ "loss": 0.5242,
2631
+ "step": 874000
2632
+ },
2633
+ {
2634
+ "epoch": 20.63,
2635
+ "learning_rate": 9.383550872778738e-05,
2636
+ "loss": 0.5278,
2637
+ "step": 876000
2638
+ },
2639
+ {
2640
+ "epoch": 20.68,
2641
+ "learning_rate": 9.336373643654661e-05,
2642
+ "loss": 0.5222,
2643
+ "step": 878000
2644
+ },
2645
+ {
2646
+ "epoch": 20.73,
2647
+ "learning_rate": 9.289196414530586e-05,
2648
+ "loss": 0.5242,
2649
+ "step": 880000
2650
+ },
2651
+ {
2652
+ "epoch": 20.77,
2653
+ "learning_rate": 9.242019185406509e-05,
2654
+ "loss": 0.5271,
2655
+ "step": 882000
2656
+ },
2657
+ {
2658
+ "epoch": 20.82,
2659
+ "learning_rate": 9.194841956282434e-05,
2660
+ "loss": 0.5267,
2661
+ "step": 884000
2662
+ },
2663
+ {
2664
+ "epoch": 20.87,
2665
+ "learning_rate": 9.147664727158358e-05,
2666
+ "loss": 0.5195,
2667
+ "step": 886000
2668
+ },
2669
+ {
2670
+ "epoch": 20.91,
2671
+ "learning_rate": 9.100487498034282e-05,
2672
+ "loss": 0.5162,
2673
+ "step": 888000
2674
+ },
2675
+ {
2676
+ "epoch": 20.96,
2677
+ "learning_rate": 9.053310268910205e-05,
2678
+ "loss": 0.5166,
2679
+ "step": 890000
2680
+ },
2681
+ {
2682
+ "epoch": 21.01,
2683
+ "learning_rate": 9.00613303978613e-05,
2684
+ "loss": 0.5266,
2685
+ "step": 892000
2686
+ },
2687
+ {
2688
+ "epoch": 21.06,
2689
+ "learning_rate": 8.958955810662053e-05,
2690
+ "loss": 0.5149,
2691
+ "step": 894000
2692
+ },
2693
+ {
2694
+ "epoch": 21.1,
2695
+ "learning_rate": 8.911778581537976e-05,
2696
+ "loss": 0.5023,
2697
+ "step": 896000
2698
+ },
2699
+ {
2700
+ "epoch": 21.15,
2701
+ "learning_rate": 8.864601352413901e-05,
2702
+ "loss": 0.5112,
2703
+ "step": 898000
2704
+ },
2705
+ {
2706
+ "epoch": 21.2,
2707
+ "learning_rate": 8.817424123289824e-05,
2708
+ "loss": 0.5084,
2709
+ "step": 900000
2710
+ },
2711
+ {
2712
+ "epoch": 21.24,
2713
+ "learning_rate": 8.770246894165747e-05,
2714
+ "loss": 0.509,
2715
+ "step": 902000
2716
+ },
2717
+ {
2718
+ "epoch": 21.29,
2719
+ "learning_rate": 8.723069665041673e-05,
2720
+ "loss": 0.5043,
2721
+ "step": 904000
2722
+ },
2723
+ {
2724
+ "epoch": 21.34,
2725
+ "learning_rate": 8.675892435917597e-05,
2726
+ "loss": 0.5188,
2727
+ "step": 906000
2728
+ },
2729
+ {
2730
+ "epoch": 21.38,
2731
+ "learning_rate": 8.628715206793521e-05,
2732
+ "loss": 0.5094,
2733
+ "step": 908000
2734
+ },
2735
+ {
2736
+ "epoch": 21.43,
2737
+ "learning_rate": 8.581537977669444e-05,
2738
+ "loss": 0.519,
2739
+ "step": 910000
2740
+ },
2741
+ {
2742
+ "epoch": 21.48,
2743
+ "learning_rate": 8.534360748545368e-05,
2744
+ "loss": 0.5108,
2745
+ "step": 912000
2746
+ },
2747
+ {
2748
+ "epoch": 21.53,
2749
+ "learning_rate": 8.487183519421292e-05,
2750
+ "loss": 0.5054,
2751
+ "step": 914000
2752
+ },
2753
+ {
2754
+ "epoch": 21.57,
2755
+ "learning_rate": 8.440006290297216e-05,
2756
+ "loss": 0.5092,
2757
+ "step": 916000
2758
+ },
2759
+ {
2760
+ "epoch": 21.62,
2761
+ "learning_rate": 8.392829061173139e-05,
2762
+ "loss": 0.5075,
2763
+ "step": 918000
2764
+ },
2765
+ {
2766
+ "epoch": 21.67,
2767
+ "learning_rate": 8.345651832049062e-05,
2768
+ "loss": 0.5024,
2769
+ "step": 920000
2770
+ },
2771
+ {
2772
+ "epoch": 21.71,
2773
+ "learning_rate": 8.298474602924988e-05,
2774
+ "loss": 0.5161,
2775
+ "step": 922000
2776
+ },
2777
+ {
2778
+ "epoch": 21.76,
2779
+ "learning_rate": 8.251297373800911e-05,
2780
+ "loss": 0.5117,
2781
+ "step": 924000
2782
+ },
2783
+ {
2784
+ "epoch": 21.81,
2785
+ "learning_rate": 8.204120144676836e-05,
2786
+ "loss": 0.5136,
2787
+ "step": 926000
2788
+ },
2789
+ {
2790
+ "epoch": 21.86,
2791
+ "learning_rate": 8.15694291555276e-05,
2792
+ "loss": 0.5062,
2793
+ "step": 928000
2794
+ },
2795
+ {
2796
+ "epoch": 21.9,
2797
+ "learning_rate": 8.109765686428683e-05,
2798
+ "loss": 0.5122,
2799
+ "step": 930000
2800
+ },
2801
+ {
2802
+ "epoch": 21.95,
2803
+ "learning_rate": 8.062588457304607e-05,
2804
+ "loss": 0.5113,
2805
+ "step": 932000
2806
+ },
2807
+ {
2808
+ "epoch": 22.0,
2809
+ "learning_rate": 8.01541122818053e-05,
2810
+ "loss": 0.5073,
2811
+ "step": 934000
2812
+ },
2813
+ {
2814
+ "epoch": 22.04,
2815
+ "learning_rate": 7.968233999056454e-05,
2816
+ "loss": 0.4986,
2817
+ "step": 936000
2818
+ },
2819
+ {
2820
+ "epoch": 22.09,
2821
+ "learning_rate": 7.921056769932378e-05,
2822
+ "loss": 0.5061,
2823
+ "step": 938000
2824
+ },
2825
+ {
2826
+ "epoch": 22.14,
2827
+ "learning_rate": 7.873879540808303e-05,
2828
+ "loss": 0.4963,
2829
+ "step": 940000
2830
+ },
2831
+ {
2832
+ "epoch": 22.19,
2833
+ "learning_rate": 7.826702311684228e-05,
2834
+ "loss": 0.4981,
2835
+ "step": 942000
2836
+ },
2837
+ {
2838
+ "epoch": 22.23,
2839
+ "learning_rate": 7.779525082560151e-05,
2840
+ "loss": 0.498,
2841
+ "step": 944000
2842
+ },
2843
+ {
2844
+ "epoch": 22.28,
2845
+ "learning_rate": 7.732347853436074e-05,
2846
+ "loss": 0.4963,
2847
+ "step": 946000
2848
+ },
2849
+ {
2850
+ "epoch": 22.33,
2851
+ "learning_rate": 7.685170624311998e-05,
2852
+ "loss": 0.4987,
2853
+ "step": 948000
2854
+ },
2855
+ {
2856
+ "epoch": 22.37,
2857
+ "learning_rate": 7.637993395187922e-05,
2858
+ "loss": 0.4999,
2859
+ "step": 950000
2860
+ },
2861
+ {
2862
+ "epoch": 22.42,
2863
+ "learning_rate": 7.590816166063845e-05,
2864
+ "loss": 0.4945,
2865
+ "step": 952000
2866
+ },
2867
+ {
2868
+ "epoch": 22.47,
2869
+ "learning_rate": 7.543638936939769e-05,
2870
+ "loss": 0.4986,
2871
+ "step": 954000
2872
+ },
2873
+ {
2874
+ "epoch": 22.52,
2875
+ "learning_rate": 7.496461707815693e-05,
2876
+ "loss": 0.4956,
2877
+ "step": 956000
2878
+ },
2879
+ {
2880
+ "epoch": 22.56,
2881
+ "learning_rate": 7.449284478691618e-05,
2882
+ "loss": 0.4958,
2883
+ "step": 958000
2884
+ },
2885
+ {
2886
+ "epoch": 22.61,
2887
+ "learning_rate": 7.402107249567541e-05,
2888
+ "loss": 0.5033,
2889
+ "step": 960000
2890
+ },
2891
+ {
2892
+ "epoch": 22.66,
2893
+ "learning_rate": 7.354930020443464e-05,
2894
+ "loss": 0.4976,
2895
+ "step": 962000
2896
+ },
2897
+ {
2898
+ "epoch": 22.7,
2899
+ "learning_rate": 7.307752791319389e-05,
2900
+ "loss": 0.5036,
2901
+ "step": 964000
2902
+ },
2903
+ {
2904
+ "epoch": 22.75,
2905
+ "learning_rate": 7.260575562195314e-05,
2906
+ "loss": 0.4973,
2907
+ "step": 966000
2908
+ },
2909
+ {
2910
+ "epoch": 22.8,
2911
+ "learning_rate": 7.213398333071237e-05,
2912
+ "loss": 0.501,
2913
+ "step": 968000
2914
+ },
2915
+ {
2916
+ "epoch": 22.85,
2917
+ "learning_rate": 7.16622110394716e-05,
2918
+ "loss": 0.5005,
2919
+ "step": 970000
2920
+ },
2921
+ {
2922
+ "epoch": 22.89,
2923
+ "learning_rate": 7.119043874823085e-05,
2924
+ "loss": 0.4991,
2925
+ "step": 972000
2926
+ },
2927
+ {
2928
+ "epoch": 22.94,
2929
+ "learning_rate": 7.07186664569901e-05,
2930
+ "loss": 0.495,
2931
+ "step": 974000
2932
+ },
2933
+ {
2934
+ "epoch": 22.99,
2935
+ "learning_rate": 7.024689416574933e-05,
2936
+ "loss": 0.5016,
2937
+ "step": 976000
2938
+ },
2939
+ {
2940
+ "epoch": 23.03,
2941
+ "learning_rate": 6.977512187450856e-05,
2942
+ "loss": 0.4966,
2943
+ "step": 978000
2944
+ },
2945
+ {
2946
+ "epoch": 23.08,
2947
+ "learning_rate": 6.93033495832678e-05,
2948
+ "loss": 0.4909,
2949
+ "step": 980000
2950
+ },
2951
+ {
2952
+ "epoch": 23.13,
2953
+ "learning_rate": 6.883157729202704e-05,
2954
+ "loss": 0.4846,
2955
+ "step": 982000
2956
+ },
2957
+ {
2958
+ "epoch": 23.17,
2959
+ "learning_rate": 6.835980500078629e-05,
2960
+ "loss": 0.4938,
2961
+ "step": 984000
2962
+ },
2963
+ {
2964
+ "epoch": 23.22,
2965
+ "learning_rate": 6.788803270954552e-05,
2966
+ "loss": 0.4967,
2967
+ "step": 986000
2968
+ },
2969
+ {
2970
+ "epoch": 23.27,
2971
+ "learning_rate": 6.741626041830475e-05,
2972
+ "loss": 0.4894,
2973
+ "step": 988000
2974
+ },
2975
+ {
2976
+ "epoch": 23.32,
2977
+ "learning_rate": 6.6944488127064e-05,
2978
+ "loss": 0.4912,
2979
+ "step": 990000
2980
+ },
2981
+ {
2982
+ "epoch": 23.36,
2983
+ "learning_rate": 6.647271583582324e-05,
2984
+ "loss": 0.4898,
2985
+ "step": 992000
2986
+ },
2987
+ {
2988
+ "epoch": 23.41,
2989
+ "learning_rate": 6.600094354458248e-05,
2990
+ "loss": 0.4854,
2991
+ "step": 994000
2992
+ },
2993
+ {
2994
+ "epoch": 23.46,
2995
+ "learning_rate": 6.552917125334171e-05,
2996
+ "loss": 0.4838,
2997
+ "step": 996000
2998
+ },
2999
+ {
3000
+ "epoch": 23.5,
3001
+ "learning_rate": 6.505739896210096e-05,
3002
+ "loss": 0.4846,
3003
+ "step": 998000
3004
+ },
3005
+ {
3006
+ "epoch": 23.55,
3007
+ "learning_rate": 6.45856266708602e-05,
3008
+ "loss": 0.4829,
3009
+ "step": 1000000
3010
+ },
3011
+ {
3012
+ "epoch": 23.6,
3013
+ "learning_rate": 6.411385437961944e-05,
3014
+ "loss": 0.4792,
3015
+ "step": 1002000
3016
+ },
3017
+ {
3018
+ "epoch": 23.65,
3019
+ "learning_rate": 6.364208208837867e-05,
3020
+ "loss": 0.486,
3021
+ "step": 1004000
3022
+ },
3023
+ {
3024
+ "epoch": 23.69,
3025
+ "learning_rate": 6.317030979713791e-05,
3026
+ "loss": 0.4946,
3027
+ "step": 1006000
3028
+ },
3029
+ {
3030
+ "epoch": 23.74,
3031
+ "learning_rate": 6.269853750589715e-05,
3032
+ "loss": 0.4825,
3033
+ "step": 1008000
3034
+ },
3035
+ {
3036
+ "epoch": 23.79,
3037
+ "learning_rate": 6.22267652146564e-05,
3038
+ "loss": 0.4845,
3039
+ "step": 1010000
3040
+ },
3041
+ {
3042
+ "epoch": 23.83,
3043
+ "learning_rate": 6.175499292341563e-05,
3044
+ "loss": 0.4875,
3045
+ "step": 1012000
3046
+ },
3047
+ {
3048
+ "epoch": 23.88,
3049
+ "learning_rate": 6.128322063217486e-05,
3050
+ "loss": 0.4935,
3051
+ "step": 1014000
3052
+ },
3053
+ {
3054
+ "epoch": 23.93,
3055
+ "learning_rate": 6.0811448340934105e-05,
3056
+ "loss": 0.4834,
3057
+ "step": 1016000
3058
+ },
3059
+ {
3060
+ "epoch": 23.98,
3061
+ "learning_rate": 6.033967604969334e-05,
3062
+ "loss": 0.4853,
3063
+ "step": 1018000
3064
+ },
3065
+ {
3066
+ "epoch": 24.02,
3067
+ "learning_rate": 5.9867903758452584e-05,
3068
+ "loss": 0.485,
3069
+ "step": 1020000
3070
+ },
3071
+ {
3072
+ "epoch": 24.07,
3073
+ "learning_rate": 5.9396131467211824e-05,
3074
+ "loss": 0.4821,
3075
+ "step": 1022000
3076
+ },
3077
+ {
3078
+ "epoch": 24.12,
3079
+ "learning_rate": 5.8924359175971057e-05,
3080
+ "loss": 0.4739,
3081
+ "step": 1024000
3082
+ },
3083
+ {
3084
+ "epoch": 24.16,
3085
+ "learning_rate": 5.8452586884730296e-05,
3086
+ "loss": 0.4824,
3087
+ "step": 1026000
3088
+ },
3089
+ {
3090
+ "epoch": 24.21,
3091
+ "learning_rate": 5.798081459348954e-05,
3092
+ "loss": 0.4823,
3093
+ "step": 1028000
3094
+ },
3095
+ {
3096
+ "epoch": 24.26,
3097
+ "learning_rate": 5.7509042302248775e-05,
3098
+ "loss": 0.4805,
3099
+ "step": 1030000
3100
+ },
3101
+ {
3102
+ "epoch": 24.31,
3103
+ "learning_rate": 5.7037270011008015e-05,
3104
+ "loss": 0.4727,
3105
+ "step": 1032000
3106
+ },
3107
+ {
3108
+ "epoch": 24.35,
3109
+ "learning_rate": 5.6565497719767254e-05,
3110
+ "loss": 0.4726,
3111
+ "step": 1034000
3112
+ },
3113
+ {
3114
+ "epoch": 24.4,
3115
+ "learning_rate": 5.609372542852649e-05,
3116
+ "loss": 0.4734,
3117
+ "step": 1036000
3118
+ },
3119
+ {
3120
+ "epoch": 24.45,
3121
+ "learning_rate": 5.562195313728573e-05,
3122
+ "loss": 0.4775,
3123
+ "step": 1038000
3124
+ },
3125
+ {
3126
+ "epoch": 24.49,
3127
+ "learning_rate": 5.515018084604497e-05,
3128
+ "loss": 0.4761,
3129
+ "step": 1040000
3130
+ },
3131
+ {
3132
+ "epoch": 24.54,
3133
+ "learning_rate": 5.467840855480421e-05,
3134
+ "loss": 0.4692,
3135
+ "step": 1042000
3136
+ },
3137
+ {
3138
+ "epoch": 24.59,
3139
+ "learning_rate": 5.4206636263563445e-05,
3140
+ "loss": 0.4826,
3141
+ "step": 1044000
3142
+ },
3143
+ {
3144
+ "epoch": 24.63,
3145
+ "learning_rate": 5.3734863972322684e-05,
3146
+ "loss": 0.4767,
3147
+ "step": 1046000
3148
+ },
3149
+ {
3150
+ "epoch": 24.68,
3151
+ "learning_rate": 5.326309168108193e-05,
3152
+ "loss": 0.4733,
3153
+ "step": 1048000
3154
+ },
3155
+ {
3156
+ "epoch": 24.73,
3157
+ "learning_rate": 5.2791319389841163e-05,
3158
+ "loss": 0.4772,
3159
+ "step": 1050000
3160
+ },
3161
+ {
3162
+ "epoch": 24.78,
3163
+ "learning_rate": 5.23195470986004e-05,
3164
+ "loss": 0.4758,
3165
+ "step": 1052000
3166
+ },
3167
+ {
3168
+ "epoch": 24.82,
3169
+ "learning_rate": 5.184777480735964e-05,
3170
+ "loss": 0.48,
3171
+ "step": 1054000
3172
+ },
3173
+ {
3174
+ "epoch": 24.87,
3175
+ "learning_rate": 5.137600251611889e-05,
3176
+ "loss": 0.477,
3177
+ "step": 1056000
3178
+ },
3179
+ {
3180
+ "epoch": 24.92,
3181
+ "learning_rate": 5.090423022487812e-05,
3182
+ "loss": 0.4715,
3183
+ "step": 1058000
3184
+ },
3185
+ {
3186
+ "epoch": 24.96,
3187
+ "learning_rate": 5.043245793363736e-05,
3188
+ "loss": 0.4696,
3189
+ "step": 1060000
3190
+ },
3191
+ {
3192
+ "epoch": 25.01,
3193
+ "learning_rate": 4.99606856423966e-05,
3194
+ "loss": 0.4682,
3195
+ "step": 1062000
3196
+ },
3197
+ {
3198
+ "epoch": 25.06,
3199
+ "learning_rate": 4.948891335115583e-05,
3200
+ "loss": 0.4725,
3201
+ "step": 1064000
3202
+ },
3203
+ {
3204
+ "epoch": 25.11,
3205
+ "learning_rate": 4.901714105991508e-05,
3206
+ "loss": 0.4655,
3207
+ "step": 1066000
3208
+ },
3209
+ {
3210
+ "epoch": 25.15,
3211
+ "learning_rate": 4.854536876867432e-05,
3212
+ "loss": 0.4629,
3213
+ "step": 1068000
3214
+ },
3215
+ {
3216
+ "epoch": 25.2,
3217
+ "learning_rate": 4.807359647743355e-05,
3218
+ "loss": 0.4654,
3219
+ "step": 1070000
3220
+ },
3221
+ {
3222
+ "epoch": 25.25,
3223
+ "learning_rate": 4.760182418619279e-05,
3224
+ "loss": 0.4663,
3225
+ "step": 1072000
3226
+ },
3227
+ {
3228
+ "epoch": 25.29,
3229
+ "learning_rate": 4.713005189495203e-05,
3230
+ "loss": 0.4682,
3231
+ "step": 1074000
3232
+ },
3233
+ {
3234
+ "epoch": 25.34,
3235
+ "learning_rate": 4.665827960371127e-05,
3236
+ "loss": 0.4647,
3237
+ "step": 1076000
3238
+ },
3239
+ {
3240
+ "epoch": 25.39,
3241
+ "learning_rate": 4.618650731247051e-05,
3242
+ "loss": 0.4722,
3243
+ "step": 1078000
3244
+ },
3245
+ {
3246
+ "epoch": 25.44,
3247
+ "learning_rate": 4.571473502122975e-05,
3248
+ "loss": 0.4604,
3249
+ "step": 1080000
3250
+ },
3251
+ {
3252
+ "epoch": 25.48,
3253
+ "learning_rate": 4.524296272998898e-05,
3254
+ "loss": 0.4585,
3255
+ "step": 1082000
3256
+ },
3257
+ {
3258
+ "epoch": 25.53,
3259
+ "learning_rate": 4.477119043874823e-05,
3260
+ "loss": 0.4619,
3261
+ "step": 1084000
3262
+ },
3263
+ {
3264
+ "epoch": 25.58,
3265
+ "learning_rate": 4.429941814750747e-05,
3266
+ "loss": 0.4661,
3267
+ "step": 1086000
3268
+ },
3269
+ {
3270
+ "epoch": 25.62,
3271
+ "learning_rate": 4.382764585626671e-05,
3272
+ "loss": 0.4724,
3273
+ "step": 1088000
3274
+ },
3275
+ {
3276
+ "epoch": 25.67,
3277
+ "learning_rate": 4.335587356502594e-05,
3278
+ "loss": 0.4667,
3279
+ "step": 1090000
3280
+ },
3281
+ {
3282
+ "epoch": 25.72,
3283
+ "learning_rate": 4.288410127378518e-05,
3284
+ "loss": 0.4692,
3285
+ "step": 1092000
3286
+ },
3287
+ {
3288
+ "epoch": 25.77,
3289
+ "learning_rate": 4.2412328982544426e-05,
3290
+ "loss": 0.4624,
3291
+ "step": 1094000
3292
+ },
3293
+ {
3294
+ "epoch": 25.81,
3295
+ "learning_rate": 4.194055669130366e-05,
3296
+ "loss": 0.4587,
3297
+ "step": 1096000
3298
+ },
3299
+ {
3300
+ "epoch": 25.86,
3301
+ "learning_rate": 4.14687844000629e-05,
3302
+ "loss": 0.4604,
3303
+ "step": 1098000
3304
+ },
3305
+ {
3306
+ "epoch": 25.91,
3307
+ "learning_rate": 4.099701210882214e-05,
3308
+ "loss": 0.4718,
3309
+ "step": 1100000
3310
+ },
3311
+ {
3312
+ "epoch": 25.95,
3313
+ "learning_rate": 4.052523981758137e-05,
3314
+ "loss": 0.4666,
3315
+ "step": 1102000
3316
+ },
3317
+ {
3318
+ "epoch": 26.0,
3319
+ "learning_rate": 4.005346752634062e-05,
3320
+ "loss": 0.4616,
3321
+ "step": 1104000
3322
+ },
3323
+ {
3324
+ "epoch": 26.05,
3325
+ "learning_rate": 3.9581695235099856e-05,
3326
+ "loss": 0.456,
3327
+ "step": 1106000
3328
+ },
3329
+ {
3330
+ "epoch": 26.1,
3331
+ "learning_rate": 3.9109922943859096e-05,
3332
+ "loss": 0.4573,
3333
+ "step": 1108000
3334
+ },
3335
+ {
3336
+ "epoch": 26.14,
3337
+ "learning_rate": 3.863815065261833e-05,
3338
+ "loss": 0.4522,
3339
+ "step": 1110000
3340
+ },
3341
+ {
3342
+ "epoch": 26.19,
3343
+ "learning_rate": 3.8166378361377575e-05,
3344
+ "loss": 0.4604,
3345
+ "step": 1112000
3346
+ },
3347
+ {
3348
+ "epoch": 26.24,
3349
+ "learning_rate": 3.7694606070136815e-05,
3350
+ "loss": 0.4539,
3351
+ "step": 1114000
3352
+ },
3353
+ {
3354
+ "epoch": 26.28,
3355
+ "learning_rate": 3.722283377889605e-05,
3356
+ "loss": 0.4485,
3357
+ "step": 1116000
3358
+ },
3359
+ {
3360
+ "epoch": 26.33,
3361
+ "learning_rate": 3.675106148765529e-05,
3362
+ "loss": 0.4596,
3363
+ "step": 1118000
3364
+ },
3365
+ {
3366
+ "epoch": 26.38,
3367
+ "learning_rate": 3.6279289196414526e-05,
3368
+ "loss": 0.4487,
3369
+ "step": 1120000
3370
+ },
3371
+ {
3372
+ "epoch": 26.42,
3373
+ "learning_rate": 3.5807516905173766e-05,
3374
+ "loss": 0.459,
3375
+ "step": 1122000
3376
+ },
3377
+ {
3378
+ "epoch": 26.47,
3379
+ "learning_rate": 3.5335744613933005e-05,
3380
+ "loss": 0.4609,
3381
+ "step": 1124000
3382
+ },
3383
+ {
3384
+ "epoch": 26.52,
3385
+ "learning_rate": 3.4863972322692245e-05,
3386
+ "loss": 0.4549,
3387
+ "step": 1126000
3388
+ },
3389
+ {
3390
+ "epoch": 26.57,
3391
+ "learning_rate": 3.4392200031451484e-05,
3392
+ "loss": 0.4553,
3393
+ "step": 1128000
3394
+ },
3395
+ {
3396
+ "epoch": 26.61,
3397
+ "learning_rate": 3.3920427740210724e-05,
3398
+ "loss": 0.4552,
3399
+ "step": 1130000
3400
+ },
3401
+ {
3402
+ "epoch": 26.66,
3403
+ "learning_rate": 3.344865544896996e-05,
3404
+ "loss": 0.4534,
3405
+ "step": 1132000
3406
+ },
3407
+ {
3408
+ "epoch": 26.71,
3409
+ "learning_rate": 3.29768831577292e-05,
3410
+ "loss": 0.4522,
3411
+ "step": 1134000
3412
+ },
3413
+ {
3414
+ "epoch": 26.75,
3415
+ "learning_rate": 3.2505110866488436e-05,
3416
+ "loss": 0.4544,
3417
+ "step": 1136000
3418
+ },
3419
+ {
3420
+ "epoch": 26.8,
3421
+ "learning_rate": 3.203333857524768e-05,
3422
+ "loss": 0.458,
3423
+ "step": 1138000
3424
+ },
3425
+ {
3426
+ "epoch": 26.85,
3427
+ "learning_rate": 3.1561566284006915e-05,
3428
+ "loss": 0.4537,
3429
+ "step": 1140000
3430
+ },
3431
+ {
3432
+ "epoch": 26.9,
3433
+ "learning_rate": 3.1089793992766154e-05,
3434
+ "loss": 0.4581,
3435
+ "step": 1142000
3436
+ },
3437
+ {
3438
+ "epoch": 26.94,
3439
+ "learning_rate": 3.0618021701525394e-05,
3440
+ "loss": 0.4525,
3441
+ "step": 1144000
3442
+ },
3443
+ {
3444
+ "epoch": 26.99,
3445
+ "learning_rate": 3.014624941028463e-05,
3446
+ "loss": 0.4562,
3447
+ "step": 1146000
3448
+ },
3449
+ {
3450
+ "epoch": 27.04,
3451
+ "learning_rate": 2.9674477119043873e-05,
3452
+ "loss": 0.4516,
3453
+ "step": 1148000
3454
+ },
3455
+ {
3456
+ "epoch": 27.08,
3457
+ "learning_rate": 2.920270482780311e-05,
3458
+ "loss": 0.4472,
3459
+ "step": 1150000
3460
+ },
3461
+ {
3462
+ "epoch": 27.13,
3463
+ "learning_rate": 2.8730932536562352e-05,
3464
+ "loss": 0.4507,
3465
+ "step": 1152000
3466
+ },
3467
+ {
3468
+ "epoch": 27.18,
3469
+ "learning_rate": 2.8259160245321588e-05,
3470
+ "loss": 0.4421,
3471
+ "step": 1154000
3472
+ },
3473
+ {
3474
+ "epoch": 27.23,
3475
+ "learning_rate": 2.778738795408083e-05,
3476
+ "loss": 0.4487,
3477
+ "step": 1156000
3478
+ },
3479
+ {
3480
+ "epoch": 27.27,
3481
+ "learning_rate": 2.7315615662840067e-05,
3482
+ "loss": 0.4528,
3483
+ "step": 1158000
3484
+ },
3485
+ {
3486
+ "epoch": 27.32,
3487
+ "learning_rate": 2.6843843371599303e-05,
3488
+ "loss": 0.4483,
3489
+ "step": 1160000
3490
+ },
3491
+ {
3492
+ "epoch": 27.37,
3493
+ "learning_rate": 2.6372071080358546e-05,
3494
+ "loss": 0.45,
3495
+ "step": 1162000
3496
+ },
3497
+ {
3498
+ "epoch": 27.41,
3499
+ "learning_rate": 2.5900298789117782e-05,
3500
+ "loss": 0.4517,
3501
+ "step": 1164000
3502
+ },
3503
+ {
3504
+ "epoch": 27.46,
3505
+ "learning_rate": 2.5428526497877022e-05,
3506
+ "loss": 0.4444,
3507
+ "step": 1166000
3508
+ },
3509
+ {
3510
+ "epoch": 27.51,
3511
+ "learning_rate": 2.495675420663626e-05,
3512
+ "loss": 0.4457,
3513
+ "step": 1168000
3514
+ },
3515
+ {
3516
+ "epoch": 27.56,
3517
+ "learning_rate": 2.44849819153955e-05,
3518
+ "loss": 0.4471,
3519
+ "step": 1170000
3520
+ },
3521
+ {
3522
+ "epoch": 27.6,
3523
+ "learning_rate": 2.401320962415474e-05,
3524
+ "loss": 0.4455,
3525
+ "step": 1172000
3526
+ },
3527
+ {
3528
+ "epoch": 27.65,
3529
+ "learning_rate": 2.3541437332913976e-05,
3530
+ "loss": 0.4466,
3531
+ "step": 1174000
3532
+ },
3533
+ {
3534
+ "epoch": 27.7,
3535
+ "learning_rate": 2.3069665041673216e-05,
3536
+ "loss": 0.4421,
3537
+ "step": 1176000
3538
+ },
3539
+ {
3540
+ "epoch": 27.74,
3541
+ "learning_rate": 2.2597892750432455e-05,
3542
+ "loss": 0.446,
3543
+ "step": 1178000
3544
+ },
3545
+ {
3546
+ "epoch": 27.79,
3547
+ "learning_rate": 2.2126120459191695e-05,
3548
+ "loss": 0.4441,
3549
+ "step": 1180000
3550
+ },
3551
+ {
3552
+ "epoch": 27.84,
3553
+ "learning_rate": 2.1654348167950935e-05,
3554
+ "loss": 0.4452,
3555
+ "step": 1182000
3556
+ },
3557
+ {
3558
+ "epoch": 27.89,
3559
+ "learning_rate": 2.1182575876710174e-05,
3560
+ "loss": 0.4418,
3561
+ "step": 1184000
3562
+ },
3563
+ {
3564
+ "epoch": 27.93,
3565
+ "learning_rate": 2.071080358546941e-05,
3566
+ "loss": 0.4396,
3567
+ "step": 1186000
3568
+ },
3569
+ {
3570
+ "epoch": 27.98,
3571
+ "learning_rate": 2.0239031294228653e-05,
3572
+ "loss": 0.4477,
3573
+ "step": 1188000
3574
+ },
3575
+ {
3576
+ "epoch": 28.03,
3577
+ "learning_rate": 1.976725900298789e-05,
3578
+ "loss": 0.4435,
3579
+ "step": 1190000
3580
+ },
3581
+ {
3582
+ "epoch": 28.07,
3583
+ "learning_rate": 1.9295486711747125e-05,
3584
+ "loss": 0.4379,
3585
+ "step": 1192000
3586
+ },
3587
+ {
3588
+ "epoch": 28.12,
3589
+ "learning_rate": 1.8823714420506368e-05,
3590
+ "loss": 0.4442,
3591
+ "step": 1194000
3592
+ },
3593
+ {
3594
+ "epoch": 28.17,
3595
+ "learning_rate": 1.8351942129265604e-05,
3596
+ "loss": 0.4329,
3597
+ "step": 1196000
3598
+ },
3599
+ {
3600
+ "epoch": 28.21,
3601
+ "learning_rate": 1.7880169838024844e-05,
3602
+ "loss": 0.4368,
3603
+ "step": 1198000
3604
+ },
3605
+ {
3606
+ "epoch": 28.26,
3607
+ "learning_rate": 1.7408397546784083e-05,
3608
+ "loss": 0.4427,
3609
+ "step": 1200000
3610
+ },
3611
+ {
3612
+ "epoch": 28.31,
3613
+ "learning_rate": 1.6936625255543323e-05,
3614
+ "loss": 0.4426,
3615
+ "step": 1202000
3616
+ },
3617
+ {
3618
+ "epoch": 28.36,
3619
+ "learning_rate": 1.6464852964302562e-05,
3620
+ "loss": 0.4374,
3621
+ "step": 1204000
3622
+ },
3623
+ {
3624
+ "epoch": 28.4,
3625
+ "learning_rate": 1.5993080673061802e-05,
3626
+ "loss": 0.4372,
3627
+ "step": 1206000
3628
+ },
3629
+ {
3630
+ "epoch": 28.45,
3631
+ "learning_rate": 1.552130838182104e-05,
3632
+ "loss": 0.4365,
3633
+ "step": 1208000
3634
+ },
3635
+ {
3636
+ "epoch": 28.5,
3637
+ "learning_rate": 1.504953609058028e-05,
3638
+ "loss": 0.4329,
3639
+ "step": 1210000
3640
+ },
3641
+ {
3642
+ "epoch": 28.54,
3643
+ "learning_rate": 1.4577763799339517e-05,
3644
+ "loss": 0.4397,
3645
+ "step": 1212000
3646
+ },
3647
+ {
3648
+ "epoch": 28.59,
3649
+ "learning_rate": 1.4105991508098757e-05,
3650
+ "loss": 0.4355,
3651
+ "step": 1214000
3652
+ },
3653
+ {
3654
+ "epoch": 28.64,
3655
+ "learning_rate": 1.3634219216857994e-05,
3656
+ "loss": 0.4386,
3657
+ "step": 1216000
3658
+ },
3659
+ {
3660
+ "epoch": 28.69,
3661
+ "learning_rate": 1.3162446925617234e-05,
3662
+ "loss": 0.4369,
3663
+ "step": 1218000
3664
+ },
3665
+ {
3666
+ "epoch": 28.73,
3667
+ "learning_rate": 1.2690674634376474e-05,
3668
+ "loss": 0.441,
3669
+ "step": 1220000
3670
+ },
3671
+ {
3672
+ "epoch": 28.78,
3673
+ "learning_rate": 1.2218902343135713e-05,
3674
+ "loss": 0.4383,
3675
+ "step": 1222000
3676
+ },
3677
+ {
3678
+ "epoch": 28.83,
3679
+ "learning_rate": 1.1747130051894953e-05,
3680
+ "loss": 0.4338,
3681
+ "step": 1224000
3682
+ },
3683
+ {
3684
+ "epoch": 28.87,
3685
+ "learning_rate": 1.1275357760654189e-05,
3686
+ "loss": 0.4373,
3687
+ "step": 1226000
3688
+ },
3689
+ {
3690
+ "epoch": 28.92,
3691
+ "learning_rate": 1.0803585469413428e-05,
3692
+ "loss": 0.4392,
3693
+ "step": 1228000
3694
+ },
3695
+ {
3696
+ "epoch": 28.97,
3697
+ "learning_rate": 1.0331813178172668e-05,
3698
+ "loss": 0.4401,
3699
+ "step": 1230000
3700
+ },
3701
+ {
3702
+ "epoch": 29.02,
3703
+ "learning_rate": 9.860040886931907e-06,
3704
+ "loss": 0.4394,
3705
+ "step": 1232000
3706
+ },
3707
+ {
3708
+ "epoch": 29.06,
3709
+ "learning_rate": 9.388268595691145e-06,
3710
+ "loss": 0.4316,
3711
+ "step": 1234000
3712
+ },
3713
+ {
3714
+ "epoch": 29.11,
3715
+ "learning_rate": 8.916496304450385e-06,
3716
+ "loss": 0.4372,
3717
+ "step": 1236000
3718
+ },
3719
+ {
3720
+ "epoch": 29.16,
3721
+ "learning_rate": 8.444724013209622e-06,
3722
+ "loss": 0.4363,
3723
+ "step": 1238000
3724
+ },
3725
+ {
3726
+ "epoch": 29.2,
3727
+ "learning_rate": 7.972951721968862e-06,
3728
+ "loss": 0.4348,
3729
+ "step": 1240000
3730
+ },
3731
+ {
3732
+ "epoch": 29.25,
3733
+ "learning_rate": 7.5011794307281015e-06,
3734
+ "loss": 0.4342,
3735
+ "step": 1242000
3736
+ },
3737
+ {
3738
+ "epoch": 29.3,
3739
+ "learning_rate": 7.02940713948734e-06,
3740
+ "loss": 0.4295,
3741
+ "step": 1244000
3742
+ },
3743
+ {
3744
+ "epoch": 29.35,
3745
+ "learning_rate": 6.557634848246579e-06,
3746
+ "loss": 0.4358,
3747
+ "step": 1246000
3748
+ },
3749
+ {
3750
+ "epoch": 29.39,
3751
+ "learning_rate": 6.085862557005818e-06,
3752
+ "loss": 0.4317,
3753
+ "step": 1248000
3754
+ },
3755
+ {
3756
+ "epoch": 29.44,
3757
+ "learning_rate": 5.614090265765056e-06,
3758
+ "loss": 0.431,
3759
+ "step": 1250000
3760
+ },
3761
+ {
3762
+ "epoch": 29.49,
3763
+ "learning_rate": 5.142317974524296e-06,
3764
+ "loss": 0.4355,
3765
+ "step": 1252000
3766
+ },
3767
+ {
3768
+ "epoch": 29.53,
3769
+ "learning_rate": 4.670545683283535e-06,
3770
+ "loss": 0.4301,
3771
+ "step": 1254000
3772
+ },
3773
+ {
3774
+ "epoch": 29.58,
3775
+ "learning_rate": 4.198773392042774e-06,
3776
+ "loss": 0.4272,
3777
+ "step": 1256000
3778
+ },
3779
+ {
3780
+ "epoch": 29.63,
3781
+ "learning_rate": 3.7270011008020125e-06,
3782
+ "loss": 0.4325,
3783
+ "step": 1258000
3784
+ },
3785
+ {
3786
+ "epoch": 29.67,
3787
+ "learning_rate": 3.255228809561251e-06,
3788
+ "loss": 0.4288,
3789
+ "step": 1260000
3790
+ },
3791
+ {
3792
+ "epoch": 29.72,
3793
+ "learning_rate": 2.7834565183204907e-06,
3794
+ "loss": 0.4246,
3795
+ "step": 1262000
3796
+ },
3797
+ {
3798
+ "epoch": 29.77,
3799
+ "learning_rate": 2.3116842270797294e-06,
3800
+ "loss": 0.437,
3801
+ "step": 1264000
3802
+ },
3803
+ {
3804
+ "epoch": 29.82,
3805
+ "learning_rate": 1.8399119358389683e-06,
3806
+ "loss": 0.4337,
3807
+ "step": 1266000
3808
+ },
3809
+ {
3810
+ "epoch": 29.86,
3811
+ "learning_rate": 1.3681396445982072e-06,
3812
+ "loss": 0.4281,
3813
+ "step": 1268000
3814
+ },
3815
+ {
3816
+ "epoch": 29.91,
3817
+ "learning_rate": 8.963673533574461e-07,
3818
+ "loss": 0.4334,
3819
+ "step": 1270000
3820
+ },
3821
+ {
3822
+ "epoch": 29.96,
3823
+ "learning_rate": 4.2459506211668494e-07,
3824
+ "loss": 0.4294,
3825
+ "step": 1272000
3826
+ }
3827
+ ],
3828
+ "logging_steps": 2000,
3829
+ "max_steps": 1273800,
3830
+ "num_input_tokens_seen": 0,
3831
+ "num_train_epochs": 30,
3832
+ "save_steps": 500,
3833
+ "total_flos": 2.5779090018764566e+21,
3834
+ "train_batch_size": 2,
3835
+ "trial_name": null,
3836
+ "trial_params": null
3837
+ }
checkpoint-1273800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cecbe2308480a0db70978cbda8460cbfae2844e623308cee5fa4aa246d04a482
3
+ size 4728
checkpoint-1273800/vocab.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "\"": 1,
3
+ ">": 2,
4
+ "[PAD]": 95,
5
+ "[UNK]": 94,
6
+ "\\": 3,
7
+ "a": 4,
8
+ "b": 5,
9
+ "c": 6,
10
+ "d": 7,
11
+ "e": 8,
12
+ "f": 9,
13
+ "g": 10,
14
+ "h": 11,
15
+ "i": 12,
16
+ "j": 13,
17
+ "k": 14,
18
+ "l": 15,
19
+ "m": 16,
20
+ "n": 17,
21
+ "o": 18,
22
+ "p": 19,
23
+ "q": 20,
24
+ "r": 21,
25
+ "s": 22,
26
+ "t": 23,
27
+ "u": 24,
28
+ "v": 25,
29
+ "w": 26,
30
+ "x": 27,
31
+ "y": 28,
32
+ "z": 29,
33
+ "|": 0,
34
+ "ç": 30,
35
+ "é": 31,
36
+ "õ": 32,
37
+ "ş": 33,
38
+ "ơ": 34,
39
+ "ư": 35,
40
+ "̀": 36,
41
+ "υ": 37,
42
+ "а": 38,
43
+ "в": 39,
44
+ "г": 40,
45
+ "д": 41,
46
+ "е": 42,
47
+ "з": 43,
48
+ "к": 44,
49
+ "л": 45,
50
+ "н": 46,
51
+ "п": 47,
52
+ "с": 48,
53
+ "ф": 49,
54
+ "ч": 50,
55
+ "ả": 51,
56
+ "ậ": 52,
57
+ "ồ": 53,
58
+ "ờ": 54,
59
+ "​": 55,
60
+ "’": 56,
61
+ "⁄": 57,
62
+ "れ": 58,
63
+ "ㅑ": 59,
64
+ "ㅛ": 60,
65
+ "각": 61,
66
+ "국": 62,
67
+ "기": 63,
68
+ "는": 64,
69
+ "데": 65,
70
+ "도": 66,
71
+ "드": 67,
72
+ "들": 68,
73
+ "랜": 69,
74
+ "로": 70,
75
+ "를": 71,
76
+ "만": 72,
77
+ "많": 73,
78
+ "부": 74,
79
+ "브": 75,
80
+ "생": 76,
81
+ "솔": 77,
82
+ "스": 78,
83
+ "앤": 79,
84
+ "에": 80,
85
+ "여": 81,
86
+ "온": 82,
87
+ "원": 83,
88
+ "전": 84,
89
+ "제": 85,
90
+ "지": 86,
91
+ "천": 87,
92
+ "초": 88,
93
+ "터": 89,
94
+ "트": 90,
95
+ "한": 91,
96
+ "행": 92,
97
+ "�": 93
98
+ }
config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
3
+ "activation_dropout": 0.0,
4
+ "adapter_attn_dim": null,
5
+ "adapter_kernel_size": 3,
6
+ "adapter_stride": 2,
7
+ "add_adapter": false,
8
+ "apply_spec_augment": true,
9
+ "architectures": [
10
+ "Wav2Vec2ForCTC"
11
+ ],
12
+ "attention_dropout": 0.05,
13
+ "bos_token_id": 1,
14
+ "classifier_proj_size": 256,
15
+ "codevector_dim": 768,
16
+ "contrastive_logits_temperature": 0.1,
17
+ "conv_bias": true,
18
+ "conv_dim": [
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512,
25
+ 512
26
+ ],
27
+ "conv_kernel": [
28
+ 10,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 3,
33
+ 2,
34
+ 2
35
+ ],
36
+ "conv_stride": [
37
+ 5,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2,
43
+ 2
44
+ ],
45
+ "ctc_loss_reduction": "mean",
46
+ "ctc_zero_infinity": true,
47
+ "diversity_loss_weight": 0.1,
48
+ "do_stable_layer_norm": true,
49
+ "eos_token_id": 2,
50
+ "feat_extract_activation": "gelu",
51
+ "feat_extract_dropout": 0.0,
52
+ "feat_extract_norm": "layer",
53
+ "feat_proj_dropout": 0.05,
54
+ "feat_quantizer_dropout": 0.0,
55
+ "final_dropout": 0.0,
56
+ "gradient_checkpointing": false,
57
+ "hidden_act": "gelu",
58
+ "hidden_dropout": 0.05,
59
+ "hidden_size": 1024,
60
+ "initializer_range": 0.02,
61
+ "intermediate_size": 4096,
62
+ "layer_norm_eps": 1e-05,
63
+ "layerdrop": 0.05,
64
+ "mask_channel_length": 10,
65
+ "mask_channel_min_space": 1,
66
+ "mask_channel_other": 0.0,
67
+ "mask_channel_prob": 0.0,
68
+ "mask_channel_selection": "static",
69
+ "mask_feature_length": 10,
70
+ "mask_feature_min_masks": 0,
71
+ "mask_feature_prob": 0.0,
72
+ "mask_time_length": 10,
73
+ "mask_time_min_masks": 2,
74
+ "mask_time_min_space": 1,
75
+ "mask_time_other": 0.0,
76
+ "mask_time_prob": 0.05,
77
+ "mask_time_selection": "static",
78
+ "model_type": "wav2vec2",
79
+ "num_adapter_layers": 3,
80
+ "num_attention_heads": 16,
81
+ "num_codevector_groups": 2,
82
+ "num_codevectors_per_group": 320,
83
+ "num_conv_pos_embedding_groups": 16,
84
+ "num_conv_pos_embeddings": 128,
85
+ "num_feat_extract_layers": 7,
86
+ "num_hidden_layers": 24,
87
+ "num_negatives": 100,
88
+ "output_hidden_size": 1024,
89
+ "pad_token_id": 95,
90
+ "proj_codevector_dim": 768,
91
+ "tdnn_dilation": [
92
+ 1,
93
+ 2,
94
+ 3,
95
+ 1,
96
+ 1
97
+ ],
98
+ "tdnn_dim": [
99
+ 512,
100
+ 512,
101
+ 512,
102
+ 512,
103
+ 1500
104
+ ],
105
+ "tdnn_kernel": [
106
+ 5,
107
+ 3,
108
+ 3,
109
+ 1,
110
+ 1
111
+ ],
112
+ "torch_dtype": "float32",
113
+ "transformers_version": "4.36.2",
114
+ "use_weighted_layer_sum": false,
115
+ "vocab_size": 98,
116
+ "xvector_output_dim": 512
117
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dafdcdb0e309cbeeae1382c1347fbaea4b82a930c87f2336462c88682bb1ef7c
3
+ size 1262209280
preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": true,
19
+ "normalized": false,
20
+ "rstrip": true,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "[UNK]",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": true,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "94": {
4
+ "content": "[UNK]",
5
+ "lstrip": true,
6
+ "normalized": false,
7
+ "rstrip": true,
8
+ "single_word": false,
9
+ "special": false
10
+ },
11
+ "95": {
12
+ "content": "[PAD]",
13
+ "lstrip": true,
14
+ "normalized": false,
15
+ "rstrip": true,
16
+ "single_word": false,
17
+ "special": false
18
+ },
19
+ "96": {
20
+ "content": "<s>",
21
+ "lstrip": false,
22
+ "normalized": true,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "97": {
28
+ "content": "</s>",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<s>",
37
+ "clean_up_tokenization_spaces": true,
38
+ "config": null,
39
+ "do_lower_case": false,
40
+ "eos_token": "</s>",
41
+ "model_max_length": 1000000000000000019884624838656,
42
+ "pad_token": "[PAD]",
43
+ "processor_class": "Wav2Vec2Processor",
44
+ "replace_word_delimiter_char": " ",
45
+ "target_lang": null,
46
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
47
+ "tokenizer_type": "wav2vec2",
48
+ "trust_remote_code": false,
49
+ "unk_token": "[UNK]",
50
+ "word_delimiter_token": "|"
51
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 30.0,
3
+ "train_loss": 0.6109381775774412,
4
+ "train_runtime": 724079.1012,
5
+ "train_samples": 679360,
6
+ "train_samples_per_second": 28.147,
7
+ "train_steps_per_second": 1.759
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,3846 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 30.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1273800,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.05,
13
+ "learning_rate": 0.0003,
14
+ "loss": 3.0982,
15
+ "step": 2000
16
+ },
17
+ {
18
+ "epoch": 0.09,
19
+ "learning_rate": 0.0002995282277087592,
20
+ "loss": 1.1999,
21
+ "step": 4000
22
+ },
23
+ {
24
+ "epoch": 0.14,
25
+ "learning_rate": 0.00029905645541751844,
26
+ "loss": 1.0995,
27
+ "step": 6000
28
+ },
29
+ {
30
+ "epoch": 0.19,
31
+ "learning_rate": 0.0002985846831262777,
32
+ "loss": 1.0716,
33
+ "step": 8000
34
+ },
35
+ {
36
+ "epoch": 0.24,
37
+ "learning_rate": 0.0002981129108350369,
38
+ "loss": 1.0664,
39
+ "step": 10000
40
+ },
41
+ {
42
+ "epoch": 0.28,
43
+ "learning_rate": 0.0002976411385437962,
44
+ "loss": 1.0223,
45
+ "step": 12000
46
+ },
47
+ {
48
+ "epoch": 0.33,
49
+ "learning_rate": 0.00029716936625255537,
50
+ "loss": 0.9955,
51
+ "step": 14000
52
+ },
53
+ {
54
+ "epoch": 0.38,
55
+ "learning_rate": 0.00029669759396131466,
56
+ "loss": 0.9925,
57
+ "step": 16000
58
+ },
59
+ {
60
+ "epoch": 0.42,
61
+ "learning_rate": 0.0002962258216700739,
62
+ "loss": 0.9951,
63
+ "step": 18000
64
+ },
65
+ {
66
+ "epoch": 0.47,
67
+ "learning_rate": 0.0002957540493788331,
68
+ "loss": 0.9811,
69
+ "step": 20000
70
+ },
71
+ {
72
+ "epoch": 0.52,
73
+ "learning_rate": 0.00029528227708759236,
74
+ "loss": 0.9727,
75
+ "step": 22000
76
+ },
77
+ {
78
+ "epoch": 0.57,
79
+ "learning_rate": 0.0002948105047963516,
80
+ "loss": 0.9655,
81
+ "step": 24000
82
+ },
83
+ {
84
+ "epoch": 0.61,
85
+ "learning_rate": 0.0002943387325051109,
86
+ "loss": 0.959,
87
+ "step": 26000
88
+ },
89
+ {
90
+ "epoch": 0.66,
91
+ "learning_rate": 0.00029386696021387005,
92
+ "loss": 0.9591,
93
+ "step": 28000
94
+ },
95
+ {
96
+ "epoch": 0.71,
97
+ "learning_rate": 0.00029339518792262934,
98
+ "loss": 0.9455,
99
+ "step": 30000
100
+ },
101
+ {
102
+ "epoch": 0.75,
103
+ "learning_rate": 0.0002929234156313886,
104
+ "loss": 0.9425,
105
+ "step": 32000
106
+ },
107
+ {
108
+ "epoch": 0.8,
109
+ "learning_rate": 0.0002924516433401478,
110
+ "loss": 0.9373,
111
+ "step": 34000
112
+ },
113
+ {
114
+ "epoch": 0.85,
115
+ "learning_rate": 0.00029197987104890704,
116
+ "loss": 0.9341,
117
+ "step": 36000
118
+ },
119
+ {
120
+ "epoch": 0.89,
121
+ "learning_rate": 0.00029150809875766627,
122
+ "loss": 0.93,
123
+ "step": 38000
124
+ },
125
+ {
126
+ "epoch": 0.94,
127
+ "learning_rate": 0.0002910363264664255,
128
+ "loss": 0.9184,
129
+ "step": 40000
130
+ },
131
+ {
132
+ "epoch": 0.99,
133
+ "learning_rate": 0.00029056455417518474,
134
+ "loss": 0.9222,
135
+ "step": 42000
136
+ },
137
+ {
138
+ "epoch": 1.04,
139
+ "learning_rate": 0.000290092781883944,
140
+ "loss": 0.8941,
141
+ "step": 44000
142
+ },
143
+ {
144
+ "epoch": 1.08,
145
+ "learning_rate": 0.0002896210095927032,
146
+ "loss": 0.8868,
147
+ "step": 46000
148
+ },
149
+ {
150
+ "epoch": 1.13,
151
+ "learning_rate": 0.0002891492373014625,
152
+ "loss": 0.8894,
153
+ "step": 48000
154
+ },
155
+ {
156
+ "epoch": 1.18,
157
+ "learning_rate": 0.0002886774650102217,
158
+ "loss": 0.8799,
159
+ "step": 50000
160
+ },
161
+ {
162
+ "epoch": 1.22,
163
+ "learning_rate": 0.00028820569271898095,
164
+ "loss": 0.8781,
165
+ "step": 52000
166
+ },
167
+ {
168
+ "epoch": 1.27,
169
+ "learning_rate": 0.0002877339204277402,
170
+ "loss": 0.8853,
171
+ "step": 54000
172
+ },
173
+ {
174
+ "epoch": 1.32,
175
+ "learning_rate": 0.0002872621481364994,
176
+ "loss": 0.8829,
177
+ "step": 56000
178
+ },
179
+ {
180
+ "epoch": 1.37,
181
+ "learning_rate": 0.00028679037584525865,
182
+ "loss": 0.8747,
183
+ "step": 58000
184
+ },
185
+ {
186
+ "epoch": 1.41,
187
+ "learning_rate": 0.0002863186035540179,
188
+ "loss": 0.8806,
189
+ "step": 60000
190
+ },
191
+ {
192
+ "epoch": 1.46,
193
+ "learning_rate": 0.00028584683126277717,
194
+ "loss": 0.8868,
195
+ "step": 62000
196
+ },
197
+ {
198
+ "epoch": 1.51,
199
+ "learning_rate": 0.00028537505897153635,
200
+ "loss": 0.8806,
201
+ "step": 64000
202
+ },
203
+ {
204
+ "epoch": 1.55,
205
+ "learning_rate": 0.00028490328668029564,
206
+ "loss": 0.8771,
207
+ "step": 66000
208
+ },
209
+ {
210
+ "epoch": 1.6,
211
+ "learning_rate": 0.00028443151438905487,
212
+ "loss": 0.8648,
213
+ "step": 68000
214
+ },
215
+ {
216
+ "epoch": 1.65,
217
+ "learning_rate": 0.0002839597420978141,
218
+ "loss": 0.8747,
219
+ "step": 70000
220
+ },
221
+ {
222
+ "epoch": 1.7,
223
+ "learning_rate": 0.00028348796980657334,
224
+ "loss": 0.8719,
225
+ "step": 72000
226
+ },
227
+ {
228
+ "epoch": 1.74,
229
+ "learning_rate": 0.00028301619751533257,
230
+ "loss": 0.8641,
231
+ "step": 74000
232
+ },
233
+ {
234
+ "epoch": 1.79,
235
+ "learning_rate": 0.0002825444252240918,
236
+ "loss": 0.8555,
237
+ "step": 76000
238
+ },
239
+ {
240
+ "epoch": 1.84,
241
+ "learning_rate": 0.00028207265293285103,
242
+ "loss": 0.8583,
243
+ "step": 78000
244
+ },
245
+ {
246
+ "epoch": 1.88,
247
+ "learning_rate": 0.0002816008806416103,
248
+ "loss": 0.8624,
249
+ "step": 80000
250
+ },
251
+ {
252
+ "epoch": 1.93,
253
+ "learning_rate": 0.0002811291083503695,
254
+ "loss": 0.8578,
255
+ "step": 82000
256
+ },
257
+ {
258
+ "epoch": 1.98,
259
+ "learning_rate": 0.0002806573360591288,
260
+ "loss": 0.8474,
261
+ "step": 84000
262
+ },
263
+ {
264
+ "epoch": 2.03,
265
+ "learning_rate": 0.000280185563767888,
266
+ "loss": 0.8366,
267
+ "step": 86000
268
+ },
269
+ {
270
+ "epoch": 2.07,
271
+ "learning_rate": 0.00027971379147664725,
272
+ "loss": 0.8226,
273
+ "step": 88000
274
+ },
275
+ {
276
+ "epoch": 2.12,
277
+ "learning_rate": 0.0002792420191854065,
278
+ "loss": 0.8267,
279
+ "step": 90000
280
+ },
281
+ {
282
+ "epoch": 2.17,
283
+ "learning_rate": 0.0002787702468941657,
284
+ "loss": 0.8231,
285
+ "step": 92000
286
+ },
287
+ {
288
+ "epoch": 2.21,
289
+ "learning_rate": 0.00027829847460292495,
290
+ "loss": 0.8284,
291
+ "step": 94000
292
+ },
293
+ {
294
+ "epoch": 2.26,
295
+ "learning_rate": 0.0002778267023116842,
296
+ "loss": 0.8153,
297
+ "step": 96000
298
+ },
299
+ {
300
+ "epoch": 2.31,
301
+ "learning_rate": 0.00027735493002044347,
302
+ "loss": 0.8241,
303
+ "step": 98000
304
+ },
305
+ {
306
+ "epoch": 2.36,
307
+ "learning_rate": 0.00027688315772920265,
308
+ "loss": 0.8246,
309
+ "step": 100000
310
+ },
311
+ {
312
+ "epoch": 2.4,
313
+ "learning_rate": 0.00027641138543796194,
314
+ "loss": 0.8219,
315
+ "step": 102000
316
+ },
317
+ {
318
+ "epoch": 2.45,
319
+ "learning_rate": 0.00027593961314672117,
320
+ "loss": 0.8244,
321
+ "step": 104000
322
+ },
323
+ {
324
+ "epoch": 2.5,
325
+ "learning_rate": 0.0002754678408554804,
326
+ "loss": 0.8201,
327
+ "step": 106000
328
+ },
329
+ {
330
+ "epoch": 2.54,
331
+ "learning_rate": 0.00027499606856423963,
332
+ "loss": 0.8193,
333
+ "step": 108000
334
+ },
335
+ {
336
+ "epoch": 2.59,
337
+ "learning_rate": 0.00027452429627299887,
338
+ "loss": 0.8225,
339
+ "step": 110000
340
+ },
341
+ {
342
+ "epoch": 2.64,
343
+ "learning_rate": 0.0002740525239817581,
344
+ "loss": 0.8105,
345
+ "step": 112000
346
+ },
347
+ {
348
+ "epoch": 2.68,
349
+ "learning_rate": 0.00027358075169051733,
350
+ "loss": 0.82,
351
+ "step": 114000
352
+ },
353
+ {
354
+ "epoch": 2.73,
355
+ "learning_rate": 0.0002731089793992766,
356
+ "loss": 0.8191,
357
+ "step": 116000
358
+ },
359
+ {
360
+ "epoch": 2.78,
361
+ "learning_rate": 0.00027263720710803585,
362
+ "loss": 0.8102,
363
+ "step": 118000
364
+ },
365
+ {
366
+ "epoch": 2.83,
367
+ "learning_rate": 0.0002721654348167951,
368
+ "loss": 0.8189,
369
+ "step": 120000
370
+ },
371
+ {
372
+ "epoch": 2.87,
373
+ "learning_rate": 0.0002716936625255543,
374
+ "loss": 0.8128,
375
+ "step": 122000
376
+ },
377
+ {
378
+ "epoch": 2.92,
379
+ "learning_rate": 0.00027122189023431355,
380
+ "loss": 0.815,
381
+ "step": 124000
382
+ },
383
+ {
384
+ "epoch": 2.97,
385
+ "learning_rate": 0.0002707501179430728,
386
+ "loss": 0.8123,
387
+ "step": 126000
388
+ },
389
+ {
390
+ "epoch": 3.01,
391
+ "learning_rate": 0.000270278345651832,
392
+ "loss": 0.7986,
393
+ "step": 128000
394
+ },
395
+ {
396
+ "epoch": 3.06,
397
+ "learning_rate": 0.00026980657336059125,
398
+ "loss": 0.7701,
399
+ "step": 130000
400
+ },
401
+ {
402
+ "epoch": 3.11,
403
+ "learning_rate": 0.0002693348010693505,
404
+ "loss": 0.7819,
405
+ "step": 132000
406
+ },
407
+ {
408
+ "epoch": 3.16,
409
+ "learning_rate": 0.00026886302877810977,
410
+ "loss": 0.7826,
411
+ "step": 134000
412
+ },
413
+ {
414
+ "epoch": 3.2,
415
+ "learning_rate": 0.000268391256486869,
416
+ "loss": 0.7868,
417
+ "step": 136000
418
+ },
419
+ {
420
+ "epoch": 3.25,
421
+ "learning_rate": 0.00026791948419562823,
422
+ "loss": 0.7843,
423
+ "step": 138000
424
+ },
425
+ {
426
+ "epoch": 3.3,
427
+ "learning_rate": 0.00026744771190438747,
428
+ "loss": 0.7841,
429
+ "step": 140000
430
+ },
431
+ {
432
+ "epoch": 3.34,
433
+ "learning_rate": 0.0002669759396131467,
434
+ "loss": 0.7845,
435
+ "step": 142000
436
+ },
437
+ {
438
+ "epoch": 3.39,
439
+ "learning_rate": 0.00026650416732190593,
440
+ "loss": 0.7859,
441
+ "step": 144000
442
+ },
443
+ {
444
+ "epoch": 3.44,
445
+ "learning_rate": 0.00026603239503066516,
446
+ "loss": 0.7746,
447
+ "step": 146000
448
+ },
449
+ {
450
+ "epoch": 3.49,
451
+ "learning_rate": 0.0002655606227394244,
452
+ "loss": 0.7749,
453
+ "step": 148000
454
+ },
455
+ {
456
+ "epoch": 3.53,
457
+ "learning_rate": 0.00026508885044818363,
458
+ "loss": 0.7747,
459
+ "step": 150000
460
+ },
461
+ {
462
+ "epoch": 3.58,
463
+ "learning_rate": 0.0002646170781569429,
464
+ "loss": 0.7816,
465
+ "step": 152000
466
+ },
467
+ {
468
+ "epoch": 3.63,
469
+ "learning_rate": 0.00026414530586570215,
470
+ "loss": 0.7807,
471
+ "step": 154000
472
+ },
473
+ {
474
+ "epoch": 3.67,
475
+ "learning_rate": 0.0002636735335744614,
476
+ "loss": 0.7817,
477
+ "step": 156000
478
+ },
479
+ {
480
+ "epoch": 3.72,
481
+ "learning_rate": 0.0002632017612832206,
482
+ "loss": 0.787,
483
+ "step": 158000
484
+ },
485
+ {
486
+ "epoch": 3.77,
487
+ "learning_rate": 0.00026272998899197985,
488
+ "loss": 0.7796,
489
+ "step": 160000
490
+ },
491
+ {
492
+ "epoch": 3.82,
493
+ "learning_rate": 0.0002622582167007391,
494
+ "loss": 0.7662,
495
+ "step": 162000
496
+ },
497
+ {
498
+ "epoch": 3.86,
499
+ "learning_rate": 0.0002617864444094983,
500
+ "loss": 0.7898,
501
+ "step": 164000
502
+ },
503
+ {
504
+ "epoch": 3.91,
505
+ "learning_rate": 0.00026131467211825755,
506
+ "loss": 0.7705,
507
+ "step": 166000
508
+ },
509
+ {
510
+ "epoch": 3.96,
511
+ "learning_rate": 0.0002608428998270168,
512
+ "loss": 0.7876,
513
+ "step": 168000
514
+ },
515
+ {
516
+ "epoch": 4.0,
517
+ "learning_rate": 0.00026037112753577607,
518
+ "loss": 0.7747,
519
+ "step": 170000
520
+ },
521
+ {
522
+ "epoch": 4.05,
523
+ "learning_rate": 0.0002598993552445353,
524
+ "loss": 0.7517,
525
+ "step": 172000
526
+ },
527
+ {
528
+ "epoch": 4.1,
529
+ "learning_rate": 0.00025942758295329453,
530
+ "loss": 0.7465,
531
+ "step": 174000
532
+ },
533
+ {
534
+ "epoch": 4.15,
535
+ "learning_rate": 0.00025895581066205376,
536
+ "loss": 0.7485,
537
+ "step": 176000
538
+ },
539
+ {
540
+ "epoch": 4.19,
541
+ "learning_rate": 0.000258484038370813,
542
+ "loss": 0.7531,
543
+ "step": 178000
544
+ },
545
+ {
546
+ "epoch": 4.24,
547
+ "learning_rate": 0.00025801226607957223,
548
+ "loss": 0.7521,
549
+ "step": 180000
550
+ },
551
+ {
552
+ "epoch": 4.29,
553
+ "learning_rate": 0.00025754049378833146,
554
+ "loss": 0.7477,
555
+ "step": 182000
556
+ },
557
+ {
558
+ "epoch": 4.33,
559
+ "learning_rate": 0.0002570687214970907,
560
+ "loss": 0.7621,
561
+ "step": 184000
562
+ },
563
+ {
564
+ "epoch": 4.38,
565
+ "learning_rate": 0.00025659694920585,
566
+ "loss": 0.7552,
567
+ "step": 186000
568
+ },
569
+ {
570
+ "epoch": 4.43,
571
+ "learning_rate": 0.0002561251769146092,
572
+ "loss": 0.7592,
573
+ "step": 188000
574
+ },
575
+ {
576
+ "epoch": 4.47,
577
+ "learning_rate": 0.00025565340462336845,
578
+ "loss": 0.7508,
579
+ "step": 190000
580
+ },
581
+ {
582
+ "epoch": 4.52,
583
+ "learning_rate": 0.0002551816323321277,
584
+ "loss": 0.7547,
585
+ "step": 192000
586
+ },
587
+ {
588
+ "epoch": 4.57,
589
+ "learning_rate": 0.0002547098600408869,
590
+ "loss": 0.7439,
591
+ "step": 194000
592
+ },
593
+ {
594
+ "epoch": 4.62,
595
+ "learning_rate": 0.00025423808774964615,
596
+ "loss": 0.762,
597
+ "step": 196000
598
+ },
599
+ {
600
+ "epoch": 4.66,
601
+ "learning_rate": 0.0002537663154584054,
602
+ "loss": 0.7507,
603
+ "step": 198000
604
+ },
605
+ {
606
+ "epoch": 4.71,
607
+ "learning_rate": 0.0002532945431671646,
608
+ "loss": 0.7553,
609
+ "step": 200000
610
+ },
611
+ {
612
+ "epoch": 4.76,
613
+ "learning_rate": 0.00025282277087592384,
614
+ "loss": 0.7498,
615
+ "step": 202000
616
+ },
617
+ {
618
+ "epoch": 4.8,
619
+ "learning_rate": 0.00025235099858468313,
620
+ "loss": 0.7466,
621
+ "step": 204000
622
+ },
623
+ {
624
+ "epoch": 4.85,
625
+ "learning_rate": 0.0002518792262934423,
626
+ "loss": 0.7496,
627
+ "step": 206000
628
+ },
629
+ {
630
+ "epoch": 4.9,
631
+ "learning_rate": 0.0002514074540022016,
632
+ "loss": 0.7406,
633
+ "step": 208000
634
+ },
635
+ {
636
+ "epoch": 4.95,
637
+ "learning_rate": 0.00025093568171096083,
638
+ "loss": 0.7447,
639
+ "step": 210000
640
+ },
641
+ {
642
+ "epoch": 4.99,
643
+ "learning_rate": 0.00025046390941972006,
644
+ "loss": 0.7532,
645
+ "step": 212000
646
+ },
647
+ {
648
+ "epoch": 5.04,
649
+ "learning_rate": 0.0002499921371284793,
650
+ "loss": 0.7339,
651
+ "step": 214000
652
+ },
653
+ {
654
+ "epoch": 5.09,
655
+ "learning_rate": 0.00024952036483723853,
656
+ "loss": 0.7214,
657
+ "step": 216000
658
+ },
659
+ {
660
+ "epoch": 5.13,
661
+ "learning_rate": 0.00024904859254599776,
662
+ "loss": 0.7227,
663
+ "step": 218000
664
+ },
665
+ {
666
+ "epoch": 5.18,
667
+ "learning_rate": 0.000248576820254757,
668
+ "loss": 0.7344,
669
+ "step": 220000
670
+ },
671
+ {
672
+ "epoch": 5.23,
673
+ "learning_rate": 0.0002481050479635163,
674
+ "loss": 0.7272,
675
+ "step": 222000
676
+ },
677
+ {
678
+ "epoch": 5.28,
679
+ "learning_rate": 0.00024763327567227546,
680
+ "loss": 0.7271,
681
+ "step": 224000
682
+ },
683
+ {
684
+ "epoch": 5.32,
685
+ "learning_rate": 0.00024716150338103475,
686
+ "loss": 0.7302,
687
+ "step": 226000
688
+ },
689
+ {
690
+ "epoch": 5.37,
691
+ "learning_rate": 0.000246689731089794,
692
+ "loss": 0.7218,
693
+ "step": 228000
694
+ },
695
+ {
696
+ "epoch": 5.42,
697
+ "learning_rate": 0.0002462179587985532,
698
+ "loss": 0.7242,
699
+ "step": 230000
700
+ },
701
+ {
702
+ "epoch": 5.46,
703
+ "learning_rate": 0.00024574618650731244,
704
+ "loss": 0.7306,
705
+ "step": 232000
706
+ },
707
+ {
708
+ "epoch": 5.51,
709
+ "learning_rate": 0.0002452744142160717,
710
+ "loss": 0.7351,
711
+ "step": 234000
712
+ },
713
+ {
714
+ "epoch": 5.56,
715
+ "learning_rate": 0.0002448026419248309,
716
+ "loss": 0.7225,
717
+ "step": 236000
718
+ },
719
+ {
720
+ "epoch": 5.61,
721
+ "learning_rate": 0.00024433086963359014,
722
+ "loss": 0.7256,
723
+ "step": 238000
724
+ },
725
+ {
726
+ "epoch": 5.65,
727
+ "learning_rate": 0.0002438590973423494,
728
+ "loss": 0.7256,
729
+ "step": 240000
730
+ },
731
+ {
732
+ "epoch": 5.7,
733
+ "learning_rate": 0.00024338732505110863,
734
+ "loss": 0.7293,
735
+ "step": 242000
736
+ },
737
+ {
738
+ "epoch": 5.75,
739
+ "learning_rate": 0.0002429155527598679,
740
+ "loss": 0.7272,
741
+ "step": 244000
742
+ },
743
+ {
744
+ "epoch": 5.79,
745
+ "learning_rate": 0.00024244378046862713,
746
+ "loss": 0.7154,
747
+ "step": 246000
748
+ },
749
+ {
750
+ "epoch": 5.84,
751
+ "learning_rate": 0.00024197200817738636,
752
+ "loss": 0.7274,
753
+ "step": 248000
754
+ },
755
+ {
756
+ "epoch": 5.89,
757
+ "learning_rate": 0.00024150023588614562,
758
+ "loss": 0.7184,
759
+ "step": 250000
760
+ },
761
+ {
762
+ "epoch": 5.93,
763
+ "learning_rate": 0.00024102846359490483,
764
+ "loss": 0.7218,
765
+ "step": 252000
766
+ },
767
+ {
768
+ "epoch": 5.98,
769
+ "learning_rate": 0.00024055669130366409,
770
+ "loss": 0.7246,
771
+ "step": 254000
772
+ },
773
+ {
774
+ "epoch": 6.03,
775
+ "learning_rate": 0.00024008491901242332,
776
+ "loss": 0.7036,
777
+ "step": 256000
778
+ },
779
+ {
780
+ "epoch": 6.08,
781
+ "learning_rate": 0.00023961314672118255,
782
+ "loss": 0.6992,
783
+ "step": 258000
784
+ },
785
+ {
786
+ "epoch": 6.12,
787
+ "learning_rate": 0.00023914137442994178,
788
+ "loss": 0.7029,
789
+ "step": 260000
790
+ },
791
+ {
792
+ "epoch": 6.17,
793
+ "learning_rate": 0.00023866960213870104,
794
+ "loss": 0.7097,
795
+ "step": 262000
796
+ },
797
+ {
798
+ "epoch": 6.22,
799
+ "learning_rate": 0.00023819782984746028,
800
+ "loss": 0.7042,
801
+ "step": 264000
802
+ },
803
+ {
804
+ "epoch": 6.26,
805
+ "learning_rate": 0.0002377260575562195,
806
+ "loss": 0.7076,
807
+ "step": 266000
808
+ },
809
+ {
810
+ "epoch": 6.31,
811
+ "learning_rate": 0.00023725428526497877,
812
+ "loss": 0.6972,
813
+ "step": 268000
814
+ },
815
+ {
816
+ "epoch": 6.36,
817
+ "learning_rate": 0.00023678251297373797,
818
+ "loss": 0.7097,
819
+ "step": 270000
820
+ },
821
+ {
822
+ "epoch": 6.41,
823
+ "learning_rate": 0.00023631074068249723,
824
+ "loss": 0.7025,
825
+ "step": 272000
826
+ },
827
+ {
828
+ "epoch": 6.45,
829
+ "learning_rate": 0.00023583896839125647,
830
+ "loss": 0.7041,
831
+ "step": 274000
832
+ },
833
+ {
834
+ "epoch": 6.5,
835
+ "learning_rate": 0.0002353671961000157,
836
+ "loss": 0.7098,
837
+ "step": 276000
838
+ },
839
+ {
840
+ "epoch": 6.55,
841
+ "learning_rate": 0.00023489542380877493,
842
+ "loss": 0.704,
843
+ "step": 278000
844
+ },
845
+ {
846
+ "epoch": 6.59,
847
+ "learning_rate": 0.0002344236515175342,
848
+ "loss": 0.7046,
849
+ "step": 280000
850
+ },
851
+ {
852
+ "epoch": 6.64,
853
+ "learning_rate": 0.00023395187922629343,
854
+ "loss": 0.701,
855
+ "step": 282000
856
+ },
857
+ {
858
+ "epoch": 6.69,
859
+ "learning_rate": 0.00023348010693505266,
860
+ "loss": 0.7107,
861
+ "step": 284000
862
+ },
863
+ {
864
+ "epoch": 6.74,
865
+ "learning_rate": 0.00023300833464381192,
866
+ "loss": 0.7107,
867
+ "step": 286000
868
+ },
869
+ {
870
+ "epoch": 6.78,
871
+ "learning_rate": 0.00023253656235257112,
872
+ "loss": 0.7051,
873
+ "step": 288000
874
+ },
875
+ {
876
+ "epoch": 6.83,
877
+ "learning_rate": 0.00023206479006133038,
878
+ "loss": 0.7055,
879
+ "step": 290000
880
+ },
881
+ {
882
+ "epoch": 6.88,
883
+ "learning_rate": 0.00023159301777008962,
884
+ "loss": 0.7004,
885
+ "step": 292000
886
+ },
887
+ {
888
+ "epoch": 6.92,
889
+ "learning_rate": 0.00023112124547884885,
890
+ "loss": 0.7051,
891
+ "step": 294000
892
+ },
893
+ {
894
+ "epoch": 6.97,
895
+ "learning_rate": 0.00023064947318760808,
896
+ "loss": 0.7025,
897
+ "step": 296000
898
+ },
899
+ {
900
+ "epoch": 7.02,
901
+ "learning_rate": 0.00023017770089636734,
902
+ "loss": 0.6976,
903
+ "step": 298000
904
+ },
905
+ {
906
+ "epoch": 7.07,
907
+ "learning_rate": 0.0002297059286051266,
908
+ "loss": 0.6785,
909
+ "step": 300000
910
+ },
911
+ {
912
+ "epoch": 7.11,
913
+ "learning_rate": 0.0002292341563138858,
914
+ "loss": 0.6773,
915
+ "step": 302000
916
+ },
917
+ {
918
+ "epoch": 7.16,
919
+ "learning_rate": 0.00022876238402264507,
920
+ "loss": 0.6763,
921
+ "step": 304000
922
+ },
923
+ {
924
+ "epoch": 7.21,
925
+ "learning_rate": 0.00022829061173140427,
926
+ "loss": 0.6774,
927
+ "step": 306000
928
+ },
929
+ {
930
+ "epoch": 7.25,
931
+ "learning_rate": 0.00022781883944016353,
932
+ "loss": 0.6786,
933
+ "step": 308000
934
+ },
935
+ {
936
+ "epoch": 7.3,
937
+ "learning_rate": 0.00022734706714892276,
938
+ "loss": 0.6885,
939
+ "step": 310000
940
+ },
941
+ {
942
+ "epoch": 7.35,
943
+ "learning_rate": 0.000226875294857682,
944
+ "loss": 0.6852,
945
+ "step": 312000
946
+ },
947
+ {
948
+ "epoch": 7.4,
949
+ "learning_rate": 0.00022640352256644123,
950
+ "loss": 0.6842,
951
+ "step": 314000
952
+ },
953
+ {
954
+ "epoch": 7.44,
955
+ "learning_rate": 0.0002259317502752005,
956
+ "loss": 0.6868,
957
+ "step": 316000
958
+ },
959
+ {
960
+ "epoch": 7.49,
961
+ "learning_rate": 0.00022545997798395975,
962
+ "loss": 0.6918,
963
+ "step": 318000
964
+ },
965
+ {
966
+ "epoch": 7.54,
967
+ "learning_rate": 0.00022498820569271896,
968
+ "loss": 0.6846,
969
+ "step": 320000
970
+ },
971
+ {
972
+ "epoch": 7.58,
973
+ "learning_rate": 0.00022451643340147822,
974
+ "loss": 0.6881,
975
+ "step": 322000
976
+ },
977
+ {
978
+ "epoch": 7.63,
979
+ "learning_rate": 0.00022404466111023742,
980
+ "loss": 0.69,
981
+ "step": 324000
982
+ },
983
+ {
984
+ "epoch": 7.68,
985
+ "learning_rate": 0.00022357288881899668,
986
+ "loss": 0.6823,
987
+ "step": 326000
988
+ },
989
+ {
990
+ "epoch": 7.72,
991
+ "learning_rate": 0.00022310111652775591,
992
+ "loss": 0.6896,
993
+ "step": 328000
994
+ },
995
+ {
996
+ "epoch": 7.77,
997
+ "learning_rate": 0.00022262934423651517,
998
+ "loss": 0.6862,
999
+ "step": 330000
1000
+ },
1001
+ {
1002
+ "epoch": 7.82,
1003
+ "learning_rate": 0.00022215757194527438,
1004
+ "loss": 0.6858,
1005
+ "step": 332000
1006
+ },
1007
+ {
1008
+ "epoch": 7.87,
1009
+ "learning_rate": 0.00022168579965403364,
1010
+ "loss": 0.691,
1011
+ "step": 334000
1012
+ },
1013
+ {
1014
+ "epoch": 7.91,
1015
+ "learning_rate": 0.00022121402736279284,
1016
+ "loss": 0.6895,
1017
+ "step": 336000
1018
+ },
1019
+ {
1020
+ "epoch": 7.96,
1021
+ "learning_rate": 0.0002207422550715521,
1022
+ "loss": 0.6924,
1023
+ "step": 338000
1024
+ },
1025
+ {
1026
+ "epoch": 8.01,
1027
+ "learning_rate": 0.00022027048278031136,
1028
+ "loss": 0.6903,
1029
+ "step": 340000
1030
+ },
1031
+ {
1032
+ "epoch": 8.05,
1033
+ "learning_rate": 0.0002197987104890706,
1034
+ "loss": 0.6637,
1035
+ "step": 342000
1036
+ },
1037
+ {
1038
+ "epoch": 8.1,
1039
+ "learning_rate": 0.00021932693819782983,
1040
+ "loss": 0.6612,
1041
+ "step": 344000
1042
+ },
1043
+ {
1044
+ "epoch": 8.15,
1045
+ "learning_rate": 0.00021885516590658906,
1046
+ "loss": 0.6665,
1047
+ "step": 346000
1048
+ },
1049
+ {
1050
+ "epoch": 8.2,
1051
+ "learning_rate": 0.00021838339361534832,
1052
+ "loss": 0.6702,
1053
+ "step": 348000
1054
+ },
1055
+ {
1056
+ "epoch": 8.24,
1057
+ "learning_rate": 0.00021791162132410753,
1058
+ "loss": 0.6667,
1059
+ "step": 350000
1060
+ },
1061
+ {
1062
+ "epoch": 8.29,
1063
+ "learning_rate": 0.0002174398490328668,
1064
+ "loss": 0.6674,
1065
+ "step": 352000
1066
+ },
1067
+ {
1068
+ "epoch": 8.34,
1069
+ "learning_rate": 0.000216968076741626,
1070
+ "loss": 0.6719,
1071
+ "step": 354000
1072
+ },
1073
+ {
1074
+ "epoch": 8.38,
1075
+ "learning_rate": 0.00021649630445038525,
1076
+ "loss": 0.6671,
1077
+ "step": 356000
1078
+ },
1079
+ {
1080
+ "epoch": 8.43,
1081
+ "learning_rate": 0.0002160245321591445,
1082
+ "loss": 0.6647,
1083
+ "step": 358000
1084
+ },
1085
+ {
1086
+ "epoch": 8.48,
1087
+ "learning_rate": 0.00021555275986790375,
1088
+ "loss": 0.6671,
1089
+ "step": 360000
1090
+ },
1091
+ {
1092
+ "epoch": 8.53,
1093
+ "learning_rate": 0.00021508098757666298,
1094
+ "loss": 0.6681,
1095
+ "step": 362000
1096
+ },
1097
+ {
1098
+ "epoch": 8.57,
1099
+ "learning_rate": 0.0002146092152854222,
1100
+ "loss": 0.6727,
1101
+ "step": 364000
1102
+ },
1103
+ {
1104
+ "epoch": 8.62,
1105
+ "learning_rate": 0.00021413744299418147,
1106
+ "loss": 0.6767,
1107
+ "step": 366000
1108
+ },
1109
+ {
1110
+ "epoch": 8.67,
1111
+ "learning_rate": 0.00021366567070294068,
1112
+ "loss": 0.6749,
1113
+ "step": 368000
1114
+ },
1115
+ {
1116
+ "epoch": 8.71,
1117
+ "learning_rate": 0.00021319389841169994,
1118
+ "loss": 0.6704,
1119
+ "step": 370000
1120
+ },
1121
+ {
1122
+ "epoch": 8.76,
1123
+ "learning_rate": 0.00021272212612045917,
1124
+ "loss": 0.6729,
1125
+ "step": 372000
1126
+ },
1127
+ {
1128
+ "epoch": 8.81,
1129
+ "learning_rate": 0.0002122503538292184,
1130
+ "loss": 0.6641,
1131
+ "step": 374000
1132
+ },
1133
+ {
1134
+ "epoch": 8.86,
1135
+ "learning_rate": 0.00021177858153797766,
1136
+ "loss": 0.6678,
1137
+ "step": 376000
1138
+ },
1139
+ {
1140
+ "epoch": 8.9,
1141
+ "learning_rate": 0.0002113068092467369,
1142
+ "loss": 0.6677,
1143
+ "step": 378000
1144
+ },
1145
+ {
1146
+ "epoch": 8.95,
1147
+ "learning_rate": 0.00021083503695549613,
1148
+ "loss": 0.6683,
1149
+ "step": 380000
1150
+ },
1151
+ {
1152
+ "epoch": 9.0,
1153
+ "learning_rate": 0.00021036326466425536,
1154
+ "loss": 0.6746,
1155
+ "step": 382000
1156
+ },
1157
+ {
1158
+ "epoch": 9.04,
1159
+ "learning_rate": 0.00020989149237301462,
1160
+ "loss": 0.6469,
1161
+ "step": 384000
1162
+ },
1163
+ {
1164
+ "epoch": 9.09,
1165
+ "learning_rate": 0.00020941972008177383,
1166
+ "loss": 0.6496,
1167
+ "step": 386000
1168
+ },
1169
+ {
1170
+ "epoch": 9.14,
1171
+ "learning_rate": 0.00020894794779053309,
1172
+ "loss": 0.6513,
1173
+ "step": 388000
1174
+ },
1175
+ {
1176
+ "epoch": 9.19,
1177
+ "learning_rate": 0.00020847617549929232,
1178
+ "loss": 0.6516,
1179
+ "step": 390000
1180
+ },
1181
+ {
1182
+ "epoch": 9.23,
1183
+ "learning_rate": 0.00020800440320805155,
1184
+ "loss": 0.6556,
1185
+ "step": 392000
1186
+ },
1187
+ {
1188
+ "epoch": 9.28,
1189
+ "learning_rate": 0.0002075326309168108,
1190
+ "loss": 0.6477,
1191
+ "step": 394000
1192
+ },
1193
+ {
1194
+ "epoch": 9.33,
1195
+ "learning_rate": 0.00020706085862557004,
1196
+ "loss": 0.6571,
1197
+ "step": 396000
1198
+ },
1199
+ {
1200
+ "epoch": 9.37,
1201
+ "learning_rate": 0.00020658908633432928,
1202
+ "loss": 0.6479,
1203
+ "step": 398000
1204
+ },
1205
+ {
1206
+ "epoch": 9.42,
1207
+ "learning_rate": 0.0002061173140430885,
1208
+ "loss": 0.65,
1209
+ "step": 400000
1210
+ },
1211
+ {
1212
+ "epoch": 9.47,
1213
+ "learning_rate": 0.00020564554175184777,
1214
+ "loss": 0.6557,
1215
+ "step": 402000
1216
+ },
1217
+ {
1218
+ "epoch": 9.51,
1219
+ "learning_rate": 0.00020517376946060697,
1220
+ "loss": 0.6644,
1221
+ "step": 404000
1222
+ },
1223
+ {
1224
+ "epoch": 9.56,
1225
+ "learning_rate": 0.00020470199716936623,
1226
+ "loss": 0.6558,
1227
+ "step": 406000
1228
+ },
1229
+ {
1230
+ "epoch": 9.61,
1231
+ "learning_rate": 0.00020423022487812547,
1232
+ "loss": 0.6503,
1233
+ "step": 408000
1234
+ },
1235
+ {
1236
+ "epoch": 9.66,
1237
+ "learning_rate": 0.0002037584525868847,
1238
+ "loss": 0.6569,
1239
+ "step": 410000
1240
+ },
1241
+ {
1242
+ "epoch": 9.7,
1243
+ "learning_rate": 0.00020328668029564396,
1244
+ "loss": 0.66,
1245
+ "step": 412000
1246
+ },
1247
+ {
1248
+ "epoch": 9.75,
1249
+ "learning_rate": 0.0002028149080044032,
1250
+ "loss": 0.6528,
1251
+ "step": 414000
1252
+ },
1253
+ {
1254
+ "epoch": 9.8,
1255
+ "learning_rate": 0.00020234313571316245,
1256
+ "loss": 0.6493,
1257
+ "step": 416000
1258
+ },
1259
+ {
1260
+ "epoch": 9.84,
1261
+ "learning_rate": 0.00020187136342192166,
1262
+ "loss": 0.6599,
1263
+ "step": 418000
1264
+ },
1265
+ {
1266
+ "epoch": 9.89,
1267
+ "learning_rate": 0.00020139959113068092,
1268
+ "loss": 0.6534,
1269
+ "step": 420000
1270
+ },
1271
+ {
1272
+ "epoch": 9.94,
1273
+ "learning_rate": 0.00020092781883944012,
1274
+ "loss": 0.6546,
1275
+ "step": 422000
1276
+ },
1277
+ {
1278
+ "epoch": 9.99,
1279
+ "learning_rate": 0.00020045604654819938,
1280
+ "loss": 0.6488,
1281
+ "step": 424000
1282
+ },
1283
+ {
1284
+ "epoch": 10.03,
1285
+ "learning_rate": 0.00019998427425695862,
1286
+ "loss": 0.643,
1287
+ "step": 426000
1288
+ },
1289
+ {
1290
+ "epoch": 10.08,
1291
+ "learning_rate": 0.00019951250196571788,
1292
+ "loss": 0.6465,
1293
+ "step": 428000
1294
+ },
1295
+ {
1296
+ "epoch": 10.13,
1297
+ "learning_rate": 0.0001990407296744771,
1298
+ "loss": 0.6324,
1299
+ "step": 430000
1300
+ },
1301
+ {
1302
+ "epoch": 10.17,
1303
+ "learning_rate": 0.00019856895738323634,
1304
+ "loss": 0.6446,
1305
+ "step": 432000
1306
+ },
1307
+ {
1308
+ "epoch": 10.22,
1309
+ "learning_rate": 0.0001980971850919956,
1310
+ "loss": 0.6283,
1311
+ "step": 434000
1312
+ },
1313
+ {
1314
+ "epoch": 10.27,
1315
+ "learning_rate": 0.0001976254128007548,
1316
+ "loss": 0.635,
1317
+ "step": 436000
1318
+ },
1319
+ {
1320
+ "epoch": 10.32,
1321
+ "learning_rate": 0.00019715364050951407,
1322
+ "loss": 0.6333,
1323
+ "step": 438000
1324
+ },
1325
+ {
1326
+ "epoch": 10.36,
1327
+ "learning_rate": 0.0001966818682182733,
1328
+ "loss": 0.6378,
1329
+ "step": 440000
1330
+ },
1331
+ {
1332
+ "epoch": 10.41,
1333
+ "learning_rate": 0.00019621009592703253,
1334
+ "loss": 0.6369,
1335
+ "step": 442000
1336
+ },
1337
+ {
1338
+ "epoch": 10.46,
1339
+ "learning_rate": 0.00019573832363579177,
1340
+ "loss": 0.6348,
1341
+ "step": 444000
1342
+ },
1343
+ {
1344
+ "epoch": 10.5,
1345
+ "learning_rate": 0.00019526655134455103,
1346
+ "loss": 0.6439,
1347
+ "step": 446000
1348
+ },
1349
+ {
1350
+ "epoch": 10.55,
1351
+ "learning_rate": 0.00019479477905331026,
1352
+ "loss": 0.6342,
1353
+ "step": 448000
1354
+ },
1355
+ {
1356
+ "epoch": 10.6,
1357
+ "learning_rate": 0.0001943230067620695,
1358
+ "loss": 0.6424,
1359
+ "step": 450000
1360
+ },
1361
+ {
1362
+ "epoch": 10.65,
1363
+ "learning_rate": 0.00019385123447082875,
1364
+ "loss": 0.6408,
1365
+ "step": 452000
1366
+ },
1367
+ {
1368
+ "epoch": 10.69,
1369
+ "learning_rate": 0.00019337946217958796,
1370
+ "loss": 0.6413,
1371
+ "step": 454000
1372
+ },
1373
+ {
1374
+ "epoch": 10.74,
1375
+ "learning_rate": 0.00019290768988834722,
1376
+ "loss": 0.643,
1377
+ "step": 456000
1378
+ },
1379
+ {
1380
+ "epoch": 10.79,
1381
+ "learning_rate": 0.00019243591759710645,
1382
+ "loss": 0.6421,
1383
+ "step": 458000
1384
+ },
1385
+ {
1386
+ "epoch": 10.83,
1387
+ "learning_rate": 0.00019196414530586568,
1388
+ "loss": 0.6393,
1389
+ "step": 460000
1390
+ },
1391
+ {
1392
+ "epoch": 10.88,
1393
+ "learning_rate": 0.00019149237301462491,
1394
+ "loss": 0.6385,
1395
+ "step": 462000
1396
+ },
1397
+ {
1398
+ "epoch": 10.93,
1399
+ "learning_rate": 0.00019102060072338417,
1400
+ "loss": 0.6387,
1401
+ "step": 464000
1402
+ },
1403
+ {
1404
+ "epoch": 10.98,
1405
+ "learning_rate": 0.00019054882843214338,
1406
+ "loss": 0.6442,
1407
+ "step": 466000
1408
+ },
1409
+ {
1410
+ "epoch": 11.02,
1411
+ "learning_rate": 0.00019007705614090264,
1412
+ "loss": 0.6294,
1413
+ "step": 468000
1414
+ },
1415
+ {
1416
+ "epoch": 11.07,
1417
+ "learning_rate": 0.0001896052838496619,
1418
+ "loss": 0.6191,
1419
+ "step": 470000
1420
+ },
1421
+ {
1422
+ "epoch": 11.12,
1423
+ "learning_rate": 0.0001891335115584211,
1424
+ "loss": 0.6226,
1425
+ "step": 472000
1426
+ },
1427
+ {
1428
+ "epoch": 11.16,
1429
+ "learning_rate": 0.00018866173926718036,
1430
+ "loss": 0.6207,
1431
+ "step": 474000
1432
+ },
1433
+ {
1434
+ "epoch": 11.21,
1435
+ "learning_rate": 0.0001881899669759396,
1436
+ "loss": 0.6282,
1437
+ "step": 476000
1438
+ },
1439
+ {
1440
+ "epoch": 11.26,
1441
+ "learning_rate": 0.00018771819468469883,
1442
+ "loss": 0.615,
1443
+ "step": 478000
1444
+ },
1445
+ {
1446
+ "epoch": 11.3,
1447
+ "learning_rate": 0.00018724642239345806,
1448
+ "loss": 0.6201,
1449
+ "step": 480000
1450
+ },
1451
+ {
1452
+ "epoch": 11.35,
1453
+ "learning_rate": 0.00018677465010221732,
1454
+ "loss": 0.6199,
1455
+ "step": 482000
1456
+ },
1457
+ {
1458
+ "epoch": 11.4,
1459
+ "learning_rate": 0.00018630287781097653,
1460
+ "loss": 0.6211,
1461
+ "step": 484000
1462
+ },
1463
+ {
1464
+ "epoch": 11.45,
1465
+ "learning_rate": 0.0001858311055197358,
1466
+ "loss": 0.6372,
1467
+ "step": 486000
1468
+ },
1469
+ {
1470
+ "epoch": 11.49,
1471
+ "learning_rate": 0.00018535933322849505,
1472
+ "loss": 0.6268,
1473
+ "step": 488000
1474
+ },
1475
+ {
1476
+ "epoch": 11.54,
1477
+ "learning_rate": 0.00018488756093725425,
1478
+ "loss": 0.6283,
1479
+ "step": 490000
1480
+ },
1481
+ {
1482
+ "epoch": 11.59,
1483
+ "learning_rate": 0.00018441578864601351,
1484
+ "loss": 0.6398,
1485
+ "step": 492000
1486
+ },
1487
+ {
1488
+ "epoch": 11.63,
1489
+ "learning_rate": 0.00018394401635477275,
1490
+ "loss": 0.6282,
1491
+ "step": 494000
1492
+ },
1493
+ {
1494
+ "epoch": 11.68,
1495
+ "learning_rate": 0.00018347224406353198,
1496
+ "loss": 0.6327,
1497
+ "step": 496000
1498
+ },
1499
+ {
1500
+ "epoch": 11.73,
1501
+ "learning_rate": 0.0001830004717722912,
1502
+ "loss": 0.6226,
1503
+ "step": 498000
1504
+ },
1505
+ {
1506
+ "epoch": 11.78,
1507
+ "learning_rate": 0.00018252869948105047,
1508
+ "loss": 0.6321,
1509
+ "step": 500000
1510
+ },
1511
+ {
1512
+ "epoch": 11.82,
1513
+ "learning_rate": 0.00018205692718980968,
1514
+ "loss": 0.6244,
1515
+ "step": 502000
1516
+ },
1517
+ {
1518
+ "epoch": 11.87,
1519
+ "learning_rate": 0.00018158515489856894,
1520
+ "loss": 0.6257,
1521
+ "step": 504000
1522
+ },
1523
+ {
1524
+ "epoch": 11.92,
1525
+ "learning_rate": 0.0001811133826073282,
1526
+ "loss": 0.6291,
1527
+ "step": 506000
1528
+ },
1529
+ {
1530
+ "epoch": 11.96,
1531
+ "learning_rate": 0.0001806416103160874,
1532
+ "loss": 0.6283,
1533
+ "step": 508000
1534
+ },
1535
+ {
1536
+ "epoch": 12.01,
1537
+ "learning_rate": 0.00018016983802484666,
1538
+ "loss": 0.6246,
1539
+ "step": 510000
1540
+ },
1541
+ {
1542
+ "epoch": 12.06,
1543
+ "learning_rate": 0.0001796980657336059,
1544
+ "loss": 0.6133,
1545
+ "step": 512000
1546
+ },
1547
+ {
1548
+ "epoch": 12.11,
1549
+ "learning_rate": 0.00017922629344236516,
1550
+ "loss": 0.6069,
1551
+ "step": 514000
1552
+ },
1553
+ {
1554
+ "epoch": 12.15,
1555
+ "learning_rate": 0.00017875452115112436,
1556
+ "loss": 0.6163,
1557
+ "step": 516000
1558
+ },
1559
+ {
1560
+ "epoch": 12.2,
1561
+ "learning_rate": 0.00017828274885988362,
1562
+ "loss": 0.6131,
1563
+ "step": 518000
1564
+ },
1565
+ {
1566
+ "epoch": 12.25,
1567
+ "learning_rate": 0.00017781097656864283,
1568
+ "loss": 0.6096,
1569
+ "step": 520000
1570
+ },
1571
+ {
1572
+ "epoch": 12.29,
1573
+ "learning_rate": 0.00017733920427740209,
1574
+ "loss": 0.6073,
1575
+ "step": 522000
1576
+ },
1577
+ {
1578
+ "epoch": 12.34,
1579
+ "learning_rate": 0.00017686743198616135,
1580
+ "loss": 0.6068,
1581
+ "step": 524000
1582
+ },
1583
+ {
1584
+ "epoch": 12.39,
1585
+ "learning_rate": 0.00017639565969492058,
1586
+ "loss": 0.6114,
1587
+ "step": 526000
1588
+ },
1589
+ {
1590
+ "epoch": 12.44,
1591
+ "learning_rate": 0.0001759238874036798,
1592
+ "loss": 0.6189,
1593
+ "step": 528000
1594
+ },
1595
+ {
1596
+ "epoch": 12.48,
1597
+ "learning_rate": 0.00017545211511243904,
1598
+ "loss": 0.6111,
1599
+ "step": 530000
1600
+ },
1601
+ {
1602
+ "epoch": 12.53,
1603
+ "learning_rate": 0.0001749803428211983,
1604
+ "loss": 0.6085,
1605
+ "step": 532000
1606
+ },
1607
+ {
1608
+ "epoch": 12.58,
1609
+ "learning_rate": 0.0001745085705299575,
1610
+ "loss": 0.6164,
1611
+ "step": 534000
1612
+ },
1613
+ {
1614
+ "epoch": 12.62,
1615
+ "learning_rate": 0.00017403679823871677,
1616
+ "loss": 0.6169,
1617
+ "step": 536000
1618
+ },
1619
+ {
1620
+ "epoch": 12.67,
1621
+ "learning_rate": 0.00017356502594747598,
1622
+ "loss": 0.615,
1623
+ "step": 538000
1624
+ },
1625
+ {
1626
+ "epoch": 12.72,
1627
+ "learning_rate": 0.00017309325365623524,
1628
+ "loss": 0.6145,
1629
+ "step": 540000
1630
+ },
1631
+ {
1632
+ "epoch": 12.76,
1633
+ "learning_rate": 0.0001726214813649945,
1634
+ "loss": 0.6149,
1635
+ "step": 542000
1636
+ },
1637
+ {
1638
+ "epoch": 12.81,
1639
+ "learning_rate": 0.00017214970907375373,
1640
+ "loss": 0.6128,
1641
+ "step": 544000
1642
+ },
1643
+ {
1644
+ "epoch": 12.86,
1645
+ "learning_rate": 0.00017167793678251296,
1646
+ "loss": 0.6123,
1647
+ "step": 546000
1648
+ },
1649
+ {
1650
+ "epoch": 12.91,
1651
+ "learning_rate": 0.0001712061644912722,
1652
+ "loss": 0.6132,
1653
+ "step": 548000
1654
+ },
1655
+ {
1656
+ "epoch": 12.95,
1657
+ "learning_rate": 0.00017073439220003145,
1658
+ "loss": 0.6154,
1659
+ "step": 550000
1660
+ },
1661
+ {
1662
+ "epoch": 13.0,
1663
+ "learning_rate": 0.00017026261990879066,
1664
+ "loss": 0.6168,
1665
+ "step": 552000
1666
+ },
1667
+ {
1668
+ "epoch": 13.05,
1669
+ "learning_rate": 0.00016979084761754992,
1670
+ "loss": 0.5926,
1671
+ "step": 554000
1672
+ },
1673
+ {
1674
+ "epoch": 13.09,
1675
+ "learning_rate": 0.00016931907532630915,
1676
+ "loss": 0.5924,
1677
+ "step": 556000
1678
+ },
1679
+ {
1680
+ "epoch": 13.14,
1681
+ "learning_rate": 0.00016884730303506838,
1682
+ "loss": 0.5949,
1683
+ "step": 558000
1684
+ },
1685
+ {
1686
+ "epoch": 13.19,
1687
+ "learning_rate": 0.00016837553074382764,
1688
+ "loss": 0.5947,
1689
+ "step": 560000
1690
+ },
1691
+ {
1692
+ "epoch": 13.24,
1693
+ "learning_rate": 0.00016790375845258688,
1694
+ "loss": 0.5971,
1695
+ "step": 562000
1696
+ },
1697
+ {
1698
+ "epoch": 13.28,
1699
+ "learning_rate": 0.0001674319861613461,
1700
+ "loss": 0.5976,
1701
+ "step": 564000
1702
+ },
1703
+ {
1704
+ "epoch": 13.33,
1705
+ "learning_rate": 0.00016696021387010534,
1706
+ "loss": 0.5984,
1707
+ "step": 566000
1708
+ },
1709
+ {
1710
+ "epoch": 13.38,
1711
+ "learning_rate": 0.0001664884415788646,
1712
+ "loss": 0.606,
1713
+ "step": 568000
1714
+ },
1715
+ {
1716
+ "epoch": 13.42,
1717
+ "learning_rate": 0.0001660166692876238,
1718
+ "loss": 0.6091,
1719
+ "step": 570000
1720
+ },
1721
+ {
1722
+ "epoch": 13.47,
1723
+ "learning_rate": 0.00016554489699638307,
1724
+ "loss": 0.5927,
1725
+ "step": 572000
1726
+ },
1727
+ {
1728
+ "epoch": 13.52,
1729
+ "learning_rate": 0.0001650731247051423,
1730
+ "loss": 0.6067,
1731
+ "step": 574000
1732
+ },
1733
+ {
1734
+ "epoch": 13.57,
1735
+ "learning_rate": 0.00016460135241390153,
1736
+ "loss": 0.6024,
1737
+ "step": 576000
1738
+ },
1739
+ {
1740
+ "epoch": 13.61,
1741
+ "learning_rate": 0.0001641295801226608,
1742
+ "loss": 0.6055,
1743
+ "step": 578000
1744
+ },
1745
+ {
1746
+ "epoch": 13.66,
1747
+ "learning_rate": 0.00016365780783142003,
1748
+ "loss": 0.6003,
1749
+ "step": 580000
1750
+ },
1751
+ {
1752
+ "epoch": 13.71,
1753
+ "learning_rate": 0.00016318603554017929,
1754
+ "loss": 0.6076,
1755
+ "step": 582000
1756
+ },
1757
+ {
1758
+ "epoch": 13.75,
1759
+ "learning_rate": 0.0001627142632489385,
1760
+ "loss": 0.606,
1761
+ "step": 584000
1762
+ },
1763
+ {
1764
+ "epoch": 13.8,
1765
+ "learning_rate": 0.00016224249095769775,
1766
+ "loss": 0.5901,
1767
+ "step": 586000
1768
+ },
1769
+ {
1770
+ "epoch": 13.85,
1771
+ "learning_rate": 0.00016177071866645696,
1772
+ "loss": 0.6084,
1773
+ "step": 588000
1774
+ },
1775
+ {
1776
+ "epoch": 13.9,
1777
+ "learning_rate": 0.00016129894637521622,
1778
+ "loss": 0.6075,
1779
+ "step": 590000
1780
+ },
1781
+ {
1782
+ "epoch": 13.94,
1783
+ "learning_rate": 0.00016082717408397545,
1784
+ "loss": 0.6085,
1785
+ "step": 592000
1786
+ },
1787
+ {
1788
+ "epoch": 13.99,
1789
+ "learning_rate": 0.00016035540179273468,
1790
+ "loss": 0.6009,
1791
+ "step": 594000
1792
+ },
1793
+ {
1794
+ "epoch": 14.04,
1795
+ "learning_rate": 0.00015988362950149391,
1796
+ "loss": 0.5894,
1797
+ "step": 596000
1798
+ },
1799
+ {
1800
+ "epoch": 14.08,
1801
+ "learning_rate": 0.00015941185721025317,
1802
+ "loss": 0.5835,
1803
+ "step": 598000
1804
+ },
1805
+ {
1806
+ "epoch": 14.13,
1807
+ "learning_rate": 0.00015894008491901243,
1808
+ "loss": 0.5859,
1809
+ "step": 600000
1810
+ },
1811
+ {
1812
+ "epoch": 14.18,
1813
+ "learning_rate": 0.00015846831262777164,
1814
+ "loss": 0.5929,
1815
+ "step": 602000
1816
+ },
1817
+ {
1818
+ "epoch": 14.23,
1819
+ "learning_rate": 0.0001579965403365309,
1820
+ "loss": 0.59,
1821
+ "step": 604000
1822
+ },
1823
+ {
1824
+ "epoch": 14.27,
1825
+ "learning_rate": 0.0001575247680452901,
1826
+ "loss": 0.582,
1827
+ "step": 606000
1828
+ },
1829
+ {
1830
+ "epoch": 14.32,
1831
+ "learning_rate": 0.00015705299575404937,
1832
+ "loss": 0.588,
1833
+ "step": 608000
1834
+ },
1835
+ {
1836
+ "epoch": 14.37,
1837
+ "learning_rate": 0.0001565812234628086,
1838
+ "loss": 0.5923,
1839
+ "step": 610000
1840
+ },
1841
+ {
1842
+ "epoch": 14.41,
1843
+ "learning_rate": 0.00015610945117156786,
1844
+ "loss": 0.5837,
1845
+ "step": 612000
1846
+ },
1847
+ {
1848
+ "epoch": 14.46,
1849
+ "learning_rate": 0.00015563767888032706,
1850
+ "loss": 0.5898,
1851
+ "step": 614000
1852
+ },
1853
+ {
1854
+ "epoch": 14.51,
1855
+ "learning_rate": 0.00015516590658908632,
1856
+ "loss": 0.5904,
1857
+ "step": 616000
1858
+ },
1859
+ {
1860
+ "epoch": 14.55,
1861
+ "learning_rate": 0.00015469413429784558,
1862
+ "loss": 0.5892,
1863
+ "step": 618000
1864
+ },
1865
+ {
1866
+ "epoch": 14.6,
1867
+ "learning_rate": 0.0001542223620066048,
1868
+ "loss": 0.5887,
1869
+ "step": 620000
1870
+ },
1871
+ {
1872
+ "epoch": 14.65,
1873
+ "learning_rate": 0.00015375058971536405,
1874
+ "loss": 0.5843,
1875
+ "step": 622000
1876
+ },
1877
+ {
1878
+ "epoch": 14.7,
1879
+ "learning_rate": 0.00015327881742412328,
1880
+ "loss": 0.5853,
1881
+ "step": 624000
1882
+ },
1883
+ {
1884
+ "epoch": 14.74,
1885
+ "learning_rate": 0.00015280704513288251,
1886
+ "loss": 0.5919,
1887
+ "step": 626000
1888
+ },
1889
+ {
1890
+ "epoch": 14.79,
1891
+ "learning_rate": 0.00015233527284164175,
1892
+ "loss": 0.5915,
1893
+ "step": 628000
1894
+ },
1895
+ {
1896
+ "epoch": 14.84,
1897
+ "learning_rate": 0.000151863500550401,
1898
+ "loss": 0.5918,
1899
+ "step": 630000
1900
+ },
1901
+ {
1902
+ "epoch": 14.88,
1903
+ "learning_rate": 0.0001513917282591602,
1904
+ "loss": 0.5932,
1905
+ "step": 632000
1906
+ },
1907
+ {
1908
+ "epoch": 14.93,
1909
+ "learning_rate": 0.00015091995596791947,
1910
+ "loss": 0.5891,
1911
+ "step": 634000
1912
+ },
1913
+ {
1914
+ "epoch": 14.98,
1915
+ "learning_rate": 0.00015044818367667873,
1916
+ "loss": 0.5888,
1917
+ "step": 636000
1918
+ },
1919
+ {
1920
+ "epoch": 15.03,
1921
+ "learning_rate": 0.00014997641138543794,
1922
+ "loss": 0.5817,
1923
+ "step": 638000
1924
+ },
1925
+ {
1926
+ "epoch": 15.07,
1927
+ "learning_rate": 0.00014950463909419717,
1928
+ "loss": 0.574,
1929
+ "step": 640000
1930
+ },
1931
+ {
1932
+ "epoch": 15.12,
1933
+ "learning_rate": 0.00014903286680295643,
1934
+ "loss": 0.5723,
1935
+ "step": 642000
1936
+ },
1937
+ {
1938
+ "epoch": 15.17,
1939
+ "learning_rate": 0.00014856109451171566,
1940
+ "loss": 0.573,
1941
+ "step": 644000
1942
+ },
1943
+ {
1944
+ "epoch": 15.21,
1945
+ "learning_rate": 0.00014808932222047492,
1946
+ "loss": 0.5758,
1947
+ "step": 646000
1948
+ },
1949
+ {
1950
+ "epoch": 15.26,
1951
+ "learning_rate": 0.00014761754992923416,
1952
+ "loss": 0.5796,
1953
+ "step": 648000
1954
+ },
1955
+ {
1956
+ "epoch": 15.31,
1957
+ "learning_rate": 0.0001471457776379934,
1958
+ "loss": 0.5722,
1959
+ "step": 650000
1960
+ },
1961
+ {
1962
+ "epoch": 15.36,
1963
+ "learning_rate": 0.00014667400534675262,
1964
+ "loss": 0.5825,
1965
+ "step": 652000
1966
+ },
1967
+ {
1968
+ "epoch": 15.4,
1969
+ "learning_rate": 0.00014620223305551185,
1970
+ "loss": 0.5744,
1971
+ "step": 654000
1972
+ },
1973
+ {
1974
+ "epoch": 15.45,
1975
+ "learning_rate": 0.0001457304607642711,
1976
+ "loss": 0.5786,
1977
+ "step": 656000
1978
+ },
1979
+ {
1980
+ "epoch": 15.5,
1981
+ "learning_rate": 0.00014525868847303032,
1982
+ "loss": 0.5808,
1983
+ "step": 658000
1984
+ },
1985
+ {
1986
+ "epoch": 15.54,
1987
+ "learning_rate": 0.00014478691618178958,
1988
+ "loss": 0.5855,
1989
+ "step": 660000
1990
+ },
1991
+ {
1992
+ "epoch": 15.59,
1993
+ "learning_rate": 0.0001443151438905488,
1994
+ "loss": 0.585,
1995
+ "step": 662000
1996
+ },
1997
+ {
1998
+ "epoch": 15.64,
1999
+ "learning_rate": 0.00014384337159930807,
2000
+ "loss": 0.5692,
2001
+ "step": 664000
2002
+ },
2003
+ {
2004
+ "epoch": 15.69,
2005
+ "learning_rate": 0.0001433715993080673,
2006
+ "loss": 0.5789,
2007
+ "step": 666000
2008
+ },
2009
+ {
2010
+ "epoch": 15.73,
2011
+ "learning_rate": 0.00014289982701682654,
2012
+ "loss": 0.5836,
2013
+ "step": 668000
2014
+ },
2015
+ {
2016
+ "epoch": 15.78,
2017
+ "learning_rate": 0.00014242805472558577,
2018
+ "loss": 0.5796,
2019
+ "step": 670000
2020
+ },
2021
+ {
2022
+ "epoch": 15.83,
2023
+ "learning_rate": 0.000141956282434345,
2024
+ "loss": 0.5756,
2025
+ "step": 672000
2026
+ },
2027
+ {
2028
+ "epoch": 15.87,
2029
+ "learning_rate": 0.00014148451014310424,
2030
+ "loss": 0.5667,
2031
+ "step": 674000
2032
+ },
2033
+ {
2034
+ "epoch": 15.92,
2035
+ "learning_rate": 0.0001410127378518635,
2036
+ "loss": 0.5845,
2037
+ "step": 676000
2038
+ },
2039
+ {
2040
+ "epoch": 15.97,
2041
+ "learning_rate": 0.00014054096556062273,
2042
+ "loss": 0.5783,
2043
+ "step": 678000
2044
+ },
2045
+ {
2046
+ "epoch": 16.02,
2047
+ "learning_rate": 0.00014006919326938196,
2048
+ "loss": 0.5749,
2049
+ "step": 680000
2050
+ },
2051
+ {
2052
+ "epoch": 16.06,
2053
+ "learning_rate": 0.00013959742097814122,
2054
+ "loss": 0.5602,
2055
+ "step": 682000
2056
+ },
2057
+ {
2058
+ "epoch": 16.11,
2059
+ "learning_rate": 0.00013912564868690045,
2060
+ "loss": 0.564,
2061
+ "step": 684000
2062
+ },
2063
+ {
2064
+ "epoch": 16.16,
2065
+ "learning_rate": 0.00013865387639565969,
2066
+ "loss": 0.5659,
2067
+ "step": 686000
2068
+ },
2069
+ {
2070
+ "epoch": 16.2,
2071
+ "learning_rate": 0.00013818210410441892,
2072
+ "loss": 0.5608,
2073
+ "step": 688000
2074
+ },
2075
+ {
2076
+ "epoch": 16.25,
2077
+ "learning_rate": 0.00013771033181317815,
2078
+ "loss": 0.5668,
2079
+ "step": 690000
2080
+ },
2081
+ {
2082
+ "epoch": 16.3,
2083
+ "learning_rate": 0.00013723855952193738,
2084
+ "loss": 0.5646,
2085
+ "step": 692000
2086
+ },
2087
+ {
2088
+ "epoch": 16.34,
2089
+ "learning_rate": 0.00013676678723069664,
2090
+ "loss": 0.5663,
2091
+ "step": 694000
2092
+ },
2093
+ {
2094
+ "epoch": 16.39,
2095
+ "learning_rate": 0.00013629501493945588,
2096
+ "loss": 0.5716,
2097
+ "step": 696000
2098
+ },
2099
+ {
2100
+ "epoch": 16.44,
2101
+ "learning_rate": 0.0001358232426482151,
2102
+ "loss": 0.5716,
2103
+ "step": 698000
2104
+ },
2105
+ {
2106
+ "epoch": 16.49,
2107
+ "learning_rate": 0.00013535147035697437,
2108
+ "loss": 0.568,
2109
+ "step": 700000
2110
+ },
2111
+ {
2112
+ "epoch": 16.53,
2113
+ "learning_rate": 0.0001348796980657336,
2114
+ "loss": 0.566,
2115
+ "step": 702000
2116
+ },
2117
+ {
2118
+ "epoch": 16.58,
2119
+ "learning_rate": 0.00013440792577449284,
2120
+ "loss": 0.5657,
2121
+ "step": 704000
2122
+ },
2123
+ {
2124
+ "epoch": 16.63,
2125
+ "learning_rate": 0.00013393615348325207,
2126
+ "loss": 0.5641,
2127
+ "step": 706000
2128
+ },
2129
+ {
2130
+ "epoch": 16.67,
2131
+ "learning_rate": 0.0001334643811920113,
2132
+ "loss": 0.5679,
2133
+ "step": 708000
2134
+ },
2135
+ {
2136
+ "epoch": 16.72,
2137
+ "learning_rate": 0.00013299260890077056,
2138
+ "loss": 0.5652,
2139
+ "step": 710000
2140
+ },
2141
+ {
2142
+ "epoch": 16.77,
2143
+ "learning_rate": 0.0001325208366095298,
2144
+ "loss": 0.5672,
2145
+ "step": 712000
2146
+ },
2147
+ {
2148
+ "epoch": 16.82,
2149
+ "learning_rate": 0.00013204906431828903,
2150
+ "loss": 0.5695,
2151
+ "step": 714000
2152
+ },
2153
+ {
2154
+ "epoch": 16.86,
2155
+ "learning_rate": 0.00013157729202704826,
2156
+ "loss": 0.5719,
2157
+ "step": 716000
2158
+ },
2159
+ {
2160
+ "epoch": 16.91,
2161
+ "learning_rate": 0.00013110551973580752,
2162
+ "loss": 0.571,
2163
+ "step": 718000
2164
+ },
2165
+ {
2166
+ "epoch": 16.96,
2167
+ "learning_rate": 0.00013063374744456675,
2168
+ "loss": 0.5684,
2169
+ "step": 720000
2170
+ },
2171
+ {
2172
+ "epoch": 17.0,
2173
+ "learning_rate": 0.00013016197515332598,
2174
+ "loss": 0.5672,
2175
+ "step": 722000
2176
+ },
2177
+ {
2178
+ "epoch": 17.05,
2179
+ "learning_rate": 0.00012969020286208522,
2180
+ "loss": 0.5566,
2181
+ "step": 724000
2182
+ },
2183
+ {
2184
+ "epoch": 17.1,
2185
+ "learning_rate": 0.00012921843057084445,
2186
+ "loss": 0.5459,
2187
+ "step": 726000
2188
+ },
2189
+ {
2190
+ "epoch": 17.15,
2191
+ "learning_rate": 0.0001287466582796037,
2192
+ "loss": 0.5522,
2193
+ "step": 728000
2194
+ },
2195
+ {
2196
+ "epoch": 17.19,
2197
+ "learning_rate": 0.00012827488598836294,
2198
+ "loss": 0.5586,
2199
+ "step": 730000
2200
+ },
2201
+ {
2202
+ "epoch": 17.24,
2203
+ "learning_rate": 0.00012780311369712217,
2204
+ "loss": 0.5551,
2205
+ "step": 732000
2206
+ },
2207
+ {
2208
+ "epoch": 17.29,
2209
+ "learning_rate": 0.0001273313414058814,
2210
+ "loss": 0.558,
2211
+ "step": 734000
2212
+ },
2213
+ {
2214
+ "epoch": 17.33,
2215
+ "learning_rate": 0.00012685956911464067,
2216
+ "loss": 0.5501,
2217
+ "step": 736000
2218
+ },
2219
+ {
2220
+ "epoch": 17.38,
2221
+ "learning_rate": 0.0001263877968233999,
2222
+ "loss": 0.5496,
2223
+ "step": 738000
2224
+ },
2225
+ {
2226
+ "epoch": 17.43,
2227
+ "learning_rate": 0.00012591602453215913,
2228
+ "loss": 0.5617,
2229
+ "step": 740000
2230
+ },
2231
+ {
2232
+ "epoch": 17.48,
2233
+ "learning_rate": 0.00012544425224091837,
2234
+ "loss": 0.5481,
2235
+ "step": 742000
2236
+ },
2237
+ {
2238
+ "epoch": 17.52,
2239
+ "learning_rate": 0.00012497247994967763,
2240
+ "loss": 0.5627,
2241
+ "step": 744000
2242
+ },
2243
+ {
2244
+ "epoch": 17.57,
2245
+ "learning_rate": 0.00012450070765843686,
2246
+ "loss": 0.5556,
2247
+ "step": 746000
2248
+ },
2249
+ {
2250
+ "epoch": 17.62,
2251
+ "learning_rate": 0.0001240289353671961,
2252
+ "loss": 0.5565,
2253
+ "step": 748000
2254
+ },
2255
+ {
2256
+ "epoch": 17.66,
2257
+ "learning_rate": 0.00012355716307595532,
2258
+ "loss": 0.5591,
2259
+ "step": 750000
2260
+ },
2261
+ {
2262
+ "epoch": 17.71,
2263
+ "learning_rate": 0.00012308539078471456,
2264
+ "loss": 0.5526,
2265
+ "step": 752000
2266
+ },
2267
+ {
2268
+ "epoch": 17.76,
2269
+ "learning_rate": 0.00012261361849347382,
2270
+ "loss": 0.5564,
2271
+ "step": 754000
2272
+ },
2273
+ {
2274
+ "epoch": 17.8,
2275
+ "learning_rate": 0.00012214184620223305,
2276
+ "loss": 0.5541,
2277
+ "step": 756000
2278
+ },
2279
+ {
2280
+ "epoch": 17.85,
2281
+ "learning_rate": 0.00012167007391099228,
2282
+ "loss": 0.5601,
2283
+ "step": 758000
2284
+ },
2285
+ {
2286
+ "epoch": 17.9,
2287
+ "learning_rate": 0.00012119830161975153,
2288
+ "loss": 0.5555,
2289
+ "step": 760000
2290
+ },
2291
+ {
2292
+ "epoch": 17.95,
2293
+ "learning_rate": 0.00012072652932851076,
2294
+ "loss": 0.5525,
2295
+ "step": 762000
2296
+ },
2297
+ {
2298
+ "epoch": 17.99,
2299
+ "learning_rate": 0.00012025475703727,
2300
+ "loss": 0.5577,
2301
+ "step": 764000
2302
+ },
2303
+ {
2304
+ "epoch": 18.04,
2305
+ "learning_rate": 0.00011978298474602924,
2306
+ "loss": 0.5403,
2307
+ "step": 766000
2308
+ },
2309
+ {
2310
+ "epoch": 18.09,
2311
+ "learning_rate": 0.00011931121245478847,
2312
+ "loss": 0.5466,
2313
+ "step": 768000
2314
+ },
2315
+ {
2316
+ "epoch": 18.13,
2317
+ "learning_rate": 0.0001188394401635477,
2318
+ "loss": 0.546,
2319
+ "step": 770000
2320
+ },
2321
+ {
2322
+ "epoch": 18.18,
2323
+ "learning_rate": 0.00011836766787230695,
2324
+ "loss": 0.5435,
2325
+ "step": 772000
2326
+ },
2327
+ {
2328
+ "epoch": 18.23,
2329
+ "learning_rate": 0.0001178958955810662,
2330
+ "loss": 0.5442,
2331
+ "step": 774000
2332
+ },
2333
+ {
2334
+ "epoch": 18.28,
2335
+ "learning_rate": 0.00011742412328982544,
2336
+ "loss": 0.5381,
2337
+ "step": 776000
2338
+ },
2339
+ {
2340
+ "epoch": 18.32,
2341
+ "learning_rate": 0.00011695235099858468,
2342
+ "loss": 0.542,
2343
+ "step": 778000
2344
+ },
2345
+ {
2346
+ "epoch": 18.37,
2347
+ "learning_rate": 0.00011648057870734391,
2348
+ "loss": 0.5458,
2349
+ "step": 780000
2350
+ },
2351
+ {
2352
+ "epoch": 18.42,
2353
+ "learning_rate": 0.00011600880641610316,
2354
+ "loss": 0.5473,
2355
+ "step": 782000
2356
+ },
2357
+ {
2358
+ "epoch": 18.46,
2359
+ "learning_rate": 0.00011553703412486239,
2360
+ "loss": 0.5511,
2361
+ "step": 784000
2362
+ },
2363
+ {
2364
+ "epoch": 18.51,
2365
+ "learning_rate": 0.00011506526183362162,
2366
+ "loss": 0.5415,
2367
+ "step": 786000
2368
+ },
2369
+ {
2370
+ "epoch": 18.56,
2371
+ "learning_rate": 0.00011459348954238087,
2372
+ "loss": 0.5431,
2373
+ "step": 788000
2374
+ },
2375
+ {
2376
+ "epoch": 18.61,
2377
+ "learning_rate": 0.0001141217172511401,
2378
+ "loss": 0.5403,
2379
+ "step": 790000
2380
+ },
2381
+ {
2382
+ "epoch": 18.65,
2383
+ "learning_rate": 0.00011364994495989935,
2384
+ "loss": 0.5402,
2385
+ "step": 792000
2386
+ },
2387
+ {
2388
+ "epoch": 18.7,
2389
+ "learning_rate": 0.00011317817266865859,
2390
+ "loss": 0.5447,
2391
+ "step": 794000
2392
+ },
2393
+ {
2394
+ "epoch": 18.75,
2395
+ "learning_rate": 0.00011270640037741783,
2396
+ "loss": 0.5485,
2397
+ "step": 796000
2398
+ },
2399
+ {
2400
+ "epoch": 18.79,
2401
+ "learning_rate": 0.00011223462808617706,
2402
+ "loss": 0.5485,
2403
+ "step": 798000
2404
+ },
2405
+ {
2406
+ "epoch": 18.84,
2407
+ "learning_rate": 0.0001117628557949363,
2408
+ "loss": 0.5441,
2409
+ "step": 800000
2410
+ },
2411
+ {
2412
+ "epoch": 18.89,
2413
+ "learning_rate": 0.00011129108350369554,
2414
+ "loss": 0.5457,
2415
+ "step": 802000
2416
+ },
2417
+ {
2418
+ "epoch": 18.94,
2419
+ "learning_rate": 0.00011081931121245477,
2420
+ "loss": 0.5416,
2421
+ "step": 804000
2422
+ },
2423
+ {
2424
+ "epoch": 18.98,
2425
+ "learning_rate": 0.00011034753892121402,
2426
+ "loss": 0.5494,
2427
+ "step": 806000
2428
+ },
2429
+ {
2430
+ "epoch": 19.03,
2431
+ "learning_rate": 0.00010987576662997325,
2432
+ "loss": 0.5347,
2433
+ "step": 808000
2434
+ },
2435
+ {
2436
+ "epoch": 19.08,
2437
+ "learning_rate": 0.00010940399433873251,
2438
+ "loss": 0.5257,
2439
+ "step": 810000
2440
+ },
2441
+ {
2442
+ "epoch": 19.12,
2443
+ "learning_rate": 0.00010893222204749174,
2444
+ "loss": 0.5278,
2445
+ "step": 812000
2446
+ },
2447
+ {
2448
+ "epoch": 19.17,
2449
+ "learning_rate": 0.00010846044975625097,
2450
+ "loss": 0.5355,
2451
+ "step": 814000
2452
+ },
2453
+ {
2454
+ "epoch": 19.22,
2455
+ "learning_rate": 0.00010798867746501022,
2456
+ "loss": 0.5378,
2457
+ "step": 816000
2458
+ },
2459
+ {
2460
+ "epoch": 19.27,
2461
+ "learning_rate": 0.00010751690517376945,
2462
+ "loss": 0.537,
2463
+ "step": 818000
2464
+ },
2465
+ {
2466
+ "epoch": 19.31,
2467
+ "learning_rate": 0.00010704513288252869,
2468
+ "loss": 0.5385,
2469
+ "step": 820000
2470
+ },
2471
+ {
2472
+ "epoch": 19.36,
2473
+ "learning_rate": 0.00010657336059128793,
2474
+ "loss": 0.5302,
2475
+ "step": 822000
2476
+ },
2477
+ {
2478
+ "epoch": 19.41,
2479
+ "learning_rate": 0.00010610158830004717,
2480
+ "loss": 0.5409,
2481
+ "step": 824000
2482
+ },
2483
+ {
2484
+ "epoch": 19.45,
2485
+ "learning_rate": 0.0001056298160088064,
2486
+ "loss": 0.5347,
2487
+ "step": 826000
2488
+ },
2489
+ {
2490
+ "epoch": 19.5,
2491
+ "learning_rate": 0.00010515804371756563,
2492
+ "loss": 0.5288,
2493
+ "step": 828000
2494
+ },
2495
+ {
2496
+ "epoch": 19.55,
2497
+ "learning_rate": 0.00010468627142632489,
2498
+ "loss": 0.5276,
2499
+ "step": 830000
2500
+ },
2501
+ {
2502
+ "epoch": 19.59,
2503
+ "learning_rate": 0.00010421449913508412,
2504
+ "loss": 0.5334,
2505
+ "step": 832000
2506
+ },
2507
+ {
2508
+ "epoch": 19.64,
2509
+ "learning_rate": 0.00010374272684384337,
2510
+ "loss": 0.5407,
2511
+ "step": 834000
2512
+ },
2513
+ {
2514
+ "epoch": 19.69,
2515
+ "learning_rate": 0.0001032709545526026,
2516
+ "loss": 0.5354,
2517
+ "step": 836000
2518
+ },
2519
+ {
2520
+ "epoch": 19.74,
2521
+ "learning_rate": 0.00010279918226136184,
2522
+ "loss": 0.5274,
2523
+ "step": 838000
2524
+ },
2525
+ {
2526
+ "epoch": 19.78,
2527
+ "learning_rate": 0.00010232740997012108,
2528
+ "loss": 0.529,
2529
+ "step": 840000
2530
+ },
2531
+ {
2532
+ "epoch": 19.83,
2533
+ "learning_rate": 0.00010185563767888031,
2534
+ "loss": 0.5327,
2535
+ "step": 842000
2536
+ },
2537
+ {
2538
+ "epoch": 19.88,
2539
+ "learning_rate": 0.00010138386538763955,
2540
+ "loss": 0.5383,
2541
+ "step": 844000
2542
+ },
2543
+ {
2544
+ "epoch": 19.92,
2545
+ "learning_rate": 0.0001009120930963988,
2546
+ "loss": 0.5352,
2547
+ "step": 846000
2548
+ },
2549
+ {
2550
+ "epoch": 19.97,
2551
+ "learning_rate": 0.00010044032080515804,
2552
+ "loss": 0.5355,
2553
+ "step": 848000
2554
+ },
2555
+ {
2556
+ "epoch": 20.02,
2557
+ "learning_rate": 9.996854851391727e-05,
2558
+ "loss": 0.5233,
2559
+ "step": 850000
2560
+ },
2561
+ {
2562
+ "epoch": 20.07,
2563
+ "learning_rate": 9.949677622267652e-05,
2564
+ "loss": 0.5255,
2565
+ "step": 852000
2566
+ },
2567
+ {
2568
+ "epoch": 20.11,
2569
+ "learning_rate": 9.902500393143575e-05,
2570
+ "loss": 0.5168,
2571
+ "step": 854000
2572
+ },
2573
+ {
2574
+ "epoch": 20.16,
2575
+ "learning_rate": 9.855323164019498e-05,
2576
+ "loss": 0.5181,
2577
+ "step": 856000
2578
+ },
2579
+ {
2580
+ "epoch": 20.21,
2581
+ "learning_rate": 9.808145934895423e-05,
2582
+ "loss": 0.5245,
2583
+ "step": 858000
2584
+ },
2585
+ {
2586
+ "epoch": 20.25,
2587
+ "learning_rate": 9.760968705771346e-05,
2588
+ "loss": 0.5191,
2589
+ "step": 860000
2590
+ },
2591
+ {
2592
+ "epoch": 20.3,
2593
+ "learning_rate": 9.71379147664727e-05,
2594
+ "loss": 0.5207,
2595
+ "step": 862000
2596
+ },
2597
+ {
2598
+ "epoch": 20.35,
2599
+ "learning_rate": 9.666614247523194e-05,
2600
+ "loss": 0.5251,
2601
+ "step": 864000
2602
+ },
2603
+ {
2604
+ "epoch": 20.4,
2605
+ "learning_rate": 9.619437018399119e-05,
2606
+ "loss": 0.5158,
2607
+ "step": 866000
2608
+ },
2609
+ {
2610
+ "epoch": 20.44,
2611
+ "learning_rate": 9.572259789275043e-05,
2612
+ "loss": 0.5178,
2613
+ "step": 868000
2614
+ },
2615
+ {
2616
+ "epoch": 20.49,
2617
+ "learning_rate": 9.525082560150967e-05,
2618
+ "loss": 0.5213,
2619
+ "step": 870000
2620
+ },
2621
+ {
2622
+ "epoch": 20.54,
2623
+ "learning_rate": 9.47790533102689e-05,
2624
+ "loss": 0.5271,
2625
+ "step": 872000
2626
+ },
2627
+ {
2628
+ "epoch": 20.58,
2629
+ "learning_rate": 9.430728101902815e-05,
2630
+ "loss": 0.5242,
2631
+ "step": 874000
2632
+ },
2633
+ {
2634
+ "epoch": 20.63,
2635
+ "learning_rate": 9.383550872778738e-05,
2636
+ "loss": 0.5278,
2637
+ "step": 876000
2638
+ },
2639
+ {
2640
+ "epoch": 20.68,
2641
+ "learning_rate": 9.336373643654661e-05,
2642
+ "loss": 0.5222,
2643
+ "step": 878000
2644
+ },
2645
+ {
2646
+ "epoch": 20.73,
2647
+ "learning_rate": 9.289196414530586e-05,
2648
+ "loss": 0.5242,
2649
+ "step": 880000
2650
+ },
2651
+ {
2652
+ "epoch": 20.77,
2653
+ "learning_rate": 9.242019185406509e-05,
2654
+ "loss": 0.5271,
2655
+ "step": 882000
2656
+ },
2657
+ {
2658
+ "epoch": 20.82,
2659
+ "learning_rate": 9.194841956282434e-05,
2660
+ "loss": 0.5267,
2661
+ "step": 884000
2662
+ },
2663
+ {
2664
+ "epoch": 20.87,
2665
+ "learning_rate": 9.147664727158358e-05,
2666
+ "loss": 0.5195,
2667
+ "step": 886000
2668
+ },
2669
+ {
2670
+ "epoch": 20.91,
2671
+ "learning_rate": 9.100487498034282e-05,
2672
+ "loss": 0.5162,
2673
+ "step": 888000
2674
+ },
2675
+ {
2676
+ "epoch": 20.96,
2677
+ "learning_rate": 9.053310268910205e-05,
2678
+ "loss": 0.5166,
2679
+ "step": 890000
2680
+ },
2681
+ {
2682
+ "epoch": 21.01,
2683
+ "learning_rate": 9.00613303978613e-05,
2684
+ "loss": 0.5266,
2685
+ "step": 892000
2686
+ },
2687
+ {
2688
+ "epoch": 21.06,
2689
+ "learning_rate": 8.958955810662053e-05,
2690
+ "loss": 0.5149,
2691
+ "step": 894000
2692
+ },
2693
+ {
2694
+ "epoch": 21.1,
2695
+ "learning_rate": 8.911778581537976e-05,
2696
+ "loss": 0.5023,
2697
+ "step": 896000
2698
+ },
2699
+ {
2700
+ "epoch": 21.15,
2701
+ "learning_rate": 8.864601352413901e-05,
2702
+ "loss": 0.5112,
2703
+ "step": 898000
2704
+ },
2705
+ {
2706
+ "epoch": 21.2,
2707
+ "learning_rate": 8.817424123289824e-05,
2708
+ "loss": 0.5084,
2709
+ "step": 900000
2710
+ },
2711
+ {
2712
+ "epoch": 21.24,
2713
+ "learning_rate": 8.770246894165747e-05,
2714
+ "loss": 0.509,
2715
+ "step": 902000
2716
+ },
2717
+ {
2718
+ "epoch": 21.29,
2719
+ "learning_rate": 8.723069665041673e-05,
2720
+ "loss": 0.5043,
2721
+ "step": 904000
2722
+ },
2723
+ {
2724
+ "epoch": 21.34,
2725
+ "learning_rate": 8.675892435917597e-05,
2726
+ "loss": 0.5188,
2727
+ "step": 906000
2728
+ },
2729
+ {
2730
+ "epoch": 21.38,
2731
+ "learning_rate": 8.628715206793521e-05,
2732
+ "loss": 0.5094,
2733
+ "step": 908000
2734
+ },
2735
+ {
2736
+ "epoch": 21.43,
2737
+ "learning_rate": 8.581537977669444e-05,
2738
+ "loss": 0.519,
2739
+ "step": 910000
2740
+ },
2741
+ {
2742
+ "epoch": 21.48,
2743
+ "learning_rate": 8.534360748545368e-05,
2744
+ "loss": 0.5108,
2745
+ "step": 912000
2746
+ },
2747
+ {
2748
+ "epoch": 21.53,
2749
+ "learning_rate": 8.487183519421292e-05,
2750
+ "loss": 0.5054,
2751
+ "step": 914000
2752
+ },
2753
+ {
2754
+ "epoch": 21.57,
2755
+ "learning_rate": 8.440006290297216e-05,
2756
+ "loss": 0.5092,
2757
+ "step": 916000
2758
+ },
2759
+ {
2760
+ "epoch": 21.62,
2761
+ "learning_rate": 8.392829061173139e-05,
2762
+ "loss": 0.5075,
2763
+ "step": 918000
2764
+ },
2765
+ {
2766
+ "epoch": 21.67,
2767
+ "learning_rate": 8.345651832049062e-05,
2768
+ "loss": 0.5024,
2769
+ "step": 920000
2770
+ },
2771
+ {
2772
+ "epoch": 21.71,
2773
+ "learning_rate": 8.298474602924988e-05,
2774
+ "loss": 0.5161,
2775
+ "step": 922000
2776
+ },
2777
+ {
2778
+ "epoch": 21.76,
2779
+ "learning_rate": 8.251297373800911e-05,
2780
+ "loss": 0.5117,
2781
+ "step": 924000
2782
+ },
2783
+ {
2784
+ "epoch": 21.81,
2785
+ "learning_rate": 8.204120144676836e-05,
2786
+ "loss": 0.5136,
2787
+ "step": 926000
2788
+ },
2789
+ {
2790
+ "epoch": 21.86,
2791
+ "learning_rate": 8.15694291555276e-05,
2792
+ "loss": 0.5062,
2793
+ "step": 928000
2794
+ },
2795
+ {
2796
+ "epoch": 21.9,
2797
+ "learning_rate": 8.109765686428683e-05,
2798
+ "loss": 0.5122,
2799
+ "step": 930000
2800
+ },
2801
+ {
2802
+ "epoch": 21.95,
2803
+ "learning_rate": 8.062588457304607e-05,
2804
+ "loss": 0.5113,
2805
+ "step": 932000
2806
+ },
2807
+ {
2808
+ "epoch": 22.0,
2809
+ "learning_rate": 8.01541122818053e-05,
2810
+ "loss": 0.5073,
2811
+ "step": 934000
2812
+ },
2813
+ {
2814
+ "epoch": 22.04,
2815
+ "learning_rate": 7.968233999056454e-05,
2816
+ "loss": 0.4986,
2817
+ "step": 936000
2818
+ },
2819
+ {
2820
+ "epoch": 22.09,
2821
+ "learning_rate": 7.921056769932378e-05,
2822
+ "loss": 0.5061,
2823
+ "step": 938000
2824
+ },
2825
+ {
2826
+ "epoch": 22.14,
2827
+ "learning_rate": 7.873879540808303e-05,
2828
+ "loss": 0.4963,
2829
+ "step": 940000
2830
+ },
2831
+ {
2832
+ "epoch": 22.19,
2833
+ "learning_rate": 7.826702311684228e-05,
2834
+ "loss": 0.4981,
2835
+ "step": 942000
2836
+ },
2837
+ {
2838
+ "epoch": 22.23,
2839
+ "learning_rate": 7.779525082560151e-05,
2840
+ "loss": 0.498,
2841
+ "step": 944000
2842
+ },
2843
+ {
2844
+ "epoch": 22.28,
2845
+ "learning_rate": 7.732347853436074e-05,
2846
+ "loss": 0.4963,
2847
+ "step": 946000
2848
+ },
2849
+ {
2850
+ "epoch": 22.33,
2851
+ "learning_rate": 7.685170624311998e-05,
2852
+ "loss": 0.4987,
2853
+ "step": 948000
2854
+ },
2855
+ {
2856
+ "epoch": 22.37,
2857
+ "learning_rate": 7.637993395187922e-05,
2858
+ "loss": 0.4999,
2859
+ "step": 950000
2860
+ },
2861
+ {
2862
+ "epoch": 22.42,
2863
+ "learning_rate": 7.590816166063845e-05,
2864
+ "loss": 0.4945,
2865
+ "step": 952000
2866
+ },
2867
+ {
2868
+ "epoch": 22.47,
2869
+ "learning_rate": 7.543638936939769e-05,
2870
+ "loss": 0.4986,
2871
+ "step": 954000
2872
+ },
2873
+ {
2874
+ "epoch": 22.52,
2875
+ "learning_rate": 7.496461707815693e-05,
2876
+ "loss": 0.4956,
2877
+ "step": 956000
2878
+ },
2879
+ {
2880
+ "epoch": 22.56,
2881
+ "learning_rate": 7.449284478691618e-05,
2882
+ "loss": 0.4958,
2883
+ "step": 958000
2884
+ },
2885
+ {
2886
+ "epoch": 22.61,
2887
+ "learning_rate": 7.402107249567541e-05,
2888
+ "loss": 0.5033,
2889
+ "step": 960000
2890
+ },
2891
+ {
2892
+ "epoch": 22.66,
2893
+ "learning_rate": 7.354930020443464e-05,
2894
+ "loss": 0.4976,
2895
+ "step": 962000
2896
+ },
2897
+ {
2898
+ "epoch": 22.7,
2899
+ "learning_rate": 7.307752791319389e-05,
2900
+ "loss": 0.5036,
2901
+ "step": 964000
2902
+ },
2903
+ {
2904
+ "epoch": 22.75,
2905
+ "learning_rate": 7.260575562195314e-05,
2906
+ "loss": 0.4973,
2907
+ "step": 966000
2908
+ },
2909
+ {
2910
+ "epoch": 22.8,
2911
+ "learning_rate": 7.213398333071237e-05,
2912
+ "loss": 0.501,
2913
+ "step": 968000
2914
+ },
2915
+ {
2916
+ "epoch": 22.85,
2917
+ "learning_rate": 7.16622110394716e-05,
2918
+ "loss": 0.5005,
2919
+ "step": 970000
2920
+ },
2921
+ {
2922
+ "epoch": 22.89,
2923
+ "learning_rate": 7.119043874823085e-05,
2924
+ "loss": 0.4991,
2925
+ "step": 972000
2926
+ },
2927
+ {
2928
+ "epoch": 22.94,
2929
+ "learning_rate": 7.07186664569901e-05,
2930
+ "loss": 0.495,
2931
+ "step": 974000
2932
+ },
2933
+ {
2934
+ "epoch": 22.99,
2935
+ "learning_rate": 7.024689416574933e-05,
2936
+ "loss": 0.5016,
2937
+ "step": 976000
2938
+ },
2939
+ {
2940
+ "epoch": 23.03,
2941
+ "learning_rate": 6.977512187450856e-05,
2942
+ "loss": 0.4966,
2943
+ "step": 978000
2944
+ },
2945
+ {
2946
+ "epoch": 23.08,
2947
+ "learning_rate": 6.93033495832678e-05,
2948
+ "loss": 0.4909,
2949
+ "step": 980000
2950
+ },
2951
+ {
2952
+ "epoch": 23.13,
2953
+ "learning_rate": 6.883157729202704e-05,
2954
+ "loss": 0.4846,
2955
+ "step": 982000
2956
+ },
2957
+ {
2958
+ "epoch": 23.17,
2959
+ "learning_rate": 6.835980500078629e-05,
2960
+ "loss": 0.4938,
2961
+ "step": 984000
2962
+ },
2963
+ {
2964
+ "epoch": 23.22,
2965
+ "learning_rate": 6.788803270954552e-05,
2966
+ "loss": 0.4967,
2967
+ "step": 986000
2968
+ },
2969
+ {
2970
+ "epoch": 23.27,
2971
+ "learning_rate": 6.741626041830475e-05,
2972
+ "loss": 0.4894,
2973
+ "step": 988000
2974
+ },
2975
+ {
2976
+ "epoch": 23.32,
2977
+ "learning_rate": 6.6944488127064e-05,
2978
+ "loss": 0.4912,
2979
+ "step": 990000
2980
+ },
2981
+ {
2982
+ "epoch": 23.36,
2983
+ "learning_rate": 6.647271583582324e-05,
2984
+ "loss": 0.4898,
2985
+ "step": 992000
2986
+ },
2987
+ {
2988
+ "epoch": 23.41,
2989
+ "learning_rate": 6.600094354458248e-05,
2990
+ "loss": 0.4854,
2991
+ "step": 994000
2992
+ },
2993
+ {
2994
+ "epoch": 23.46,
2995
+ "learning_rate": 6.552917125334171e-05,
2996
+ "loss": 0.4838,
2997
+ "step": 996000
2998
+ },
2999
+ {
3000
+ "epoch": 23.5,
3001
+ "learning_rate": 6.505739896210096e-05,
3002
+ "loss": 0.4846,
3003
+ "step": 998000
3004
+ },
3005
+ {
3006
+ "epoch": 23.55,
3007
+ "learning_rate": 6.45856266708602e-05,
3008
+ "loss": 0.4829,
3009
+ "step": 1000000
3010
+ },
3011
+ {
3012
+ "epoch": 23.6,
3013
+ "learning_rate": 6.411385437961944e-05,
3014
+ "loss": 0.4792,
3015
+ "step": 1002000
3016
+ },
3017
+ {
3018
+ "epoch": 23.65,
3019
+ "learning_rate": 6.364208208837867e-05,
3020
+ "loss": 0.486,
3021
+ "step": 1004000
3022
+ },
3023
+ {
3024
+ "epoch": 23.69,
3025
+ "learning_rate": 6.317030979713791e-05,
3026
+ "loss": 0.4946,
3027
+ "step": 1006000
3028
+ },
3029
+ {
3030
+ "epoch": 23.74,
3031
+ "learning_rate": 6.269853750589715e-05,
3032
+ "loss": 0.4825,
3033
+ "step": 1008000
3034
+ },
3035
+ {
3036
+ "epoch": 23.79,
3037
+ "learning_rate": 6.22267652146564e-05,
3038
+ "loss": 0.4845,
3039
+ "step": 1010000
3040
+ },
3041
+ {
3042
+ "epoch": 23.83,
3043
+ "learning_rate": 6.175499292341563e-05,
3044
+ "loss": 0.4875,
3045
+ "step": 1012000
3046
+ },
3047
+ {
3048
+ "epoch": 23.88,
3049
+ "learning_rate": 6.128322063217486e-05,
3050
+ "loss": 0.4935,
3051
+ "step": 1014000
3052
+ },
3053
+ {
3054
+ "epoch": 23.93,
3055
+ "learning_rate": 6.0811448340934105e-05,
3056
+ "loss": 0.4834,
3057
+ "step": 1016000
3058
+ },
3059
+ {
3060
+ "epoch": 23.98,
3061
+ "learning_rate": 6.033967604969334e-05,
3062
+ "loss": 0.4853,
3063
+ "step": 1018000
3064
+ },
3065
+ {
3066
+ "epoch": 24.02,
3067
+ "learning_rate": 5.9867903758452584e-05,
3068
+ "loss": 0.485,
3069
+ "step": 1020000
3070
+ },
3071
+ {
3072
+ "epoch": 24.07,
3073
+ "learning_rate": 5.9396131467211824e-05,
3074
+ "loss": 0.4821,
3075
+ "step": 1022000
3076
+ },
3077
+ {
3078
+ "epoch": 24.12,
3079
+ "learning_rate": 5.8924359175971057e-05,
3080
+ "loss": 0.4739,
3081
+ "step": 1024000
3082
+ },
3083
+ {
3084
+ "epoch": 24.16,
3085
+ "learning_rate": 5.8452586884730296e-05,
3086
+ "loss": 0.4824,
3087
+ "step": 1026000
3088
+ },
3089
+ {
3090
+ "epoch": 24.21,
3091
+ "learning_rate": 5.798081459348954e-05,
3092
+ "loss": 0.4823,
3093
+ "step": 1028000
3094
+ },
3095
+ {
3096
+ "epoch": 24.26,
3097
+ "learning_rate": 5.7509042302248775e-05,
3098
+ "loss": 0.4805,
3099
+ "step": 1030000
3100
+ },
3101
+ {
3102
+ "epoch": 24.31,
3103
+ "learning_rate": 5.7037270011008015e-05,
3104
+ "loss": 0.4727,
3105
+ "step": 1032000
3106
+ },
3107
+ {
3108
+ "epoch": 24.35,
3109
+ "learning_rate": 5.6565497719767254e-05,
3110
+ "loss": 0.4726,
3111
+ "step": 1034000
3112
+ },
3113
+ {
3114
+ "epoch": 24.4,
3115
+ "learning_rate": 5.609372542852649e-05,
3116
+ "loss": 0.4734,
3117
+ "step": 1036000
3118
+ },
3119
+ {
3120
+ "epoch": 24.45,
3121
+ "learning_rate": 5.562195313728573e-05,
3122
+ "loss": 0.4775,
3123
+ "step": 1038000
3124
+ },
3125
+ {
3126
+ "epoch": 24.49,
3127
+ "learning_rate": 5.515018084604497e-05,
3128
+ "loss": 0.4761,
3129
+ "step": 1040000
3130
+ },
3131
+ {
3132
+ "epoch": 24.54,
3133
+ "learning_rate": 5.467840855480421e-05,
3134
+ "loss": 0.4692,
3135
+ "step": 1042000
3136
+ },
3137
+ {
3138
+ "epoch": 24.59,
3139
+ "learning_rate": 5.4206636263563445e-05,
3140
+ "loss": 0.4826,
3141
+ "step": 1044000
3142
+ },
3143
+ {
3144
+ "epoch": 24.63,
3145
+ "learning_rate": 5.3734863972322684e-05,
3146
+ "loss": 0.4767,
3147
+ "step": 1046000
3148
+ },
3149
+ {
3150
+ "epoch": 24.68,
3151
+ "learning_rate": 5.326309168108193e-05,
3152
+ "loss": 0.4733,
3153
+ "step": 1048000
3154
+ },
3155
+ {
3156
+ "epoch": 24.73,
3157
+ "learning_rate": 5.2791319389841163e-05,
3158
+ "loss": 0.4772,
3159
+ "step": 1050000
3160
+ },
3161
+ {
3162
+ "epoch": 24.78,
3163
+ "learning_rate": 5.23195470986004e-05,
3164
+ "loss": 0.4758,
3165
+ "step": 1052000
3166
+ },
3167
+ {
3168
+ "epoch": 24.82,
3169
+ "learning_rate": 5.184777480735964e-05,
3170
+ "loss": 0.48,
3171
+ "step": 1054000
3172
+ },
3173
+ {
3174
+ "epoch": 24.87,
3175
+ "learning_rate": 5.137600251611889e-05,
3176
+ "loss": 0.477,
3177
+ "step": 1056000
3178
+ },
3179
+ {
3180
+ "epoch": 24.92,
3181
+ "learning_rate": 5.090423022487812e-05,
3182
+ "loss": 0.4715,
3183
+ "step": 1058000
3184
+ },
3185
+ {
3186
+ "epoch": 24.96,
3187
+ "learning_rate": 5.043245793363736e-05,
3188
+ "loss": 0.4696,
3189
+ "step": 1060000
3190
+ },
3191
+ {
3192
+ "epoch": 25.01,
3193
+ "learning_rate": 4.99606856423966e-05,
3194
+ "loss": 0.4682,
3195
+ "step": 1062000
3196
+ },
3197
+ {
3198
+ "epoch": 25.06,
3199
+ "learning_rate": 4.948891335115583e-05,
3200
+ "loss": 0.4725,
3201
+ "step": 1064000
3202
+ },
3203
+ {
3204
+ "epoch": 25.11,
3205
+ "learning_rate": 4.901714105991508e-05,
3206
+ "loss": 0.4655,
3207
+ "step": 1066000
3208
+ },
3209
+ {
3210
+ "epoch": 25.15,
3211
+ "learning_rate": 4.854536876867432e-05,
3212
+ "loss": 0.4629,
3213
+ "step": 1068000
3214
+ },
3215
+ {
3216
+ "epoch": 25.2,
3217
+ "learning_rate": 4.807359647743355e-05,
3218
+ "loss": 0.4654,
3219
+ "step": 1070000
3220
+ },
3221
+ {
3222
+ "epoch": 25.25,
3223
+ "learning_rate": 4.760182418619279e-05,
3224
+ "loss": 0.4663,
3225
+ "step": 1072000
3226
+ },
3227
+ {
3228
+ "epoch": 25.29,
3229
+ "learning_rate": 4.713005189495203e-05,
3230
+ "loss": 0.4682,
3231
+ "step": 1074000
3232
+ },
3233
+ {
3234
+ "epoch": 25.34,
3235
+ "learning_rate": 4.665827960371127e-05,
3236
+ "loss": 0.4647,
3237
+ "step": 1076000
3238
+ },
3239
+ {
3240
+ "epoch": 25.39,
3241
+ "learning_rate": 4.618650731247051e-05,
3242
+ "loss": 0.4722,
3243
+ "step": 1078000
3244
+ },
3245
+ {
3246
+ "epoch": 25.44,
3247
+ "learning_rate": 4.571473502122975e-05,
3248
+ "loss": 0.4604,
3249
+ "step": 1080000
3250
+ },
3251
+ {
3252
+ "epoch": 25.48,
3253
+ "learning_rate": 4.524296272998898e-05,
3254
+ "loss": 0.4585,
3255
+ "step": 1082000
3256
+ },
3257
+ {
3258
+ "epoch": 25.53,
3259
+ "learning_rate": 4.477119043874823e-05,
3260
+ "loss": 0.4619,
3261
+ "step": 1084000
3262
+ },
3263
+ {
3264
+ "epoch": 25.58,
3265
+ "learning_rate": 4.429941814750747e-05,
3266
+ "loss": 0.4661,
3267
+ "step": 1086000
3268
+ },
3269
+ {
3270
+ "epoch": 25.62,
3271
+ "learning_rate": 4.382764585626671e-05,
3272
+ "loss": 0.4724,
3273
+ "step": 1088000
3274
+ },
3275
+ {
3276
+ "epoch": 25.67,
3277
+ "learning_rate": 4.335587356502594e-05,
3278
+ "loss": 0.4667,
3279
+ "step": 1090000
3280
+ },
3281
+ {
3282
+ "epoch": 25.72,
3283
+ "learning_rate": 4.288410127378518e-05,
3284
+ "loss": 0.4692,
3285
+ "step": 1092000
3286
+ },
3287
+ {
3288
+ "epoch": 25.77,
3289
+ "learning_rate": 4.2412328982544426e-05,
3290
+ "loss": 0.4624,
3291
+ "step": 1094000
3292
+ },
3293
+ {
3294
+ "epoch": 25.81,
3295
+ "learning_rate": 4.194055669130366e-05,
3296
+ "loss": 0.4587,
3297
+ "step": 1096000
3298
+ },
3299
+ {
3300
+ "epoch": 25.86,
3301
+ "learning_rate": 4.14687844000629e-05,
3302
+ "loss": 0.4604,
3303
+ "step": 1098000
3304
+ },
3305
+ {
3306
+ "epoch": 25.91,
3307
+ "learning_rate": 4.099701210882214e-05,
3308
+ "loss": 0.4718,
3309
+ "step": 1100000
3310
+ },
3311
+ {
3312
+ "epoch": 25.95,
3313
+ "learning_rate": 4.052523981758137e-05,
3314
+ "loss": 0.4666,
3315
+ "step": 1102000
3316
+ },
3317
+ {
3318
+ "epoch": 26.0,
3319
+ "learning_rate": 4.005346752634062e-05,
3320
+ "loss": 0.4616,
3321
+ "step": 1104000
3322
+ },
3323
+ {
3324
+ "epoch": 26.05,
3325
+ "learning_rate": 3.9581695235099856e-05,
3326
+ "loss": 0.456,
3327
+ "step": 1106000
3328
+ },
3329
+ {
3330
+ "epoch": 26.1,
3331
+ "learning_rate": 3.9109922943859096e-05,
3332
+ "loss": 0.4573,
3333
+ "step": 1108000
3334
+ },
3335
+ {
3336
+ "epoch": 26.14,
3337
+ "learning_rate": 3.863815065261833e-05,
3338
+ "loss": 0.4522,
3339
+ "step": 1110000
3340
+ },
3341
+ {
3342
+ "epoch": 26.19,
3343
+ "learning_rate": 3.8166378361377575e-05,
3344
+ "loss": 0.4604,
3345
+ "step": 1112000
3346
+ },
3347
+ {
3348
+ "epoch": 26.24,
3349
+ "learning_rate": 3.7694606070136815e-05,
3350
+ "loss": 0.4539,
3351
+ "step": 1114000
3352
+ },
3353
+ {
3354
+ "epoch": 26.28,
3355
+ "learning_rate": 3.722283377889605e-05,
3356
+ "loss": 0.4485,
3357
+ "step": 1116000
3358
+ },
3359
+ {
3360
+ "epoch": 26.33,
3361
+ "learning_rate": 3.675106148765529e-05,
3362
+ "loss": 0.4596,
3363
+ "step": 1118000
3364
+ },
3365
+ {
3366
+ "epoch": 26.38,
3367
+ "learning_rate": 3.6279289196414526e-05,
3368
+ "loss": 0.4487,
3369
+ "step": 1120000
3370
+ },
3371
+ {
3372
+ "epoch": 26.42,
3373
+ "learning_rate": 3.5807516905173766e-05,
3374
+ "loss": 0.459,
3375
+ "step": 1122000
3376
+ },
3377
+ {
3378
+ "epoch": 26.47,
3379
+ "learning_rate": 3.5335744613933005e-05,
3380
+ "loss": 0.4609,
3381
+ "step": 1124000
3382
+ },
3383
+ {
3384
+ "epoch": 26.52,
3385
+ "learning_rate": 3.4863972322692245e-05,
3386
+ "loss": 0.4549,
3387
+ "step": 1126000
3388
+ },
3389
+ {
3390
+ "epoch": 26.57,
3391
+ "learning_rate": 3.4392200031451484e-05,
3392
+ "loss": 0.4553,
3393
+ "step": 1128000
3394
+ },
3395
+ {
3396
+ "epoch": 26.61,
3397
+ "learning_rate": 3.3920427740210724e-05,
3398
+ "loss": 0.4552,
3399
+ "step": 1130000
3400
+ },
3401
+ {
3402
+ "epoch": 26.66,
3403
+ "learning_rate": 3.344865544896996e-05,
3404
+ "loss": 0.4534,
3405
+ "step": 1132000
3406
+ },
3407
+ {
3408
+ "epoch": 26.71,
3409
+ "learning_rate": 3.29768831577292e-05,
3410
+ "loss": 0.4522,
3411
+ "step": 1134000
3412
+ },
3413
+ {
3414
+ "epoch": 26.75,
3415
+ "learning_rate": 3.2505110866488436e-05,
3416
+ "loss": 0.4544,
3417
+ "step": 1136000
3418
+ },
3419
+ {
3420
+ "epoch": 26.8,
3421
+ "learning_rate": 3.203333857524768e-05,
3422
+ "loss": 0.458,
3423
+ "step": 1138000
3424
+ },
3425
+ {
3426
+ "epoch": 26.85,
3427
+ "learning_rate": 3.1561566284006915e-05,
3428
+ "loss": 0.4537,
3429
+ "step": 1140000
3430
+ },
3431
+ {
3432
+ "epoch": 26.9,
3433
+ "learning_rate": 3.1089793992766154e-05,
3434
+ "loss": 0.4581,
3435
+ "step": 1142000
3436
+ },
3437
+ {
3438
+ "epoch": 26.94,
3439
+ "learning_rate": 3.0618021701525394e-05,
3440
+ "loss": 0.4525,
3441
+ "step": 1144000
3442
+ },
3443
+ {
3444
+ "epoch": 26.99,
3445
+ "learning_rate": 3.014624941028463e-05,
3446
+ "loss": 0.4562,
3447
+ "step": 1146000
3448
+ },
3449
+ {
3450
+ "epoch": 27.04,
3451
+ "learning_rate": 2.9674477119043873e-05,
3452
+ "loss": 0.4516,
3453
+ "step": 1148000
3454
+ },
3455
+ {
3456
+ "epoch": 27.08,
3457
+ "learning_rate": 2.920270482780311e-05,
3458
+ "loss": 0.4472,
3459
+ "step": 1150000
3460
+ },
3461
+ {
3462
+ "epoch": 27.13,
3463
+ "learning_rate": 2.8730932536562352e-05,
3464
+ "loss": 0.4507,
3465
+ "step": 1152000
3466
+ },
3467
+ {
3468
+ "epoch": 27.18,
3469
+ "learning_rate": 2.8259160245321588e-05,
3470
+ "loss": 0.4421,
3471
+ "step": 1154000
3472
+ },
3473
+ {
3474
+ "epoch": 27.23,
3475
+ "learning_rate": 2.778738795408083e-05,
3476
+ "loss": 0.4487,
3477
+ "step": 1156000
3478
+ },
3479
+ {
3480
+ "epoch": 27.27,
3481
+ "learning_rate": 2.7315615662840067e-05,
3482
+ "loss": 0.4528,
3483
+ "step": 1158000
3484
+ },
3485
+ {
3486
+ "epoch": 27.32,
3487
+ "learning_rate": 2.6843843371599303e-05,
3488
+ "loss": 0.4483,
3489
+ "step": 1160000
3490
+ },
3491
+ {
3492
+ "epoch": 27.37,
3493
+ "learning_rate": 2.6372071080358546e-05,
3494
+ "loss": 0.45,
3495
+ "step": 1162000
3496
+ },
3497
+ {
3498
+ "epoch": 27.41,
3499
+ "learning_rate": 2.5900298789117782e-05,
3500
+ "loss": 0.4517,
3501
+ "step": 1164000
3502
+ },
3503
+ {
3504
+ "epoch": 27.46,
3505
+ "learning_rate": 2.5428526497877022e-05,
3506
+ "loss": 0.4444,
3507
+ "step": 1166000
3508
+ },
3509
+ {
3510
+ "epoch": 27.51,
3511
+ "learning_rate": 2.495675420663626e-05,
3512
+ "loss": 0.4457,
3513
+ "step": 1168000
3514
+ },
3515
+ {
3516
+ "epoch": 27.56,
3517
+ "learning_rate": 2.44849819153955e-05,
3518
+ "loss": 0.4471,
3519
+ "step": 1170000
3520
+ },
3521
+ {
3522
+ "epoch": 27.6,
3523
+ "learning_rate": 2.401320962415474e-05,
3524
+ "loss": 0.4455,
3525
+ "step": 1172000
3526
+ },
3527
+ {
3528
+ "epoch": 27.65,
3529
+ "learning_rate": 2.3541437332913976e-05,
3530
+ "loss": 0.4466,
3531
+ "step": 1174000
3532
+ },
3533
+ {
3534
+ "epoch": 27.7,
3535
+ "learning_rate": 2.3069665041673216e-05,
3536
+ "loss": 0.4421,
3537
+ "step": 1176000
3538
+ },
3539
+ {
3540
+ "epoch": 27.74,
3541
+ "learning_rate": 2.2597892750432455e-05,
3542
+ "loss": 0.446,
3543
+ "step": 1178000
3544
+ },
3545
+ {
3546
+ "epoch": 27.79,
3547
+ "learning_rate": 2.2126120459191695e-05,
3548
+ "loss": 0.4441,
3549
+ "step": 1180000
3550
+ },
3551
+ {
3552
+ "epoch": 27.84,
3553
+ "learning_rate": 2.1654348167950935e-05,
3554
+ "loss": 0.4452,
3555
+ "step": 1182000
3556
+ },
3557
+ {
3558
+ "epoch": 27.89,
3559
+ "learning_rate": 2.1182575876710174e-05,
3560
+ "loss": 0.4418,
3561
+ "step": 1184000
3562
+ },
3563
+ {
3564
+ "epoch": 27.93,
3565
+ "learning_rate": 2.071080358546941e-05,
3566
+ "loss": 0.4396,
3567
+ "step": 1186000
3568
+ },
3569
+ {
3570
+ "epoch": 27.98,
3571
+ "learning_rate": 2.0239031294228653e-05,
3572
+ "loss": 0.4477,
3573
+ "step": 1188000
3574
+ },
3575
+ {
3576
+ "epoch": 28.03,
3577
+ "learning_rate": 1.976725900298789e-05,
3578
+ "loss": 0.4435,
3579
+ "step": 1190000
3580
+ },
3581
+ {
3582
+ "epoch": 28.07,
3583
+ "learning_rate": 1.9295486711747125e-05,
3584
+ "loss": 0.4379,
3585
+ "step": 1192000
3586
+ },
3587
+ {
3588
+ "epoch": 28.12,
3589
+ "learning_rate": 1.8823714420506368e-05,
3590
+ "loss": 0.4442,
3591
+ "step": 1194000
3592
+ },
3593
+ {
3594
+ "epoch": 28.17,
3595
+ "learning_rate": 1.8351942129265604e-05,
3596
+ "loss": 0.4329,
3597
+ "step": 1196000
3598
+ },
3599
+ {
3600
+ "epoch": 28.21,
3601
+ "learning_rate": 1.7880169838024844e-05,
3602
+ "loss": 0.4368,
3603
+ "step": 1198000
3604
+ },
3605
+ {
3606
+ "epoch": 28.26,
3607
+ "learning_rate": 1.7408397546784083e-05,
3608
+ "loss": 0.4427,
3609
+ "step": 1200000
3610
+ },
3611
+ {
3612
+ "epoch": 28.31,
3613
+ "learning_rate": 1.6936625255543323e-05,
3614
+ "loss": 0.4426,
3615
+ "step": 1202000
3616
+ },
3617
+ {
3618
+ "epoch": 28.36,
3619
+ "learning_rate": 1.6464852964302562e-05,
3620
+ "loss": 0.4374,
3621
+ "step": 1204000
3622
+ },
3623
+ {
3624
+ "epoch": 28.4,
3625
+ "learning_rate": 1.5993080673061802e-05,
3626
+ "loss": 0.4372,
3627
+ "step": 1206000
3628
+ },
3629
+ {
3630
+ "epoch": 28.45,
3631
+ "learning_rate": 1.552130838182104e-05,
3632
+ "loss": 0.4365,
3633
+ "step": 1208000
3634
+ },
3635
+ {
3636
+ "epoch": 28.5,
3637
+ "learning_rate": 1.504953609058028e-05,
3638
+ "loss": 0.4329,
3639
+ "step": 1210000
3640
+ },
3641
+ {
3642
+ "epoch": 28.54,
3643
+ "learning_rate": 1.4577763799339517e-05,
3644
+ "loss": 0.4397,
3645
+ "step": 1212000
3646
+ },
3647
+ {
3648
+ "epoch": 28.59,
3649
+ "learning_rate": 1.4105991508098757e-05,
3650
+ "loss": 0.4355,
3651
+ "step": 1214000
3652
+ },
3653
+ {
3654
+ "epoch": 28.64,
3655
+ "learning_rate": 1.3634219216857994e-05,
3656
+ "loss": 0.4386,
3657
+ "step": 1216000
3658
+ },
3659
+ {
3660
+ "epoch": 28.69,
3661
+ "learning_rate": 1.3162446925617234e-05,
3662
+ "loss": 0.4369,
3663
+ "step": 1218000
3664
+ },
3665
+ {
3666
+ "epoch": 28.73,
3667
+ "learning_rate": 1.2690674634376474e-05,
3668
+ "loss": 0.441,
3669
+ "step": 1220000
3670
+ },
3671
+ {
3672
+ "epoch": 28.78,
3673
+ "learning_rate": 1.2218902343135713e-05,
3674
+ "loss": 0.4383,
3675
+ "step": 1222000
3676
+ },
3677
+ {
3678
+ "epoch": 28.83,
3679
+ "learning_rate": 1.1747130051894953e-05,
3680
+ "loss": 0.4338,
3681
+ "step": 1224000
3682
+ },
3683
+ {
3684
+ "epoch": 28.87,
3685
+ "learning_rate": 1.1275357760654189e-05,
3686
+ "loss": 0.4373,
3687
+ "step": 1226000
3688
+ },
3689
+ {
3690
+ "epoch": 28.92,
3691
+ "learning_rate": 1.0803585469413428e-05,
3692
+ "loss": 0.4392,
3693
+ "step": 1228000
3694
+ },
3695
+ {
3696
+ "epoch": 28.97,
3697
+ "learning_rate": 1.0331813178172668e-05,
3698
+ "loss": 0.4401,
3699
+ "step": 1230000
3700
+ },
3701
+ {
3702
+ "epoch": 29.02,
3703
+ "learning_rate": 9.860040886931907e-06,
3704
+ "loss": 0.4394,
3705
+ "step": 1232000
3706
+ },
3707
+ {
3708
+ "epoch": 29.06,
3709
+ "learning_rate": 9.388268595691145e-06,
3710
+ "loss": 0.4316,
3711
+ "step": 1234000
3712
+ },
3713
+ {
3714
+ "epoch": 29.11,
3715
+ "learning_rate": 8.916496304450385e-06,
3716
+ "loss": 0.4372,
3717
+ "step": 1236000
3718
+ },
3719
+ {
3720
+ "epoch": 29.16,
3721
+ "learning_rate": 8.444724013209622e-06,
3722
+ "loss": 0.4363,
3723
+ "step": 1238000
3724
+ },
3725
+ {
3726
+ "epoch": 29.2,
3727
+ "learning_rate": 7.972951721968862e-06,
3728
+ "loss": 0.4348,
3729
+ "step": 1240000
3730
+ },
3731
+ {
3732
+ "epoch": 29.25,
3733
+ "learning_rate": 7.5011794307281015e-06,
3734
+ "loss": 0.4342,
3735
+ "step": 1242000
3736
+ },
3737
+ {
3738
+ "epoch": 29.3,
3739
+ "learning_rate": 7.02940713948734e-06,
3740
+ "loss": 0.4295,
3741
+ "step": 1244000
3742
+ },
3743
+ {
3744
+ "epoch": 29.35,
3745
+ "learning_rate": 6.557634848246579e-06,
3746
+ "loss": 0.4358,
3747
+ "step": 1246000
3748
+ },
3749
+ {
3750
+ "epoch": 29.39,
3751
+ "learning_rate": 6.085862557005818e-06,
3752
+ "loss": 0.4317,
3753
+ "step": 1248000
3754
+ },
3755
+ {
3756
+ "epoch": 29.44,
3757
+ "learning_rate": 5.614090265765056e-06,
3758
+ "loss": 0.431,
3759
+ "step": 1250000
3760
+ },
3761
+ {
3762
+ "epoch": 29.49,
3763
+ "learning_rate": 5.142317974524296e-06,
3764
+ "loss": 0.4355,
3765
+ "step": 1252000
3766
+ },
3767
+ {
3768
+ "epoch": 29.53,
3769
+ "learning_rate": 4.670545683283535e-06,
3770
+ "loss": 0.4301,
3771
+ "step": 1254000
3772
+ },
3773
+ {
3774
+ "epoch": 29.58,
3775
+ "learning_rate": 4.198773392042774e-06,
3776
+ "loss": 0.4272,
3777
+ "step": 1256000
3778
+ },
3779
+ {
3780
+ "epoch": 29.63,
3781
+ "learning_rate": 3.7270011008020125e-06,
3782
+ "loss": 0.4325,
3783
+ "step": 1258000
3784
+ },
3785
+ {
3786
+ "epoch": 29.67,
3787
+ "learning_rate": 3.255228809561251e-06,
3788
+ "loss": 0.4288,
3789
+ "step": 1260000
3790
+ },
3791
+ {
3792
+ "epoch": 29.72,
3793
+ "learning_rate": 2.7834565183204907e-06,
3794
+ "loss": 0.4246,
3795
+ "step": 1262000
3796
+ },
3797
+ {
3798
+ "epoch": 29.77,
3799
+ "learning_rate": 2.3116842270797294e-06,
3800
+ "loss": 0.437,
3801
+ "step": 1264000
3802
+ },
3803
+ {
3804
+ "epoch": 29.82,
3805
+ "learning_rate": 1.8399119358389683e-06,
3806
+ "loss": 0.4337,
3807
+ "step": 1266000
3808
+ },
3809
+ {
3810
+ "epoch": 29.86,
3811
+ "learning_rate": 1.3681396445982072e-06,
3812
+ "loss": 0.4281,
3813
+ "step": 1268000
3814
+ },
3815
+ {
3816
+ "epoch": 29.91,
3817
+ "learning_rate": 8.963673533574461e-07,
3818
+ "loss": 0.4334,
3819
+ "step": 1270000
3820
+ },
3821
+ {
3822
+ "epoch": 29.96,
3823
+ "learning_rate": 4.2459506211668494e-07,
3824
+ "loss": 0.4294,
3825
+ "step": 1272000
3826
+ },
3827
+ {
3828
+ "epoch": 30.0,
3829
+ "step": 1273800,
3830
+ "total_flos": 2.5779090018764566e+21,
3831
+ "train_loss": 0.6109381775774412,
3832
+ "train_runtime": 724079.1012,
3833
+ "train_samples_per_second": 28.147,
3834
+ "train_steps_per_second": 1.759
3835
+ }
3836
+ ],
3837
+ "logging_steps": 2000,
3838
+ "max_steps": 1273800,
3839
+ "num_input_tokens_seen": 0,
3840
+ "num_train_epochs": 30,
3841
+ "save_steps": 500,
3842
+ "total_flos": 2.5779090018764566e+21,
3843
+ "train_batch_size": 2,
3844
+ "trial_name": null,
3845
+ "trial_params": null
3846
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cecbe2308480a0db70978cbda8460cbfae2844e623308cee5fa4aa246d04a482
3
+ size 4728
vocab.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "\"": 1,
3
+ ">": 2,
4
+ "[PAD]": 95,
5
+ "[UNK]": 94,
6
+ "\\": 3,
7
+ "a": 4,
8
+ "b": 5,
9
+ "c": 6,
10
+ "d": 7,
11
+ "e": 8,
12
+ "f": 9,
13
+ "g": 10,
14
+ "h": 11,
15
+ "i": 12,
16
+ "j": 13,
17
+ "k": 14,
18
+ "l": 15,
19
+ "m": 16,
20
+ "n": 17,
21
+ "o": 18,
22
+ "p": 19,
23
+ "q": 20,
24
+ "r": 21,
25
+ "s": 22,
26
+ "t": 23,
27
+ "u": 24,
28
+ "v": 25,
29
+ "w": 26,
30
+ "x": 27,
31
+ "y": 28,
32
+ "z": 29,
33
+ "|": 0,
34
+ "ç": 30,
35
+ "é": 31,
36
+ "õ": 32,
37
+ "ş": 33,
38
+ "ơ": 34,
39
+ "ư": 35,
40
+ "̀": 36,
41
+ "υ": 37,
42
+ "а": 38,
43
+ "в": 39,
44
+ "г": 40,
45
+ "д": 41,
46
+ "е": 42,
47
+ "з": 43,
48
+ "к": 44,
49
+ "л": 45,
50
+ "н": 46,
51
+ "п": 47,
52
+ "с": 48,
53
+ "ф": 49,
54
+ "ч": 50,
55
+ "ả": 51,
56
+ "ậ": 52,
57
+ "ồ": 53,
58
+ "ờ": 54,
59
+ "​": 55,
60
+ "’": 56,
61
+ "⁄": 57,
62
+ "れ": 58,
63
+ "ㅑ": 59,
64
+ "ㅛ": 60,
65
+ "각": 61,
66
+ "국": 62,
67
+ "기": 63,
68
+ "는": 64,
69
+ "데": 65,
70
+ "도": 66,
71
+ "드": 67,
72
+ "들": 68,
73
+ "랜": 69,
74
+ "로": 70,
75
+ "를": 71,
76
+ "만": 72,
77
+ "많": 73,
78
+ "부": 74,
79
+ "브": 75,
80
+ "생": 76,
81
+ "솔": 77,
82
+ "스": 78,
83
+ "앤": 79,
84
+ "에": 80,
85
+ "여": 81,
86
+ "온": 82,
87
+ "원": 83,
88
+ "전": 84,
89
+ "제": 85,
90
+ "지": 86,
91
+ "천": 87,
92
+ "초": 88,
93
+ "터": 89,
94
+ "트": 90,
95
+ "한": 91,
96
+ "행": 92,
97
+ "�": 93
98
+ }