Hugging Face's logo

AlienKevin
/
canto_ocr

Model card Files
xet
Community
2
AlienKevin commited on
Commit
8f1ccbe
·
1 Parent(s): 18f80d0

Obtain 82% accuracy on more than 8000 sentences

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +2 -2
  2. checkpoint/trocr-custdata-8000/checkpoint-5000/config.json +180 -0
  3. checkpoint/trocr-custdata-8000/checkpoint-5000/optimizer.pt +3 -0
  4. checkpoint/{trocr-custdata/checkpoint-1000 → trocr-custdata-8000/checkpoint-5000}/preprocessor_config.json +0 -0
  5. checkpoint/trocr-custdata-8000/checkpoint-5000/pytorch_model.bin +3 -0
  6. checkpoint/trocr-custdata-8000/checkpoint-5000/scheduler.pt +3 -0
  7. checkpoint/trocr-custdata-8000/checkpoint-5000/trainer_state.json +3066 -0
  8. checkpoint/trocr-custdata-8000/checkpoint-5000/training_args.bin +3 -0
  9. checkpoint/trocr-custdata-8000/checkpoint-6000/config.json +180 -0
  10. checkpoint/trocr-custdata-8000/checkpoint-6000/optimizer.pt +3 -0
  11. checkpoint/{trocr-custdata/checkpoint-2000 → trocr-custdata-8000/checkpoint-6000}/preprocessor_config.json +0 -0
  12. checkpoint/trocr-custdata-8000/checkpoint-6000/pytorch_model.bin +3 -0
  13. checkpoint/trocr-custdata-8000/checkpoint-6000/scheduler.pt +3 -0
  14. checkpoint/trocr-custdata-8000/checkpoint-6000/trainer_state.json +3676 -0
  15. checkpoint/trocr-custdata-8000/checkpoint-6000/training_args.bin +3 -0
  16. checkpoint/trocr-custdata-8000/checkpoint-7000/config.json +180 -0
  17. checkpoint/trocr-custdata-8000/checkpoint-7000/optimizer.pt +3 -0
  18. checkpoint/{trocr-custdata/checkpoint-3000 → trocr-custdata-8000/checkpoint-7000}/preprocessor_config.json +0 -0
  19. checkpoint/trocr-custdata-8000/checkpoint-7000/pytorch_model.bin +3 -0
  20. checkpoint/trocr-custdata-8000/checkpoint-7000/scheduler.pt +3 -0
  21. checkpoint/trocr-custdata-8000/checkpoint-7000/trainer_state.json +4286 -0
  22. checkpoint/trocr-custdata-8000/checkpoint-7000/training_args.bin +3 -0
  23. checkpoint/trocr-custdata-8000/checkpoint-8000/config.json +180 -0
  24. checkpoint/trocr-custdata-8000/checkpoint-8000/optimizer.pt +3 -0
  25. checkpoint/{trocr-custdata/checkpoint-4000 → trocr-custdata-8000/checkpoint-8000}/preprocessor_config.json +0 -0
  26. checkpoint/trocr-custdata-8000/checkpoint-8000/pytorch_model.bin +3 -0
  27. checkpoint/trocr-custdata-8000/checkpoint-8000/scheduler.pt +3 -0
  28. checkpoint/trocr-custdata-8000/checkpoint-8000/trainer_state.json +0 -0
  29. checkpoint/trocr-custdata-8000/checkpoint-8000/training_args.bin +3 -0
  30. checkpoint/trocr-custdata-8000/checkpoint-9000/config.json +180 -0
  31. checkpoint/trocr-custdata-8000/checkpoint-9000/optimizer.pt +3 -0
  32. checkpoint/trocr-custdata-8000/checkpoint-9000/preprocessor_config.json +22 -0
  33. checkpoint/trocr-custdata-8000/checkpoint-9000/pytorch_model.bin +3 -0
  34. checkpoint/trocr-custdata-8000/checkpoint-9000/scheduler.pt +3 -0
  35. checkpoint/trocr-custdata-8000/checkpoint-9000/trainer_state.json +0 -0
  36. checkpoint/trocr-custdata-8000/checkpoint-9000/training_args.bin +3 -0
  37. checkpoint/trocr-custdata-8000/last/config.json +180 -0
  38. checkpoint/{trocr-custdata → trocr-custdata-8000}/last/merges.txt +0 -0
  39. checkpoint/{trocr-custdata → trocr-custdata-8000}/last/preprocessor_config.json +0 -0
  40. checkpoint/trocr-custdata-8000/last/pytorch_model.bin +3 -0
  41. checkpoint/{trocr-custdata → trocr-custdata-8000}/last/special_tokens_map.json +0 -0
  42. checkpoint/trocr-custdata-8000/last/tokenizer.json +3408 -0
  43. checkpoint/{trocr-custdata → trocr-custdata-8000}/last/tokenizer_config.json +0 -0
  44. checkpoint/trocr-custdata-8000/last/training_args.bin +3 -0
  45. checkpoint/trocr-custdata-8000/last/vocab.json +1 -0
  46. checkpoint/trocr-custdata/checkpoint-1000/config.json +0 -180
  47. checkpoint/trocr-custdata/checkpoint-1000/optimizer.pt +0 -3
  48. checkpoint/trocr-custdata/checkpoint-1000/pytorch_model.bin +0 -3
  49. checkpoint/trocr-custdata/checkpoint-1000/scheduler.pt +0 -3
  50. checkpoint/trocr-custdata/checkpoint-1000/trainer_state.json +0 -626
README.md CHANGED
@@ -21,7 +21,7 @@ docker run --gpus all -it -v /tmp/trocr-chinese:/trocr-chinese trocr-chinese:lat
21
  8. Generate custom vocab:
22
  ```
23
  python gen_vocab.py \
24
- --dataset_path "dataset/*.txt" \
25
  --cust_vocab ./cust-data/vocab.txt
26
  ```
27
  9. Download pretrained weights from https://pan.baidu.com/s/1rARdfadQlQGKGHa3de82BA, password: 0o65
@@ -35,7 +35,7 @@ To enable M1 GPU support, install the dev version of transformers by running `pi
35
  In Dec 21, 2022, the dev version that's working for me is `transformers-4.26.0.dev0`. Later stable releases may have M1 GPU support built-in so you don't need to install the dev version.
36
  If you are running the whole procedure again, remember to reinstall the older transformers version as instructed in step 10. Otherwise, the weights initialized will not be in the correct format and you will see miserable accuracy rate, likely due to breaking changes involving how tokenization is done.
37
  ```
38
- python train.py --cust_data_init_weights_path ./cust-data/weights --checkpoint_path ./checkpoint/trocr-custdata --dataset_path "./dataset/*.jpg" --per_device_train_batch_size 8
39
  ```
40
 
41
  ## 训练
 
21
  8. Generate custom vocab:
22
  ```
23
  python gen_vocab.py \
24
+ --dataset_path "dataset/*/*.txt" \
25
  --cust_vocab ./cust-data/vocab.txt
26
  ```
27
  9. Download pretrained weights from https://pan.baidu.com/s/1rARdfadQlQGKGHa3de82BA, password: 0o65
 
35
  In Dec 21, 2022, the dev version that's working for me is `transformers-4.26.0.dev0`. Later stable releases may have M1 GPU support built-in so you don't need to install the dev version.
36
  If you are running the whole procedure again, remember to reinstall the older transformers version as instructed in step 10. Otherwise, the weights initialized will not be in the correct format and you will see miserable accuracy rate, likely due to breaking changes involving how tokenization is done.
37
  ```
38
+ python train.py --cust_data_init_weights_path ./cust-data/weights --checkpoint_path ./checkpoint/trocr-custdata --dataset_path "./dataset/*/*.jpg" --per_device_train_batch_size 8
39
  ```
40
 
41
  ## 训练
checkpoint/trocr-custdata-8000/checkpoint-5000/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/trocr-custdata-8000/checkpoint-5000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65851b8f1512fe492b4db754e60363a55f28000469438978682e3736cea00c26
3
+ size 243062533
checkpoint/{trocr-custdata/checkpoint-1000 → trocr-custdata-8000/checkpoint-5000}/preprocessor_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/checkpoint-5000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8af687b725c2365115846f3881465541ed682f38d3076fcd3c1aa44b68b1127
3
+ size 122142261
checkpoint/trocr-custdata-8000/checkpoint-5000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f067598c41ba885b676b605d72cfb9c3ac9493bb8694381ea51b16ff5422c6d8
3
+ size 627
checkpoint/trocr-custdata-8000/checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,3066 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 5.2246603970741905,
5
+ "global_step": 5000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 4.994775339602926e-05,
13
+ "loss": 0.3717,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 4.989550679205852e-05,
19
+ "loss": 0.2368,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 4.9843260188087774e-05,
25
+ "loss": 0.1928,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 4.979101358411704e-05,
31
+ "loss": 0.1648,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 4.9738766980146295e-05,
37
+ "loss": 0.1645,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 4.968652037617555e-05,
43
+ "loss": 0.1151,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 4.963427377220481e-05,
49
+ "loss": 0.1206,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 4.958202716823407e-05,
55
+ "loss": 0.1289,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 4.9529780564263324e-05,
61
+ "loss": 0.1123,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 4.947753396029258e-05,
67
+ "loss": 0.0992,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.11,
72
+ "learning_rate": 4.9425287356321845e-05,
73
+ "loss": 0.097,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.13,
78
+ "learning_rate": 4.93730407523511e-05,
79
+ "loss": 0.1028,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.14,
84
+ "learning_rate": 4.932079414838036e-05,
85
+ "loss": 0.087,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.15,
90
+ "learning_rate": 4.9268547544409617e-05,
91
+ "loss": 0.0907,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.16,
96
+ "learning_rate": 4.9216300940438874e-05,
97
+ "loss": 0.0866,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.17,
102
+ "learning_rate": 4.916405433646813e-05,
103
+ "loss": 0.085,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.18,
108
+ "learning_rate": 4.911180773249739e-05,
109
+ "loss": 0.0846,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.19,
114
+ "learning_rate": 4.905956112852665e-05,
115
+ "loss": 0.0791,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.2,
120
+ "learning_rate": 4.900731452455591e-05,
121
+ "loss": 0.073,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.21,
126
+ "learning_rate": 4.895506792058516e-05,
127
+ "loss": 0.0861,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.22,
132
+ "learning_rate": 4.8902821316614424e-05,
133
+ "loss": 0.0828,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.23,
138
+ "learning_rate": 4.885057471264368e-05,
139
+ "loss": 0.0711,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.24,
144
+ "learning_rate": 4.879832810867294e-05,
145
+ "loss": 0.069,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.25,
150
+ "learning_rate": 4.8746081504702195e-05,
151
+ "loss": 0.0647,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.26,
156
+ "learning_rate": 4.869383490073145e-05,
157
+ "loss": 0.0862,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.27,
162
+ "learning_rate": 4.8641588296760716e-05,
163
+ "loss": 0.0628,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.28,
168
+ "learning_rate": 4.858934169278997e-05,
169
+ "loss": 0.0697,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.29,
174
+ "learning_rate": 4.853709508881923e-05,
175
+ "loss": 0.0685,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.3,
180
+ "learning_rate": 4.848484848484849e-05,
181
+ "loss": 0.0602,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.31,
186
+ "learning_rate": 4.8432601880877745e-05,
187
+ "loss": 0.0576,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.32,
192
+ "learning_rate": 4.8380355276907e-05,
193
+ "loss": 0.068,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.33,
198
+ "learning_rate": 4.832810867293626e-05,
199
+ "loss": 0.0645,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.34,
204
+ "learning_rate": 4.827586206896552e-05,
205
+ "loss": 0.0785,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.36,
210
+ "learning_rate": 4.8223615464994774e-05,
211
+ "loss": 0.0606,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.37,
216
+ "learning_rate": 4.817136886102404e-05,
217
+ "loss": 0.0682,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.38,
222
+ "learning_rate": 4.8119122257053295e-05,
223
+ "loss": 0.0682,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.39,
228
+ "learning_rate": 4.806687565308255e-05,
229
+ "loss": 0.0605,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.4,
234
+ "learning_rate": 4.801462904911181e-05,
235
+ "loss": 0.0578,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.41,
240
+ "learning_rate": 4.7962382445141066e-05,
241
+ "loss": 0.0562,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.42,
246
+ "learning_rate": 4.791013584117033e-05,
247
+ "loss": 0.0724,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.43,
252
+ "learning_rate": 4.785788923719958e-05,
253
+ "loss": 0.056,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.44,
258
+ "learning_rate": 4.7805642633228845e-05,
259
+ "loss": 0.0509,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.45,
264
+ "learning_rate": 4.77533960292581e-05,
265
+ "loss": 0.0516,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.46,
270
+ "learning_rate": 4.770114942528736e-05,
271
+ "loss": 0.0588,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.47,
276
+ "learning_rate": 4.7648902821316616e-05,
277
+ "loss": 0.0527,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.48,
282
+ "learning_rate": 4.7596656217345873e-05,
283
+ "loss": 0.0614,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.49,
288
+ "learning_rate": 4.754440961337514e-05,
289
+ "loss": 0.0464,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.5,
294
+ "learning_rate": 4.749216300940439e-05,
295
+ "loss": 0.0515,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.51,
300
+ "learning_rate": 4.743991640543365e-05,
301
+ "loss": 0.0456,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.52,
306
+ "learning_rate": 4.738766980146291e-05,
307
+ "loss": 0.0583,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.53,
312
+ "learning_rate": 4.7335423197492166e-05,
313
+ "loss": 0.0595,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 0.54,
318
+ "learning_rate": 4.728317659352142e-05,
319
+ "loss": 0.0547,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 0.55,
324
+ "learning_rate": 4.723092998955068e-05,
325
+ "loss": 0.0451,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 0.56,
330
+ "learning_rate": 4.7178683385579944e-05,
331
+ "loss": 0.0395,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 0.57,
336
+ "learning_rate": 4.7126436781609195e-05,
337
+ "loss": 0.0493,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 0.59,
342
+ "learning_rate": 4.707419017763845e-05,
343
+ "loss": 0.0664,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 0.6,
348
+ "learning_rate": 4.7021943573667716e-05,
349
+ "loss": 0.0445,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 0.61,
354
+ "learning_rate": 4.696969696969697e-05,
355
+ "loss": 0.0443,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 0.62,
360
+ "learning_rate": 4.691745036572623e-05,
361
+ "loss": 0.0461,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 0.63,
366
+ "learning_rate": 4.686520376175549e-05,
367
+ "loss": 0.0415,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 0.64,
372
+ "learning_rate": 4.6812957157784745e-05,
373
+ "loss": 0.0483,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 0.65,
378
+ "learning_rate": 4.6760710553814e-05,
379
+ "loss": 0.0424,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 0.66,
384
+ "learning_rate": 4.670846394984326e-05,
385
+ "loss": 0.0327,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 0.67,
390
+ "learning_rate": 4.665621734587252e-05,
391
+ "loss": 0.0421,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 0.68,
396
+ "learning_rate": 4.660397074190178e-05,
397
+ "loss": 0.0491,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 0.69,
402
+ "learning_rate": 4.655172413793104e-05,
403
+ "loss": 0.0501,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 0.7,
408
+ "learning_rate": 4.6499477533960295e-05,
409
+ "loss": 0.0447,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 0.71,
414
+ "learning_rate": 4.644723092998955e-05,
415
+ "loss": 0.0449,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 0.72,
420
+ "learning_rate": 4.639498432601881e-05,
421
+ "loss": 0.0478,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 0.73,
426
+ "learning_rate": 4.6342737722048066e-05,
427
+ "loss": 0.0505,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 0.74,
432
+ "learning_rate": 4.629049111807733e-05,
433
+ "loss": 0.0394,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 0.75,
438
+ "learning_rate": 4.623824451410659e-05,
439
+ "loss": 0.0406,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 0.76,
444
+ "learning_rate": 4.6185997910135844e-05,
445
+ "loss": 0.0347,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 0.77,
450
+ "learning_rate": 4.61337513061651e-05,
451
+ "loss": 0.0445,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 0.78,
456
+ "learning_rate": 4.608150470219436e-05,
457
+ "loss": 0.0428,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 0.79,
462
+ "learning_rate": 4.6029258098223616e-05,
463
+ "loss": 0.0466,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 0.8,
468
+ "learning_rate": 4.597701149425287e-05,
469
+ "loss": 0.0395,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 0.82,
474
+ "learning_rate": 4.592476489028214e-05,
475
+ "loss": 0.0387,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 0.83,
480
+ "learning_rate": 4.5872518286311394e-05,
481
+ "loss": 0.0395,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 0.84,
486
+ "learning_rate": 4.582027168234065e-05,
487
+ "loss": 0.0305,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 0.85,
492
+ "learning_rate": 4.576802507836991e-05,
493
+ "loss": 0.0434,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 0.86,
498
+ "learning_rate": 4.5715778474399166e-05,
499
+ "loss": 0.0395,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 0.87,
504
+ "learning_rate": 4.566353187042842e-05,
505
+ "loss": 0.0374,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 0.88,
510
+ "learning_rate": 4.561128526645768e-05,
511
+ "loss": 0.0321,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 0.89,
516
+ "learning_rate": 4.5559038662486944e-05,
517
+ "loss": 0.0394,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 0.9,
522
+ "learning_rate": 4.55067920585162e-05,
523
+ "loss": 0.0383,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 0.91,
528
+ "learning_rate": 4.545454545454546e-05,
529
+ "loss": 0.0388,
530
+ "step": 870
531
+ },
532
+ {
533
+ "epoch": 0.92,
534
+ "learning_rate": 4.5402298850574716e-05,
535
+ "loss": 0.035,
536
+ "step": 880
537
+ },
538
+ {
539
+ "epoch": 0.93,
540
+ "learning_rate": 4.535005224660397e-05,
541
+ "loss": 0.0414,
542
+ "step": 890
543
+ },
544
+ {
545
+ "epoch": 0.94,
546
+ "learning_rate": 4.529780564263323e-05,
547
+ "loss": 0.0439,
548
+ "step": 900
549
+ },
550
+ {
551
+ "epoch": 0.95,
552
+ "learning_rate": 4.524555903866249e-05,
553
+ "loss": 0.0337,
554
+ "step": 910
555
+ },
556
+ {
557
+ "epoch": 0.96,
558
+ "learning_rate": 4.519331243469175e-05,
559
+ "loss": 0.0384,
560
+ "step": 920
561
+ },
562
+ {
563
+ "epoch": 0.97,
564
+ "learning_rate": 4.514106583072101e-05,
565
+ "loss": 0.0353,
566
+ "step": 930
567
+ },
568
+ {
569
+ "epoch": 0.98,
570
+ "learning_rate": 4.508881922675026e-05,
571
+ "loss": 0.0351,
572
+ "step": 940
573
+ },
574
+ {
575
+ "epoch": 0.99,
576
+ "learning_rate": 4.503657262277952e-05,
577
+ "loss": 0.0324,
578
+ "step": 950
579
+ },
580
+ {
581
+ "epoch": 1.0,
582
+ "learning_rate": 4.498432601880878e-05,
583
+ "loss": 0.0296,
584
+ "step": 960
585
+ },
586
+ {
587
+ "epoch": 1.01,
588
+ "learning_rate": 4.493207941483804e-05,
589
+ "loss": 0.0262,
590
+ "step": 970
591
+ },
592
+ {
593
+ "epoch": 1.02,
594
+ "learning_rate": 4.4879832810867294e-05,
595
+ "loss": 0.0274,
596
+ "step": 980
597
+ },
598
+ {
599
+ "epoch": 1.03,
600
+ "learning_rate": 4.482758620689655e-05,
601
+ "loss": 0.0278,
602
+ "step": 990
603
+ },
604
+ {
605
+ "epoch": 1.04,
606
+ "learning_rate": 4.4775339602925815e-05,
607
+ "loss": 0.0205,
608
+ "step": 1000
609
+ },
610
+ {
611
+ "epoch": 1.04,
612
+ "eval_acc": 0.5359801475533992,
613
+ "eval_cer": 0.06462110646211065,
614
+ "eval_loss": 0.0336129367351532,
615
+ "eval_runtime": 135.8305,
616
+ "eval_samples_per_second": 2.967,
617
+ "eval_steps_per_second": 0.375,
618
+ "step": 1000
619
+ },
620
+ {
621
+ "epoch": 1.06,
622
+ "learning_rate": 4.4723092998955066e-05,
623
+ "loss": 0.0264,
624
+ "step": 1010
625
+ },
626
+ {
627
+ "epoch": 1.07,
628
+ "learning_rate": 4.467084639498433e-05,
629
+ "loss": 0.0218,
630
+ "step": 1020
631
+ },
632
+ {
633
+ "epoch": 1.08,
634
+ "learning_rate": 4.461859979101359e-05,
635
+ "loss": 0.0269,
636
+ "step": 1030
637
+ },
638
+ {
639
+ "epoch": 1.09,
640
+ "learning_rate": 4.4566353187042844e-05,
641
+ "loss": 0.0242,
642
+ "step": 1040
643
+ },
644
+ {
645
+ "epoch": 1.1,
646
+ "learning_rate": 4.45141065830721e-05,
647
+ "loss": 0.03,
648
+ "step": 1050
649
+ },
650
+ {
651
+ "epoch": 1.11,
652
+ "learning_rate": 4.446185997910136e-05,
653
+ "loss": 0.0214,
654
+ "step": 1060
655
+ },
656
+ {
657
+ "epoch": 1.12,
658
+ "learning_rate": 4.440961337513062e-05,
659
+ "loss": 0.0301,
660
+ "step": 1070
661
+ },
662
+ {
663
+ "epoch": 1.13,
664
+ "learning_rate": 4.435736677115987e-05,
665
+ "loss": 0.0217,
666
+ "step": 1080
667
+ },
668
+ {
669
+ "epoch": 1.14,
670
+ "learning_rate": 4.430512016718914e-05,
671
+ "loss": 0.0323,
672
+ "step": 1090
673
+ },
674
+ {
675
+ "epoch": 1.15,
676
+ "learning_rate": 4.4252873563218394e-05,
677
+ "loss": 0.0258,
678
+ "step": 1100
679
+ },
680
+ {
681
+ "epoch": 1.16,
682
+ "learning_rate": 4.420062695924765e-05,
683
+ "loss": 0.0293,
684
+ "step": 1110
685
+ },
686
+ {
687
+ "epoch": 1.17,
688
+ "learning_rate": 4.414838035527691e-05,
689
+ "loss": 0.0265,
690
+ "step": 1120
691
+ },
692
+ {
693
+ "epoch": 1.18,
694
+ "learning_rate": 4.4096133751306166e-05,
695
+ "loss": 0.0237,
696
+ "step": 1130
697
+ },
698
+ {
699
+ "epoch": 1.19,
700
+ "learning_rate": 4.404388714733543e-05,
701
+ "loss": 0.0275,
702
+ "step": 1140
703
+ },
704
+ {
705
+ "epoch": 1.2,
706
+ "learning_rate": 4.399164054336468e-05,
707
+ "loss": 0.0219,
708
+ "step": 1150
709
+ },
710
+ {
711
+ "epoch": 1.21,
712
+ "learning_rate": 4.3939393939393944e-05,
713
+ "loss": 0.0334,
714
+ "step": 1160
715
+ },
716
+ {
717
+ "epoch": 1.22,
718
+ "learning_rate": 4.38871473354232e-05,
719
+ "loss": 0.0183,
720
+ "step": 1170
721
+ },
722
+ {
723
+ "epoch": 1.23,
724
+ "learning_rate": 4.383490073145246e-05,
725
+ "loss": 0.0241,
726
+ "step": 1180
727
+ },
728
+ {
729
+ "epoch": 1.24,
730
+ "learning_rate": 4.3782654127481716e-05,
731
+ "loss": 0.0233,
732
+ "step": 1190
733
+ },
734
+ {
735
+ "epoch": 1.25,
736
+ "learning_rate": 4.373040752351097e-05,
737
+ "loss": 0.025,
738
+ "step": 1200
739
+ },
740
+ {
741
+ "epoch": 1.26,
742
+ "learning_rate": 4.367816091954024e-05,
743
+ "loss": 0.0228,
744
+ "step": 1210
745
+ },
746
+ {
747
+ "epoch": 1.27,
748
+ "learning_rate": 4.362591431556949e-05,
749
+ "loss": 0.0182,
750
+ "step": 1220
751
+ },
752
+ {
753
+ "epoch": 1.29,
754
+ "learning_rate": 4.357366771159875e-05,
755
+ "loss": 0.0234,
756
+ "step": 1230
757
+ },
758
+ {
759
+ "epoch": 1.3,
760
+ "learning_rate": 4.352142110762801e-05,
761
+ "loss": 0.019,
762
+ "step": 1240
763
+ },
764
+ {
765
+ "epoch": 1.31,
766
+ "learning_rate": 4.346917450365726e-05,
767
+ "loss": 0.0191,
768
+ "step": 1250
769
+ },
770
+ {
771
+ "epoch": 1.32,
772
+ "learning_rate": 4.341692789968652e-05,
773
+ "loss": 0.0177,
774
+ "step": 1260
775
+ },
776
+ {
777
+ "epoch": 1.33,
778
+ "learning_rate": 4.336468129571578e-05,
779
+ "loss": 0.0214,
780
+ "step": 1270
781
+ },
782
+ {
783
+ "epoch": 1.34,
784
+ "learning_rate": 4.3312434691745044e-05,
785
+ "loss": 0.0252,
786
+ "step": 1280
787
+ },
788
+ {
789
+ "epoch": 1.35,
790
+ "learning_rate": 4.3260188087774294e-05,
791
+ "loss": 0.0233,
792
+ "step": 1290
793
+ },
794
+ {
795
+ "epoch": 1.36,
796
+ "learning_rate": 4.320794148380355e-05,
797
+ "loss": 0.0184,
798
+ "step": 1300
799
+ },
800
+ {
801
+ "epoch": 1.37,
802
+ "learning_rate": 4.3155694879832815e-05,
803
+ "loss": 0.0212,
804
+ "step": 1310
805
+ },
806
+ {
807
+ "epoch": 1.38,
808
+ "learning_rate": 4.3103448275862066e-05,
809
+ "loss": 0.0186,
810
+ "step": 1320
811
+ },
812
+ {
813
+ "epoch": 1.39,
814
+ "learning_rate": 4.305120167189133e-05,
815
+ "loss": 0.0179,
816
+ "step": 1330
817
+ },
818
+ {
819
+ "epoch": 1.4,
820
+ "learning_rate": 4.299895506792059e-05,
821
+ "loss": 0.0184,
822
+ "step": 1340
823
+ },
824
+ {
825
+ "epoch": 1.41,
826
+ "learning_rate": 4.294670846394985e-05,
827
+ "loss": 0.0208,
828
+ "step": 1350
829
+ },
830
+ {
831
+ "epoch": 1.42,
832
+ "learning_rate": 4.28944618599791e-05,
833
+ "loss": 0.0218,
834
+ "step": 1360
835
+ },
836
+ {
837
+ "epoch": 1.43,
838
+ "learning_rate": 4.284221525600836e-05,
839
+ "loss": 0.0203,
840
+ "step": 1370
841
+ },
842
+ {
843
+ "epoch": 1.44,
844
+ "learning_rate": 4.278996865203762e-05,
845
+ "loss": 0.022,
846
+ "step": 1380
847
+ },
848
+ {
849
+ "epoch": 1.45,
850
+ "learning_rate": 4.273772204806687e-05,
851
+ "loss": 0.0247,
852
+ "step": 1390
853
+ },
854
+ {
855
+ "epoch": 1.46,
856
+ "learning_rate": 4.268547544409614e-05,
857
+ "loss": 0.0195,
858
+ "step": 1400
859
+ },
860
+ {
861
+ "epoch": 1.47,
862
+ "learning_rate": 4.2633228840125394e-05,
863
+ "loss": 0.0236,
864
+ "step": 1410
865
+ },
866
+ {
867
+ "epoch": 1.48,
868
+ "learning_rate": 4.258098223615465e-05,
869
+ "loss": 0.0182,
870
+ "step": 1420
871
+ },
872
+ {
873
+ "epoch": 1.49,
874
+ "learning_rate": 4.252873563218391e-05,
875
+ "loss": 0.0238,
876
+ "step": 1430
877
+ },
878
+ {
879
+ "epoch": 1.5,
880
+ "learning_rate": 4.2476489028213165e-05,
881
+ "loss": 0.0244,
882
+ "step": 1440
883
+ },
884
+ {
885
+ "epoch": 1.52,
886
+ "learning_rate": 4.242424242424243e-05,
887
+ "loss": 0.0366,
888
+ "step": 1450
889
+ },
890
+ {
891
+ "epoch": 1.53,
892
+ "learning_rate": 4.2371995820271687e-05,
893
+ "loss": 0.0159,
894
+ "step": 1460
895
+ },
896
+ {
897
+ "epoch": 1.54,
898
+ "learning_rate": 4.2319749216300944e-05,
899
+ "loss": 0.0146,
900
+ "step": 1470
901
+ },
902
+ {
903
+ "epoch": 1.55,
904
+ "learning_rate": 4.22675026123302e-05,
905
+ "loss": 0.0219,
906
+ "step": 1480
907
+ },
908
+ {
909
+ "epoch": 1.56,
910
+ "learning_rate": 4.221525600835946e-05,
911
+ "loss": 0.0176,
912
+ "step": 1490
913
+ },
914
+ {
915
+ "epoch": 1.57,
916
+ "learning_rate": 4.2163009404388715e-05,
917
+ "loss": 0.0207,
918
+ "step": 1500
919
+ },
920
+ {
921
+ "epoch": 1.58,
922
+ "learning_rate": 4.211076280041797e-05,
923
+ "loss": 0.0141,
924
+ "step": 1510
925
+ },
926
+ {
927
+ "epoch": 1.59,
928
+ "learning_rate": 4.2058516196447236e-05,
929
+ "loss": 0.023,
930
+ "step": 1520
931
+ },
932
+ {
933
+ "epoch": 1.6,
934
+ "learning_rate": 4.2006269592476494e-05,
935
+ "loss": 0.0154,
936
+ "step": 1530
937
+ },
938
+ {
939
+ "epoch": 1.61,
940
+ "learning_rate": 4.195402298850575e-05,
941
+ "loss": 0.0208,
942
+ "step": 1540
943
+ },
944
+ {
945
+ "epoch": 1.62,
946
+ "learning_rate": 4.190177638453501e-05,
947
+ "loss": 0.0283,
948
+ "step": 1550
949
+ },
950
+ {
951
+ "epoch": 1.63,
952
+ "learning_rate": 4.1849529780564265e-05,
953
+ "loss": 0.0238,
954
+ "step": 1560
955
+ },
956
+ {
957
+ "epoch": 1.64,
958
+ "learning_rate": 4.179728317659352e-05,
959
+ "loss": 0.0261,
960
+ "step": 1570
961
+ },
962
+ {
963
+ "epoch": 1.65,
964
+ "learning_rate": 4.174503657262278e-05,
965
+ "loss": 0.0206,
966
+ "step": 1580
967
+ },
968
+ {
969
+ "epoch": 1.66,
970
+ "learning_rate": 4.1692789968652043e-05,
971
+ "loss": 0.0239,
972
+ "step": 1590
973
+ },
974
+ {
975
+ "epoch": 1.67,
976
+ "learning_rate": 4.16405433646813e-05,
977
+ "loss": 0.0256,
978
+ "step": 1600
979
+ },
980
+ {
981
+ "epoch": 1.68,
982
+ "learning_rate": 4.158829676071056e-05,
983
+ "loss": 0.0216,
984
+ "step": 1610
985
+ },
986
+ {
987
+ "epoch": 1.69,
988
+ "learning_rate": 4.1536050156739815e-05,
989
+ "loss": 0.0188,
990
+ "step": 1620
991
+ },
992
+ {
993
+ "epoch": 1.7,
994
+ "learning_rate": 4.148380355276907e-05,
995
+ "loss": 0.018,
996
+ "step": 1630
997
+ },
998
+ {
999
+ "epoch": 1.71,
1000
+ "learning_rate": 4.143155694879833e-05,
1001
+ "loss": 0.0149,
1002
+ "step": 1640
1003
+ },
1004
+ {
1005
+ "epoch": 1.72,
1006
+ "learning_rate": 4.1379310344827587e-05,
1007
+ "loss": 0.0158,
1008
+ "step": 1650
1009
+ },
1010
+ {
1011
+ "epoch": 1.73,
1012
+ "learning_rate": 4.132706374085685e-05,
1013
+ "loss": 0.0216,
1014
+ "step": 1660
1015
+ },
1016
+ {
1017
+ "epoch": 1.75,
1018
+ "learning_rate": 4.127481713688611e-05,
1019
+ "loss": 0.0249,
1020
+ "step": 1670
1021
+ },
1022
+ {
1023
+ "epoch": 1.76,
1024
+ "learning_rate": 4.122257053291536e-05,
1025
+ "loss": 0.02,
1026
+ "step": 1680
1027
+ },
1028
+ {
1029
+ "epoch": 1.77,
1030
+ "learning_rate": 4.117032392894462e-05,
1031
+ "loss": 0.0188,
1032
+ "step": 1690
1033
+ },
1034
+ {
1035
+ "epoch": 1.78,
1036
+ "learning_rate": 4.111807732497388e-05,
1037
+ "loss": 0.0194,
1038
+ "step": 1700
1039
+ },
1040
+ {
1041
+ "epoch": 1.79,
1042
+ "learning_rate": 4.1065830721003136e-05,
1043
+ "loss": 0.0258,
1044
+ "step": 1710
1045
+ },
1046
+ {
1047
+ "epoch": 1.8,
1048
+ "learning_rate": 4.1013584117032394e-05,
1049
+ "loss": 0.0187,
1050
+ "step": 1720
1051
+ },
1052
+ {
1053
+ "epoch": 1.81,
1054
+ "learning_rate": 4.096133751306165e-05,
1055
+ "loss": 0.0195,
1056
+ "step": 1730
1057
+ },
1058
+ {
1059
+ "epoch": 1.82,
1060
+ "learning_rate": 4.0909090909090915e-05,
1061
+ "loss": 0.0218,
1062
+ "step": 1740
1063
+ },
1064
+ {
1065
+ "epoch": 1.83,
1066
+ "learning_rate": 4.0856844305120165e-05,
1067
+ "loss": 0.0182,
1068
+ "step": 1750
1069
+ },
1070
+ {
1071
+ "epoch": 1.84,
1072
+ "learning_rate": 4.080459770114943e-05,
1073
+ "loss": 0.0227,
1074
+ "step": 1760
1075
+ },
1076
+ {
1077
+ "epoch": 1.85,
1078
+ "learning_rate": 4.0752351097178686e-05,
1079
+ "loss": 0.0267,
1080
+ "step": 1770
1081
+ },
1082
+ {
1083
+ "epoch": 1.86,
1084
+ "learning_rate": 4.0700104493207943e-05,
1085
+ "loss": 0.0187,
1086
+ "step": 1780
1087
+ },
1088
+ {
1089
+ "epoch": 1.87,
1090
+ "learning_rate": 4.06478578892372e-05,
1091
+ "loss": 0.0251,
1092
+ "step": 1790
1093
+ },
1094
+ {
1095
+ "epoch": 1.88,
1096
+ "learning_rate": 4.059561128526646e-05,
1097
+ "loss": 0.0153,
1098
+ "step": 1800
1099
+ },
1100
+ {
1101
+ "epoch": 1.89,
1102
+ "learning_rate": 4.054336468129572e-05,
1103
+ "loss": 0.0301,
1104
+ "step": 1810
1105
+ },
1106
+ {
1107
+ "epoch": 1.9,
1108
+ "learning_rate": 4.049111807732497e-05,
1109
+ "loss": 0.0157,
1110
+ "step": 1820
1111
+ },
1112
+ {
1113
+ "epoch": 1.91,
1114
+ "learning_rate": 4.0438871473354236e-05,
1115
+ "loss": 0.0204,
1116
+ "step": 1830
1117
+ },
1118
+ {
1119
+ "epoch": 1.92,
1120
+ "learning_rate": 4.038662486938349e-05,
1121
+ "loss": 0.0203,
1122
+ "step": 1840
1123
+ },
1124
+ {
1125
+ "epoch": 1.93,
1126
+ "learning_rate": 4.033437826541275e-05,
1127
+ "loss": 0.0138,
1128
+ "step": 1850
1129
+ },
1130
+ {
1131
+ "epoch": 1.94,
1132
+ "learning_rate": 4.028213166144201e-05,
1133
+ "loss": 0.0231,
1134
+ "step": 1860
1135
+ },
1136
+ {
1137
+ "epoch": 1.95,
1138
+ "learning_rate": 4.0229885057471265e-05,
1139
+ "loss": 0.0288,
1140
+ "step": 1870
1141
+ },
1142
+ {
1143
+ "epoch": 1.96,
1144
+ "learning_rate": 4.017763845350053e-05,
1145
+ "loss": 0.0189,
1146
+ "step": 1880
1147
+ },
1148
+ {
1149
+ "epoch": 1.97,
1150
+ "learning_rate": 4.012539184952978e-05,
1151
+ "loss": 0.0229,
1152
+ "step": 1890
1153
+ },
1154
+ {
1155
+ "epoch": 1.99,
1156
+ "learning_rate": 4.007314524555904e-05,
1157
+ "loss": 0.0171,
1158
+ "step": 1900
1159
+ },
1160
+ {
1161
+ "epoch": 2.0,
1162
+ "learning_rate": 4.00208986415883e-05,
1163
+ "loss": 0.02,
1164
+ "step": 1910
1165
+ },
1166
+ {
1167
+ "epoch": 2.01,
1168
+ "learning_rate": 3.996865203761756e-05,
1169
+ "loss": 0.014,
1170
+ "step": 1920
1171
+ },
1172
+ {
1173
+ "epoch": 2.02,
1174
+ "learning_rate": 3.9916405433646815e-05,
1175
+ "loss": 0.0164,
1176
+ "step": 1930
1177
+ },
1178
+ {
1179
+ "epoch": 2.03,
1180
+ "learning_rate": 3.986415882967607e-05,
1181
+ "loss": 0.0148,
1182
+ "step": 1940
1183
+ },
1184
+ {
1185
+ "epoch": 2.04,
1186
+ "learning_rate": 3.9811912225705336e-05,
1187
+ "loss": 0.0135,
1188
+ "step": 1950
1189
+ },
1190
+ {
1191
+ "epoch": 2.05,
1192
+ "learning_rate": 3.9759665621734586e-05,
1193
+ "loss": 0.015,
1194
+ "step": 1960
1195
+ },
1196
+ {
1197
+ "epoch": 2.06,
1198
+ "learning_rate": 3.970741901776385e-05,
1199
+ "loss": 0.0183,
1200
+ "step": 1970
1201
+ },
1202
+ {
1203
+ "epoch": 2.07,
1204
+ "learning_rate": 3.965517241379311e-05,
1205
+ "loss": 0.0105,
1206
+ "step": 1980
1207
+ },
1208
+ {
1209
+ "epoch": 2.08,
1210
+ "learning_rate": 3.960292580982236e-05,
1211
+ "loss": 0.0152,
1212
+ "step": 1990
1213
+ },
1214
+ {
1215
+ "epoch": 2.09,
1216
+ "learning_rate": 3.955067920585162e-05,
1217
+ "loss": 0.0063,
1218
+ "step": 2000
1219
+ },
1220
+ {
1221
+ "epoch": 2.09,
1222
+ "eval_acc": 0.660049626153723,
1223
+ "eval_cer": 0.04602510460251046,
1224
+ "eval_loss": 0.023699596524238586,
1225
+ "eval_runtime": 127.5932,
1226
+ "eval_samples_per_second": 3.158,
1227
+ "eval_steps_per_second": 0.4,
1228
+ "step": 2000
1229
+ },
1230
+ {
1231
+ "epoch": 2.1,
1232
+ "learning_rate": 3.949843260188088e-05,
1233
+ "loss": 0.0147,
1234
+ "step": 2010
1235
+ },
1236
+ {
1237
+ "epoch": 2.11,
1238
+ "learning_rate": 3.944618599791014e-05,
1239
+ "loss": 0.0126,
1240
+ "step": 2020
1241
+ },
1242
+ {
1243
+ "epoch": 2.12,
1244
+ "learning_rate": 3.939393939393939e-05,
1245
+ "loss": 0.0143,
1246
+ "step": 2030
1247
+ },
1248
+ {
1249
+ "epoch": 2.13,
1250
+ "learning_rate": 3.934169278996865e-05,
1251
+ "loss": 0.0147,
1252
+ "step": 2040
1253
+ },
1254
+ {
1255
+ "epoch": 2.14,
1256
+ "learning_rate": 3.9289446185997915e-05,
1257
+ "loss": 0.0134,
1258
+ "step": 2050
1259
+ },
1260
+ {
1261
+ "epoch": 2.15,
1262
+ "learning_rate": 3.9237199582027165e-05,
1263
+ "loss": 0.013,
1264
+ "step": 2060
1265
+ },
1266
+ {
1267
+ "epoch": 2.16,
1268
+ "learning_rate": 3.918495297805643e-05,
1269
+ "loss": 0.0217,
1270
+ "step": 2070
1271
+ },
1272
+ {
1273
+ "epoch": 2.17,
1274
+ "learning_rate": 3.9132706374085686e-05,
1275
+ "loss": 0.0174,
1276
+ "step": 2080
1277
+ },
1278
+ {
1279
+ "epoch": 2.18,
1280
+ "learning_rate": 3.908045977011495e-05,
1281
+ "loss": 0.0116,
1282
+ "step": 2090
1283
+ },
1284
+ {
1285
+ "epoch": 2.19,
1286
+ "learning_rate": 3.90282131661442e-05,
1287
+ "loss": 0.0169,
1288
+ "step": 2100
1289
+ },
1290
+ {
1291
+ "epoch": 2.2,
1292
+ "learning_rate": 3.897596656217346e-05,
1293
+ "loss": 0.0199,
1294
+ "step": 2110
1295
+ },
1296
+ {
1297
+ "epoch": 2.22,
1298
+ "learning_rate": 3.892371995820272e-05,
1299
+ "loss": 0.0124,
1300
+ "step": 2120
1301
+ },
1302
+ {
1303
+ "epoch": 2.23,
1304
+ "learning_rate": 3.887147335423197e-05,
1305
+ "loss": 0.0114,
1306
+ "step": 2130
1307
+ },
1308
+ {
1309
+ "epoch": 2.24,
1310
+ "learning_rate": 3.8819226750261236e-05,
1311
+ "loss": 0.0128,
1312
+ "step": 2140
1313
+ },
1314
+ {
1315
+ "epoch": 2.25,
1316
+ "learning_rate": 3.876698014629049e-05,
1317
+ "loss": 0.0097,
1318
+ "step": 2150
1319
+ },
1320
+ {
1321
+ "epoch": 2.26,
1322
+ "learning_rate": 3.871473354231975e-05,
1323
+ "loss": 0.0132,
1324
+ "step": 2160
1325
+ },
1326
+ {
1327
+ "epoch": 2.27,
1328
+ "learning_rate": 3.866248693834901e-05,
1329
+ "loss": 0.0133,
1330
+ "step": 2170
1331
+ },
1332
+ {
1333
+ "epoch": 2.28,
1334
+ "learning_rate": 3.8610240334378265e-05,
1335
+ "loss": 0.0094,
1336
+ "step": 2180
1337
+ },
1338
+ {
1339
+ "epoch": 2.29,
1340
+ "learning_rate": 3.855799373040753e-05,
1341
+ "loss": 0.0185,
1342
+ "step": 2190
1343
+ },
1344
+ {
1345
+ "epoch": 2.3,
1346
+ "learning_rate": 3.850574712643678e-05,
1347
+ "loss": 0.017,
1348
+ "step": 2200
1349
+ },
1350
+ {
1351
+ "epoch": 2.31,
1352
+ "learning_rate": 3.845350052246604e-05,
1353
+ "loss": 0.0128,
1354
+ "step": 2210
1355
+ },
1356
+ {
1357
+ "epoch": 2.32,
1358
+ "learning_rate": 3.84012539184953e-05,
1359
+ "loss": 0.0121,
1360
+ "step": 2220
1361
+ },
1362
+ {
1363
+ "epoch": 2.33,
1364
+ "learning_rate": 3.834900731452456e-05,
1365
+ "loss": 0.015,
1366
+ "step": 2230
1367
+ },
1368
+ {
1369
+ "epoch": 2.34,
1370
+ "learning_rate": 3.8296760710553815e-05,
1371
+ "loss": 0.0077,
1372
+ "step": 2240
1373
+ },
1374
+ {
1375
+ "epoch": 2.35,
1376
+ "learning_rate": 3.824451410658307e-05,
1377
+ "loss": 0.0118,
1378
+ "step": 2250
1379
+ },
1380
+ {
1381
+ "epoch": 2.36,
1382
+ "learning_rate": 3.8192267502612336e-05,
1383
+ "loss": 0.0133,
1384
+ "step": 2260
1385
+ },
1386
+ {
1387
+ "epoch": 2.37,
1388
+ "learning_rate": 3.8140020898641586e-05,
1389
+ "loss": 0.0156,
1390
+ "step": 2270
1391
+ },
1392
+ {
1393
+ "epoch": 2.38,
1394
+ "learning_rate": 3.808777429467085e-05,
1395
+ "loss": 0.0131,
1396
+ "step": 2280
1397
+ },
1398
+ {
1399
+ "epoch": 2.39,
1400
+ "learning_rate": 3.803552769070011e-05,
1401
+ "loss": 0.01,
1402
+ "step": 2290
1403
+ },
1404
+ {
1405
+ "epoch": 2.4,
1406
+ "learning_rate": 3.7983281086729364e-05,
1407
+ "loss": 0.0166,
1408
+ "step": 2300
1409
+ },
1410
+ {
1411
+ "epoch": 2.41,
1412
+ "learning_rate": 3.793103448275862e-05,
1413
+ "loss": 0.0125,
1414
+ "step": 2310
1415
+ },
1416
+ {
1417
+ "epoch": 2.42,
1418
+ "learning_rate": 3.787878787878788e-05,
1419
+ "loss": 0.0092,
1420
+ "step": 2320
1421
+ },
1422
+ {
1423
+ "epoch": 2.43,
1424
+ "learning_rate": 3.782654127481714e-05,
1425
+ "loss": 0.0105,
1426
+ "step": 2330
1427
+ },
1428
+ {
1429
+ "epoch": 2.45,
1430
+ "learning_rate": 3.777429467084639e-05,
1431
+ "loss": 0.0138,
1432
+ "step": 2340
1433
+ },
1434
+ {
1435
+ "epoch": 2.46,
1436
+ "learning_rate": 3.772204806687566e-05,
1437
+ "loss": 0.0136,
1438
+ "step": 2350
1439
+ },
1440
+ {
1441
+ "epoch": 2.47,
1442
+ "learning_rate": 3.7669801462904914e-05,
1443
+ "loss": 0.0153,
1444
+ "step": 2360
1445
+ },
1446
+ {
1447
+ "epoch": 2.48,
1448
+ "learning_rate": 3.761755485893417e-05,
1449
+ "loss": 0.0094,
1450
+ "step": 2370
1451
+ },
1452
+ {
1453
+ "epoch": 2.49,
1454
+ "learning_rate": 3.756530825496343e-05,
1455
+ "loss": 0.0081,
1456
+ "step": 2380
1457
+ },
1458
+ {
1459
+ "epoch": 2.5,
1460
+ "learning_rate": 3.7513061650992686e-05,
1461
+ "loss": 0.0169,
1462
+ "step": 2390
1463
+ },
1464
+ {
1465
+ "epoch": 2.51,
1466
+ "learning_rate": 3.746081504702195e-05,
1467
+ "loss": 0.0086,
1468
+ "step": 2400
1469
+ },
1470
+ {
1471
+ "epoch": 2.52,
1472
+ "learning_rate": 3.740856844305121e-05,
1473
+ "loss": 0.0124,
1474
+ "step": 2410
1475
+ },
1476
+ {
1477
+ "epoch": 2.53,
1478
+ "learning_rate": 3.735632183908046e-05,
1479
+ "loss": 0.0143,
1480
+ "step": 2420
1481
+ },
1482
+ {
1483
+ "epoch": 2.54,
1484
+ "learning_rate": 3.730407523510972e-05,
1485
+ "loss": 0.0092,
1486
+ "step": 2430
1487
+ },
1488
+ {
1489
+ "epoch": 2.55,
1490
+ "learning_rate": 3.725182863113898e-05,
1491
+ "loss": 0.0147,
1492
+ "step": 2440
1493
+ },
1494
+ {
1495
+ "epoch": 2.56,
1496
+ "learning_rate": 3.7199582027168236e-05,
1497
+ "loss": 0.009,
1498
+ "step": 2450
1499
+ },
1500
+ {
1501
+ "epoch": 2.57,
1502
+ "learning_rate": 3.714733542319749e-05,
1503
+ "loss": 0.0083,
1504
+ "step": 2460
1505
+ },
1506
+ {
1507
+ "epoch": 2.58,
1508
+ "learning_rate": 3.709508881922675e-05,
1509
+ "loss": 0.0146,
1510
+ "step": 2470
1511
+ },
1512
+ {
1513
+ "epoch": 2.59,
1514
+ "learning_rate": 3.7042842215256014e-05,
1515
+ "loss": 0.0142,
1516
+ "step": 2480
1517
+ },
1518
+ {
1519
+ "epoch": 2.6,
1520
+ "learning_rate": 3.6990595611285264e-05,
1521
+ "loss": 0.0097,
1522
+ "step": 2490
1523
+ },
1524
+ {
1525
+ "epoch": 2.61,
1526
+ "learning_rate": 3.693834900731453e-05,
1527
+ "loss": 0.0067,
1528
+ "step": 2500
1529
+ },
1530
+ {
1531
+ "epoch": 2.62,
1532
+ "learning_rate": 3.6886102403343786e-05,
1533
+ "loss": 0.0085,
1534
+ "step": 2510
1535
+ },
1536
+ {
1537
+ "epoch": 2.63,
1538
+ "learning_rate": 3.683385579937304e-05,
1539
+ "loss": 0.01,
1540
+ "step": 2520
1541
+ },
1542
+ {
1543
+ "epoch": 2.64,
1544
+ "learning_rate": 3.67816091954023e-05,
1545
+ "loss": 0.0128,
1546
+ "step": 2530
1547
+ },
1548
+ {
1549
+ "epoch": 2.65,
1550
+ "learning_rate": 3.672936259143156e-05,
1551
+ "loss": 0.0113,
1552
+ "step": 2540
1553
+ },
1554
+ {
1555
+ "epoch": 2.66,
1556
+ "learning_rate": 3.667711598746082e-05,
1557
+ "loss": 0.0098,
1558
+ "step": 2550
1559
+ },
1560
+ {
1561
+ "epoch": 2.68,
1562
+ "learning_rate": 3.662486938349007e-05,
1563
+ "loss": 0.0159,
1564
+ "step": 2560
1565
+ },
1566
+ {
1567
+ "epoch": 2.69,
1568
+ "learning_rate": 3.6572622779519335e-05,
1569
+ "loss": 0.0188,
1570
+ "step": 2570
1571
+ },
1572
+ {
1573
+ "epoch": 2.7,
1574
+ "learning_rate": 3.652037617554859e-05,
1575
+ "loss": 0.0149,
1576
+ "step": 2580
1577
+ },
1578
+ {
1579
+ "epoch": 2.71,
1580
+ "learning_rate": 3.646812957157785e-05,
1581
+ "loss": 0.0159,
1582
+ "step": 2590
1583
+ },
1584
+ {
1585
+ "epoch": 2.72,
1586
+ "learning_rate": 3.641588296760711e-05,
1587
+ "loss": 0.0093,
1588
+ "step": 2600
1589
+ },
1590
+ {
1591
+ "epoch": 2.73,
1592
+ "learning_rate": 3.6363636363636364e-05,
1593
+ "loss": 0.0115,
1594
+ "step": 2610
1595
+ },
1596
+ {
1597
+ "epoch": 2.74,
1598
+ "learning_rate": 3.631138975966563e-05,
1599
+ "loss": 0.0138,
1600
+ "step": 2620
1601
+ },
1602
+ {
1603
+ "epoch": 2.75,
1604
+ "learning_rate": 3.625914315569488e-05,
1605
+ "loss": 0.011,
1606
+ "step": 2630
1607
+ },
1608
+ {
1609
+ "epoch": 2.76,
1610
+ "learning_rate": 3.620689655172414e-05,
1611
+ "loss": 0.0144,
1612
+ "step": 2640
1613
+ },
1614
+ {
1615
+ "epoch": 2.77,
1616
+ "learning_rate": 3.61546499477534e-05,
1617
+ "loss": 0.0098,
1618
+ "step": 2650
1619
+ },
1620
+ {
1621
+ "epoch": 2.78,
1622
+ "learning_rate": 3.610240334378266e-05,
1623
+ "loss": 0.0132,
1624
+ "step": 2660
1625
+ },
1626
+ {
1627
+ "epoch": 2.79,
1628
+ "learning_rate": 3.6050156739811914e-05,
1629
+ "loss": 0.0125,
1630
+ "step": 2670
1631
+ },
1632
+ {
1633
+ "epoch": 2.8,
1634
+ "learning_rate": 3.599791013584117e-05,
1635
+ "loss": 0.0118,
1636
+ "step": 2680
1637
+ },
1638
+ {
1639
+ "epoch": 2.81,
1640
+ "learning_rate": 3.5945663531870435e-05,
1641
+ "loss": 0.0131,
1642
+ "step": 2690
1643
+ },
1644
+ {
1645
+ "epoch": 2.82,
1646
+ "learning_rate": 3.5893416927899686e-05,
1647
+ "loss": 0.0081,
1648
+ "step": 2700
1649
+ },
1650
+ {
1651
+ "epoch": 2.83,
1652
+ "learning_rate": 3.584117032392895e-05,
1653
+ "loss": 0.0115,
1654
+ "step": 2710
1655
+ },
1656
+ {
1657
+ "epoch": 2.84,
1658
+ "learning_rate": 3.578892371995821e-05,
1659
+ "loss": 0.0109,
1660
+ "step": 2720
1661
+ },
1662
+ {
1663
+ "epoch": 2.85,
1664
+ "learning_rate": 3.573667711598746e-05,
1665
+ "loss": 0.0149,
1666
+ "step": 2730
1667
+ },
1668
+ {
1669
+ "epoch": 2.86,
1670
+ "learning_rate": 3.568443051201672e-05,
1671
+ "loss": 0.0093,
1672
+ "step": 2740
1673
+ },
1674
+ {
1675
+ "epoch": 2.87,
1676
+ "learning_rate": 3.563218390804598e-05,
1677
+ "loss": 0.0124,
1678
+ "step": 2750
1679
+ },
1680
+ {
1681
+ "epoch": 2.88,
1682
+ "learning_rate": 3.557993730407524e-05,
1683
+ "loss": 0.0094,
1684
+ "step": 2760
1685
+ },
1686
+ {
1687
+ "epoch": 2.89,
1688
+ "learning_rate": 3.552769070010449e-05,
1689
+ "loss": 0.0105,
1690
+ "step": 2770
1691
+ },
1692
+ {
1693
+ "epoch": 2.9,
1694
+ "learning_rate": 3.547544409613375e-05,
1695
+ "loss": 0.0221,
1696
+ "step": 2780
1697
+ },
1698
+ {
1699
+ "epoch": 2.92,
1700
+ "learning_rate": 3.5423197492163014e-05,
1701
+ "loss": 0.0088,
1702
+ "step": 2790
1703
+ },
1704
+ {
1705
+ "epoch": 2.93,
1706
+ "learning_rate": 3.5370950888192264e-05,
1707
+ "loss": 0.0077,
1708
+ "step": 2800
1709
+ },
1710
+ {
1711
+ "epoch": 2.94,
1712
+ "learning_rate": 3.531870428422153e-05,
1713
+ "loss": 0.0106,
1714
+ "step": 2810
1715
+ },
1716
+ {
1717
+ "epoch": 2.95,
1718
+ "learning_rate": 3.5266457680250785e-05,
1719
+ "loss": 0.0113,
1720
+ "step": 2820
1721
+ },
1722
+ {
1723
+ "epoch": 2.96,
1724
+ "learning_rate": 3.521421107628005e-05,
1725
+ "loss": 0.012,
1726
+ "step": 2830
1727
+ },
1728
+ {
1729
+ "epoch": 2.97,
1730
+ "learning_rate": 3.51619644723093e-05,
1731
+ "loss": 0.0094,
1732
+ "step": 2840
1733
+ },
1734
+ {
1735
+ "epoch": 2.98,
1736
+ "learning_rate": 3.510971786833856e-05,
1737
+ "loss": 0.0088,
1738
+ "step": 2850
1739
+ },
1740
+ {
1741
+ "epoch": 2.99,
1742
+ "learning_rate": 3.505747126436782e-05,
1743
+ "loss": 0.0162,
1744
+ "step": 2860
1745
+ },
1746
+ {
1747
+ "epoch": 3.0,
1748
+ "learning_rate": 3.500522466039707e-05,
1749
+ "loss": 0.0104,
1750
+ "step": 2870
1751
+ },
1752
+ {
1753
+ "epoch": 3.01,
1754
+ "learning_rate": 3.4952978056426335e-05,
1755
+ "loss": 0.0045,
1756
+ "step": 2880
1757
+ },
1758
+ {
1759
+ "epoch": 3.02,
1760
+ "learning_rate": 3.490073145245559e-05,
1761
+ "loss": 0.0076,
1762
+ "step": 2890
1763
+ },
1764
+ {
1765
+ "epoch": 3.03,
1766
+ "learning_rate": 3.484848484848485e-05,
1767
+ "loss": 0.0092,
1768
+ "step": 2900
1769
+ },
1770
+ {
1771
+ "epoch": 3.04,
1772
+ "learning_rate": 3.479623824451411e-05,
1773
+ "loss": 0.0088,
1774
+ "step": 2910
1775
+ },
1776
+ {
1777
+ "epoch": 3.05,
1778
+ "learning_rate": 3.4743991640543364e-05,
1779
+ "loss": 0.0068,
1780
+ "step": 2920
1781
+ },
1782
+ {
1783
+ "epoch": 3.06,
1784
+ "learning_rate": 3.469174503657263e-05,
1785
+ "loss": 0.0037,
1786
+ "step": 2930
1787
+ },
1788
+ {
1789
+ "epoch": 3.07,
1790
+ "learning_rate": 3.463949843260188e-05,
1791
+ "loss": 0.008,
1792
+ "step": 2940
1793
+ },
1794
+ {
1795
+ "epoch": 3.08,
1796
+ "learning_rate": 3.458725182863114e-05,
1797
+ "loss": 0.0053,
1798
+ "step": 2950
1799
+ },
1800
+ {
1801
+ "epoch": 3.09,
1802
+ "learning_rate": 3.45350052246604e-05,
1803
+ "loss": 0.0085,
1804
+ "step": 2960
1805
+ },
1806
+ {
1807
+ "epoch": 3.1,
1808
+ "learning_rate": 3.4482758620689657e-05,
1809
+ "loss": 0.0122,
1810
+ "step": 2970
1811
+ },
1812
+ {
1813
+ "epoch": 3.11,
1814
+ "learning_rate": 3.4430512016718914e-05,
1815
+ "loss": 0.0096,
1816
+ "step": 2980
1817
+ },
1818
+ {
1819
+ "epoch": 3.12,
1820
+ "learning_rate": 3.437826541274817e-05,
1821
+ "loss": 0.0079,
1822
+ "step": 2990
1823
+ },
1824
+ {
1825
+ "epoch": 3.13,
1826
+ "learning_rate": 3.4326018808777435e-05,
1827
+ "loss": 0.01,
1828
+ "step": 3000
1829
+ },
1830
+ {
1831
+ "epoch": 3.13,
1832
+ "eval_acc": 0.7444168716019433,
1833
+ "eval_cer": 0.031845653184565316,
1834
+ "eval_loss": 0.017891723662614822,
1835
+ "eval_runtime": 127.5496,
1836
+ "eval_samples_per_second": 3.16,
1837
+ "eval_steps_per_second": 0.4,
1838
+ "step": 3000
1839
+ },
1840
+ {
1841
+ "epoch": 3.15,
1842
+ "learning_rate": 3.4273772204806685e-05,
1843
+ "loss": 0.0105,
1844
+ "step": 3010
1845
+ },
1846
+ {
1847
+ "epoch": 3.16,
1848
+ "learning_rate": 3.422152560083595e-05,
1849
+ "loss": 0.0091,
1850
+ "step": 3020
1851
+ },
1852
+ {
1853
+ "epoch": 3.17,
1854
+ "learning_rate": 3.4169278996865206e-05,
1855
+ "loss": 0.005,
1856
+ "step": 3030
1857
+ },
1858
+ {
1859
+ "epoch": 3.18,
1860
+ "learning_rate": 3.4117032392894464e-05,
1861
+ "loss": 0.0069,
1862
+ "step": 3040
1863
+ },
1864
+ {
1865
+ "epoch": 3.19,
1866
+ "learning_rate": 3.406478578892372e-05,
1867
+ "loss": 0.0066,
1868
+ "step": 3050
1869
+ },
1870
+ {
1871
+ "epoch": 3.2,
1872
+ "learning_rate": 3.401253918495298e-05,
1873
+ "loss": 0.0065,
1874
+ "step": 3060
1875
+ },
1876
+ {
1877
+ "epoch": 3.21,
1878
+ "learning_rate": 3.396029258098224e-05,
1879
+ "loss": 0.0098,
1880
+ "step": 3070
1881
+ },
1882
+ {
1883
+ "epoch": 3.22,
1884
+ "learning_rate": 3.390804597701149e-05,
1885
+ "loss": 0.0089,
1886
+ "step": 3080
1887
+ },
1888
+ {
1889
+ "epoch": 3.23,
1890
+ "learning_rate": 3.3855799373040756e-05,
1891
+ "loss": 0.0082,
1892
+ "step": 3090
1893
+ },
1894
+ {
1895
+ "epoch": 3.24,
1896
+ "learning_rate": 3.3803552769070014e-05,
1897
+ "loss": 0.0089,
1898
+ "step": 3100
1899
+ },
1900
+ {
1901
+ "epoch": 3.25,
1902
+ "learning_rate": 3.375130616509927e-05,
1903
+ "loss": 0.0097,
1904
+ "step": 3110
1905
+ },
1906
+ {
1907
+ "epoch": 3.26,
1908
+ "learning_rate": 3.369905956112853e-05,
1909
+ "loss": 0.0079,
1910
+ "step": 3120
1911
+ },
1912
+ {
1913
+ "epoch": 3.27,
1914
+ "learning_rate": 3.3646812957157785e-05,
1915
+ "loss": 0.0079,
1916
+ "step": 3130
1917
+ },
1918
+ {
1919
+ "epoch": 3.28,
1920
+ "learning_rate": 3.359456635318705e-05,
1921
+ "loss": 0.0062,
1922
+ "step": 3140
1923
+ },
1924
+ {
1925
+ "epoch": 3.29,
1926
+ "learning_rate": 3.35423197492163e-05,
1927
+ "loss": 0.0095,
1928
+ "step": 3150
1929
+ },
1930
+ {
1931
+ "epoch": 3.3,
1932
+ "learning_rate": 3.3490073145245557e-05,
1933
+ "loss": 0.0089,
1934
+ "step": 3160
1935
+ },
1936
+ {
1937
+ "epoch": 3.31,
1938
+ "learning_rate": 3.343782654127482e-05,
1939
+ "loss": 0.0107,
1940
+ "step": 3170
1941
+ },
1942
+ {
1943
+ "epoch": 3.32,
1944
+ "learning_rate": 3.338557993730408e-05,
1945
+ "loss": 0.0135,
1946
+ "step": 3180
1947
+ },
1948
+ {
1949
+ "epoch": 3.33,
1950
+ "learning_rate": 3.3333333333333335e-05,
1951
+ "loss": 0.006,
1952
+ "step": 3190
1953
+ },
1954
+ {
1955
+ "epoch": 3.34,
1956
+ "learning_rate": 3.328108672936259e-05,
1957
+ "loss": 0.0078,
1958
+ "step": 3200
1959
+ },
1960
+ {
1961
+ "epoch": 3.35,
1962
+ "learning_rate": 3.322884012539185e-05,
1963
+ "loss": 0.005,
1964
+ "step": 3210
1965
+ },
1966
+ {
1967
+ "epoch": 3.36,
1968
+ "learning_rate": 3.3176593521421106e-05,
1969
+ "loss": 0.0073,
1970
+ "step": 3220
1971
+ },
1972
+ {
1973
+ "epoch": 3.38,
1974
+ "learning_rate": 3.3124346917450364e-05,
1975
+ "loss": 0.0084,
1976
+ "step": 3230
1977
+ },
1978
+ {
1979
+ "epoch": 3.39,
1980
+ "learning_rate": 3.307210031347963e-05,
1981
+ "loss": 0.0077,
1982
+ "step": 3240
1983
+ },
1984
+ {
1985
+ "epoch": 3.4,
1986
+ "learning_rate": 3.3019853709508885e-05,
1987
+ "loss": 0.0065,
1988
+ "step": 3250
1989
+ },
1990
+ {
1991
+ "epoch": 3.41,
1992
+ "learning_rate": 3.296760710553814e-05,
1993
+ "loss": 0.0052,
1994
+ "step": 3260
1995
+ },
1996
+ {
1997
+ "epoch": 3.42,
1998
+ "learning_rate": 3.29153605015674e-05,
1999
+ "loss": 0.0078,
2000
+ "step": 3270
2001
+ },
2002
+ {
2003
+ "epoch": 3.43,
2004
+ "learning_rate": 3.2863113897596656e-05,
2005
+ "loss": 0.0056,
2006
+ "step": 3280
2007
+ },
2008
+ {
2009
+ "epoch": 3.44,
2010
+ "learning_rate": 3.2810867293625914e-05,
2011
+ "loss": 0.0084,
2012
+ "step": 3290
2013
+ },
2014
+ {
2015
+ "epoch": 3.45,
2016
+ "learning_rate": 3.275862068965517e-05,
2017
+ "loss": 0.0077,
2018
+ "step": 3300
2019
+ },
2020
+ {
2021
+ "epoch": 3.46,
2022
+ "learning_rate": 3.2706374085684435e-05,
2023
+ "loss": 0.0071,
2024
+ "step": 3310
2025
+ },
2026
+ {
2027
+ "epoch": 3.47,
2028
+ "learning_rate": 3.265412748171369e-05,
2029
+ "loss": 0.0084,
2030
+ "step": 3320
2031
+ },
2032
+ {
2033
+ "epoch": 3.48,
2034
+ "learning_rate": 3.260188087774295e-05,
2035
+ "loss": 0.0079,
2036
+ "step": 3330
2037
+ },
2038
+ {
2039
+ "epoch": 3.49,
2040
+ "learning_rate": 3.2549634273772206e-05,
2041
+ "loss": 0.0058,
2042
+ "step": 3340
2043
+ },
2044
+ {
2045
+ "epoch": 3.5,
2046
+ "learning_rate": 3.2497387669801463e-05,
2047
+ "loss": 0.0074,
2048
+ "step": 3350
2049
+ },
2050
+ {
2051
+ "epoch": 3.51,
2052
+ "learning_rate": 3.244514106583073e-05,
2053
+ "loss": 0.007,
2054
+ "step": 3360
2055
+ },
2056
+ {
2057
+ "epoch": 3.52,
2058
+ "learning_rate": 3.239289446185998e-05,
2059
+ "loss": 0.0089,
2060
+ "step": 3370
2061
+ },
2062
+ {
2063
+ "epoch": 3.53,
2064
+ "learning_rate": 3.234064785788924e-05,
2065
+ "loss": 0.0059,
2066
+ "step": 3380
2067
+ },
2068
+ {
2069
+ "epoch": 3.54,
2070
+ "learning_rate": 3.22884012539185e-05,
2071
+ "loss": 0.0082,
2072
+ "step": 3390
2073
+ },
2074
+ {
2075
+ "epoch": 3.55,
2076
+ "learning_rate": 3.2236154649947756e-05,
2077
+ "loss": 0.0107,
2078
+ "step": 3400
2079
+ },
2080
+ {
2081
+ "epoch": 3.56,
2082
+ "learning_rate": 3.218390804597701e-05,
2083
+ "loss": 0.0063,
2084
+ "step": 3410
2085
+ },
2086
+ {
2087
+ "epoch": 3.57,
2088
+ "learning_rate": 3.213166144200627e-05,
2089
+ "loss": 0.0059,
2090
+ "step": 3420
2091
+ },
2092
+ {
2093
+ "epoch": 3.58,
2094
+ "learning_rate": 3.2079414838035534e-05,
2095
+ "loss": 0.0074,
2096
+ "step": 3430
2097
+ },
2098
+ {
2099
+ "epoch": 3.59,
2100
+ "learning_rate": 3.2027168234064785e-05,
2101
+ "loss": 0.011,
2102
+ "step": 3440
2103
+ },
2104
+ {
2105
+ "epoch": 3.61,
2106
+ "learning_rate": 3.197492163009405e-05,
2107
+ "loss": 0.0107,
2108
+ "step": 3450
2109
+ },
2110
+ {
2111
+ "epoch": 3.62,
2112
+ "learning_rate": 3.1922675026123306e-05,
2113
+ "loss": 0.0071,
2114
+ "step": 3460
2115
+ },
2116
+ {
2117
+ "epoch": 3.63,
2118
+ "learning_rate": 3.1870428422152556e-05,
2119
+ "loss": 0.0073,
2120
+ "step": 3470
2121
+ },
2122
+ {
2123
+ "epoch": 3.64,
2124
+ "learning_rate": 3.181818181818182e-05,
2125
+ "loss": 0.0079,
2126
+ "step": 3480
2127
+ },
2128
+ {
2129
+ "epoch": 3.65,
2130
+ "learning_rate": 3.176593521421108e-05,
2131
+ "loss": 0.0068,
2132
+ "step": 3490
2133
+ },
2134
+ {
2135
+ "epoch": 3.66,
2136
+ "learning_rate": 3.171368861024034e-05,
2137
+ "loss": 0.0085,
2138
+ "step": 3500
2139
+ },
2140
+ {
2141
+ "epoch": 3.67,
2142
+ "learning_rate": 3.166144200626959e-05,
2143
+ "loss": 0.0087,
2144
+ "step": 3510
2145
+ },
2146
+ {
2147
+ "epoch": 3.68,
2148
+ "learning_rate": 3.160919540229885e-05,
2149
+ "loss": 0.005,
2150
+ "step": 3520
2151
+ },
2152
+ {
2153
+ "epoch": 3.69,
2154
+ "learning_rate": 3.155694879832811e-05,
2155
+ "loss": 0.0051,
2156
+ "step": 3530
2157
+ },
2158
+ {
2159
+ "epoch": 3.7,
2160
+ "learning_rate": 3.1504702194357363e-05,
2161
+ "loss": 0.0076,
2162
+ "step": 3540
2163
+ },
2164
+ {
2165
+ "epoch": 3.71,
2166
+ "learning_rate": 3.145245559038663e-05,
2167
+ "loss": 0.0054,
2168
+ "step": 3550
2169
+ },
2170
+ {
2171
+ "epoch": 3.72,
2172
+ "learning_rate": 3.1400208986415885e-05,
2173
+ "loss": 0.0083,
2174
+ "step": 3560
2175
+ },
2176
+ {
2177
+ "epoch": 3.73,
2178
+ "learning_rate": 3.134796238244515e-05,
2179
+ "loss": 0.0076,
2180
+ "step": 3570
2181
+ },
2182
+ {
2183
+ "epoch": 3.74,
2184
+ "learning_rate": 3.12957157784744e-05,
2185
+ "loss": 0.0094,
2186
+ "step": 3580
2187
+ },
2188
+ {
2189
+ "epoch": 3.75,
2190
+ "learning_rate": 3.1243469174503656e-05,
2191
+ "loss": 0.0083,
2192
+ "step": 3590
2193
+ },
2194
+ {
2195
+ "epoch": 3.76,
2196
+ "learning_rate": 3.119122257053292e-05,
2197
+ "loss": 0.0073,
2198
+ "step": 3600
2199
+ },
2200
+ {
2201
+ "epoch": 3.77,
2202
+ "learning_rate": 3.113897596656217e-05,
2203
+ "loss": 0.0063,
2204
+ "step": 3610
2205
+ },
2206
+ {
2207
+ "epoch": 3.78,
2208
+ "learning_rate": 3.1086729362591434e-05,
2209
+ "loss": 0.0086,
2210
+ "step": 3620
2211
+ },
2212
+ {
2213
+ "epoch": 3.79,
2214
+ "learning_rate": 3.103448275862069e-05,
2215
+ "loss": 0.0118,
2216
+ "step": 3630
2217
+ },
2218
+ {
2219
+ "epoch": 3.8,
2220
+ "learning_rate": 3.098223615464995e-05,
2221
+ "loss": 0.0088,
2222
+ "step": 3640
2223
+ },
2224
+ {
2225
+ "epoch": 3.81,
2226
+ "learning_rate": 3.0929989550679206e-05,
2227
+ "loss": 0.0064,
2228
+ "step": 3650
2229
+ },
2230
+ {
2231
+ "epoch": 3.82,
2232
+ "learning_rate": 3.087774294670846e-05,
2233
+ "loss": 0.0065,
2234
+ "step": 3660
2235
+ },
2236
+ {
2237
+ "epoch": 3.83,
2238
+ "learning_rate": 3.082549634273773e-05,
2239
+ "loss": 0.0057,
2240
+ "step": 3670
2241
+ },
2242
+ {
2243
+ "epoch": 3.85,
2244
+ "learning_rate": 3.077324973876698e-05,
2245
+ "loss": 0.0129,
2246
+ "step": 3680
2247
+ },
2248
+ {
2249
+ "epoch": 3.86,
2250
+ "learning_rate": 3.072100313479624e-05,
2251
+ "loss": 0.0119,
2252
+ "step": 3690
2253
+ },
2254
+ {
2255
+ "epoch": 3.87,
2256
+ "learning_rate": 3.06687565308255e-05,
2257
+ "loss": 0.0074,
2258
+ "step": 3700
2259
+ },
2260
+ {
2261
+ "epoch": 3.88,
2262
+ "learning_rate": 3.0616509926854756e-05,
2263
+ "loss": 0.0058,
2264
+ "step": 3710
2265
+ },
2266
+ {
2267
+ "epoch": 3.89,
2268
+ "learning_rate": 3.056426332288401e-05,
2269
+ "loss": 0.0049,
2270
+ "step": 3720
2271
+ },
2272
+ {
2273
+ "epoch": 3.9,
2274
+ "learning_rate": 3.0512016718913274e-05,
2275
+ "loss": 0.0075,
2276
+ "step": 3730
2277
+ },
2278
+ {
2279
+ "epoch": 3.91,
2280
+ "learning_rate": 3.045977011494253e-05,
2281
+ "loss": 0.0087,
2282
+ "step": 3740
2283
+ },
2284
+ {
2285
+ "epoch": 3.92,
2286
+ "learning_rate": 3.0407523510971785e-05,
2287
+ "loss": 0.0071,
2288
+ "step": 3750
2289
+ },
2290
+ {
2291
+ "epoch": 3.93,
2292
+ "learning_rate": 3.0355276907001045e-05,
2293
+ "loss": 0.0101,
2294
+ "step": 3760
2295
+ },
2296
+ {
2297
+ "epoch": 3.94,
2298
+ "learning_rate": 3.0303030303030306e-05,
2299
+ "loss": 0.008,
2300
+ "step": 3770
2301
+ },
2302
+ {
2303
+ "epoch": 3.95,
2304
+ "learning_rate": 3.0250783699059566e-05,
2305
+ "loss": 0.0077,
2306
+ "step": 3780
2307
+ },
2308
+ {
2309
+ "epoch": 3.96,
2310
+ "learning_rate": 3.019853709508882e-05,
2311
+ "loss": 0.0114,
2312
+ "step": 3790
2313
+ },
2314
+ {
2315
+ "epoch": 3.97,
2316
+ "learning_rate": 3.0146290491118077e-05,
2317
+ "loss": 0.0067,
2318
+ "step": 3800
2319
+ },
2320
+ {
2321
+ "epoch": 3.98,
2322
+ "learning_rate": 3.0094043887147338e-05,
2323
+ "loss": 0.0067,
2324
+ "step": 3810
2325
+ },
2326
+ {
2327
+ "epoch": 3.99,
2328
+ "learning_rate": 3.004179728317659e-05,
2329
+ "loss": 0.0099,
2330
+ "step": 3820
2331
+ },
2332
+ {
2333
+ "epoch": 4.0,
2334
+ "learning_rate": 2.9989550679205852e-05,
2335
+ "loss": 0.0078,
2336
+ "step": 3830
2337
+ },
2338
+ {
2339
+ "epoch": 4.01,
2340
+ "learning_rate": 2.9937304075235113e-05,
2341
+ "loss": 0.0042,
2342
+ "step": 3840
2343
+ },
2344
+ {
2345
+ "epoch": 4.02,
2346
+ "learning_rate": 2.988505747126437e-05,
2347
+ "loss": 0.0056,
2348
+ "step": 3850
2349
+ },
2350
+ {
2351
+ "epoch": 4.03,
2352
+ "learning_rate": 2.9832810867293627e-05,
2353
+ "loss": 0.0043,
2354
+ "step": 3860
2355
+ },
2356
+ {
2357
+ "epoch": 4.04,
2358
+ "learning_rate": 2.9780564263322884e-05,
2359
+ "loss": 0.0042,
2360
+ "step": 3870
2361
+ },
2362
+ {
2363
+ "epoch": 4.05,
2364
+ "learning_rate": 2.9728317659352145e-05,
2365
+ "loss": 0.0062,
2366
+ "step": 3880
2367
+ },
2368
+ {
2369
+ "epoch": 4.06,
2370
+ "learning_rate": 2.96760710553814e-05,
2371
+ "loss": 0.0057,
2372
+ "step": 3890
2373
+ },
2374
+ {
2375
+ "epoch": 4.08,
2376
+ "learning_rate": 2.962382445141066e-05,
2377
+ "loss": 0.0069,
2378
+ "step": 3900
2379
+ },
2380
+ {
2381
+ "epoch": 4.09,
2382
+ "learning_rate": 2.957157784743992e-05,
2383
+ "loss": 0.0072,
2384
+ "step": 3910
2385
+ },
2386
+ {
2387
+ "epoch": 4.1,
2388
+ "learning_rate": 2.9519331243469177e-05,
2389
+ "loss": 0.0072,
2390
+ "step": 3920
2391
+ },
2392
+ {
2393
+ "epoch": 4.11,
2394
+ "learning_rate": 2.946708463949843e-05,
2395
+ "loss": 0.0032,
2396
+ "step": 3930
2397
+ },
2398
+ {
2399
+ "epoch": 4.12,
2400
+ "learning_rate": 2.941483803552769e-05,
2401
+ "loss": 0.0089,
2402
+ "step": 3940
2403
+ },
2404
+ {
2405
+ "epoch": 4.13,
2406
+ "learning_rate": 2.9362591431556952e-05,
2407
+ "loss": 0.0043,
2408
+ "step": 3950
2409
+ },
2410
+ {
2411
+ "epoch": 4.14,
2412
+ "learning_rate": 2.9310344827586206e-05,
2413
+ "loss": 0.0063,
2414
+ "step": 3960
2415
+ },
2416
+ {
2417
+ "epoch": 4.15,
2418
+ "learning_rate": 2.9258098223615466e-05,
2419
+ "loss": 0.0045,
2420
+ "step": 3970
2421
+ },
2422
+ {
2423
+ "epoch": 4.16,
2424
+ "learning_rate": 2.9205851619644723e-05,
2425
+ "loss": 0.0058,
2426
+ "step": 3980
2427
+ },
2428
+ {
2429
+ "epoch": 4.17,
2430
+ "learning_rate": 2.9153605015673984e-05,
2431
+ "loss": 0.005,
2432
+ "step": 3990
2433
+ },
2434
+ {
2435
+ "epoch": 4.18,
2436
+ "learning_rate": 2.9101358411703238e-05,
2437
+ "loss": 0.0044,
2438
+ "step": 4000
2439
+ },
2440
+ {
2441
+ "epoch": 4.18,
2442
+ "eval_acc": 0.769230767322008,
2443
+ "eval_cer": 0.0299860529986053,
2444
+ "eval_loss": 0.01652107760310173,
2445
+ "eval_runtime": 126.6833,
2446
+ "eval_samples_per_second": 3.181,
2447
+ "eval_steps_per_second": 0.403,
2448
+ "step": 4000
2449
+ },
2450
+ {
2451
+ "epoch": 4.19,
2452
+ "learning_rate": 2.90491118077325e-05,
2453
+ "loss": 0.0068,
2454
+ "step": 4010
2455
+ },
2456
+ {
2457
+ "epoch": 4.2,
2458
+ "learning_rate": 2.899686520376176e-05,
2459
+ "loss": 0.0039,
2460
+ "step": 4020
2461
+ },
2462
+ {
2463
+ "epoch": 4.21,
2464
+ "learning_rate": 2.8944618599791013e-05,
2465
+ "loss": 0.01,
2466
+ "step": 4030
2467
+ },
2468
+ {
2469
+ "epoch": 4.22,
2470
+ "learning_rate": 2.8892371995820273e-05,
2471
+ "loss": 0.0046,
2472
+ "step": 4040
2473
+ },
2474
+ {
2475
+ "epoch": 4.23,
2476
+ "learning_rate": 2.884012539184953e-05,
2477
+ "loss": 0.0096,
2478
+ "step": 4050
2479
+ },
2480
+ {
2481
+ "epoch": 4.24,
2482
+ "learning_rate": 2.878787878787879e-05,
2483
+ "loss": 0.006,
2484
+ "step": 4060
2485
+ },
2486
+ {
2487
+ "epoch": 4.25,
2488
+ "learning_rate": 2.8735632183908045e-05,
2489
+ "loss": 0.0052,
2490
+ "step": 4070
2491
+ },
2492
+ {
2493
+ "epoch": 4.26,
2494
+ "learning_rate": 2.8683385579937305e-05,
2495
+ "loss": 0.005,
2496
+ "step": 4080
2497
+ },
2498
+ {
2499
+ "epoch": 4.27,
2500
+ "learning_rate": 2.8631138975966566e-05,
2501
+ "loss": 0.004,
2502
+ "step": 4090
2503
+ },
2504
+ {
2505
+ "epoch": 4.28,
2506
+ "learning_rate": 2.857889237199582e-05,
2507
+ "loss": 0.0045,
2508
+ "step": 4100
2509
+ },
2510
+ {
2511
+ "epoch": 4.29,
2512
+ "learning_rate": 2.8526645768025077e-05,
2513
+ "loss": 0.0037,
2514
+ "step": 4110
2515
+ },
2516
+ {
2517
+ "epoch": 4.31,
2518
+ "learning_rate": 2.8474399164054338e-05,
2519
+ "loss": 0.0125,
2520
+ "step": 4120
2521
+ },
2522
+ {
2523
+ "epoch": 4.32,
2524
+ "learning_rate": 2.8422152560083598e-05,
2525
+ "loss": 0.0057,
2526
+ "step": 4130
2527
+ },
2528
+ {
2529
+ "epoch": 4.33,
2530
+ "learning_rate": 2.8369905956112852e-05,
2531
+ "loss": 0.0057,
2532
+ "step": 4140
2533
+ },
2534
+ {
2535
+ "epoch": 4.34,
2536
+ "learning_rate": 2.8317659352142113e-05,
2537
+ "loss": 0.0062,
2538
+ "step": 4150
2539
+ },
2540
+ {
2541
+ "epoch": 4.35,
2542
+ "learning_rate": 2.8265412748171373e-05,
2543
+ "loss": 0.0074,
2544
+ "step": 4160
2545
+ },
2546
+ {
2547
+ "epoch": 4.36,
2548
+ "learning_rate": 2.8213166144200627e-05,
2549
+ "loss": 0.0034,
2550
+ "step": 4170
2551
+ },
2552
+ {
2553
+ "epoch": 4.37,
2554
+ "learning_rate": 2.8160919540229884e-05,
2555
+ "loss": 0.0055,
2556
+ "step": 4180
2557
+ },
2558
+ {
2559
+ "epoch": 4.38,
2560
+ "learning_rate": 2.8108672936259145e-05,
2561
+ "loss": 0.0055,
2562
+ "step": 4190
2563
+ },
2564
+ {
2565
+ "epoch": 4.39,
2566
+ "learning_rate": 2.8056426332288405e-05,
2567
+ "loss": 0.0041,
2568
+ "step": 4200
2569
+ },
2570
+ {
2571
+ "epoch": 4.4,
2572
+ "learning_rate": 2.800417972831766e-05,
2573
+ "loss": 0.0089,
2574
+ "step": 4210
2575
+ },
2576
+ {
2577
+ "epoch": 4.41,
2578
+ "learning_rate": 2.795193312434692e-05,
2579
+ "loss": 0.0045,
2580
+ "step": 4220
2581
+ },
2582
+ {
2583
+ "epoch": 4.42,
2584
+ "learning_rate": 2.7899686520376177e-05,
2585
+ "loss": 0.0056,
2586
+ "step": 4230
2587
+ },
2588
+ {
2589
+ "epoch": 4.43,
2590
+ "learning_rate": 2.7847439916405434e-05,
2591
+ "loss": 0.0036,
2592
+ "step": 4240
2593
+ },
2594
+ {
2595
+ "epoch": 4.44,
2596
+ "learning_rate": 2.779519331243469e-05,
2597
+ "loss": 0.0036,
2598
+ "step": 4250
2599
+ },
2600
+ {
2601
+ "epoch": 4.45,
2602
+ "learning_rate": 2.774294670846395e-05,
2603
+ "loss": 0.004,
2604
+ "step": 4260
2605
+ },
2606
+ {
2607
+ "epoch": 4.46,
2608
+ "learning_rate": 2.7690700104493212e-05,
2609
+ "loss": 0.0049,
2610
+ "step": 4270
2611
+ },
2612
+ {
2613
+ "epoch": 4.47,
2614
+ "learning_rate": 2.7638453500522466e-05,
2615
+ "loss": 0.0042,
2616
+ "step": 4280
2617
+ },
2618
+ {
2619
+ "epoch": 4.48,
2620
+ "learning_rate": 2.7586206896551727e-05,
2621
+ "loss": 0.0066,
2622
+ "step": 4290
2623
+ },
2624
+ {
2625
+ "epoch": 4.49,
2626
+ "learning_rate": 2.7533960292580984e-05,
2627
+ "loss": 0.0054,
2628
+ "step": 4300
2629
+ },
2630
+ {
2631
+ "epoch": 4.5,
2632
+ "learning_rate": 2.7481713688610244e-05,
2633
+ "loss": 0.0041,
2634
+ "step": 4310
2635
+ },
2636
+ {
2637
+ "epoch": 4.51,
2638
+ "learning_rate": 2.7429467084639498e-05,
2639
+ "loss": 0.0034,
2640
+ "step": 4320
2641
+ },
2642
+ {
2643
+ "epoch": 4.52,
2644
+ "learning_rate": 2.737722048066876e-05,
2645
+ "loss": 0.0065,
2646
+ "step": 4330
2647
+ },
2648
+ {
2649
+ "epoch": 4.54,
2650
+ "learning_rate": 2.732497387669802e-05,
2651
+ "loss": 0.0062,
2652
+ "step": 4340
2653
+ },
2654
+ {
2655
+ "epoch": 4.55,
2656
+ "learning_rate": 2.7272727272727273e-05,
2657
+ "loss": 0.0068,
2658
+ "step": 4350
2659
+ },
2660
+ {
2661
+ "epoch": 4.56,
2662
+ "learning_rate": 2.722048066875653e-05,
2663
+ "loss": 0.0045,
2664
+ "step": 4360
2665
+ },
2666
+ {
2667
+ "epoch": 4.57,
2668
+ "learning_rate": 2.716823406478579e-05,
2669
+ "loss": 0.0089,
2670
+ "step": 4370
2671
+ },
2672
+ {
2673
+ "epoch": 4.58,
2674
+ "learning_rate": 2.711598746081505e-05,
2675
+ "loss": 0.0046,
2676
+ "step": 4380
2677
+ },
2678
+ {
2679
+ "epoch": 4.59,
2680
+ "learning_rate": 2.7063740856844305e-05,
2681
+ "loss": 0.0061,
2682
+ "step": 4390
2683
+ },
2684
+ {
2685
+ "epoch": 4.6,
2686
+ "learning_rate": 2.7011494252873566e-05,
2687
+ "loss": 0.0036,
2688
+ "step": 4400
2689
+ },
2690
+ {
2691
+ "epoch": 4.61,
2692
+ "learning_rate": 2.6959247648902823e-05,
2693
+ "loss": 0.0048,
2694
+ "step": 4410
2695
+ },
2696
+ {
2697
+ "epoch": 4.62,
2698
+ "learning_rate": 2.690700104493208e-05,
2699
+ "loss": 0.0028,
2700
+ "step": 4420
2701
+ },
2702
+ {
2703
+ "epoch": 4.63,
2704
+ "learning_rate": 2.6854754440961337e-05,
2705
+ "loss": 0.0037,
2706
+ "step": 4430
2707
+ },
2708
+ {
2709
+ "epoch": 4.64,
2710
+ "learning_rate": 2.6802507836990598e-05,
2711
+ "loss": 0.0039,
2712
+ "step": 4440
2713
+ },
2714
+ {
2715
+ "epoch": 4.65,
2716
+ "learning_rate": 2.675026123301986e-05,
2717
+ "loss": 0.0077,
2718
+ "step": 4450
2719
+ },
2720
+ {
2721
+ "epoch": 4.66,
2722
+ "learning_rate": 2.6698014629049112e-05,
2723
+ "loss": 0.0035,
2724
+ "step": 4460
2725
+ },
2726
+ {
2727
+ "epoch": 4.67,
2728
+ "learning_rate": 2.6645768025078373e-05,
2729
+ "loss": 0.008,
2730
+ "step": 4470
2731
+ },
2732
+ {
2733
+ "epoch": 4.68,
2734
+ "learning_rate": 2.659352142110763e-05,
2735
+ "loss": 0.0053,
2736
+ "step": 4480
2737
+ },
2738
+ {
2739
+ "epoch": 4.69,
2740
+ "learning_rate": 2.6541274817136884e-05,
2741
+ "loss": 0.0031,
2742
+ "step": 4490
2743
+ },
2744
+ {
2745
+ "epoch": 4.7,
2746
+ "learning_rate": 2.6489028213166144e-05,
2747
+ "loss": 0.0038,
2748
+ "step": 4500
2749
+ },
2750
+ {
2751
+ "epoch": 4.71,
2752
+ "learning_rate": 2.6436781609195405e-05,
2753
+ "loss": 0.0042,
2754
+ "step": 4510
2755
+ },
2756
+ {
2757
+ "epoch": 4.72,
2758
+ "learning_rate": 2.6384535005224666e-05,
2759
+ "loss": 0.0073,
2760
+ "step": 4520
2761
+ },
2762
+ {
2763
+ "epoch": 4.73,
2764
+ "learning_rate": 2.633228840125392e-05,
2765
+ "loss": 0.0056,
2766
+ "step": 4530
2767
+ },
2768
+ {
2769
+ "epoch": 4.74,
2770
+ "learning_rate": 2.6280041797283177e-05,
2771
+ "loss": 0.0039,
2772
+ "step": 4540
2773
+ },
2774
+ {
2775
+ "epoch": 4.75,
2776
+ "learning_rate": 2.6227795193312437e-05,
2777
+ "loss": 0.0045,
2778
+ "step": 4550
2779
+ },
2780
+ {
2781
+ "epoch": 4.76,
2782
+ "learning_rate": 2.617554858934169e-05,
2783
+ "loss": 0.0065,
2784
+ "step": 4560
2785
+ },
2786
+ {
2787
+ "epoch": 4.78,
2788
+ "learning_rate": 2.612330198537095e-05,
2789
+ "loss": 0.0039,
2790
+ "step": 4570
2791
+ },
2792
+ {
2793
+ "epoch": 4.79,
2794
+ "learning_rate": 2.6071055381400212e-05,
2795
+ "loss": 0.0043,
2796
+ "step": 4580
2797
+ },
2798
+ {
2799
+ "epoch": 4.8,
2800
+ "learning_rate": 2.601880877742947e-05,
2801
+ "loss": 0.0033,
2802
+ "step": 4590
2803
+ },
2804
+ {
2805
+ "epoch": 4.81,
2806
+ "learning_rate": 2.5966562173458726e-05,
2807
+ "loss": 0.0035,
2808
+ "step": 4600
2809
+ },
2810
+ {
2811
+ "epoch": 4.82,
2812
+ "learning_rate": 2.5914315569487984e-05,
2813
+ "loss": 0.0046,
2814
+ "step": 4610
2815
+ },
2816
+ {
2817
+ "epoch": 4.83,
2818
+ "learning_rate": 2.5862068965517244e-05,
2819
+ "loss": 0.0026,
2820
+ "step": 4620
2821
+ },
2822
+ {
2823
+ "epoch": 4.84,
2824
+ "learning_rate": 2.5809822361546498e-05,
2825
+ "loss": 0.0031,
2826
+ "step": 4630
2827
+ },
2828
+ {
2829
+ "epoch": 4.85,
2830
+ "learning_rate": 2.575757575757576e-05,
2831
+ "loss": 0.0042,
2832
+ "step": 4640
2833
+ },
2834
+ {
2835
+ "epoch": 4.86,
2836
+ "learning_rate": 2.570532915360502e-05,
2837
+ "loss": 0.0037,
2838
+ "step": 4650
2839
+ },
2840
+ {
2841
+ "epoch": 4.87,
2842
+ "learning_rate": 2.5653082549634276e-05,
2843
+ "loss": 0.0034,
2844
+ "step": 4660
2845
+ },
2846
+ {
2847
+ "epoch": 4.88,
2848
+ "learning_rate": 2.560083594566353e-05,
2849
+ "loss": 0.006,
2850
+ "step": 4670
2851
+ },
2852
+ {
2853
+ "epoch": 4.89,
2854
+ "learning_rate": 2.554858934169279e-05,
2855
+ "loss": 0.005,
2856
+ "step": 4680
2857
+ },
2858
+ {
2859
+ "epoch": 4.9,
2860
+ "learning_rate": 2.549634273772205e-05,
2861
+ "loss": 0.0033,
2862
+ "step": 4690
2863
+ },
2864
+ {
2865
+ "epoch": 4.91,
2866
+ "learning_rate": 2.5444096133751305e-05,
2867
+ "loss": 0.0061,
2868
+ "step": 4700
2869
+ },
2870
+ {
2871
+ "epoch": 4.92,
2872
+ "learning_rate": 2.5391849529780566e-05,
2873
+ "loss": 0.0059,
2874
+ "step": 4710
2875
+ },
2876
+ {
2877
+ "epoch": 4.93,
2878
+ "learning_rate": 2.5339602925809823e-05,
2879
+ "loss": 0.0046,
2880
+ "step": 4720
2881
+ },
2882
+ {
2883
+ "epoch": 4.94,
2884
+ "learning_rate": 2.5287356321839083e-05,
2885
+ "loss": 0.0062,
2886
+ "step": 4730
2887
+ },
2888
+ {
2889
+ "epoch": 4.95,
2890
+ "learning_rate": 2.5235109717868337e-05,
2891
+ "loss": 0.0033,
2892
+ "step": 4740
2893
+ },
2894
+ {
2895
+ "epoch": 4.96,
2896
+ "learning_rate": 2.5182863113897598e-05,
2897
+ "loss": 0.0065,
2898
+ "step": 4750
2899
+ },
2900
+ {
2901
+ "epoch": 4.97,
2902
+ "learning_rate": 2.5130616509926858e-05,
2903
+ "loss": 0.0048,
2904
+ "step": 4760
2905
+ },
2906
+ {
2907
+ "epoch": 4.98,
2908
+ "learning_rate": 2.5078369905956112e-05,
2909
+ "loss": 0.0043,
2910
+ "step": 4770
2911
+ },
2912
+ {
2913
+ "epoch": 4.99,
2914
+ "learning_rate": 2.5026123301985373e-05,
2915
+ "loss": 0.0055,
2916
+ "step": 4780
2917
+ },
2918
+ {
2919
+ "epoch": 5.01,
2920
+ "learning_rate": 2.497387669801463e-05,
2921
+ "loss": 0.0039,
2922
+ "step": 4790
2923
+ },
2924
+ {
2925
+ "epoch": 5.02,
2926
+ "learning_rate": 2.4921630094043887e-05,
2927
+ "loss": 0.0025,
2928
+ "step": 4800
2929
+ },
2930
+ {
2931
+ "epoch": 5.03,
2932
+ "learning_rate": 2.4869383490073148e-05,
2933
+ "loss": 0.0039,
2934
+ "step": 4810
2935
+ },
2936
+ {
2937
+ "epoch": 5.04,
2938
+ "learning_rate": 2.4817136886102405e-05,
2939
+ "loss": 0.0048,
2940
+ "step": 4820
2941
+ },
2942
+ {
2943
+ "epoch": 5.05,
2944
+ "learning_rate": 2.4764890282131662e-05,
2945
+ "loss": 0.005,
2946
+ "step": 4830
2947
+ },
2948
+ {
2949
+ "epoch": 5.06,
2950
+ "learning_rate": 2.4712643678160922e-05,
2951
+ "loss": 0.0039,
2952
+ "step": 4840
2953
+ },
2954
+ {
2955
+ "epoch": 5.07,
2956
+ "learning_rate": 2.466039707419018e-05,
2957
+ "loss": 0.0096,
2958
+ "step": 4850
2959
+ },
2960
+ {
2961
+ "epoch": 5.08,
2962
+ "learning_rate": 2.4608150470219437e-05,
2963
+ "loss": 0.0046,
2964
+ "step": 4860
2965
+ },
2966
+ {
2967
+ "epoch": 5.09,
2968
+ "learning_rate": 2.4555903866248694e-05,
2969
+ "loss": 0.0026,
2970
+ "step": 4870
2971
+ },
2972
+ {
2973
+ "epoch": 5.1,
2974
+ "learning_rate": 2.4503657262277955e-05,
2975
+ "loss": 0.0043,
2976
+ "step": 4880
2977
+ },
2978
+ {
2979
+ "epoch": 5.11,
2980
+ "learning_rate": 2.4451410658307212e-05,
2981
+ "loss": 0.0031,
2982
+ "step": 4890
2983
+ },
2984
+ {
2985
+ "epoch": 5.12,
2986
+ "learning_rate": 2.439916405433647e-05,
2987
+ "loss": 0.005,
2988
+ "step": 4900
2989
+ },
2990
+ {
2991
+ "epoch": 5.13,
2992
+ "learning_rate": 2.4346917450365726e-05,
2993
+ "loss": 0.002,
2994
+ "step": 4910
2995
+ },
2996
+ {
2997
+ "epoch": 5.14,
2998
+ "learning_rate": 2.4294670846394983e-05,
2999
+ "loss": 0.0021,
3000
+ "step": 4920
3001
+ },
3002
+ {
3003
+ "epoch": 5.15,
3004
+ "learning_rate": 2.4242424242424244e-05,
3005
+ "loss": 0.0052,
3006
+ "step": 4930
3007
+ },
3008
+ {
3009
+ "epoch": 5.16,
3010
+ "learning_rate": 2.41901776384535e-05,
3011
+ "loss": 0.006,
3012
+ "step": 4940
3013
+ },
3014
+ {
3015
+ "epoch": 5.17,
3016
+ "learning_rate": 2.413793103448276e-05,
3017
+ "loss": 0.0028,
3018
+ "step": 4950
3019
+ },
3020
+ {
3021
+ "epoch": 5.18,
3022
+ "learning_rate": 2.408568443051202e-05,
3023
+ "loss": 0.0026,
3024
+ "step": 4960
3025
+ },
3026
+ {
3027
+ "epoch": 5.19,
3028
+ "learning_rate": 2.4033437826541276e-05,
3029
+ "loss": 0.0033,
3030
+ "step": 4970
3031
+ },
3032
+ {
3033
+ "epoch": 5.2,
3034
+ "learning_rate": 2.3981191222570533e-05,
3035
+ "loss": 0.003,
3036
+ "step": 4980
3037
+ },
3038
+ {
3039
+ "epoch": 5.21,
3040
+ "learning_rate": 2.392894461859979e-05,
3041
+ "loss": 0.0051,
3042
+ "step": 4990
3043
+ },
3044
+ {
3045
+ "epoch": 5.22,
3046
+ "learning_rate": 2.387669801462905e-05,
3047
+ "loss": 0.0062,
3048
+ "step": 5000
3049
+ },
3050
+ {
3051
+ "epoch": 5.22,
3052
+ "eval_acc": 0.7915632734700663,
3053
+ "eval_cer": 0.026731752673175267,
3054
+ "eval_loss": 0.01478270161896944,
3055
+ "eval_runtime": 128.2616,
3056
+ "eval_samples_per_second": 3.142,
3057
+ "eval_steps_per_second": 0.398,
3058
+ "step": 5000
3059
+ }
3060
+ ],
3061
+ "max_steps": 9570,
3062
+ "num_train_epochs": 10,
3063
+ "total_flos": 3.133753627068334e+18,
3064
+ "trial_name": null,
3065
+ "trial_params": null
3066
+ }
checkpoint/trocr-custdata-8000/checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/checkpoint-6000/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/trocr-custdata-8000/checkpoint-6000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb391be28b1fc9620d729287f5e0f0c73fb0635d8bfc48d5ba52bba7d32bea59
3
+ size 243062533
checkpoint/{trocr-custdata/checkpoint-2000 → trocr-custdata-8000/checkpoint-6000}/preprocessor_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/checkpoint-6000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a81037b1926bb383febee12b498020415177c2220108790e664f39467d772a70
3
+ size 122142261
checkpoint/trocr-custdata-8000/checkpoint-6000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:648f1e9246591c436800f6ff4d35a73535e498abf8c7ef9b3c9861cba337dcdf
3
+ size 627
checkpoint/trocr-custdata-8000/checkpoint-6000/trainer_state.json ADDED
@@ -0,0 +1,3676 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 6.269592476489028,
5
+ "global_step": 6000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 4.994775339602926e-05,
13
+ "loss": 0.3717,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 4.989550679205852e-05,
19
+ "loss": 0.2368,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 4.9843260188087774e-05,
25
+ "loss": 0.1928,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 4.979101358411704e-05,
31
+ "loss": 0.1648,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 4.9738766980146295e-05,
37
+ "loss": 0.1645,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 4.968652037617555e-05,
43
+ "loss": 0.1151,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 4.963427377220481e-05,
49
+ "loss": 0.1206,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 4.958202716823407e-05,
55
+ "loss": 0.1289,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 4.9529780564263324e-05,
61
+ "loss": 0.1123,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 4.947753396029258e-05,
67
+ "loss": 0.0992,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.11,
72
+ "learning_rate": 4.9425287356321845e-05,
73
+ "loss": 0.097,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.13,
78
+ "learning_rate": 4.93730407523511e-05,
79
+ "loss": 0.1028,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.14,
84
+ "learning_rate": 4.932079414838036e-05,
85
+ "loss": 0.087,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.15,
90
+ "learning_rate": 4.9268547544409617e-05,
91
+ "loss": 0.0907,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.16,
96
+ "learning_rate": 4.9216300940438874e-05,
97
+ "loss": 0.0866,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.17,
102
+ "learning_rate": 4.916405433646813e-05,
103
+ "loss": 0.085,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.18,
108
+ "learning_rate": 4.911180773249739e-05,
109
+ "loss": 0.0846,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.19,
114
+ "learning_rate": 4.905956112852665e-05,
115
+ "loss": 0.0791,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.2,
120
+ "learning_rate": 4.900731452455591e-05,
121
+ "loss": 0.073,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.21,
126
+ "learning_rate": 4.895506792058516e-05,
127
+ "loss": 0.0861,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.22,
132
+ "learning_rate": 4.8902821316614424e-05,
133
+ "loss": 0.0828,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.23,
138
+ "learning_rate": 4.885057471264368e-05,
139
+ "loss": 0.0711,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.24,
144
+ "learning_rate": 4.879832810867294e-05,
145
+ "loss": 0.069,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.25,
150
+ "learning_rate": 4.8746081504702195e-05,
151
+ "loss": 0.0647,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.26,
156
+ "learning_rate": 4.869383490073145e-05,
157
+ "loss": 0.0862,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.27,
162
+ "learning_rate": 4.8641588296760716e-05,
163
+ "loss": 0.0628,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.28,
168
+ "learning_rate": 4.858934169278997e-05,
169
+ "loss": 0.0697,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.29,
174
+ "learning_rate": 4.853709508881923e-05,
175
+ "loss": 0.0685,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.3,
180
+ "learning_rate": 4.848484848484849e-05,
181
+ "loss": 0.0602,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.31,
186
+ "learning_rate": 4.8432601880877745e-05,
187
+ "loss": 0.0576,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.32,
192
+ "learning_rate": 4.8380355276907e-05,
193
+ "loss": 0.068,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.33,
198
+ "learning_rate": 4.832810867293626e-05,
199
+ "loss": 0.0645,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.34,
204
+ "learning_rate": 4.827586206896552e-05,
205
+ "loss": 0.0785,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.36,
210
+ "learning_rate": 4.8223615464994774e-05,
211
+ "loss": 0.0606,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.37,
216
+ "learning_rate": 4.817136886102404e-05,
217
+ "loss": 0.0682,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.38,
222
+ "learning_rate": 4.8119122257053295e-05,
223
+ "loss": 0.0682,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.39,
228
+ "learning_rate": 4.806687565308255e-05,
229
+ "loss": 0.0605,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.4,
234
+ "learning_rate": 4.801462904911181e-05,
235
+ "loss": 0.0578,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.41,
240
+ "learning_rate": 4.7962382445141066e-05,
241
+ "loss": 0.0562,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.42,
246
+ "learning_rate": 4.791013584117033e-05,
247
+ "loss": 0.0724,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.43,
252
+ "learning_rate": 4.785788923719958e-05,
253
+ "loss": 0.056,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.44,
258
+ "learning_rate": 4.7805642633228845e-05,
259
+ "loss": 0.0509,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.45,
264
+ "learning_rate": 4.77533960292581e-05,
265
+ "loss": 0.0516,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.46,
270
+ "learning_rate": 4.770114942528736e-05,
271
+ "loss": 0.0588,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.47,
276
+ "learning_rate": 4.7648902821316616e-05,
277
+ "loss": 0.0527,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.48,
282
+ "learning_rate": 4.7596656217345873e-05,
283
+ "loss": 0.0614,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.49,
288
+ "learning_rate": 4.754440961337514e-05,
289
+ "loss": 0.0464,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.5,
294
+ "learning_rate": 4.749216300940439e-05,
295
+ "loss": 0.0515,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.51,
300
+ "learning_rate": 4.743991640543365e-05,
301
+ "loss": 0.0456,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.52,
306
+ "learning_rate": 4.738766980146291e-05,
307
+ "loss": 0.0583,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.53,
312
+ "learning_rate": 4.7335423197492166e-05,
313
+ "loss": 0.0595,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 0.54,
318
+ "learning_rate": 4.728317659352142e-05,
319
+ "loss": 0.0547,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 0.55,
324
+ "learning_rate": 4.723092998955068e-05,
325
+ "loss": 0.0451,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 0.56,
330
+ "learning_rate": 4.7178683385579944e-05,
331
+ "loss": 0.0395,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 0.57,
336
+ "learning_rate": 4.7126436781609195e-05,
337
+ "loss": 0.0493,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 0.59,
342
+ "learning_rate": 4.707419017763845e-05,
343
+ "loss": 0.0664,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 0.6,
348
+ "learning_rate": 4.7021943573667716e-05,
349
+ "loss": 0.0445,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 0.61,
354
+ "learning_rate": 4.696969696969697e-05,
355
+ "loss": 0.0443,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 0.62,
360
+ "learning_rate": 4.691745036572623e-05,
361
+ "loss": 0.0461,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 0.63,
366
+ "learning_rate": 4.686520376175549e-05,
367
+ "loss": 0.0415,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 0.64,
372
+ "learning_rate": 4.6812957157784745e-05,
373
+ "loss": 0.0483,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 0.65,
378
+ "learning_rate": 4.6760710553814e-05,
379
+ "loss": 0.0424,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 0.66,
384
+ "learning_rate": 4.670846394984326e-05,
385
+ "loss": 0.0327,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 0.67,
390
+ "learning_rate": 4.665621734587252e-05,
391
+ "loss": 0.0421,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 0.68,
396
+ "learning_rate": 4.660397074190178e-05,
397
+ "loss": 0.0491,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 0.69,
402
+ "learning_rate": 4.655172413793104e-05,
403
+ "loss": 0.0501,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 0.7,
408
+ "learning_rate": 4.6499477533960295e-05,
409
+ "loss": 0.0447,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 0.71,
414
+ "learning_rate": 4.644723092998955e-05,
415
+ "loss": 0.0449,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 0.72,
420
+ "learning_rate": 4.639498432601881e-05,
421
+ "loss": 0.0478,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 0.73,
426
+ "learning_rate": 4.6342737722048066e-05,
427
+ "loss": 0.0505,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 0.74,
432
+ "learning_rate": 4.629049111807733e-05,
433
+ "loss": 0.0394,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 0.75,
438
+ "learning_rate": 4.623824451410659e-05,
439
+ "loss": 0.0406,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 0.76,
444
+ "learning_rate": 4.6185997910135844e-05,
445
+ "loss": 0.0347,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 0.77,
450
+ "learning_rate": 4.61337513061651e-05,
451
+ "loss": 0.0445,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 0.78,
456
+ "learning_rate": 4.608150470219436e-05,
457
+ "loss": 0.0428,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 0.79,
462
+ "learning_rate": 4.6029258098223616e-05,
463
+ "loss": 0.0466,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 0.8,
468
+ "learning_rate": 4.597701149425287e-05,
469
+ "loss": 0.0395,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 0.82,
474
+ "learning_rate": 4.592476489028214e-05,
475
+ "loss": 0.0387,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 0.83,
480
+ "learning_rate": 4.5872518286311394e-05,
481
+ "loss": 0.0395,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 0.84,
486
+ "learning_rate": 4.582027168234065e-05,
487
+ "loss": 0.0305,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 0.85,
492
+ "learning_rate": 4.576802507836991e-05,
493
+ "loss": 0.0434,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 0.86,
498
+ "learning_rate": 4.5715778474399166e-05,
499
+ "loss": 0.0395,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 0.87,
504
+ "learning_rate": 4.566353187042842e-05,
505
+ "loss": 0.0374,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 0.88,
510
+ "learning_rate": 4.561128526645768e-05,
511
+ "loss": 0.0321,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 0.89,
516
+ "learning_rate": 4.5559038662486944e-05,
517
+ "loss": 0.0394,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 0.9,
522
+ "learning_rate": 4.55067920585162e-05,
523
+ "loss": 0.0383,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 0.91,
528
+ "learning_rate": 4.545454545454546e-05,
529
+ "loss": 0.0388,
530
+ "step": 870
531
+ },
532
+ {
533
+ "epoch": 0.92,
534
+ "learning_rate": 4.5402298850574716e-05,
535
+ "loss": 0.035,
536
+ "step": 880
537
+ },
538
+ {
539
+ "epoch": 0.93,
540
+ "learning_rate": 4.535005224660397e-05,
541
+ "loss": 0.0414,
542
+ "step": 890
543
+ },
544
+ {
545
+ "epoch": 0.94,
546
+ "learning_rate": 4.529780564263323e-05,
547
+ "loss": 0.0439,
548
+ "step": 900
549
+ },
550
+ {
551
+ "epoch": 0.95,
552
+ "learning_rate": 4.524555903866249e-05,
553
+ "loss": 0.0337,
554
+ "step": 910
555
+ },
556
+ {
557
+ "epoch": 0.96,
558
+ "learning_rate": 4.519331243469175e-05,
559
+ "loss": 0.0384,
560
+ "step": 920
561
+ },
562
+ {
563
+ "epoch": 0.97,
564
+ "learning_rate": 4.514106583072101e-05,
565
+ "loss": 0.0353,
566
+ "step": 930
567
+ },
568
+ {
569
+ "epoch": 0.98,
570
+ "learning_rate": 4.508881922675026e-05,
571
+ "loss": 0.0351,
572
+ "step": 940
573
+ },
574
+ {
575
+ "epoch": 0.99,
576
+ "learning_rate": 4.503657262277952e-05,
577
+ "loss": 0.0324,
578
+ "step": 950
579
+ },
580
+ {
581
+ "epoch": 1.0,
582
+ "learning_rate": 4.498432601880878e-05,
583
+ "loss": 0.0296,
584
+ "step": 960
585
+ },
586
+ {
587
+ "epoch": 1.01,
588
+ "learning_rate": 4.493207941483804e-05,
589
+ "loss": 0.0262,
590
+ "step": 970
591
+ },
592
+ {
593
+ "epoch": 1.02,
594
+ "learning_rate": 4.4879832810867294e-05,
595
+ "loss": 0.0274,
596
+ "step": 980
597
+ },
598
+ {
599
+ "epoch": 1.03,
600
+ "learning_rate": 4.482758620689655e-05,
601
+ "loss": 0.0278,
602
+ "step": 990
603
+ },
604
+ {
605
+ "epoch": 1.04,
606
+ "learning_rate": 4.4775339602925815e-05,
607
+ "loss": 0.0205,
608
+ "step": 1000
609
+ },
610
+ {
611
+ "epoch": 1.04,
612
+ "eval_acc": 0.5359801475533992,
613
+ "eval_cer": 0.06462110646211065,
614
+ "eval_loss": 0.0336129367351532,
615
+ "eval_runtime": 135.8305,
616
+ "eval_samples_per_second": 2.967,
617
+ "eval_steps_per_second": 0.375,
618
+ "step": 1000
619
+ },
620
+ {
621
+ "epoch": 1.06,
622
+ "learning_rate": 4.4723092998955066e-05,
623
+ "loss": 0.0264,
624
+ "step": 1010
625
+ },
626
+ {
627
+ "epoch": 1.07,
628
+ "learning_rate": 4.467084639498433e-05,
629
+ "loss": 0.0218,
630
+ "step": 1020
631
+ },
632
+ {
633
+ "epoch": 1.08,
634
+ "learning_rate": 4.461859979101359e-05,
635
+ "loss": 0.0269,
636
+ "step": 1030
637
+ },
638
+ {
639
+ "epoch": 1.09,
640
+ "learning_rate": 4.4566353187042844e-05,
641
+ "loss": 0.0242,
642
+ "step": 1040
643
+ },
644
+ {
645
+ "epoch": 1.1,
646
+ "learning_rate": 4.45141065830721e-05,
647
+ "loss": 0.03,
648
+ "step": 1050
649
+ },
650
+ {
651
+ "epoch": 1.11,
652
+ "learning_rate": 4.446185997910136e-05,
653
+ "loss": 0.0214,
654
+ "step": 1060
655
+ },
656
+ {
657
+ "epoch": 1.12,
658
+ "learning_rate": 4.440961337513062e-05,
659
+ "loss": 0.0301,
660
+ "step": 1070
661
+ },
662
+ {
663
+ "epoch": 1.13,
664
+ "learning_rate": 4.435736677115987e-05,
665
+ "loss": 0.0217,
666
+ "step": 1080
667
+ },
668
+ {
669
+ "epoch": 1.14,
670
+ "learning_rate": 4.430512016718914e-05,
671
+ "loss": 0.0323,
672
+ "step": 1090
673
+ },
674
+ {
675
+ "epoch": 1.15,
676
+ "learning_rate": 4.4252873563218394e-05,
677
+ "loss": 0.0258,
678
+ "step": 1100
679
+ },
680
+ {
681
+ "epoch": 1.16,
682
+ "learning_rate": 4.420062695924765e-05,
683
+ "loss": 0.0293,
684
+ "step": 1110
685
+ },
686
+ {
687
+ "epoch": 1.17,
688
+ "learning_rate": 4.414838035527691e-05,
689
+ "loss": 0.0265,
690
+ "step": 1120
691
+ },
692
+ {
693
+ "epoch": 1.18,
694
+ "learning_rate": 4.4096133751306166e-05,
695
+ "loss": 0.0237,
696
+ "step": 1130
697
+ },
698
+ {
699
+ "epoch": 1.19,
700
+ "learning_rate": 4.404388714733543e-05,
701
+ "loss": 0.0275,
702
+ "step": 1140
703
+ },
704
+ {
705
+ "epoch": 1.2,
706
+ "learning_rate": 4.399164054336468e-05,
707
+ "loss": 0.0219,
708
+ "step": 1150
709
+ },
710
+ {
711
+ "epoch": 1.21,
712
+ "learning_rate": 4.3939393939393944e-05,
713
+ "loss": 0.0334,
714
+ "step": 1160
715
+ },
716
+ {
717
+ "epoch": 1.22,
718
+ "learning_rate": 4.38871473354232e-05,
719
+ "loss": 0.0183,
720
+ "step": 1170
721
+ },
722
+ {
723
+ "epoch": 1.23,
724
+ "learning_rate": 4.383490073145246e-05,
725
+ "loss": 0.0241,
726
+ "step": 1180
727
+ },
728
+ {
729
+ "epoch": 1.24,
730
+ "learning_rate": 4.3782654127481716e-05,
731
+ "loss": 0.0233,
732
+ "step": 1190
733
+ },
734
+ {
735
+ "epoch": 1.25,
736
+ "learning_rate": 4.373040752351097e-05,
737
+ "loss": 0.025,
738
+ "step": 1200
739
+ },
740
+ {
741
+ "epoch": 1.26,
742
+ "learning_rate": 4.367816091954024e-05,
743
+ "loss": 0.0228,
744
+ "step": 1210
745
+ },
746
+ {
747
+ "epoch": 1.27,
748
+ "learning_rate": 4.362591431556949e-05,
749
+ "loss": 0.0182,
750
+ "step": 1220
751
+ },
752
+ {
753
+ "epoch": 1.29,
754
+ "learning_rate": 4.357366771159875e-05,
755
+ "loss": 0.0234,
756
+ "step": 1230
757
+ },
758
+ {
759
+ "epoch": 1.3,
760
+ "learning_rate": 4.352142110762801e-05,
761
+ "loss": 0.019,
762
+ "step": 1240
763
+ },
764
+ {
765
+ "epoch": 1.31,
766
+ "learning_rate": 4.346917450365726e-05,
767
+ "loss": 0.0191,
768
+ "step": 1250
769
+ },
770
+ {
771
+ "epoch": 1.32,
772
+ "learning_rate": 4.341692789968652e-05,
773
+ "loss": 0.0177,
774
+ "step": 1260
775
+ },
776
+ {
777
+ "epoch": 1.33,
778
+ "learning_rate": 4.336468129571578e-05,
779
+ "loss": 0.0214,
780
+ "step": 1270
781
+ },
782
+ {
783
+ "epoch": 1.34,
784
+ "learning_rate": 4.3312434691745044e-05,
785
+ "loss": 0.0252,
786
+ "step": 1280
787
+ },
788
+ {
789
+ "epoch": 1.35,
790
+ "learning_rate": 4.3260188087774294e-05,
791
+ "loss": 0.0233,
792
+ "step": 1290
793
+ },
794
+ {
795
+ "epoch": 1.36,
796
+ "learning_rate": 4.320794148380355e-05,
797
+ "loss": 0.0184,
798
+ "step": 1300
799
+ },
800
+ {
801
+ "epoch": 1.37,
802
+ "learning_rate": 4.3155694879832815e-05,
803
+ "loss": 0.0212,
804
+ "step": 1310
805
+ },
806
+ {
807
+ "epoch": 1.38,
808
+ "learning_rate": 4.3103448275862066e-05,
809
+ "loss": 0.0186,
810
+ "step": 1320
811
+ },
812
+ {
813
+ "epoch": 1.39,
814
+ "learning_rate": 4.305120167189133e-05,
815
+ "loss": 0.0179,
816
+ "step": 1330
817
+ },
818
+ {
819
+ "epoch": 1.4,
820
+ "learning_rate": 4.299895506792059e-05,
821
+ "loss": 0.0184,
822
+ "step": 1340
823
+ },
824
+ {
825
+ "epoch": 1.41,
826
+ "learning_rate": 4.294670846394985e-05,
827
+ "loss": 0.0208,
828
+ "step": 1350
829
+ },
830
+ {
831
+ "epoch": 1.42,
832
+ "learning_rate": 4.28944618599791e-05,
833
+ "loss": 0.0218,
834
+ "step": 1360
835
+ },
836
+ {
837
+ "epoch": 1.43,
838
+ "learning_rate": 4.284221525600836e-05,
839
+ "loss": 0.0203,
840
+ "step": 1370
841
+ },
842
+ {
843
+ "epoch": 1.44,
844
+ "learning_rate": 4.278996865203762e-05,
845
+ "loss": 0.022,
846
+ "step": 1380
847
+ },
848
+ {
849
+ "epoch": 1.45,
850
+ "learning_rate": 4.273772204806687e-05,
851
+ "loss": 0.0247,
852
+ "step": 1390
853
+ },
854
+ {
855
+ "epoch": 1.46,
856
+ "learning_rate": 4.268547544409614e-05,
857
+ "loss": 0.0195,
858
+ "step": 1400
859
+ },
860
+ {
861
+ "epoch": 1.47,
862
+ "learning_rate": 4.2633228840125394e-05,
863
+ "loss": 0.0236,
864
+ "step": 1410
865
+ },
866
+ {
867
+ "epoch": 1.48,
868
+ "learning_rate": 4.258098223615465e-05,
869
+ "loss": 0.0182,
870
+ "step": 1420
871
+ },
872
+ {
873
+ "epoch": 1.49,
874
+ "learning_rate": 4.252873563218391e-05,
875
+ "loss": 0.0238,
876
+ "step": 1430
877
+ },
878
+ {
879
+ "epoch": 1.5,
880
+ "learning_rate": 4.2476489028213165e-05,
881
+ "loss": 0.0244,
882
+ "step": 1440
883
+ },
884
+ {
885
+ "epoch": 1.52,
886
+ "learning_rate": 4.242424242424243e-05,
887
+ "loss": 0.0366,
888
+ "step": 1450
889
+ },
890
+ {
891
+ "epoch": 1.53,
892
+ "learning_rate": 4.2371995820271687e-05,
893
+ "loss": 0.0159,
894
+ "step": 1460
895
+ },
896
+ {
897
+ "epoch": 1.54,
898
+ "learning_rate": 4.2319749216300944e-05,
899
+ "loss": 0.0146,
900
+ "step": 1470
901
+ },
902
+ {
903
+ "epoch": 1.55,
904
+ "learning_rate": 4.22675026123302e-05,
905
+ "loss": 0.0219,
906
+ "step": 1480
907
+ },
908
+ {
909
+ "epoch": 1.56,
910
+ "learning_rate": 4.221525600835946e-05,
911
+ "loss": 0.0176,
912
+ "step": 1490
913
+ },
914
+ {
915
+ "epoch": 1.57,
916
+ "learning_rate": 4.2163009404388715e-05,
917
+ "loss": 0.0207,
918
+ "step": 1500
919
+ },
920
+ {
921
+ "epoch": 1.58,
922
+ "learning_rate": 4.211076280041797e-05,
923
+ "loss": 0.0141,
924
+ "step": 1510
925
+ },
926
+ {
927
+ "epoch": 1.59,
928
+ "learning_rate": 4.2058516196447236e-05,
929
+ "loss": 0.023,
930
+ "step": 1520
931
+ },
932
+ {
933
+ "epoch": 1.6,
934
+ "learning_rate": 4.2006269592476494e-05,
935
+ "loss": 0.0154,
936
+ "step": 1530
937
+ },
938
+ {
939
+ "epoch": 1.61,
940
+ "learning_rate": 4.195402298850575e-05,
941
+ "loss": 0.0208,
942
+ "step": 1540
943
+ },
944
+ {
945
+ "epoch": 1.62,
946
+ "learning_rate": 4.190177638453501e-05,
947
+ "loss": 0.0283,
948
+ "step": 1550
949
+ },
950
+ {
951
+ "epoch": 1.63,
952
+ "learning_rate": 4.1849529780564265e-05,
953
+ "loss": 0.0238,
954
+ "step": 1560
955
+ },
956
+ {
957
+ "epoch": 1.64,
958
+ "learning_rate": 4.179728317659352e-05,
959
+ "loss": 0.0261,
960
+ "step": 1570
961
+ },
962
+ {
963
+ "epoch": 1.65,
964
+ "learning_rate": 4.174503657262278e-05,
965
+ "loss": 0.0206,
966
+ "step": 1580
967
+ },
968
+ {
969
+ "epoch": 1.66,
970
+ "learning_rate": 4.1692789968652043e-05,
971
+ "loss": 0.0239,
972
+ "step": 1590
973
+ },
974
+ {
975
+ "epoch": 1.67,
976
+ "learning_rate": 4.16405433646813e-05,
977
+ "loss": 0.0256,
978
+ "step": 1600
979
+ },
980
+ {
981
+ "epoch": 1.68,
982
+ "learning_rate": 4.158829676071056e-05,
983
+ "loss": 0.0216,
984
+ "step": 1610
985
+ },
986
+ {
987
+ "epoch": 1.69,
988
+ "learning_rate": 4.1536050156739815e-05,
989
+ "loss": 0.0188,
990
+ "step": 1620
991
+ },
992
+ {
993
+ "epoch": 1.7,
994
+ "learning_rate": 4.148380355276907e-05,
995
+ "loss": 0.018,
996
+ "step": 1630
997
+ },
998
+ {
999
+ "epoch": 1.71,
1000
+ "learning_rate": 4.143155694879833e-05,
1001
+ "loss": 0.0149,
1002
+ "step": 1640
1003
+ },
1004
+ {
1005
+ "epoch": 1.72,
1006
+ "learning_rate": 4.1379310344827587e-05,
1007
+ "loss": 0.0158,
1008
+ "step": 1650
1009
+ },
1010
+ {
1011
+ "epoch": 1.73,
1012
+ "learning_rate": 4.132706374085685e-05,
1013
+ "loss": 0.0216,
1014
+ "step": 1660
1015
+ },
1016
+ {
1017
+ "epoch": 1.75,
1018
+ "learning_rate": 4.127481713688611e-05,
1019
+ "loss": 0.0249,
1020
+ "step": 1670
1021
+ },
1022
+ {
1023
+ "epoch": 1.76,
1024
+ "learning_rate": 4.122257053291536e-05,
1025
+ "loss": 0.02,
1026
+ "step": 1680
1027
+ },
1028
+ {
1029
+ "epoch": 1.77,
1030
+ "learning_rate": 4.117032392894462e-05,
1031
+ "loss": 0.0188,
1032
+ "step": 1690
1033
+ },
1034
+ {
1035
+ "epoch": 1.78,
1036
+ "learning_rate": 4.111807732497388e-05,
1037
+ "loss": 0.0194,
1038
+ "step": 1700
1039
+ },
1040
+ {
1041
+ "epoch": 1.79,
1042
+ "learning_rate": 4.1065830721003136e-05,
1043
+ "loss": 0.0258,
1044
+ "step": 1710
1045
+ },
1046
+ {
1047
+ "epoch": 1.8,
1048
+ "learning_rate": 4.1013584117032394e-05,
1049
+ "loss": 0.0187,
1050
+ "step": 1720
1051
+ },
1052
+ {
1053
+ "epoch": 1.81,
1054
+ "learning_rate": 4.096133751306165e-05,
1055
+ "loss": 0.0195,
1056
+ "step": 1730
1057
+ },
1058
+ {
1059
+ "epoch": 1.82,
1060
+ "learning_rate": 4.0909090909090915e-05,
1061
+ "loss": 0.0218,
1062
+ "step": 1740
1063
+ },
1064
+ {
1065
+ "epoch": 1.83,
1066
+ "learning_rate": 4.0856844305120165e-05,
1067
+ "loss": 0.0182,
1068
+ "step": 1750
1069
+ },
1070
+ {
1071
+ "epoch": 1.84,
1072
+ "learning_rate": 4.080459770114943e-05,
1073
+ "loss": 0.0227,
1074
+ "step": 1760
1075
+ },
1076
+ {
1077
+ "epoch": 1.85,
1078
+ "learning_rate": 4.0752351097178686e-05,
1079
+ "loss": 0.0267,
1080
+ "step": 1770
1081
+ },
1082
+ {
1083
+ "epoch": 1.86,
1084
+ "learning_rate": 4.0700104493207943e-05,
1085
+ "loss": 0.0187,
1086
+ "step": 1780
1087
+ },
1088
+ {
1089
+ "epoch": 1.87,
1090
+ "learning_rate": 4.06478578892372e-05,
1091
+ "loss": 0.0251,
1092
+ "step": 1790
1093
+ },
1094
+ {
1095
+ "epoch": 1.88,
1096
+ "learning_rate": 4.059561128526646e-05,
1097
+ "loss": 0.0153,
1098
+ "step": 1800
1099
+ },
1100
+ {
1101
+ "epoch": 1.89,
1102
+ "learning_rate": 4.054336468129572e-05,
1103
+ "loss": 0.0301,
1104
+ "step": 1810
1105
+ },
1106
+ {
1107
+ "epoch": 1.9,
1108
+ "learning_rate": 4.049111807732497e-05,
1109
+ "loss": 0.0157,
1110
+ "step": 1820
1111
+ },
1112
+ {
1113
+ "epoch": 1.91,
1114
+ "learning_rate": 4.0438871473354236e-05,
1115
+ "loss": 0.0204,
1116
+ "step": 1830
1117
+ },
1118
+ {
1119
+ "epoch": 1.92,
1120
+ "learning_rate": 4.038662486938349e-05,
1121
+ "loss": 0.0203,
1122
+ "step": 1840
1123
+ },
1124
+ {
1125
+ "epoch": 1.93,
1126
+ "learning_rate": 4.033437826541275e-05,
1127
+ "loss": 0.0138,
1128
+ "step": 1850
1129
+ },
1130
+ {
1131
+ "epoch": 1.94,
1132
+ "learning_rate": 4.028213166144201e-05,
1133
+ "loss": 0.0231,
1134
+ "step": 1860
1135
+ },
1136
+ {
1137
+ "epoch": 1.95,
1138
+ "learning_rate": 4.0229885057471265e-05,
1139
+ "loss": 0.0288,
1140
+ "step": 1870
1141
+ },
1142
+ {
1143
+ "epoch": 1.96,
1144
+ "learning_rate": 4.017763845350053e-05,
1145
+ "loss": 0.0189,
1146
+ "step": 1880
1147
+ },
1148
+ {
1149
+ "epoch": 1.97,
1150
+ "learning_rate": 4.012539184952978e-05,
1151
+ "loss": 0.0229,
1152
+ "step": 1890
1153
+ },
1154
+ {
1155
+ "epoch": 1.99,
1156
+ "learning_rate": 4.007314524555904e-05,
1157
+ "loss": 0.0171,
1158
+ "step": 1900
1159
+ },
1160
+ {
1161
+ "epoch": 2.0,
1162
+ "learning_rate": 4.00208986415883e-05,
1163
+ "loss": 0.02,
1164
+ "step": 1910
1165
+ },
1166
+ {
1167
+ "epoch": 2.01,
1168
+ "learning_rate": 3.996865203761756e-05,
1169
+ "loss": 0.014,
1170
+ "step": 1920
1171
+ },
1172
+ {
1173
+ "epoch": 2.02,
1174
+ "learning_rate": 3.9916405433646815e-05,
1175
+ "loss": 0.0164,
1176
+ "step": 1930
1177
+ },
1178
+ {
1179
+ "epoch": 2.03,
1180
+ "learning_rate": 3.986415882967607e-05,
1181
+ "loss": 0.0148,
1182
+ "step": 1940
1183
+ },
1184
+ {
1185
+ "epoch": 2.04,
1186
+ "learning_rate": 3.9811912225705336e-05,
1187
+ "loss": 0.0135,
1188
+ "step": 1950
1189
+ },
1190
+ {
1191
+ "epoch": 2.05,
1192
+ "learning_rate": 3.9759665621734586e-05,
1193
+ "loss": 0.015,
1194
+ "step": 1960
1195
+ },
1196
+ {
1197
+ "epoch": 2.06,
1198
+ "learning_rate": 3.970741901776385e-05,
1199
+ "loss": 0.0183,
1200
+ "step": 1970
1201
+ },
1202
+ {
1203
+ "epoch": 2.07,
1204
+ "learning_rate": 3.965517241379311e-05,
1205
+ "loss": 0.0105,
1206
+ "step": 1980
1207
+ },
1208
+ {
1209
+ "epoch": 2.08,
1210
+ "learning_rate": 3.960292580982236e-05,
1211
+ "loss": 0.0152,
1212
+ "step": 1990
1213
+ },
1214
+ {
1215
+ "epoch": 2.09,
1216
+ "learning_rate": 3.955067920585162e-05,
1217
+ "loss": 0.0063,
1218
+ "step": 2000
1219
+ },
1220
+ {
1221
+ "epoch": 2.09,
1222
+ "eval_acc": 0.660049626153723,
1223
+ "eval_cer": 0.04602510460251046,
1224
+ "eval_loss": 0.023699596524238586,
1225
+ "eval_runtime": 127.5932,
1226
+ "eval_samples_per_second": 3.158,
1227
+ "eval_steps_per_second": 0.4,
1228
+ "step": 2000
1229
+ },
1230
+ {
1231
+ "epoch": 2.1,
1232
+ "learning_rate": 3.949843260188088e-05,
1233
+ "loss": 0.0147,
1234
+ "step": 2010
1235
+ },
1236
+ {
1237
+ "epoch": 2.11,
1238
+ "learning_rate": 3.944618599791014e-05,
1239
+ "loss": 0.0126,
1240
+ "step": 2020
1241
+ },
1242
+ {
1243
+ "epoch": 2.12,
1244
+ "learning_rate": 3.939393939393939e-05,
1245
+ "loss": 0.0143,
1246
+ "step": 2030
1247
+ },
1248
+ {
1249
+ "epoch": 2.13,
1250
+ "learning_rate": 3.934169278996865e-05,
1251
+ "loss": 0.0147,
1252
+ "step": 2040
1253
+ },
1254
+ {
1255
+ "epoch": 2.14,
1256
+ "learning_rate": 3.9289446185997915e-05,
1257
+ "loss": 0.0134,
1258
+ "step": 2050
1259
+ },
1260
+ {
1261
+ "epoch": 2.15,
1262
+ "learning_rate": 3.9237199582027165e-05,
1263
+ "loss": 0.013,
1264
+ "step": 2060
1265
+ },
1266
+ {
1267
+ "epoch": 2.16,
1268
+ "learning_rate": 3.918495297805643e-05,
1269
+ "loss": 0.0217,
1270
+ "step": 2070
1271
+ },
1272
+ {
1273
+ "epoch": 2.17,
1274
+ "learning_rate": 3.9132706374085686e-05,
1275
+ "loss": 0.0174,
1276
+ "step": 2080
1277
+ },
1278
+ {
1279
+ "epoch": 2.18,
1280
+ "learning_rate": 3.908045977011495e-05,
1281
+ "loss": 0.0116,
1282
+ "step": 2090
1283
+ },
1284
+ {
1285
+ "epoch": 2.19,
1286
+ "learning_rate": 3.90282131661442e-05,
1287
+ "loss": 0.0169,
1288
+ "step": 2100
1289
+ },
1290
+ {
1291
+ "epoch": 2.2,
1292
+ "learning_rate": 3.897596656217346e-05,
1293
+ "loss": 0.0199,
1294
+ "step": 2110
1295
+ },
1296
+ {
1297
+ "epoch": 2.22,
1298
+ "learning_rate": 3.892371995820272e-05,
1299
+ "loss": 0.0124,
1300
+ "step": 2120
1301
+ },
1302
+ {
1303
+ "epoch": 2.23,
1304
+ "learning_rate": 3.887147335423197e-05,
1305
+ "loss": 0.0114,
1306
+ "step": 2130
1307
+ },
1308
+ {
1309
+ "epoch": 2.24,
1310
+ "learning_rate": 3.8819226750261236e-05,
1311
+ "loss": 0.0128,
1312
+ "step": 2140
1313
+ },
1314
+ {
1315
+ "epoch": 2.25,
1316
+ "learning_rate": 3.876698014629049e-05,
1317
+ "loss": 0.0097,
1318
+ "step": 2150
1319
+ },
1320
+ {
1321
+ "epoch": 2.26,
1322
+ "learning_rate": 3.871473354231975e-05,
1323
+ "loss": 0.0132,
1324
+ "step": 2160
1325
+ },
1326
+ {
1327
+ "epoch": 2.27,
1328
+ "learning_rate": 3.866248693834901e-05,
1329
+ "loss": 0.0133,
1330
+ "step": 2170
1331
+ },
1332
+ {
1333
+ "epoch": 2.28,
1334
+ "learning_rate": 3.8610240334378265e-05,
1335
+ "loss": 0.0094,
1336
+ "step": 2180
1337
+ },
1338
+ {
1339
+ "epoch": 2.29,
1340
+ "learning_rate": 3.855799373040753e-05,
1341
+ "loss": 0.0185,
1342
+ "step": 2190
1343
+ },
1344
+ {
1345
+ "epoch": 2.3,
1346
+ "learning_rate": 3.850574712643678e-05,
1347
+ "loss": 0.017,
1348
+ "step": 2200
1349
+ },
1350
+ {
1351
+ "epoch": 2.31,
1352
+ "learning_rate": 3.845350052246604e-05,
1353
+ "loss": 0.0128,
1354
+ "step": 2210
1355
+ },
1356
+ {
1357
+ "epoch": 2.32,
1358
+ "learning_rate": 3.84012539184953e-05,
1359
+ "loss": 0.0121,
1360
+ "step": 2220
1361
+ },
1362
+ {
1363
+ "epoch": 2.33,
1364
+ "learning_rate": 3.834900731452456e-05,
1365
+ "loss": 0.015,
1366
+ "step": 2230
1367
+ },
1368
+ {
1369
+ "epoch": 2.34,
1370
+ "learning_rate": 3.8296760710553815e-05,
1371
+ "loss": 0.0077,
1372
+ "step": 2240
1373
+ },
1374
+ {
1375
+ "epoch": 2.35,
1376
+ "learning_rate": 3.824451410658307e-05,
1377
+ "loss": 0.0118,
1378
+ "step": 2250
1379
+ },
1380
+ {
1381
+ "epoch": 2.36,
1382
+ "learning_rate": 3.8192267502612336e-05,
1383
+ "loss": 0.0133,
1384
+ "step": 2260
1385
+ },
1386
+ {
1387
+ "epoch": 2.37,
1388
+ "learning_rate": 3.8140020898641586e-05,
1389
+ "loss": 0.0156,
1390
+ "step": 2270
1391
+ },
1392
+ {
1393
+ "epoch": 2.38,
1394
+ "learning_rate": 3.808777429467085e-05,
1395
+ "loss": 0.0131,
1396
+ "step": 2280
1397
+ },
1398
+ {
1399
+ "epoch": 2.39,
1400
+ "learning_rate": 3.803552769070011e-05,
1401
+ "loss": 0.01,
1402
+ "step": 2290
1403
+ },
1404
+ {
1405
+ "epoch": 2.4,
1406
+ "learning_rate": 3.7983281086729364e-05,
1407
+ "loss": 0.0166,
1408
+ "step": 2300
1409
+ },
1410
+ {
1411
+ "epoch": 2.41,
1412
+ "learning_rate": 3.793103448275862e-05,
1413
+ "loss": 0.0125,
1414
+ "step": 2310
1415
+ },
1416
+ {
1417
+ "epoch": 2.42,
1418
+ "learning_rate": 3.787878787878788e-05,
1419
+ "loss": 0.0092,
1420
+ "step": 2320
1421
+ },
1422
+ {
1423
+ "epoch": 2.43,
1424
+ "learning_rate": 3.782654127481714e-05,
1425
+ "loss": 0.0105,
1426
+ "step": 2330
1427
+ },
1428
+ {
1429
+ "epoch": 2.45,
1430
+ "learning_rate": 3.777429467084639e-05,
1431
+ "loss": 0.0138,
1432
+ "step": 2340
1433
+ },
1434
+ {
1435
+ "epoch": 2.46,
1436
+ "learning_rate": 3.772204806687566e-05,
1437
+ "loss": 0.0136,
1438
+ "step": 2350
1439
+ },
1440
+ {
1441
+ "epoch": 2.47,
1442
+ "learning_rate": 3.7669801462904914e-05,
1443
+ "loss": 0.0153,
1444
+ "step": 2360
1445
+ },
1446
+ {
1447
+ "epoch": 2.48,
1448
+ "learning_rate": 3.761755485893417e-05,
1449
+ "loss": 0.0094,
1450
+ "step": 2370
1451
+ },
1452
+ {
1453
+ "epoch": 2.49,
1454
+ "learning_rate": 3.756530825496343e-05,
1455
+ "loss": 0.0081,
1456
+ "step": 2380
1457
+ },
1458
+ {
1459
+ "epoch": 2.5,
1460
+ "learning_rate": 3.7513061650992686e-05,
1461
+ "loss": 0.0169,
1462
+ "step": 2390
1463
+ },
1464
+ {
1465
+ "epoch": 2.51,
1466
+ "learning_rate": 3.746081504702195e-05,
1467
+ "loss": 0.0086,
1468
+ "step": 2400
1469
+ },
1470
+ {
1471
+ "epoch": 2.52,
1472
+ "learning_rate": 3.740856844305121e-05,
1473
+ "loss": 0.0124,
1474
+ "step": 2410
1475
+ },
1476
+ {
1477
+ "epoch": 2.53,
1478
+ "learning_rate": 3.735632183908046e-05,
1479
+ "loss": 0.0143,
1480
+ "step": 2420
1481
+ },
1482
+ {
1483
+ "epoch": 2.54,
1484
+ "learning_rate": 3.730407523510972e-05,
1485
+ "loss": 0.0092,
1486
+ "step": 2430
1487
+ },
1488
+ {
1489
+ "epoch": 2.55,
1490
+ "learning_rate": 3.725182863113898e-05,
1491
+ "loss": 0.0147,
1492
+ "step": 2440
1493
+ },
1494
+ {
1495
+ "epoch": 2.56,
1496
+ "learning_rate": 3.7199582027168236e-05,
1497
+ "loss": 0.009,
1498
+ "step": 2450
1499
+ },
1500
+ {
1501
+ "epoch": 2.57,
1502
+ "learning_rate": 3.714733542319749e-05,
1503
+ "loss": 0.0083,
1504
+ "step": 2460
1505
+ },
1506
+ {
1507
+ "epoch": 2.58,
1508
+ "learning_rate": 3.709508881922675e-05,
1509
+ "loss": 0.0146,
1510
+ "step": 2470
1511
+ },
1512
+ {
1513
+ "epoch": 2.59,
1514
+ "learning_rate": 3.7042842215256014e-05,
1515
+ "loss": 0.0142,
1516
+ "step": 2480
1517
+ },
1518
+ {
1519
+ "epoch": 2.6,
1520
+ "learning_rate": 3.6990595611285264e-05,
1521
+ "loss": 0.0097,
1522
+ "step": 2490
1523
+ },
1524
+ {
1525
+ "epoch": 2.61,
1526
+ "learning_rate": 3.693834900731453e-05,
1527
+ "loss": 0.0067,
1528
+ "step": 2500
1529
+ },
1530
+ {
1531
+ "epoch": 2.62,
1532
+ "learning_rate": 3.6886102403343786e-05,
1533
+ "loss": 0.0085,
1534
+ "step": 2510
1535
+ },
1536
+ {
1537
+ "epoch": 2.63,
1538
+ "learning_rate": 3.683385579937304e-05,
1539
+ "loss": 0.01,
1540
+ "step": 2520
1541
+ },
1542
+ {
1543
+ "epoch": 2.64,
1544
+ "learning_rate": 3.67816091954023e-05,
1545
+ "loss": 0.0128,
1546
+ "step": 2530
1547
+ },
1548
+ {
1549
+ "epoch": 2.65,
1550
+ "learning_rate": 3.672936259143156e-05,
1551
+ "loss": 0.0113,
1552
+ "step": 2540
1553
+ },
1554
+ {
1555
+ "epoch": 2.66,
1556
+ "learning_rate": 3.667711598746082e-05,
1557
+ "loss": 0.0098,
1558
+ "step": 2550
1559
+ },
1560
+ {
1561
+ "epoch": 2.68,
1562
+ "learning_rate": 3.662486938349007e-05,
1563
+ "loss": 0.0159,
1564
+ "step": 2560
1565
+ },
1566
+ {
1567
+ "epoch": 2.69,
1568
+ "learning_rate": 3.6572622779519335e-05,
1569
+ "loss": 0.0188,
1570
+ "step": 2570
1571
+ },
1572
+ {
1573
+ "epoch": 2.7,
1574
+ "learning_rate": 3.652037617554859e-05,
1575
+ "loss": 0.0149,
1576
+ "step": 2580
1577
+ },
1578
+ {
1579
+ "epoch": 2.71,
1580
+ "learning_rate": 3.646812957157785e-05,
1581
+ "loss": 0.0159,
1582
+ "step": 2590
1583
+ },
1584
+ {
1585
+ "epoch": 2.72,
1586
+ "learning_rate": 3.641588296760711e-05,
1587
+ "loss": 0.0093,
1588
+ "step": 2600
1589
+ },
1590
+ {
1591
+ "epoch": 2.73,
1592
+ "learning_rate": 3.6363636363636364e-05,
1593
+ "loss": 0.0115,
1594
+ "step": 2610
1595
+ },
1596
+ {
1597
+ "epoch": 2.74,
1598
+ "learning_rate": 3.631138975966563e-05,
1599
+ "loss": 0.0138,
1600
+ "step": 2620
1601
+ },
1602
+ {
1603
+ "epoch": 2.75,
1604
+ "learning_rate": 3.625914315569488e-05,
1605
+ "loss": 0.011,
1606
+ "step": 2630
1607
+ },
1608
+ {
1609
+ "epoch": 2.76,
1610
+ "learning_rate": 3.620689655172414e-05,
1611
+ "loss": 0.0144,
1612
+ "step": 2640
1613
+ },
1614
+ {
1615
+ "epoch": 2.77,
1616
+ "learning_rate": 3.61546499477534e-05,
1617
+ "loss": 0.0098,
1618
+ "step": 2650
1619
+ },
1620
+ {
1621
+ "epoch": 2.78,
1622
+ "learning_rate": 3.610240334378266e-05,
1623
+ "loss": 0.0132,
1624
+ "step": 2660
1625
+ },
1626
+ {
1627
+ "epoch": 2.79,
1628
+ "learning_rate": 3.6050156739811914e-05,
1629
+ "loss": 0.0125,
1630
+ "step": 2670
1631
+ },
1632
+ {
1633
+ "epoch": 2.8,
1634
+ "learning_rate": 3.599791013584117e-05,
1635
+ "loss": 0.0118,
1636
+ "step": 2680
1637
+ },
1638
+ {
1639
+ "epoch": 2.81,
1640
+ "learning_rate": 3.5945663531870435e-05,
1641
+ "loss": 0.0131,
1642
+ "step": 2690
1643
+ },
1644
+ {
1645
+ "epoch": 2.82,
1646
+ "learning_rate": 3.5893416927899686e-05,
1647
+ "loss": 0.0081,
1648
+ "step": 2700
1649
+ },
1650
+ {
1651
+ "epoch": 2.83,
1652
+ "learning_rate": 3.584117032392895e-05,
1653
+ "loss": 0.0115,
1654
+ "step": 2710
1655
+ },
1656
+ {
1657
+ "epoch": 2.84,
1658
+ "learning_rate": 3.578892371995821e-05,
1659
+ "loss": 0.0109,
1660
+ "step": 2720
1661
+ },
1662
+ {
1663
+ "epoch": 2.85,
1664
+ "learning_rate": 3.573667711598746e-05,
1665
+ "loss": 0.0149,
1666
+ "step": 2730
1667
+ },
1668
+ {
1669
+ "epoch": 2.86,
1670
+ "learning_rate": 3.568443051201672e-05,
1671
+ "loss": 0.0093,
1672
+ "step": 2740
1673
+ },
1674
+ {
1675
+ "epoch": 2.87,
1676
+ "learning_rate": 3.563218390804598e-05,
1677
+ "loss": 0.0124,
1678
+ "step": 2750
1679
+ },
1680
+ {
1681
+ "epoch": 2.88,
1682
+ "learning_rate": 3.557993730407524e-05,
1683
+ "loss": 0.0094,
1684
+ "step": 2760
1685
+ },
1686
+ {
1687
+ "epoch": 2.89,
1688
+ "learning_rate": 3.552769070010449e-05,
1689
+ "loss": 0.0105,
1690
+ "step": 2770
1691
+ },
1692
+ {
1693
+ "epoch": 2.9,
1694
+ "learning_rate": 3.547544409613375e-05,
1695
+ "loss": 0.0221,
1696
+ "step": 2780
1697
+ },
1698
+ {
1699
+ "epoch": 2.92,
1700
+ "learning_rate": 3.5423197492163014e-05,
1701
+ "loss": 0.0088,
1702
+ "step": 2790
1703
+ },
1704
+ {
1705
+ "epoch": 2.93,
1706
+ "learning_rate": 3.5370950888192264e-05,
1707
+ "loss": 0.0077,
1708
+ "step": 2800
1709
+ },
1710
+ {
1711
+ "epoch": 2.94,
1712
+ "learning_rate": 3.531870428422153e-05,
1713
+ "loss": 0.0106,
1714
+ "step": 2810
1715
+ },
1716
+ {
1717
+ "epoch": 2.95,
1718
+ "learning_rate": 3.5266457680250785e-05,
1719
+ "loss": 0.0113,
1720
+ "step": 2820
1721
+ },
1722
+ {
1723
+ "epoch": 2.96,
1724
+ "learning_rate": 3.521421107628005e-05,
1725
+ "loss": 0.012,
1726
+ "step": 2830
1727
+ },
1728
+ {
1729
+ "epoch": 2.97,
1730
+ "learning_rate": 3.51619644723093e-05,
1731
+ "loss": 0.0094,
1732
+ "step": 2840
1733
+ },
1734
+ {
1735
+ "epoch": 2.98,
1736
+ "learning_rate": 3.510971786833856e-05,
1737
+ "loss": 0.0088,
1738
+ "step": 2850
1739
+ },
1740
+ {
1741
+ "epoch": 2.99,
1742
+ "learning_rate": 3.505747126436782e-05,
1743
+ "loss": 0.0162,
1744
+ "step": 2860
1745
+ },
1746
+ {
1747
+ "epoch": 3.0,
1748
+ "learning_rate": 3.500522466039707e-05,
1749
+ "loss": 0.0104,
1750
+ "step": 2870
1751
+ },
1752
+ {
1753
+ "epoch": 3.01,
1754
+ "learning_rate": 3.4952978056426335e-05,
1755
+ "loss": 0.0045,
1756
+ "step": 2880
1757
+ },
1758
+ {
1759
+ "epoch": 3.02,
1760
+ "learning_rate": 3.490073145245559e-05,
1761
+ "loss": 0.0076,
1762
+ "step": 2890
1763
+ },
1764
+ {
1765
+ "epoch": 3.03,
1766
+ "learning_rate": 3.484848484848485e-05,
1767
+ "loss": 0.0092,
1768
+ "step": 2900
1769
+ },
1770
+ {
1771
+ "epoch": 3.04,
1772
+ "learning_rate": 3.479623824451411e-05,
1773
+ "loss": 0.0088,
1774
+ "step": 2910
1775
+ },
1776
+ {
1777
+ "epoch": 3.05,
1778
+ "learning_rate": 3.4743991640543364e-05,
1779
+ "loss": 0.0068,
1780
+ "step": 2920
1781
+ },
1782
+ {
1783
+ "epoch": 3.06,
1784
+ "learning_rate": 3.469174503657263e-05,
1785
+ "loss": 0.0037,
1786
+ "step": 2930
1787
+ },
1788
+ {
1789
+ "epoch": 3.07,
1790
+ "learning_rate": 3.463949843260188e-05,
1791
+ "loss": 0.008,
1792
+ "step": 2940
1793
+ },
1794
+ {
1795
+ "epoch": 3.08,
1796
+ "learning_rate": 3.458725182863114e-05,
1797
+ "loss": 0.0053,
1798
+ "step": 2950
1799
+ },
1800
+ {
1801
+ "epoch": 3.09,
1802
+ "learning_rate": 3.45350052246604e-05,
1803
+ "loss": 0.0085,
1804
+ "step": 2960
1805
+ },
1806
+ {
1807
+ "epoch": 3.1,
1808
+ "learning_rate": 3.4482758620689657e-05,
1809
+ "loss": 0.0122,
1810
+ "step": 2970
1811
+ },
1812
+ {
1813
+ "epoch": 3.11,
1814
+ "learning_rate": 3.4430512016718914e-05,
1815
+ "loss": 0.0096,
1816
+ "step": 2980
1817
+ },
1818
+ {
1819
+ "epoch": 3.12,
1820
+ "learning_rate": 3.437826541274817e-05,
1821
+ "loss": 0.0079,
1822
+ "step": 2990
1823
+ },
1824
+ {
1825
+ "epoch": 3.13,
1826
+ "learning_rate": 3.4326018808777435e-05,
1827
+ "loss": 0.01,
1828
+ "step": 3000
1829
+ },
1830
+ {
1831
+ "epoch": 3.13,
1832
+ "eval_acc": 0.7444168716019433,
1833
+ "eval_cer": 0.031845653184565316,
1834
+ "eval_loss": 0.017891723662614822,
1835
+ "eval_runtime": 127.5496,
1836
+ "eval_samples_per_second": 3.16,
1837
+ "eval_steps_per_second": 0.4,
1838
+ "step": 3000
1839
+ },
1840
+ {
1841
+ "epoch": 3.15,
1842
+ "learning_rate": 3.4273772204806685e-05,
1843
+ "loss": 0.0105,
1844
+ "step": 3010
1845
+ },
1846
+ {
1847
+ "epoch": 3.16,
1848
+ "learning_rate": 3.422152560083595e-05,
1849
+ "loss": 0.0091,
1850
+ "step": 3020
1851
+ },
1852
+ {
1853
+ "epoch": 3.17,
1854
+ "learning_rate": 3.4169278996865206e-05,
1855
+ "loss": 0.005,
1856
+ "step": 3030
1857
+ },
1858
+ {
1859
+ "epoch": 3.18,
1860
+ "learning_rate": 3.4117032392894464e-05,
1861
+ "loss": 0.0069,
1862
+ "step": 3040
1863
+ },
1864
+ {
1865
+ "epoch": 3.19,
1866
+ "learning_rate": 3.406478578892372e-05,
1867
+ "loss": 0.0066,
1868
+ "step": 3050
1869
+ },
1870
+ {
1871
+ "epoch": 3.2,
1872
+ "learning_rate": 3.401253918495298e-05,
1873
+ "loss": 0.0065,
1874
+ "step": 3060
1875
+ },
1876
+ {
1877
+ "epoch": 3.21,
1878
+ "learning_rate": 3.396029258098224e-05,
1879
+ "loss": 0.0098,
1880
+ "step": 3070
1881
+ },
1882
+ {
1883
+ "epoch": 3.22,
1884
+ "learning_rate": 3.390804597701149e-05,
1885
+ "loss": 0.0089,
1886
+ "step": 3080
1887
+ },
1888
+ {
1889
+ "epoch": 3.23,
1890
+ "learning_rate": 3.3855799373040756e-05,
1891
+ "loss": 0.0082,
1892
+ "step": 3090
1893
+ },
1894
+ {
1895
+ "epoch": 3.24,
1896
+ "learning_rate": 3.3803552769070014e-05,
1897
+ "loss": 0.0089,
1898
+ "step": 3100
1899
+ },
1900
+ {
1901
+ "epoch": 3.25,
1902
+ "learning_rate": 3.375130616509927e-05,
1903
+ "loss": 0.0097,
1904
+ "step": 3110
1905
+ },
1906
+ {
1907
+ "epoch": 3.26,
1908
+ "learning_rate": 3.369905956112853e-05,
1909
+ "loss": 0.0079,
1910
+ "step": 3120
1911
+ },
1912
+ {
1913
+ "epoch": 3.27,
1914
+ "learning_rate": 3.3646812957157785e-05,
1915
+ "loss": 0.0079,
1916
+ "step": 3130
1917
+ },
1918
+ {
1919
+ "epoch": 3.28,
1920
+ "learning_rate": 3.359456635318705e-05,
1921
+ "loss": 0.0062,
1922
+ "step": 3140
1923
+ },
1924
+ {
1925
+ "epoch": 3.29,
1926
+ "learning_rate": 3.35423197492163e-05,
1927
+ "loss": 0.0095,
1928
+ "step": 3150
1929
+ },
1930
+ {
1931
+ "epoch": 3.3,
1932
+ "learning_rate": 3.3490073145245557e-05,
1933
+ "loss": 0.0089,
1934
+ "step": 3160
1935
+ },
1936
+ {
1937
+ "epoch": 3.31,
1938
+ "learning_rate": 3.343782654127482e-05,
1939
+ "loss": 0.0107,
1940
+ "step": 3170
1941
+ },
1942
+ {
1943
+ "epoch": 3.32,
1944
+ "learning_rate": 3.338557993730408e-05,
1945
+ "loss": 0.0135,
1946
+ "step": 3180
1947
+ },
1948
+ {
1949
+ "epoch": 3.33,
1950
+ "learning_rate": 3.3333333333333335e-05,
1951
+ "loss": 0.006,
1952
+ "step": 3190
1953
+ },
1954
+ {
1955
+ "epoch": 3.34,
1956
+ "learning_rate": 3.328108672936259e-05,
1957
+ "loss": 0.0078,
1958
+ "step": 3200
1959
+ },
1960
+ {
1961
+ "epoch": 3.35,
1962
+ "learning_rate": 3.322884012539185e-05,
1963
+ "loss": 0.005,
1964
+ "step": 3210
1965
+ },
1966
+ {
1967
+ "epoch": 3.36,
1968
+ "learning_rate": 3.3176593521421106e-05,
1969
+ "loss": 0.0073,
1970
+ "step": 3220
1971
+ },
1972
+ {
1973
+ "epoch": 3.38,
1974
+ "learning_rate": 3.3124346917450364e-05,
1975
+ "loss": 0.0084,
1976
+ "step": 3230
1977
+ },
1978
+ {
1979
+ "epoch": 3.39,
1980
+ "learning_rate": 3.307210031347963e-05,
1981
+ "loss": 0.0077,
1982
+ "step": 3240
1983
+ },
1984
+ {
1985
+ "epoch": 3.4,
1986
+ "learning_rate": 3.3019853709508885e-05,
1987
+ "loss": 0.0065,
1988
+ "step": 3250
1989
+ },
1990
+ {
1991
+ "epoch": 3.41,
1992
+ "learning_rate": 3.296760710553814e-05,
1993
+ "loss": 0.0052,
1994
+ "step": 3260
1995
+ },
1996
+ {
1997
+ "epoch": 3.42,
1998
+ "learning_rate": 3.29153605015674e-05,
1999
+ "loss": 0.0078,
2000
+ "step": 3270
2001
+ },
2002
+ {
2003
+ "epoch": 3.43,
2004
+ "learning_rate": 3.2863113897596656e-05,
2005
+ "loss": 0.0056,
2006
+ "step": 3280
2007
+ },
2008
+ {
2009
+ "epoch": 3.44,
2010
+ "learning_rate": 3.2810867293625914e-05,
2011
+ "loss": 0.0084,
2012
+ "step": 3290
2013
+ },
2014
+ {
2015
+ "epoch": 3.45,
2016
+ "learning_rate": 3.275862068965517e-05,
2017
+ "loss": 0.0077,
2018
+ "step": 3300
2019
+ },
2020
+ {
2021
+ "epoch": 3.46,
2022
+ "learning_rate": 3.2706374085684435e-05,
2023
+ "loss": 0.0071,
2024
+ "step": 3310
2025
+ },
2026
+ {
2027
+ "epoch": 3.47,
2028
+ "learning_rate": 3.265412748171369e-05,
2029
+ "loss": 0.0084,
2030
+ "step": 3320
2031
+ },
2032
+ {
2033
+ "epoch": 3.48,
2034
+ "learning_rate": 3.260188087774295e-05,
2035
+ "loss": 0.0079,
2036
+ "step": 3330
2037
+ },
2038
+ {
2039
+ "epoch": 3.49,
2040
+ "learning_rate": 3.2549634273772206e-05,
2041
+ "loss": 0.0058,
2042
+ "step": 3340
2043
+ },
2044
+ {
2045
+ "epoch": 3.5,
2046
+ "learning_rate": 3.2497387669801463e-05,
2047
+ "loss": 0.0074,
2048
+ "step": 3350
2049
+ },
2050
+ {
2051
+ "epoch": 3.51,
2052
+ "learning_rate": 3.244514106583073e-05,
2053
+ "loss": 0.007,
2054
+ "step": 3360
2055
+ },
2056
+ {
2057
+ "epoch": 3.52,
2058
+ "learning_rate": 3.239289446185998e-05,
2059
+ "loss": 0.0089,
2060
+ "step": 3370
2061
+ },
2062
+ {
2063
+ "epoch": 3.53,
2064
+ "learning_rate": 3.234064785788924e-05,
2065
+ "loss": 0.0059,
2066
+ "step": 3380
2067
+ },
2068
+ {
2069
+ "epoch": 3.54,
2070
+ "learning_rate": 3.22884012539185e-05,
2071
+ "loss": 0.0082,
2072
+ "step": 3390
2073
+ },
2074
+ {
2075
+ "epoch": 3.55,
2076
+ "learning_rate": 3.2236154649947756e-05,
2077
+ "loss": 0.0107,
2078
+ "step": 3400
2079
+ },
2080
+ {
2081
+ "epoch": 3.56,
2082
+ "learning_rate": 3.218390804597701e-05,
2083
+ "loss": 0.0063,
2084
+ "step": 3410
2085
+ },
2086
+ {
2087
+ "epoch": 3.57,
2088
+ "learning_rate": 3.213166144200627e-05,
2089
+ "loss": 0.0059,
2090
+ "step": 3420
2091
+ },
2092
+ {
2093
+ "epoch": 3.58,
2094
+ "learning_rate": 3.2079414838035534e-05,
2095
+ "loss": 0.0074,
2096
+ "step": 3430
2097
+ },
2098
+ {
2099
+ "epoch": 3.59,
2100
+ "learning_rate": 3.2027168234064785e-05,
2101
+ "loss": 0.011,
2102
+ "step": 3440
2103
+ },
2104
+ {
2105
+ "epoch": 3.61,
2106
+ "learning_rate": 3.197492163009405e-05,
2107
+ "loss": 0.0107,
2108
+ "step": 3450
2109
+ },
2110
+ {
2111
+ "epoch": 3.62,
2112
+ "learning_rate": 3.1922675026123306e-05,
2113
+ "loss": 0.0071,
2114
+ "step": 3460
2115
+ },
2116
+ {
2117
+ "epoch": 3.63,
2118
+ "learning_rate": 3.1870428422152556e-05,
2119
+ "loss": 0.0073,
2120
+ "step": 3470
2121
+ },
2122
+ {
2123
+ "epoch": 3.64,
2124
+ "learning_rate": 3.181818181818182e-05,
2125
+ "loss": 0.0079,
2126
+ "step": 3480
2127
+ },
2128
+ {
2129
+ "epoch": 3.65,
2130
+ "learning_rate": 3.176593521421108e-05,
2131
+ "loss": 0.0068,
2132
+ "step": 3490
2133
+ },
2134
+ {
2135
+ "epoch": 3.66,
2136
+ "learning_rate": 3.171368861024034e-05,
2137
+ "loss": 0.0085,
2138
+ "step": 3500
2139
+ },
2140
+ {
2141
+ "epoch": 3.67,
2142
+ "learning_rate": 3.166144200626959e-05,
2143
+ "loss": 0.0087,
2144
+ "step": 3510
2145
+ },
2146
+ {
2147
+ "epoch": 3.68,
2148
+ "learning_rate": 3.160919540229885e-05,
2149
+ "loss": 0.005,
2150
+ "step": 3520
2151
+ },
2152
+ {
2153
+ "epoch": 3.69,
2154
+ "learning_rate": 3.155694879832811e-05,
2155
+ "loss": 0.0051,
2156
+ "step": 3530
2157
+ },
2158
+ {
2159
+ "epoch": 3.7,
2160
+ "learning_rate": 3.1504702194357363e-05,
2161
+ "loss": 0.0076,
2162
+ "step": 3540
2163
+ },
2164
+ {
2165
+ "epoch": 3.71,
2166
+ "learning_rate": 3.145245559038663e-05,
2167
+ "loss": 0.0054,
2168
+ "step": 3550
2169
+ },
2170
+ {
2171
+ "epoch": 3.72,
2172
+ "learning_rate": 3.1400208986415885e-05,
2173
+ "loss": 0.0083,
2174
+ "step": 3560
2175
+ },
2176
+ {
2177
+ "epoch": 3.73,
2178
+ "learning_rate": 3.134796238244515e-05,
2179
+ "loss": 0.0076,
2180
+ "step": 3570
2181
+ },
2182
+ {
2183
+ "epoch": 3.74,
2184
+ "learning_rate": 3.12957157784744e-05,
2185
+ "loss": 0.0094,
2186
+ "step": 3580
2187
+ },
2188
+ {
2189
+ "epoch": 3.75,
2190
+ "learning_rate": 3.1243469174503656e-05,
2191
+ "loss": 0.0083,
2192
+ "step": 3590
2193
+ },
2194
+ {
2195
+ "epoch": 3.76,
2196
+ "learning_rate": 3.119122257053292e-05,
2197
+ "loss": 0.0073,
2198
+ "step": 3600
2199
+ },
2200
+ {
2201
+ "epoch": 3.77,
2202
+ "learning_rate": 3.113897596656217e-05,
2203
+ "loss": 0.0063,
2204
+ "step": 3610
2205
+ },
2206
+ {
2207
+ "epoch": 3.78,
2208
+ "learning_rate": 3.1086729362591434e-05,
2209
+ "loss": 0.0086,
2210
+ "step": 3620
2211
+ },
2212
+ {
2213
+ "epoch": 3.79,
2214
+ "learning_rate": 3.103448275862069e-05,
2215
+ "loss": 0.0118,
2216
+ "step": 3630
2217
+ },
2218
+ {
2219
+ "epoch": 3.8,
2220
+ "learning_rate": 3.098223615464995e-05,
2221
+ "loss": 0.0088,
2222
+ "step": 3640
2223
+ },
2224
+ {
2225
+ "epoch": 3.81,
2226
+ "learning_rate": 3.0929989550679206e-05,
2227
+ "loss": 0.0064,
2228
+ "step": 3650
2229
+ },
2230
+ {
2231
+ "epoch": 3.82,
2232
+ "learning_rate": 3.087774294670846e-05,
2233
+ "loss": 0.0065,
2234
+ "step": 3660
2235
+ },
2236
+ {
2237
+ "epoch": 3.83,
2238
+ "learning_rate": 3.082549634273773e-05,
2239
+ "loss": 0.0057,
2240
+ "step": 3670
2241
+ },
2242
+ {
2243
+ "epoch": 3.85,
2244
+ "learning_rate": 3.077324973876698e-05,
2245
+ "loss": 0.0129,
2246
+ "step": 3680
2247
+ },
2248
+ {
2249
+ "epoch": 3.86,
2250
+ "learning_rate": 3.072100313479624e-05,
2251
+ "loss": 0.0119,
2252
+ "step": 3690
2253
+ },
2254
+ {
2255
+ "epoch": 3.87,
2256
+ "learning_rate": 3.06687565308255e-05,
2257
+ "loss": 0.0074,
2258
+ "step": 3700
2259
+ },
2260
+ {
2261
+ "epoch": 3.88,
2262
+ "learning_rate": 3.0616509926854756e-05,
2263
+ "loss": 0.0058,
2264
+ "step": 3710
2265
+ },
2266
+ {
2267
+ "epoch": 3.89,
2268
+ "learning_rate": 3.056426332288401e-05,
2269
+ "loss": 0.0049,
2270
+ "step": 3720
2271
+ },
2272
+ {
2273
+ "epoch": 3.9,
2274
+ "learning_rate": 3.0512016718913274e-05,
2275
+ "loss": 0.0075,
2276
+ "step": 3730
2277
+ },
2278
+ {
2279
+ "epoch": 3.91,
2280
+ "learning_rate": 3.045977011494253e-05,
2281
+ "loss": 0.0087,
2282
+ "step": 3740
2283
+ },
2284
+ {
2285
+ "epoch": 3.92,
2286
+ "learning_rate": 3.0407523510971785e-05,
2287
+ "loss": 0.0071,
2288
+ "step": 3750
2289
+ },
2290
+ {
2291
+ "epoch": 3.93,
2292
+ "learning_rate": 3.0355276907001045e-05,
2293
+ "loss": 0.0101,
2294
+ "step": 3760
2295
+ },
2296
+ {
2297
+ "epoch": 3.94,
2298
+ "learning_rate": 3.0303030303030306e-05,
2299
+ "loss": 0.008,
2300
+ "step": 3770
2301
+ },
2302
+ {
2303
+ "epoch": 3.95,
2304
+ "learning_rate": 3.0250783699059566e-05,
2305
+ "loss": 0.0077,
2306
+ "step": 3780
2307
+ },
2308
+ {
2309
+ "epoch": 3.96,
2310
+ "learning_rate": 3.019853709508882e-05,
2311
+ "loss": 0.0114,
2312
+ "step": 3790
2313
+ },
2314
+ {
2315
+ "epoch": 3.97,
2316
+ "learning_rate": 3.0146290491118077e-05,
2317
+ "loss": 0.0067,
2318
+ "step": 3800
2319
+ },
2320
+ {
2321
+ "epoch": 3.98,
2322
+ "learning_rate": 3.0094043887147338e-05,
2323
+ "loss": 0.0067,
2324
+ "step": 3810
2325
+ },
2326
+ {
2327
+ "epoch": 3.99,
2328
+ "learning_rate": 3.004179728317659e-05,
2329
+ "loss": 0.0099,
2330
+ "step": 3820
2331
+ },
2332
+ {
2333
+ "epoch": 4.0,
2334
+ "learning_rate": 2.9989550679205852e-05,
2335
+ "loss": 0.0078,
2336
+ "step": 3830
2337
+ },
2338
+ {
2339
+ "epoch": 4.01,
2340
+ "learning_rate": 2.9937304075235113e-05,
2341
+ "loss": 0.0042,
2342
+ "step": 3840
2343
+ },
2344
+ {
2345
+ "epoch": 4.02,
2346
+ "learning_rate": 2.988505747126437e-05,
2347
+ "loss": 0.0056,
2348
+ "step": 3850
2349
+ },
2350
+ {
2351
+ "epoch": 4.03,
2352
+ "learning_rate": 2.9832810867293627e-05,
2353
+ "loss": 0.0043,
2354
+ "step": 3860
2355
+ },
2356
+ {
2357
+ "epoch": 4.04,
2358
+ "learning_rate": 2.9780564263322884e-05,
2359
+ "loss": 0.0042,
2360
+ "step": 3870
2361
+ },
2362
+ {
2363
+ "epoch": 4.05,
2364
+ "learning_rate": 2.9728317659352145e-05,
2365
+ "loss": 0.0062,
2366
+ "step": 3880
2367
+ },
2368
+ {
2369
+ "epoch": 4.06,
2370
+ "learning_rate": 2.96760710553814e-05,
2371
+ "loss": 0.0057,
2372
+ "step": 3890
2373
+ },
2374
+ {
2375
+ "epoch": 4.08,
2376
+ "learning_rate": 2.962382445141066e-05,
2377
+ "loss": 0.0069,
2378
+ "step": 3900
2379
+ },
2380
+ {
2381
+ "epoch": 4.09,
2382
+ "learning_rate": 2.957157784743992e-05,
2383
+ "loss": 0.0072,
2384
+ "step": 3910
2385
+ },
2386
+ {
2387
+ "epoch": 4.1,
2388
+ "learning_rate": 2.9519331243469177e-05,
2389
+ "loss": 0.0072,
2390
+ "step": 3920
2391
+ },
2392
+ {
2393
+ "epoch": 4.11,
2394
+ "learning_rate": 2.946708463949843e-05,
2395
+ "loss": 0.0032,
2396
+ "step": 3930
2397
+ },
2398
+ {
2399
+ "epoch": 4.12,
2400
+ "learning_rate": 2.941483803552769e-05,
2401
+ "loss": 0.0089,
2402
+ "step": 3940
2403
+ },
2404
+ {
2405
+ "epoch": 4.13,
2406
+ "learning_rate": 2.9362591431556952e-05,
2407
+ "loss": 0.0043,
2408
+ "step": 3950
2409
+ },
2410
+ {
2411
+ "epoch": 4.14,
2412
+ "learning_rate": 2.9310344827586206e-05,
2413
+ "loss": 0.0063,
2414
+ "step": 3960
2415
+ },
2416
+ {
2417
+ "epoch": 4.15,
2418
+ "learning_rate": 2.9258098223615466e-05,
2419
+ "loss": 0.0045,
2420
+ "step": 3970
2421
+ },
2422
+ {
2423
+ "epoch": 4.16,
2424
+ "learning_rate": 2.9205851619644723e-05,
2425
+ "loss": 0.0058,
2426
+ "step": 3980
2427
+ },
2428
+ {
2429
+ "epoch": 4.17,
2430
+ "learning_rate": 2.9153605015673984e-05,
2431
+ "loss": 0.005,
2432
+ "step": 3990
2433
+ },
2434
+ {
2435
+ "epoch": 4.18,
2436
+ "learning_rate": 2.9101358411703238e-05,
2437
+ "loss": 0.0044,
2438
+ "step": 4000
2439
+ },
2440
+ {
2441
+ "epoch": 4.18,
2442
+ "eval_acc": 0.769230767322008,
2443
+ "eval_cer": 0.0299860529986053,
2444
+ "eval_loss": 0.01652107760310173,
2445
+ "eval_runtime": 126.6833,
2446
+ "eval_samples_per_second": 3.181,
2447
+ "eval_steps_per_second": 0.403,
2448
+ "step": 4000
2449
+ },
2450
+ {
2451
+ "epoch": 4.19,
2452
+ "learning_rate": 2.90491118077325e-05,
2453
+ "loss": 0.0068,
2454
+ "step": 4010
2455
+ },
2456
+ {
2457
+ "epoch": 4.2,
2458
+ "learning_rate": 2.899686520376176e-05,
2459
+ "loss": 0.0039,
2460
+ "step": 4020
2461
+ },
2462
+ {
2463
+ "epoch": 4.21,
2464
+ "learning_rate": 2.8944618599791013e-05,
2465
+ "loss": 0.01,
2466
+ "step": 4030
2467
+ },
2468
+ {
2469
+ "epoch": 4.22,
2470
+ "learning_rate": 2.8892371995820273e-05,
2471
+ "loss": 0.0046,
2472
+ "step": 4040
2473
+ },
2474
+ {
2475
+ "epoch": 4.23,
2476
+ "learning_rate": 2.884012539184953e-05,
2477
+ "loss": 0.0096,
2478
+ "step": 4050
2479
+ },
2480
+ {
2481
+ "epoch": 4.24,
2482
+ "learning_rate": 2.878787878787879e-05,
2483
+ "loss": 0.006,
2484
+ "step": 4060
2485
+ },
2486
+ {
2487
+ "epoch": 4.25,
2488
+ "learning_rate": 2.8735632183908045e-05,
2489
+ "loss": 0.0052,
2490
+ "step": 4070
2491
+ },
2492
+ {
2493
+ "epoch": 4.26,
2494
+ "learning_rate": 2.8683385579937305e-05,
2495
+ "loss": 0.005,
2496
+ "step": 4080
2497
+ },
2498
+ {
2499
+ "epoch": 4.27,
2500
+ "learning_rate": 2.8631138975966566e-05,
2501
+ "loss": 0.004,
2502
+ "step": 4090
2503
+ },
2504
+ {
2505
+ "epoch": 4.28,
2506
+ "learning_rate": 2.857889237199582e-05,
2507
+ "loss": 0.0045,
2508
+ "step": 4100
2509
+ },
2510
+ {
2511
+ "epoch": 4.29,
2512
+ "learning_rate": 2.8526645768025077e-05,
2513
+ "loss": 0.0037,
2514
+ "step": 4110
2515
+ },
2516
+ {
2517
+ "epoch": 4.31,
2518
+ "learning_rate": 2.8474399164054338e-05,
2519
+ "loss": 0.0125,
2520
+ "step": 4120
2521
+ },
2522
+ {
2523
+ "epoch": 4.32,
2524
+ "learning_rate": 2.8422152560083598e-05,
2525
+ "loss": 0.0057,
2526
+ "step": 4130
2527
+ },
2528
+ {
2529
+ "epoch": 4.33,
2530
+ "learning_rate": 2.8369905956112852e-05,
2531
+ "loss": 0.0057,
2532
+ "step": 4140
2533
+ },
2534
+ {
2535
+ "epoch": 4.34,
2536
+ "learning_rate": 2.8317659352142113e-05,
2537
+ "loss": 0.0062,
2538
+ "step": 4150
2539
+ },
2540
+ {
2541
+ "epoch": 4.35,
2542
+ "learning_rate": 2.8265412748171373e-05,
2543
+ "loss": 0.0074,
2544
+ "step": 4160
2545
+ },
2546
+ {
2547
+ "epoch": 4.36,
2548
+ "learning_rate": 2.8213166144200627e-05,
2549
+ "loss": 0.0034,
2550
+ "step": 4170
2551
+ },
2552
+ {
2553
+ "epoch": 4.37,
2554
+ "learning_rate": 2.8160919540229884e-05,
2555
+ "loss": 0.0055,
2556
+ "step": 4180
2557
+ },
2558
+ {
2559
+ "epoch": 4.38,
2560
+ "learning_rate": 2.8108672936259145e-05,
2561
+ "loss": 0.0055,
2562
+ "step": 4190
2563
+ },
2564
+ {
2565
+ "epoch": 4.39,
2566
+ "learning_rate": 2.8056426332288405e-05,
2567
+ "loss": 0.0041,
2568
+ "step": 4200
2569
+ },
2570
+ {
2571
+ "epoch": 4.4,
2572
+ "learning_rate": 2.800417972831766e-05,
2573
+ "loss": 0.0089,
2574
+ "step": 4210
2575
+ },
2576
+ {
2577
+ "epoch": 4.41,
2578
+ "learning_rate": 2.795193312434692e-05,
2579
+ "loss": 0.0045,
2580
+ "step": 4220
2581
+ },
2582
+ {
2583
+ "epoch": 4.42,
2584
+ "learning_rate": 2.7899686520376177e-05,
2585
+ "loss": 0.0056,
2586
+ "step": 4230
2587
+ },
2588
+ {
2589
+ "epoch": 4.43,
2590
+ "learning_rate": 2.7847439916405434e-05,
2591
+ "loss": 0.0036,
2592
+ "step": 4240
2593
+ },
2594
+ {
2595
+ "epoch": 4.44,
2596
+ "learning_rate": 2.779519331243469e-05,
2597
+ "loss": 0.0036,
2598
+ "step": 4250
2599
+ },
2600
+ {
2601
+ "epoch": 4.45,
2602
+ "learning_rate": 2.774294670846395e-05,
2603
+ "loss": 0.004,
2604
+ "step": 4260
2605
+ },
2606
+ {
2607
+ "epoch": 4.46,
2608
+ "learning_rate": 2.7690700104493212e-05,
2609
+ "loss": 0.0049,
2610
+ "step": 4270
2611
+ },
2612
+ {
2613
+ "epoch": 4.47,
2614
+ "learning_rate": 2.7638453500522466e-05,
2615
+ "loss": 0.0042,
2616
+ "step": 4280
2617
+ },
2618
+ {
2619
+ "epoch": 4.48,
2620
+ "learning_rate": 2.7586206896551727e-05,
2621
+ "loss": 0.0066,
2622
+ "step": 4290
2623
+ },
2624
+ {
2625
+ "epoch": 4.49,
2626
+ "learning_rate": 2.7533960292580984e-05,
2627
+ "loss": 0.0054,
2628
+ "step": 4300
2629
+ },
2630
+ {
2631
+ "epoch": 4.5,
2632
+ "learning_rate": 2.7481713688610244e-05,
2633
+ "loss": 0.0041,
2634
+ "step": 4310
2635
+ },
2636
+ {
2637
+ "epoch": 4.51,
2638
+ "learning_rate": 2.7429467084639498e-05,
2639
+ "loss": 0.0034,
2640
+ "step": 4320
2641
+ },
2642
+ {
2643
+ "epoch": 4.52,
2644
+ "learning_rate": 2.737722048066876e-05,
2645
+ "loss": 0.0065,
2646
+ "step": 4330
2647
+ },
2648
+ {
2649
+ "epoch": 4.54,
2650
+ "learning_rate": 2.732497387669802e-05,
2651
+ "loss": 0.0062,
2652
+ "step": 4340
2653
+ },
2654
+ {
2655
+ "epoch": 4.55,
2656
+ "learning_rate": 2.7272727272727273e-05,
2657
+ "loss": 0.0068,
2658
+ "step": 4350
2659
+ },
2660
+ {
2661
+ "epoch": 4.56,
2662
+ "learning_rate": 2.722048066875653e-05,
2663
+ "loss": 0.0045,
2664
+ "step": 4360
2665
+ },
2666
+ {
2667
+ "epoch": 4.57,
2668
+ "learning_rate": 2.716823406478579e-05,
2669
+ "loss": 0.0089,
2670
+ "step": 4370
2671
+ },
2672
+ {
2673
+ "epoch": 4.58,
2674
+ "learning_rate": 2.711598746081505e-05,
2675
+ "loss": 0.0046,
2676
+ "step": 4380
2677
+ },
2678
+ {
2679
+ "epoch": 4.59,
2680
+ "learning_rate": 2.7063740856844305e-05,
2681
+ "loss": 0.0061,
2682
+ "step": 4390
2683
+ },
2684
+ {
2685
+ "epoch": 4.6,
2686
+ "learning_rate": 2.7011494252873566e-05,
2687
+ "loss": 0.0036,
2688
+ "step": 4400
2689
+ },
2690
+ {
2691
+ "epoch": 4.61,
2692
+ "learning_rate": 2.6959247648902823e-05,
2693
+ "loss": 0.0048,
2694
+ "step": 4410
2695
+ },
2696
+ {
2697
+ "epoch": 4.62,
2698
+ "learning_rate": 2.690700104493208e-05,
2699
+ "loss": 0.0028,
2700
+ "step": 4420
2701
+ },
2702
+ {
2703
+ "epoch": 4.63,
2704
+ "learning_rate": 2.6854754440961337e-05,
2705
+ "loss": 0.0037,
2706
+ "step": 4430
2707
+ },
2708
+ {
2709
+ "epoch": 4.64,
2710
+ "learning_rate": 2.6802507836990598e-05,
2711
+ "loss": 0.0039,
2712
+ "step": 4440
2713
+ },
2714
+ {
2715
+ "epoch": 4.65,
2716
+ "learning_rate": 2.675026123301986e-05,
2717
+ "loss": 0.0077,
2718
+ "step": 4450
2719
+ },
2720
+ {
2721
+ "epoch": 4.66,
2722
+ "learning_rate": 2.6698014629049112e-05,
2723
+ "loss": 0.0035,
2724
+ "step": 4460
2725
+ },
2726
+ {
2727
+ "epoch": 4.67,
2728
+ "learning_rate": 2.6645768025078373e-05,
2729
+ "loss": 0.008,
2730
+ "step": 4470
2731
+ },
2732
+ {
2733
+ "epoch": 4.68,
2734
+ "learning_rate": 2.659352142110763e-05,
2735
+ "loss": 0.0053,
2736
+ "step": 4480
2737
+ },
2738
+ {
2739
+ "epoch": 4.69,
2740
+ "learning_rate": 2.6541274817136884e-05,
2741
+ "loss": 0.0031,
2742
+ "step": 4490
2743
+ },
2744
+ {
2745
+ "epoch": 4.7,
2746
+ "learning_rate": 2.6489028213166144e-05,
2747
+ "loss": 0.0038,
2748
+ "step": 4500
2749
+ },
2750
+ {
2751
+ "epoch": 4.71,
2752
+ "learning_rate": 2.6436781609195405e-05,
2753
+ "loss": 0.0042,
2754
+ "step": 4510
2755
+ },
2756
+ {
2757
+ "epoch": 4.72,
2758
+ "learning_rate": 2.6384535005224666e-05,
2759
+ "loss": 0.0073,
2760
+ "step": 4520
2761
+ },
2762
+ {
2763
+ "epoch": 4.73,
2764
+ "learning_rate": 2.633228840125392e-05,
2765
+ "loss": 0.0056,
2766
+ "step": 4530
2767
+ },
2768
+ {
2769
+ "epoch": 4.74,
2770
+ "learning_rate": 2.6280041797283177e-05,
2771
+ "loss": 0.0039,
2772
+ "step": 4540
2773
+ },
2774
+ {
2775
+ "epoch": 4.75,
2776
+ "learning_rate": 2.6227795193312437e-05,
2777
+ "loss": 0.0045,
2778
+ "step": 4550
2779
+ },
2780
+ {
2781
+ "epoch": 4.76,
2782
+ "learning_rate": 2.617554858934169e-05,
2783
+ "loss": 0.0065,
2784
+ "step": 4560
2785
+ },
2786
+ {
2787
+ "epoch": 4.78,
2788
+ "learning_rate": 2.612330198537095e-05,
2789
+ "loss": 0.0039,
2790
+ "step": 4570
2791
+ },
2792
+ {
2793
+ "epoch": 4.79,
2794
+ "learning_rate": 2.6071055381400212e-05,
2795
+ "loss": 0.0043,
2796
+ "step": 4580
2797
+ },
2798
+ {
2799
+ "epoch": 4.8,
2800
+ "learning_rate": 2.601880877742947e-05,
2801
+ "loss": 0.0033,
2802
+ "step": 4590
2803
+ },
2804
+ {
2805
+ "epoch": 4.81,
2806
+ "learning_rate": 2.5966562173458726e-05,
2807
+ "loss": 0.0035,
2808
+ "step": 4600
2809
+ },
2810
+ {
2811
+ "epoch": 4.82,
2812
+ "learning_rate": 2.5914315569487984e-05,
2813
+ "loss": 0.0046,
2814
+ "step": 4610
2815
+ },
2816
+ {
2817
+ "epoch": 4.83,
2818
+ "learning_rate": 2.5862068965517244e-05,
2819
+ "loss": 0.0026,
2820
+ "step": 4620
2821
+ },
2822
+ {
2823
+ "epoch": 4.84,
2824
+ "learning_rate": 2.5809822361546498e-05,
2825
+ "loss": 0.0031,
2826
+ "step": 4630
2827
+ },
2828
+ {
2829
+ "epoch": 4.85,
2830
+ "learning_rate": 2.575757575757576e-05,
2831
+ "loss": 0.0042,
2832
+ "step": 4640
2833
+ },
2834
+ {
2835
+ "epoch": 4.86,
2836
+ "learning_rate": 2.570532915360502e-05,
2837
+ "loss": 0.0037,
2838
+ "step": 4650
2839
+ },
2840
+ {
2841
+ "epoch": 4.87,
2842
+ "learning_rate": 2.5653082549634276e-05,
2843
+ "loss": 0.0034,
2844
+ "step": 4660
2845
+ },
2846
+ {
2847
+ "epoch": 4.88,
2848
+ "learning_rate": 2.560083594566353e-05,
2849
+ "loss": 0.006,
2850
+ "step": 4670
2851
+ },
2852
+ {
2853
+ "epoch": 4.89,
2854
+ "learning_rate": 2.554858934169279e-05,
2855
+ "loss": 0.005,
2856
+ "step": 4680
2857
+ },
2858
+ {
2859
+ "epoch": 4.9,
2860
+ "learning_rate": 2.549634273772205e-05,
2861
+ "loss": 0.0033,
2862
+ "step": 4690
2863
+ },
2864
+ {
2865
+ "epoch": 4.91,
2866
+ "learning_rate": 2.5444096133751305e-05,
2867
+ "loss": 0.0061,
2868
+ "step": 4700
2869
+ },
2870
+ {
2871
+ "epoch": 4.92,
2872
+ "learning_rate": 2.5391849529780566e-05,
2873
+ "loss": 0.0059,
2874
+ "step": 4710
2875
+ },
2876
+ {
2877
+ "epoch": 4.93,
2878
+ "learning_rate": 2.5339602925809823e-05,
2879
+ "loss": 0.0046,
2880
+ "step": 4720
2881
+ },
2882
+ {
2883
+ "epoch": 4.94,
2884
+ "learning_rate": 2.5287356321839083e-05,
2885
+ "loss": 0.0062,
2886
+ "step": 4730
2887
+ },
2888
+ {
2889
+ "epoch": 4.95,
2890
+ "learning_rate": 2.5235109717868337e-05,
2891
+ "loss": 0.0033,
2892
+ "step": 4740
2893
+ },
2894
+ {
2895
+ "epoch": 4.96,
2896
+ "learning_rate": 2.5182863113897598e-05,
2897
+ "loss": 0.0065,
2898
+ "step": 4750
2899
+ },
2900
+ {
2901
+ "epoch": 4.97,
2902
+ "learning_rate": 2.5130616509926858e-05,
2903
+ "loss": 0.0048,
2904
+ "step": 4760
2905
+ },
2906
+ {
2907
+ "epoch": 4.98,
2908
+ "learning_rate": 2.5078369905956112e-05,
2909
+ "loss": 0.0043,
2910
+ "step": 4770
2911
+ },
2912
+ {
2913
+ "epoch": 4.99,
2914
+ "learning_rate": 2.5026123301985373e-05,
2915
+ "loss": 0.0055,
2916
+ "step": 4780
2917
+ },
2918
+ {
2919
+ "epoch": 5.01,
2920
+ "learning_rate": 2.497387669801463e-05,
2921
+ "loss": 0.0039,
2922
+ "step": 4790
2923
+ },
2924
+ {
2925
+ "epoch": 5.02,
2926
+ "learning_rate": 2.4921630094043887e-05,
2927
+ "loss": 0.0025,
2928
+ "step": 4800
2929
+ },
2930
+ {
2931
+ "epoch": 5.03,
2932
+ "learning_rate": 2.4869383490073148e-05,
2933
+ "loss": 0.0039,
2934
+ "step": 4810
2935
+ },
2936
+ {
2937
+ "epoch": 5.04,
2938
+ "learning_rate": 2.4817136886102405e-05,
2939
+ "loss": 0.0048,
2940
+ "step": 4820
2941
+ },
2942
+ {
2943
+ "epoch": 5.05,
2944
+ "learning_rate": 2.4764890282131662e-05,
2945
+ "loss": 0.005,
2946
+ "step": 4830
2947
+ },
2948
+ {
2949
+ "epoch": 5.06,
2950
+ "learning_rate": 2.4712643678160922e-05,
2951
+ "loss": 0.0039,
2952
+ "step": 4840
2953
+ },
2954
+ {
2955
+ "epoch": 5.07,
2956
+ "learning_rate": 2.466039707419018e-05,
2957
+ "loss": 0.0096,
2958
+ "step": 4850
2959
+ },
2960
+ {
2961
+ "epoch": 5.08,
2962
+ "learning_rate": 2.4608150470219437e-05,
2963
+ "loss": 0.0046,
2964
+ "step": 4860
2965
+ },
2966
+ {
2967
+ "epoch": 5.09,
2968
+ "learning_rate": 2.4555903866248694e-05,
2969
+ "loss": 0.0026,
2970
+ "step": 4870
2971
+ },
2972
+ {
2973
+ "epoch": 5.1,
2974
+ "learning_rate": 2.4503657262277955e-05,
2975
+ "loss": 0.0043,
2976
+ "step": 4880
2977
+ },
2978
+ {
2979
+ "epoch": 5.11,
2980
+ "learning_rate": 2.4451410658307212e-05,
2981
+ "loss": 0.0031,
2982
+ "step": 4890
2983
+ },
2984
+ {
2985
+ "epoch": 5.12,
2986
+ "learning_rate": 2.439916405433647e-05,
2987
+ "loss": 0.005,
2988
+ "step": 4900
2989
+ },
2990
+ {
2991
+ "epoch": 5.13,
2992
+ "learning_rate": 2.4346917450365726e-05,
2993
+ "loss": 0.002,
2994
+ "step": 4910
2995
+ },
2996
+ {
2997
+ "epoch": 5.14,
2998
+ "learning_rate": 2.4294670846394983e-05,
2999
+ "loss": 0.0021,
3000
+ "step": 4920
3001
+ },
3002
+ {
3003
+ "epoch": 5.15,
3004
+ "learning_rate": 2.4242424242424244e-05,
3005
+ "loss": 0.0052,
3006
+ "step": 4930
3007
+ },
3008
+ {
3009
+ "epoch": 5.16,
3010
+ "learning_rate": 2.41901776384535e-05,
3011
+ "loss": 0.006,
3012
+ "step": 4940
3013
+ },
3014
+ {
3015
+ "epoch": 5.17,
3016
+ "learning_rate": 2.413793103448276e-05,
3017
+ "loss": 0.0028,
3018
+ "step": 4950
3019
+ },
3020
+ {
3021
+ "epoch": 5.18,
3022
+ "learning_rate": 2.408568443051202e-05,
3023
+ "loss": 0.0026,
3024
+ "step": 4960
3025
+ },
3026
+ {
3027
+ "epoch": 5.19,
3028
+ "learning_rate": 2.4033437826541276e-05,
3029
+ "loss": 0.0033,
3030
+ "step": 4970
3031
+ },
3032
+ {
3033
+ "epoch": 5.2,
3034
+ "learning_rate": 2.3981191222570533e-05,
3035
+ "loss": 0.003,
3036
+ "step": 4980
3037
+ },
3038
+ {
3039
+ "epoch": 5.21,
3040
+ "learning_rate": 2.392894461859979e-05,
3041
+ "loss": 0.0051,
3042
+ "step": 4990
3043
+ },
3044
+ {
3045
+ "epoch": 5.22,
3046
+ "learning_rate": 2.387669801462905e-05,
3047
+ "loss": 0.0062,
3048
+ "step": 5000
3049
+ },
3050
+ {
3051
+ "epoch": 5.22,
3052
+ "eval_acc": 0.7915632734700663,
3053
+ "eval_cer": 0.026731752673175267,
3054
+ "eval_loss": 0.01478270161896944,
3055
+ "eval_runtime": 128.2616,
3056
+ "eval_samples_per_second": 3.142,
3057
+ "eval_steps_per_second": 0.398,
3058
+ "step": 5000
3059
+ },
3060
+ {
3061
+ "epoch": 5.24,
3062
+ "learning_rate": 2.3824451410658308e-05,
3063
+ "loss": 0.0038,
3064
+ "step": 5010
3065
+ },
3066
+ {
3067
+ "epoch": 5.25,
3068
+ "learning_rate": 2.377220480668757e-05,
3069
+ "loss": 0.0061,
3070
+ "step": 5020
3071
+ },
3072
+ {
3073
+ "epoch": 5.26,
3074
+ "learning_rate": 2.3719958202716826e-05,
3075
+ "loss": 0.0035,
3076
+ "step": 5030
3077
+ },
3078
+ {
3079
+ "epoch": 5.27,
3080
+ "learning_rate": 2.3667711598746083e-05,
3081
+ "loss": 0.0034,
3082
+ "step": 5040
3083
+ },
3084
+ {
3085
+ "epoch": 5.28,
3086
+ "learning_rate": 2.361546499477534e-05,
3087
+ "loss": 0.0033,
3088
+ "step": 5050
3089
+ },
3090
+ {
3091
+ "epoch": 5.29,
3092
+ "learning_rate": 2.3563218390804597e-05,
3093
+ "loss": 0.0036,
3094
+ "step": 5060
3095
+ },
3096
+ {
3097
+ "epoch": 5.3,
3098
+ "learning_rate": 2.3510971786833858e-05,
3099
+ "loss": 0.004,
3100
+ "step": 5070
3101
+ },
3102
+ {
3103
+ "epoch": 5.31,
3104
+ "learning_rate": 2.3458725182863115e-05,
3105
+ "loss": 0.0023,
3106
+ "step": 5080
3107
+ },
3108
+ {
3109
+ "epoch": 5.32,
3110
+ "learning_rate": 2.3406478578892372e-05,
3111
+ "loss": 0.0024,
3112
+ "step": 5090
3113
+ },
3114
+ {
3115
+ "epoch": 5.33,
3116
+ "learning_rate": 2.335423197492163e-05,
3117
+ "loss": 0.0026,
3118
+ "step": 5100
3119
+ },
3120
+ {
3121
+ "epoch": 5.34,
3122
+ "learning_rate": 2.330198537095089e-05,
3123
+ "loss": 0.0031,
3124
+ "step": 5110
3125
+ },
3126
+ {
3127
+ "epoch": 5.35,
3128
+ "learning_rate": 2.3249738766980147e-05,
3129
+ "loss": 0.0032,
3130
+ "step": 5120
3131
+ },
3132
+ {
3133
+ "epoch": 5.36,
3134
+ "learning_rate": 2.3197492163009404e-05,
3135
+ "loss": 0.0032,
3136
+ "step": 5130
3137
+ },
3138
+ {
3139
+ "epoch": 5.37,
3140
+ "learning_rate": 2.3145245559038665e-05,
3141
+ "loss": 0.0043,
3142
+ "step": 5140
3143
+ },
3144
+ {
3145
+ "epoch": 5.38,
3146
+ "learning_rate": 2.3092998955067922e-05,
3147
+ "loss": 0.0039,
3148
+ "step": 5150
3149
+ },
3150
+ {
3151
+ "epoch": 5.39,
3152
+ "learning_rate": 2.304075235109718e-05,
3153
+ "loss": 0.0026,
3154
+ "step": 5160
3155
+ },
3156
+ {
3157
+ "epoch": 5.4,
3158
+ "learning_rate": 2.2988505747126437e-05,
3159
+ "loss": 0.0032,
3160
+ "step": 5170
3161
+ },
3162
+ {
3163
+ "epoch": 5.41,
3164
+ "learning_rate": 2.2936259143155697e-05,
3165
+ "loss": 0.0043,
3166
+ "step": 5180
3167
+ },
3168
+ {
3169
+ "epoch": 5.42,
3170
+ "learning_rate": 2.2884012539184954e-05,
3171
+ "loss": 0.0037,
3172
+ "step": 5190
3173
+ },
3174
+ {
3175
+ "epoch": 5.43,
3176
+ "learning_rate": 2.283176593521421e-05,
3177
+ "loss": 0.0038,
3178
+ "step": 5200
3179
+ },
3180
+ {
3181
+ "epoch": 5.44,
3182
+ "learning_rate": 2.2779519331243472e-05,
3183
+ "loss": 0.0038,
3184
+ "step": 5210
3185
+ },
3186
+ {
3187
+ "epoch": 5.45,
3188
+ "learning_rate": 2.272727272727273e-05,
3189
+ "loss": 0.004,
3190
+ "step": 5220
3191
+ },
3192
+ {
3193
+ "epoch": 5.46,
3194
+ "learning_rate": 2.2675026123301986e-05,
3195
+ "loss": 0.0025,
3196
+ "step": 5230
3197
+ },
3198
+ {
3199
+ "epoch": 5.48,
3200
+ "learning_rate": 2.2622779519331244e-05,
3201
+ "loss": 0.0023,
3202
+ "step": 5240
3203
+ },
3204
+ {
3205
+ "epoch": 5.49,
3206
+ "learning_rate": 2.2570532915360504e-05,
3207
+ "loss": 0.0024,
3208
+ "step": 5250
3209
+ },
3210
+ {
3211
+ "epoch": 5.5,
3212
+ "learning_rate": 2.251828631138976e-05,
3213
+ "loss": 0.0021,
3214
+ "step": 5260
3215
+ },
3216
+ {
3217
+ "epoch": 5.51,
3218
+ "learning_rate": 2.246603970741902e-05,
3219
+ "loss": 0.0045,
3220
+ "step": 5270
3221
+ },
3222
+ {
3223
+ "epoch": 5.52,
3224
+ "learning_rate": 2.2413793103448276e-05,
3225
+ "loss": 0.0026,
3226
+ "step": 5280
3227
+ },
3228
+ {
3229
+ "epoch": 5.53,
3230
+ "learning_rate": 2.2361546499477533e-05,
3231
+ "loss": 0.0044,
3232
+ "step": 5290
3233
+ },
3234
+ {
3235
+ "epoch": 5.54,
3236
+ "learning_rate": 2.2309299895506794e-05,
3237
+ "loss": 0.0046,
3238
+ "step": 5300
3239
+ },
3240
+ {
3241
+ "epoch": 5.55,
3242
+ "learning_rate": 2.225705329153605e-05,
3243
+ "loss": 0.0038,
3244
+ "step": 5310
3245
+ },
3246
+ {
3247
+ "epoch": 5.56,
3248
+ "learning_rate": 2.220480668756531e-05,
3249
+ "loss": 0.009,
3250
+ "step": 5320
3251
+ },
3252
+ {
3253
+ "epoch": 5.57,
3254
+ "learning_rate": 2.215256008359457e-05,
3255
+ "loss": 0.0048,
3256
+ "step": 5330
3257
+ },
3258
+ {
3259
+ "epoch": 5.58,
3260
+ "learning_rate": 2.2100313479623826e-05,
3261
+ "loss": 0.0033,
3262
+ "step": 5340
3263
+ },
3264
+ {
3265
+ "epoch": 5.59,
3266
+ "learning_rate": 2.2048066875653083e-05,
3267
+ "loss": 0.0024,
3268
+ "step": 5350
3269
+ },
3270
+ {
3271
+ "epoch": 5.6,
3272
+ "learning_rate": 2.199582027168234e-05,
3273
+ "loss": 0.002,
3274
+ "step": 5360
3275
+ },
3276
+ {
3277
+ "epoch": 5.61,
3278
+ "learning_rate": 2.19435736677116e-05,
3279
+ "loss": 0.0032,
3280
+ "step": 5370
3281
+ },
3282
+ {
3283
+ "epoch": 5.62,
3284
+ "learning_rate": 2.1891327063740858e-05,
3285
+ "loss": 0.0055,
3286
+ "step": 5380
3287
+ },
3288
+ {
3289
+ "epoch": 5.63,
3290
+ "learning_rate": 2.183908045977012e-05,
3291
+ "loss": 0.0059,
3292
+ "step": 5390
3293
+ },
3294
+ {
3295
+ "epoch": 5.64,
3296
+ "learning_rate": 2.1786833855799376e-05,
3297
+ "loss": 0.0024,
3298
+ "step": 5400
3299
+ },
3300
+ {
3301
+ "epoch": 5.65,
3302
+ "learning_rate": 2.173458725182863e-05,
3303
+ "loss": 0.0039,
3304
+ "step": 5410
3305
+ },
3306
+ {
3307
+ "epoch": 5.66,
3308
+ "learning_rate": 2.168234064785789e-05,
3309
+ "loss": 0.002,
3310
+ "step": 5420
3311
+ },
3312
+ {
3313
+ "epoch": 5.67,
3314
+ "learning_rate": 2.1630094043887147e-05,
3315
+ "loss": 0.0061,
3316
+ "step": 5430
3317
+ },
3318
+ {
3319
+ "epoch": 5.68,
3320
+ "learning_rate": 2.1577847439916408e-05,
3321
+ "loss": 0.0056,
3322
+ "step": 5440
3323
+ },
3324
+ {
3325
+ "epoch": 5.69,
3326
+ "learning_rate": 2.1525600835945665e-05,
3327
+ "loss": 0.0032,
3328
+ "step": 5450
3329
+ },
3330
+ {
3331
+ "epoch": 5.71,
3332
+ "learning_rate": 2.1473354231974925e-05,
3333
+ "loss": 0.0029,
3334
+ "step": 5460
3335
+ },
3336
+ {
3337
+ "epoch": 5.72,
3338
+ "learning_rate": 2.142110762800418e-05,
3339
+ "loss": 0.0017,
3340
+ "step": 5470
3341
+ },
3342
+ {
3343
+ "epoch": 5.73,
3344
+ "learning_rate": 2.1368861024033436e-05,
3345
+ "loss": 0.005,
3346
+ "step": 5480
3347
+ },
3348
+ {
3349
+ "epoch": 5.74,
3350
+ "learning_rate": 2.1316614420062697e-05,
3351
+ "loss": 0.0051,
3352
+ "step": 5490
3353
+ },
3354
+ {
3355
+ "epoch": 5.75,
3356
+ "learning_rate": 2.1264367816091954e-05,
3357
+ "loss": 0.0047,
3358
+ "step": 5500
3359
+ },
3360
+ {
3361
+ "epoch": 5.76,
3362
+ "learning_rate": 2.1212121212121215e-05,
3363
+ "loss": 0.0032,
3364
+ "step": 5510
3365
+ },
3366
+ {
3367
+ "epoch": 5.77,
3368
+ "learning_rate": 2.1159874608150472e-05,
3369
+ "loss": 0.004,
3370
+ "step": 5520
3371
+ },
3372
+ {
3373
+ "epoch": 5.78,
3374
+ "learning_rate": 2.110762800417973e-05,
3375
+ "loss": 0.0059,
3376
+ "step": 5530
3377
+ },
3378
+ {
3379
+ "epoch": 5.79,
3380
+ "learning_rate": 2.1055381400208986e-05,
3381
+ "loss": 0.0045,
3382
+ "step": 5540
3383
+ },
3384
+ {
3385
+ "epoch": 5.8,
3386
+ "learning_rate": 2.1003134796238247e-05,
3387
+ "loss": 0.0051,
3388
+ "step": 5550
3389
+ },
3390
+ {
3391
+ "epoch": 5.81,
3392
+ "learning_rate": 2.0950888192267504e-05,
3393
+ "loss": 0.005,
3394
+ "step": 5560
3395
+ },
3396
+ {
3397
+ "epoch": 5.82,
3398
+ "learning_rate": 2.089864158829676e-05,
3399
+ "loss": 0.0039,
3400
+ "step": 5570
3401
+ },
3402
+ {
3403
+ "epoch": 5.83,
3404
+ "learning_rate": 2.0846394984326022e-05,
3405
+ "loss": 0.0036,
3406
+ "step": 5580
3407
+ },
3408
+ {
3409
+ "epoch": 5.84,
3410
+ "learning_rate": 2.079414838035528e-05,
3411
+ "loss": 0.0041,
3412
+ "step": 5590
3413
+ },
3414
+ {
3415
+ "epoch": 5.85,
3416
+ "learning_rate": 2.0741901776384536e-05,
3417
+ "loss": 0.002,
3418
+ "step": 5600
3419
+ },
3420
+ {
3421
+ "epoch": 5.86,
3422
+ "learning_rate": 2.0689655172413793e-05,
3423
+ "loss": 0.0067,
3424
+ "step": 5610
3425
+ },
3426
+ {
3427
+ "epoch": 5.87,
3428
+ "learning_rate": 2.0637408568443054e-05,
3429
+ "loss": 0.0061,
3430
+ "step": 5620
3431
+ },
3432
+ {
3433
+ "epoch": 5.88,
3434
+ "learning_rate": 2.058516196447231e-05,
3435
+ "loss": 0.0041,
3436
+ "step": 5630
3437
+ },
3438
+ {
3439
+ "epoch": 5.89,
3440
+ "learning_rate": 2.0532915360501568e-05,
3441
+ "loss": 0.0063,
3442
+ "step": 5640
3443
+ },
3444
+ {
3445
+ "epoch": 5.9,
3446
+ "learning_rate": 2.0480668756530825e-05,
3447
+ "loss": 0.0036,
3448
+ "step": 5650
3449
+ },
3450
+ {
3451
+ "epoch": 5.91,
3452
+ "learning_rate": 2.0428422152560083e-05,
3453
+ "loss": 0.0034,
3454
+ "step": 5660
3455
+ },
3456
+ {
3457
+ "epoch": 5.92,
3458
+ "learning_rate": 2.0376175548589343e-05,
3459
+ "loss": 0.0036,
3460
+ "step": 5670
3461
+ },
3462
+ {
3463
+ "epoch": 5.94,
3464
+ "learning_rate": 2.03239289446186e-05,
3465
+ "loss": 0.0021,
3466
+ "step": 5680
3467
+ },
3468
+ {
3469
+ "epoch": 5.95,
3470
+ "learning_rate": 2.027168234064786e-05,
3471
+ "loss": 0.0027,
3472
+ "step": 5690
3473
+ },
3474
+ {
3475
+ "epoch": 5.96,
3476
+ "learning_rate": 2.0219435736677118e-05,
3477
+ "loss": 0.004,
3478
+ "step": 5700
3479
+ },
3480
+ {
3481
+ "epoch": 5.97,
3482
+ "learning_rate": 2.0167189132706375e-05,
3483
+ "loss": 0.0031,
3484
+ "step": 5710
3485
+ },
3486
+ {
3487
+ "epoch": 5.98,
3488
+ "learning_rate": 2.0114942528735632e-05,
3489
+ "loss": 0.0022,
3490
+ "step": 5720
3491
+ },
3492
+ {
3493
+ "epoch": 5.99,
3494
+ "learning_rate": 2.006269592476489e-05,
3495
+ "loss": 0.0029,
3496
+ "step": 5730
3497
+ },
3498
+ {
3499
+ "epoch": 6.0,
3500
+ "learning_rate": 2.001044932079415e-05,
3501
+ "loss": 0.0062,
3502
+ "step": 5740
3503
+ },
3504
+ {
3505
+ "epoch": 6.01,
3506
+ "learning_rate": 1.9958202716823407e-05,
3507
+ "loss": 0.0016,
3508
+ "step": 5750
3509
+ },
3510
+ {
3511
+ "epoch": 6.02,
3512
+ "learning_rate": 1.9905956112852668e-05,
3513
+ "loss": 0.0019,
3514
+ "step": 5760
3515
+ },
3516
+ {
3517
+ "epoch": 6.03,
3518
+ "learning_rate": 1.9853709508881925e-05,
3519
+ "loss": 0.003,
3520
+ "step": 5770
3521
+ },
3522
+ {
3523
+ "epoch": 6.04,
3524
+ "learning_rate": 1.980146290491118e-05,
3525
+ "loss": 0.002,
3526
+ "step": 5780
3527
+ },
3528
+ {
3529
+ "epoch": 6.05,
3530
+ "learning_rate": 1.974921630094044e-05,
3531
+ "loss": 0.003,
3532
+ "step": 5790
3533
+ },
3534
+ {
3535
+ "epoch": 6.06,
3536
+ "learning_rate": 1.9696969696969697e-05,
3537
+ "loss": 0.0033,
3538
+ "step": 5800
3539
+ },
3540
+ {
3541
+ "epoch": 6.07,
3542
+ "learning_rate": 1.9644723092998957e-05,
3543
+ "loss": 0.0016,
3544
+ "step": 5810
3545
+ },
3546
+ {
3547
+ "epoch": 6.08,
3548
+ "learning_rate": 1.9592476489028214e-05,
3549
+ "loss": 0.004,
3550
+ "step": 5820
3551
+ },
3552
+ {
3553
+ "epoch": 6.09,
3554
+ "learning_rate": 1.9540229885057475e-05,
3555
+ "loss": 0.0025,
3556
+ "step": 5830
3557
+ },
3558
+ {
3559
+ "epoch": 6.1,
3560
+ "learning_rate": 1.948798328108673e-05,
3561
+ "loss": 0.0025,
3562
+ "step": 5840
3563
+ },
3564
+ {
3565
+ "epoch": 6.11,
3566
+ "learning_rate": 1.9435736677115986e-05,
3567
+ "loss": 0.0031,
3568
+ "step": 5850
3569
+ },
3570
+ {
3571
+ "epoch": 6.12,
3572
+ "learning_rate": 1.9383490073145247e-05,
3573
+ "loss": 0.002,
3574
+ "step": 5860
3575
+ },
3576
+ {
3577
+ "epoch": 6.13,
3578
+ "learning_rate": 1.9331243469174504e-05,
3579
+ "loss": 0.0027,
3580
+ "step": 5870
3581
+ },
3582
+ {
3583
+ "epoch": 6.14,
3584
+ "learning_rate": 1.9278996865203764e-05,
3585
+ "loss": 0.0019,
3586
+ "step": 5880
3587
+ },
3588
+ {
3589
+ "epoch": 6.15,
3590
+ "learning_rate": 1.922675026123302e-05,
3591
+ "loss": 0.0033,
3592
+ "step": 5890
3593
+ },
3594
+ {
3595
+ "epoch": 6.17,
3596
+ "learning_rate": 1.917450365726228e-05,
3597
+ "loss": 0.0027,
3598
+ "step": 5900
3599
+ },
3600
+ {
3601
+ "epoch": 6.18,
3602
+ "learning_rate": 1.9122257053291536e-05,
3603
+ "loss": 0.0048,
3604
+ "step": 5910
3605
+ },
3606
+ {
3607
+ "epoch": 6.19,
3608
+ "learning_rate": 1.9070010449320793e-05,
3609
+ "loss": 0.0016,
3610
+ "step": 5920
3611
+ },
3612
+ {
3613
+ "epoch": 6.2,
3614
+ "learning_rate": 1.9017763845350054e-05,
3615
+ "loss": 0.002,
3616
+ "step": 5930
3617
+ },
3618
+ {
3619
+ "epoch": 6.21,
3620
+ "learning_rate": 1.896551724137931e-05,
3621
+ "loss": 0.0027,
3622
+ "step": 5940
3623
+ },
3624
+ {
3625
+ "epoch": 6.22,
3626
+ "learning_rate": 1.891327063740857e-05,
3627
+ "loss": 0.0021,
3628
+ "step": 5950
3629
+ },
3630
+ {
3631
+ "epoch": 6.23,
3632
+ "learning_rate": 1.886102403343783e-05,
3633
+ "loss": 0.0068,
3634
+ "step": 5960
3635
+ },
3636
+ {
3637
+ "epoch": 6.24,
3638
+ "learning_rate": 1.8808777429467086e-05,
3639
+ "loss": 0.0017,
3640
+ "step": 5970
3641
+ },
3642
+ {
3643
+ "epoch": 6.25,
3644
+ "learning_rate": 1.8756530825496343e-05,
3645
+ "loss": 0.0013,
3646
+ "step": 5980
3647
+ },
3648
+ {
3649
+ "epoch": 6.26,
3650
+ "learning_rate": 1.8704284221525603e-05,
3651
+ "loss": 0.0033,
3652
+ "step": 5990
3653
+ },
3654
+ {
3655
+ "epoch": 6.27,
3656
+ "learning_rate": 1.865203761755486e-05,
3657
+ "loss": 0.0024,
3658
+ "step": 6000
3659
+ },
3660
+ {
3661
+ "epoch": 6.27,
3662
+ "eval_acc": 0.8064516109021052,
3663
+ "eval_cer": 0.02487215248721525,
3664
+ "eval_loss": 0.014062131755053997,
3665
+ "eval_runtime": 122.3207,
3666
+ "eval_samples_per_second": 3.295,
3667
+ "eval_steps_per_second": 0.417,
3668
+ "step": 6000
3669
+ }
3670
+ ],
3671
+ "max_steps": 9570,
3672
+ "num_train_epochs": 10,
3673
+ "total_flos": 3.760504352482001e+18,
3674
+ "trial_name": null,
3675
+ "trial_params": null
3676
+ }
checkpoint/trocr-custdata-8000/checkpoint-6000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/checkpoint-7000/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/trocr-custdata-8000/checkpoint-7000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c60c24a1a5e79a55db3312c60ff461392665d15afcacddfb21a9d2a606091206
3
+ size 243062533
checkpoint/{trocr-custdata/checkpoint-3000 → trocr-custdata-8000/checkpoint-7000}/preprocessor_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/checkpoint-7000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd9c92560dcc5cae8002aaa12068ae8baed2c848538286a28a8da5956728f4e5
3
+ size 122142261
checkpoint/trocr-custdata-8000/checkpoint-7000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47d9efe0aae719d738be54c064ec776715032ad8da06f4b50124d393a6fca69c
3
+ size 627
checkpoint/trocr-custdata-8000/checkpoint-7000/trainer_state.json ADDED
@@ -0,0 +1,4286 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 7.3145245559038665,
5
+ "global_step": 7000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 4.994775339602926e-05,
13
+ "loss": 0.3717,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 4.989550679205852e-05,
19
+ "loss": 0.2368,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 4.9843260188087774e-05,
25
+ "loss": 0.1928,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 4.979101358411704e-05,
31
+ "loss": 0.1648,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 4.9738766980146295e-05,
37
+ "loss": 0.1645,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 4.968652037617555e-05,
43
+ "loss": 0.1151,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 4.963427377220481e-05,
49
+ "loss": 0.1206,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 4.958202716823407e-05,
55
+ "loss": 0.1289,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 4.9529780564263324e-05,
61
+ "loss": 0.1123,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 4.947753396029258e-05,
67
+ "loss": 0.0992,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.11,
72
+ "learning_rate": 4.9425287356321845e-05,
73
+ "loss": 0.097,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.13,
78
+ "learning_rate": 4.93730407523511e-05,
79
+ "loss": 0.1028,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.14,
84
+ "learning_rate": 4.932079414838036e-05,
85
+ "loss": 0.087,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.15,
90
+ "learning_rate": 4.9268547544409617e-05,
91
+ "loss": 0.0907,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.16,
96
+ "learning_rate": 4.9216300940438874e-05,
97
+ "loss": 0.0866,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.17,
102
+ "learning_rate": 4.916405433646813e-05,
103
+ "loss": 0.085,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.18,
108
+ "learning_rate": 4.911180773249739e-05,
109
+ "loss": 0.0846,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.19,
114
+ "learning_rate": 4.905956112852665e-05,
115
+ "loss": 0.0791,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.2,
120
+ "learning_rate": 4.900731452455591e-05,
121
+ "loss": 0.073,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.21,
126
+ "learning_rate": 4.895506792058516e-05,
127
+ "loss": 0.0861,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.22,
132
+ "learning_rate": 4.8902821316614424e-05,
133
+ "loss": 0.0828,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.23,
138
+ "learning_rate": 4.885057471264368e-05,
139
+ "loss": 0.0711,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.24,
144
+ "learning_rate": 4.879832810867294e-05,
145
+ "loss": 0.069,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.25,
150
+ "learning_rate": 4.8746081504702195e-05,
151
+ "loss": 0.0647,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.26,
156
+ "learning_rate": 4.869383490073145e-05,
157
+ "loss": 0.0862,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.27,
162
+ "learning_rate": 4.8641588296760716e-05,
163
+ "loss": 0.0628,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.28,
168
+ "learning_rate": 4.858934169278997e-05,
169
+ "loss": 0.0697,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.29,
174
+ "learning_rate": 4.853709508881923e-05,
175
+ "loss": 0.0685,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.3,
180
+ "learning_rate": 4.848484848484849e-05,
181
+ "loss": 0.0602,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.31,
186
+ "learning_rate": 4.8432601880877745e-05,
187
+ "loss": 0.0576,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.32,
192
+ "learning_rate": 4.8380355276907e-05,
193
+ "loss": 0.068,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.33,
198
+ "learning_rate": 4.832810867293626e-05,
199
+ "loss": 0.0645,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.34,
204
+ "learning_rate": 4.827586206896552e-05,
205
+ "loss": 0.0785,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.36,
210
+ "learning_rate": 4.8223615464994774e-05,
211
+ "loss": 0.0606,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.37,
216
+ "learning_rate": 4.817136886102404e-05,
217
+ "loss": 0.0682,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.38,
222
+ "learning_rate": 4.8119122257053295e-05,
223
+ "loss": 0.0682,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.39,
228
+ "learning_rate": 4.806687565308255e-05,
229
+ "loss": 0.0605,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.4,
234
+ "learning_rate": 4.801462904911181e-05,
235
+ "loss": 0.0578,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.41,
240
+ "learning_rate": 4.7962382445141066e-05,
241
+ "loss": 0.0562,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.42,
246
+ "learning_rate": 4.791013584117033e-05,
247
+ "loss": 0.0724,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.43,
252
+ "learning_rate": 4.785788923719958e-05,
253
+ "loss": 0.056,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.44,
258
+ "learning_rate": 4.7805642633228845e-05,
259
+ "loss": 0.0509,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.45,
264
+ "learning_rate": 4.77533960292581e-05,
265
+ "loss": 0.0516,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.46,
270
+ "learning_rate": 4.770114942528736e-05,
271
+ "loss": 0.0588,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.47,
276
+ "learning_rate": 4.7648902821316616e-05,
277
+ "loss": 0.0527,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.48,
282
+ "learning_rate": 4.7596656217345873e-05,
283
+ "loss": 0.0614,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.49,
288
+ "learning_rate": 4.754440961337514e-05,
289
+ "loss": 0.0464,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.5,
294
+ "learning_rate": 4.749216300940439e-05,
295
+ "loss": 0.0515,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.51,
300
+ "learning_rate": 4.743991640543365e-05,
301
+ "loss": 0.0456,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.52,
306
+ "learning_rate": 4.738766980146291e-05,
307
+ "loss": 0.0583,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.53,
312
+ "learning_rate": 4.7335423197492166e-05,
313
+ "loss": 0.0595,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 0.54,
318
+ "learning_rate": 4.728317659352142e-05,
319
+ "loss": 0.0547,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 0.55,
324
+ "learning_rate": 4.723092998955068e-05,
325
+ "loss": 0.0451,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 0.56,
330
+ "learning_rate": 4.7178683385579944e-05,
331
+ "loss": 0.0395,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 0.57,
336
+ "learning_rate": 4.7126436781609195e-05,
337
+ "loss": 0.0493,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 0.59,
342
+ "learning_rate": 4.707419017763845e-05,
343
+ "loss": 0.0664,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 0.6,
348
+ "learning_rate": 4.7021943573667716e-05,
349
+ "loss": 0.0445,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 0.61,
354
+ "learning_rate": 4.696969696969697e-05,
355
+ "loss": 0.0443,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 0.62,
360
+ "learning_rate": 4.691745036572623e-05,
361
+ "loss": 0.0461,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 0.63,
366
+ "learning_rate": 4.686520376175549e-05,
367
+ "loss": 0.0415,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 0.64,
372
+ "learning_rate": 4.6812957157784745e-05,
373
+ "loss": 0.0483,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 0.65,
378
+ "learning_rate": 4.6760710553814e-05,
379
+ "loss": 0.0424,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 0.66,
384
+ "learning_rate": 4.670846394984326e-05,
385
+ "loss": 0.0327,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 0.67,
390
+ "learning_rate": 4.665621734587252e-05,
391
+ "loss": 0.0421,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 0.68,
396
+ "learning_rate": 4.660397074190178e-05,
397
+ "loss": 0.0491,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 0.69,
402
+ "learning_rate": 4.655172413793104e-05,
403
+ "loss": 0.0501,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 0.7,
408
+ "learning_rate": 4.6499477533960295e-05,
409
+ "loss": 0.0447,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 0.71,
414
+ "learning_rate": 4.644723092998955e-05,
415
+ "loss": 0.0449,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 0.72,
420
+ "learning_rate": 4.639498432601881e-05,
421
+ "loss": 0.0478,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 0.73,
426
+ "learning_rate": 4.6342737722048066e-05,
427
+ "loss": 0.0505,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 0.74,
432
+ "learning_rate": 4.629049111807733e-05,
433
+ "loss": 0.0394,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 0.75,
438
+ "learning_rate": 4.623824451410659e-05,
439
+ "loss": 0.0406,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 0.76,
444
+ "learning_rate": 4.6185997910135844e-05,
445
+ "loss": 0.0347,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 0.77,
450
+ "learning_rate": 4.61337513061651e-05,
451
+ "loss": 0.0445,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 0.78,
456
+ "learning_rate": 4.608150470219436e-05,
457
+ "loss": 0.0428,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 0.79,
462
+ "learning_rate": 4.6029258098223616e-05,
463
+ "loss": 0.0466,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 0.8,
468
+ "learning_rate": 4.597701149425287e-05,
469
+ "loss": 0.0395,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 0.82,
474
+ "learning_rate": 4.592476489028214e-05,
475
+ "loss": 0.0387,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 0.83,
480
+ "learning_rate": 4.5872518286311394e-05,
481
+ "loss": 0.0395,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 0.84,
486
+ "learning_rate": 4.582027168234065e-05,
487
+ "loss": 0.0305,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 0.85,
492
+ "learning_rate": 4.576802507836991e-05,
493
+ "loss": 0.0434,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 0.86,
498
+ "learning_rate": 4.5715778474399166e-05,
499
+ "loss": 0.0395,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 0.87,
504
+ "learning_rate": 4.566353187042842e-05,
505
+ "loss": 0.0374,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 0.88,
510
+ "learning_rate": 4.561128526645768e-05,
511
+ "loss": 0.0321,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 0.89,
516
+ "learning_rate": 4.5559038662486944e-05,
517
+ "loss": 0.0394,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 0.9,
522
+ "learning_rate": 4.55067920585162e-05,
523
+ "loss": 0.0383,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 0.91,
528
+ "learning_rate": 4.545454545454546e-05,
529
+ "loss": 0.0388,
530
+ "step": 870
531
+ },
532
+ {
533
+ "epoch": 0.92,
534
+ "learning_rate": 4.5402298850574716e-05,
535
+ "loss": 0.035,
536
+ "step": 880
537
+ },
538
+ {
539
+ "epoch": 0.93,
540
+ "learning_rate": 4.535005224660397e-05,
541
+ "loss": 0.0414,
542
+ "step": 890
543
+ },
544
+ {
545
+ "epoch": 0.94,
546
+ "learning_rate": 4.529780564263323e-05,
547
+ "loss": 0.0439,
548
+ "step": 900
549
+ },
550
+ {
551
+ "epoch": 0.95,
552
+ "learning_rate": 4.524555903866249e-05,
553
+ "loss": 0.0337,
554
+ "step": 910
555
+ },
556
+ {
557
+ "epoch": 0.96,
558
+ "learning_rate": 4.519331243469175e-05,
559
+ "loss": 0.0384,
560
+ "step": 920
561
+ },
562
+ {
563
+ "epoch": 0.97,
564
+ "learning_rate": 4.514106583072101e-05,
565
+ "loss": 0.0353,
566
+ "step": 930
567
+ },
568
+ {
569
+ "epoch": 0.98,
570
+ "learning_rate": 4.508881922675026e-05,
571
+ "loss": 0.0351,
572
+ "step": 940
573
+ },
574
+ {
575
+ "epoch": 0.99,
576
+ "learning_rate": 4.503657262277952e-05,
577
+ "loss": 0.0324,
578
+ "step": 950
579
+ },
580
+ {
581
+ "epoch": 1.0,
582
+ "learning_rate": 4.498432601880878e-05,
583
+ "loss": 0.0296,
584
+ "step": 960
585
+ },
586
+ {
587
+ "epoch": 1.01,
588
+ "learning_rate": 4.493207941483804e-05,
589
+ "loss": 0.0262,
590
+ "step": 970
591
+ },
592
+ {
593
+ "epoch": 1.02,
594
+ "learning_rate": 4.4879832810867294e-05,
595
+ "loss": 0.0274,
596
+ "step": 980
597
+ },
598
+ {
599
+ "epoch": 1.03,
600
+ "learning_rate": 4.482758620689655e-05,
601
+ "loss": 0.0278,
602
+ "step": 990
603
+ },
604
+ {
605
+ "epoch": 1.04,
606
+ "learning_rate": 4.4775339602925815e-05,
607
+ "loss": 0.0205,
608
+ "step": 1000
609
+ },
610
+ {
611
+ "epoch": 1.04,
612
+ "eval_acc": 0.5359801475533992,
613
+ "eval_cer": 0.06462110646211065,
614
+ "eval_loss": 0.0336129367351532,
615
+ "eval_runtime": 135.8305,
616
+ "eval_samples_per_second": 2.967,
617
+ "eval_steps_per_second": 0.375,
618
+ "step": 1000
619
+ },
620
+ {
621
+ "epoch": 1.06,
622
+ "learning_rate": 4.4723092998955066e-05,
623
+ "loss": 0.0264,
624
+ "step": 1010
625
+ },
626
+ {
627
+ "epoch": 1.07,
628
+ "learning_rate": 4.467084639498433e-05,
629
+ "loss": 0.0218,
630
+ "step": 1020
631
+ },
632
+ {
633
+ "epoch": 1.08,
634
+ "learning_rate": 4.461859979101359e-05,
635
+ "loss": 0.0269,
636
+ "step": 1030
637
+ },
638
+ {
639
+ "epoch": 1.09,
640
+ "learning_rate": 4.4566353187042844e-05,
641
+ "loss": 0.0242,
642
+ "step": 1040
643
+ },
644
+ {
645
+ "epoch": 1.1,
646
+ "learning_rate": 4.45141065830721e-05,
647
+ "loss": 0.03,
648
+ "step": 1050
649
+ },
650
+ {
651
+ "epoch": 1.11,
652
+ "learning_rate": 4.446185997910136e-05,
653
+ "loss": 0.0214,
654
+ "step": 1060
655
+ },
656
+ {
657
+ "epoch": 1.12,
658
+ "learning_rate": 4.440961337513062e-05,
659
+ "loss": 0.0301,
660
+ "step": 1070
661
+ },
662
+ {
663
+ "epoch": 1.13,
664
+ "learning_rate": 4.435736677115987e-05,
665
+ "loss": 0.0217,
666
+ "step": 1080
667
+ },
668
+ {
669
+ "epoch": 1.14,
670
+ "learning_rate": 4.430512016718914e-05,
671
+ "loss": 0.0323,
672
+ "step": 1090
673
+ },
674
+ {
675
+ "epoch": 1.15,
676
+ "learning_rate": 4.4252873563218394e-05,
677
+ "loss": 0.0258,
678
+ "step": 1100
679
+ },
680
+ {
681
+ "epoch": 1.16,
682
+ "learning_rate": 4.420062695924765e-05,
683
+ "loss": 0.0293,
684
+ "step": 1110
685
+ },
686
+ {
687
+ "epoch": 1.17,
688
+ "learning_rate": 4.414838035527691e-05,
689
+ "loss": 0.0265,
690
+ "step": 1120
691
+ },
692
+ {
693
+ "epoch": 1.18,
694
+ "learning_rate": 4.4096133751306166e-05,
695
+ "loss": 0.0237,
696
+ "step": 1130
697
+ },
698
+ {
699
+ "epoch": 1.19,
700
+ "learning_rate": 4.404388714733543e-05,
701
+ "loss": 0.0275,
702
+ "step": 1140
703
+ },
704
+ {
705
+ "epoch": 1.2,
706
+ "learning_rate": 4.399164054336468e-05,
707
+ "loss": 0.0219,
708
+ "step": 1150
709
+ },
710
+ {
711
+ "epoch": 1.21,
712
+ "learning_rate": 4.3939393939393944e-05,
713
+ "loss": 0.0334,
714
+ "step": 1160
715
+ },
716
+ {
717
+ "epoch": 1.22,
718
+ "learning_rate": 4.38871473354232e-05,
719
+ "loss": 0.0183,
720
+ "step": 1170
721
+ },
722
+ {
723
+ "epoch": 1.23,
724
+ "learning_rate": 4.383490073145246e-05,
725
+ "loss": 0.0241,
726
+ "step": 1180
727
+ },
728
+ {
729
+ "epoch": 1.24,
730
+ "learning_rate": 4.3782654127481716e-05,
731
+ "loss": 0.0233,
732
+ "step": 1190
733
+ },
734
+ {
735
+ "epoch": 1.25,
736
+ "learning_rate": 4.373040752351097e-05,
737
+ "loss": 0.025,
738
+ "step": 1200
739
+ },
740
+ {
741
+ "epoch": 1.26,
742
+ "learning_rate": 4.367816091954024e-05,
743
+ "loss": 0.0228,
744
+ "step": 1210
745
+ },
746
+ {
747
+ "epoch": 1.27,
748
+ "learning_rate": 4.362591431556949e-05,
749
+ "loss": 0.0182,
750
+ "step": 1220
751
+ },
752
+ {
753
+ "epoch": 1.29,
754
+ "learning_rate": 4.357366771159875e-05,
755
+ "loss": 0.0234,
756
+ "step": 1230
757
+ },
758
+ {
759
+ "epoch": 1.3,
760
+ "learning_rate": 4.352142110762801e-05,
761
+ "loss": 0.019,
762
+ "step": 1240
763
+ },
764
+ {
765
+ "epoch": 1.31,
766
+ "learning_rate": 4.346917450365726e-05,
767
+ "loss": 0.0191,
768
+ "step": 1250
769
+ },
770
+ {
771
+ "epoch": 1.32,
772
+ "learning_rate": 4.341692789968652e-05,
773
+ "loss": 0.0177,
774
+ "step": 1260
775
+ },
776
+ {
777
+ "epoch": 1.33,
778
+ "learning_rate": 4.336468129571578e-05,
779
+ "loss": 0.0214,
780
+ "step": 1270
781
+ },
782
+ {
783
+ "epoch": 1.34,
784
+ "learning_rate": 4.3312434691745044e-05,
785
+ "loss": 0.0252,
786
+ "step": 1280
787
+ },
788
+ {
789
+ "epoch": 1.35,
790
+ "learning_rate": 4.3260188087774294e-05,
791
+ "loss": 0.0233,
792
+ "step": 1290
793
+ },
794
+ {
795
+ "epoch": 1.36,
796
+ "learning_rate": 4.320794148380355e-05,
797
+ "loss": 0.0184,
798
+ "step": 1300
799
+ },
800
+ {
801
+ "epoch": 1.37,
802
+ "learning_rate": 4.3155694879832815e-05,
803
+ "loss": 0.0212,
804
+ "step": 1310
805
+ },
806
+ {
807
+ "epoch": 1.38,
808
+ "learning_rate": 4.3103448275862066e-05,
809
+ "loss": 0.0186,
810
+ "step": 1320
811
+ },
812
+ {
813
+ "epoch": 1.39,
814
+ "learning_rate": 4.305120167189133e-05,
815
+ "loss": 0.0179,
816
+ "step": 1330
817
+ },
818
+ {
819
+ "epoch": 1.4,
820
+ "learning_rate": 4.299895506792059e-05,
821
+ "loss": 0.0184,
822
+ "step": 1340
823
+ },
824
+ {
825
+ "epoch": 1.41,
826
+ "learning_rate": 4.294670846394985e-05,
827
+ "loss": 0.0208,
828
+ "step": 1350
829
+ },
830
+ {
831
+ "epoch": 1.42,
832
+ "learning_rate": 4.28944618599791e-05,
833
+ "loss": 0.0218,
834
+ "step": 1360
835
+ },
836
+ {
837
+ "epoch": 1.43,
838
+ "learning_rate": 4.284221525600836e-05,
839
+ "loss": 0.0203,
840
+ "step": 1370
841
+ },
842
+ {
843
+ "epoch": 1.44,
844
+ "learning_rate": 4.278996865203762e-05,
845
+ "loss": 0.022,
846
+ "step": 1380
847
+ },
848
+ {
849
+ "epoch": 1.45,
850
+ "learning_rate": 4.273772204806687e-05,
851
+ "loss": 0.0247,
852
+ "step": 1390
853
+ },
854
+ {
855
+ "epoch": 1.46,
856
+ "learning_rate": 4.268547544409614e-05,
857
+ "loss": 0.0195,
858
+ "step": 1400
859
+ },
860
+ {
861
+ "epoch": 1.47,
862
+ "learning_rate": 4.2633228840125394e-05,
863
+ "loss": 0.0236,
864
+ "step": 1410
865
+ },
866
+ {
867
+ "epoch": 1.48,
868
+ "learning_rate": 4.258098223615465e-05,
869
+ "loss": 0.0182,
870
+ "step": 1420
871
+ },
872
+ {
873
+ "epoch": 1.49,
874
+ "learning_rate": 4.252873563218391e-05,
875
+ "loss": 0.0238,
876
+ "step": 1430
877
+ },
878
+ {
879
+ "epoch": 1.5,
880
+ "learning_rate": 4.2476489028213165e-05,
881
+ "loss": 0.0244,
882
+ "step": 1440
883
+ },
884
+ {
885
+ "epoch": 1.52,
886
+ "learning_rate": 4.242424242424243e-05,
887
+ "loss": 0.0366,
888
+ "step": 1450
889
+ },
890
+ {
891
+ "epoch": 1.53,
892
+ "learning_rate": 4.2371995820271687e-05,
893
+ "loss": 0.0159,
894
+ "step": 1460
895
+ },
896
+ {
897
+ "epoch": 1.54,
898
+ "learning_rate": 4.2319749216300944e-05,
899
+ "loss": 0.0146,
900
+ "step": 1470
901
+ },
902
+ {
903
+ "epoch": 1.55,
904
+ "learning_rate": 4.22675026123302e-05,
905
+ "loss": 0.0219,
906
+ "step": 1480
907
+ },
908
+ {
909
+ "epoch": 1.56,
910
+ "learning_rate": 4.221525600835946e-05,
911
+ "loss": 0.0176,
912
+ "step": 1490
913
+ },
914
+ {
915
+ "epoch": 1.57,
916
+ "learning_rate": 4.2163009404388715e-05,
917
+ "loss": 0.0207,
918
+ "step": 1500
919
+ },
920
+ {
921
+ "epoch": 1.58,
922
+ "learning_rate": 4.211076280041797e-05,
923
+ "loss": 0.0141,
924
+ "step": 1510
925
+ },
926
+ {
927
+ "epoch": 1.59,
928
+ "learning_rate": 4.2058516196447236e-05,
929
+ "loss": 0.023,
930
+ "step": 1520
931
+ },
932
+ {
933
+ "epoch": 1.6,
934
+ "learning_rate": 4.2006269592476494e-05,
935
+ "loss": 0.0154,
936
+ "step": 1530
937
+ },
938
+ {
939
+ "epoch": 1.61,
940
+ "learning_rate": 4.195402298850575e-05,
941
+ "loss": 0.0208,
942
+ "step": 1540
943
+ },
944
+ {
945
+ "epoch": 1.62,
946
+ "learning_rate": 4.190177638453501e-05,
947
+ "loss": 0.0283,
948
+ "step": 1550
949
+ },
950
+ {
951
+ "epoch": 1.63,
952
+ "learning_rate": 4.1849529780564265e-05,
953
+ "loss": 0.0238,
954
+ "step": 1560
955
+ },
956
+ {
957
+ "epoch": 1.64,
958
+ "learning_rate": 4.179728317659352e-05,
959
+ "loss": 0.0261,
960
+ "step": 1570
961
+ },
962
+ {
963
+ "epoch": 1.65,
964
+ "learning_rate": 4.174503657262278e-05,
965
+ "loss": 0.0206,
966
+ "step": 1580
967
+ },
968
+ {
969
+ "epoch": 1.66,
970
+ "learning_rate": 4.1692789968652043e-05,
971
+ "loss": 0.0239,
972
+ "step": 1590
973
+ },
974
+ {
975
+ "epoch": 1.67,
976
+ "learning_rate": 4.16405433646813e-05,
977
+ "loss": 0.0256,
978
+ "step": 1600
979
+ },
980
+ {
981
+ "epoch": 1.68,
982
+ "learning_rate": 4.158829676071056e-05,
983
+ "loss": 0.0216,
984
+ "step": 1610
985
+ },
986
+ {
987
+ "epoch": 1.69,
988
+ "learning_rate": 4.1536050156739815e-05,
989
+ "loss": 0.0188,
990
+ "step": 1620
991
+ },
992
+ {
993
+ "epoch": 1.7,
994
+ "learning_rate": 4.148380355276907e-05,
995
+ "loss": 0.018,
996
+ "step": 1630
997
+ },
998
+ {
999
+ "epoch": 1.71,
1000
+ "learning_rate": 4.143155694879833e-05,
1001
+ "loss": 0.0149,
1002
+ "step": 1640
1003
+ },
1004
+ {
1005
+ "epoch": 1.72,
1006
+ "learning_rate": 4.1379310344827587e-05,
1007
+ "loss": 0.0158,
1008
+ "step": 1650
1009
+ },
1010
+ {
1011
+ "epoch": 1.73,
1012
+ "learning_rate": 4.132706374085685e-05,
1013
+ "loss": 0.0216,
1014
+ "step": 1660
1015
+ },
1016
+ {
1017
+ "epoch": 1.75,
1018
+ "learning_rate": 4.127481713688611e-05,
1019
+ "loss": 0.0249,
1020
+ "step": 1670
1021
+ },
1022
+ {
1023
+ "epoch": 1.76,
1024
+ "learning_rate": 4.122257053291536e-05,
1025
+ "loss": 0.02,
1026
+ "step": 1680
1027
+ },
1028
+ {
1029
+ "epoch": 1.77,
1030
+ "learning_rate": 4.117032392894462e-05,
1031
+ "loss": 0.0188,
1032
+ "step": 1690
1033
+ },
1034
+ {
1035
+ "epoch": 1.78,
1036
+ "learning_rate": 4.111807732497388e-05,
1037
+ "loss": 0.0194,
1038
+ "step": 1700
1039
+ },
1040
+ {
1041
+ "epoch": 1.79,
1042
+ "learning_rate": 4.1065830721003136e-05,
1043
+ "loss": 0.0258,
1044
+ "step": 1710
1045
+ },
1046
+ {
1047
+ "epoch": 1.8,
1048
+ "learning_rate": 4.1013584117032394e-05,
1049
+ "loss": 0.0187,
1050
+ "step": 1720
1051
+ },
1052
+ {
1053
+ "epoch": 1.81,
1054
+ "learning_rate": 4.096133751306165e-05,
1055
+ "loss": 0.0195,
1056
+ "step": 1730
1057
+ },
1058
+ {
1059
+ "epoch": 1.82,
1060
+ "learning_rate": 4.0909090909090915e-05,
1061
+ "loss": 0.0218,
1062
+ "step": 1740
1063
+ },
1064
+ {
1065
+ "epoch": 1.83,
1066
+ "learning_rate": 4.0856844305120165e-05,
1067
+ "loss": 0.0182,
1068
+ "step": 1750
1069
+ },
1070
+ {
1071
+ "epoch": 1.84,
1072
+ "learning_rate": 4.080459770114943e-05,
1073
+ "loss": 0.0227,
1074
+ "step": 1760
1075
+ },
1076
+ {
1077
+ "epoch": 1.85,
1078
+ "learning_rate": 4.0752351097178686e-05,
1079
+ "loss": 0.0267,
1080
+ "step": 1770
1081
+ },
1082
+ {
1083
+ "epoch": 1.86,
1084
+ "learning_rate": 4.0700104493207943e-05,
1085
+ "loss": 0.0187,
1086
+ "step": 1780
1087
+ },
1088
+ {
1089
+ "epoch": 1.87,
1090
+ "learning_rate": 4.06478578892372e-05,
1091
+ "loss": 0.0251,
1092
+ "step": 1790
1093
+ },
1094
+ {
1095
+ "epoch": 1.88,
1096
+ "learning_rate": 4.059561128526646e-05,
1097
+ "loss": 0.0153,
1098
+ "step": 1800
1099
+ },
1100
+ {
1101
+ "epoch": 1.89,
1102
+ "learning_rate": 4.054336468129572e-05,
1103
+ "loss": 0.0301,
1104
+ "step": 1810
1105
+ },
1106
+ {
1107
+ "epoch": 1.9,
1108
+ "learning_rate": 4.049111807732497e-05,
1109
+ "loss": 0.0157,
1110
+ "step": 1820
1111
+ },
1112
+ {
1113
+ "epoch": 1.91,
1114
+ "learning_rate": 4.0438871473354236e-05,
1115
+ "loss": 0.0204,
1116
+ "step": 1830
1117
+ },
1118
+ {
1119
+ "epoch": 1.92,
1120
+ "learning_rate": 4.038662486938349e-05,
1121
+ "loss": 0.0203,
1122
+ "step": 1840
1123
+ },
1124
+ {
1125
+ "epoch": 1.93,
1126
+ "learning_rate": 4.033437826541275e-05,
1127
+ "loss": 0.0138,
1128
+ "step": 1850
1129
+ },
1130
+ {
1131
+ "epoch": 1.94,
1132
+ "learning_rate": 4.028213166144201e-05,
1133
+ "loss": 0.0231,
1134
+ "step": 1860
1135
+ },
1136
+ {
1137
+ "epoch": 1.95,
1138
+ "learning_rate": 4.0229885057471265e-05,
1139
+ "loss": 0.0288,
1140
+ "step": 1870
1141
+ },
1142
+ {
1143
+ "epoch": 1.96,
1144
+ "learning_rate": 4.017763845350053e-05,
1145
+ "loss": 0.0189,
1146
+ "step": 1880
1147
+ },
1148
+ {
1149
+ "epoch": 1.97,
1150
+ "learning_rate": 4.012539184952978e-05,
1151
+ "loss": 0.0229,
1152
+ "step": 1890
1153
+ },
1154
+ {
1155
+ "epoch": 1.99,
1156
+ "learning_rate": 4.007314524555904e-05,
1157
+ "loss": 0.0171,
1158
+ "step": 1900
1159
+ },
1160
+ {
1161
+ "epoch": 2.0,
1162
+ "learning_rate": 4.00208986415883e-05,
1163
+ "loss": 0.02,
1164
+ "step": 1910
1165
+ },
1166
+ {
1167
+ "epoch": 2.01,
1168
+ "learning_rate": 3.996865203761756e-05,
1169
+ "loss": 0.014,
1170
+ "step": 1920
1171
+ },
1172
+ {
1173
+ "epoch": 2.02,
1174
+ "learning_rate": 3.9916405433646815e-05,
1175
+ "loss": 0.0164,
1176
+ "step": 1930
1177
+ },
1178
+ {
1179
+ "epoch": 2.03,
1180
+ "learning_rate": 3.986415882967607e-05,
1181
+ "loss": 0.0148,
1182
+ "step": 1940
1183
+ },
1184
+ {
1185
+ "epoch": 2.04,
1186
+ "learning_rate": 3.9811912225705336e-05,
1187
+ "loss": 0.0135,
1188
+ "step": 1950
1189
+ },
1190
+ {
1191
+ "epoch": 2.05,
1192
+ "learning_rate": 3.9759665621734586e-05,
1193
+ "loss": 0.015,
1194
+ "step": 1960
1195
+ },
1196
+ {
1197
+ "epoch": 2.06,
1198
+ "learning_rate": 3.970741901776385e-05,
1199
+ "loss": 0.0183,
1200
+ "step": 1970
1201
+ },
1202
+ {
1203
+ "epoch": 2.07,
1204
+ "learning_rate": 3.965517241379311e-05,
1205
+ "loss": 0.0105,
1206
+ "step": 1980
1207
+ },
1208
+ {
1209
+ "epoch": 2.08,
1210
+ "learning_rate": 3.960292580982236e-05,
1211
+ "loss": 0.0152,
1212
+ "step": 1990
1213
+ },
1214
+ {
1215
+ "epoch": 2.09,
1216
+ "learning_rate": 3.955067920585162e-05,
1217
+ "loss": 0.0063,
1218
+ "step": 2000
1219
+ },
1220
+ {
1221
+ "epoch": 2.09,
1222
+ "eval_acc": 0.660049626153723,
1223
+ "eval_cer": 0.04602510460251046,
1224
+ "eval_loss": 0.023699596524238586,
1225
+ "eval_runtime": 127.5932,
1226
+ "eval_samples_per_second": 3.158,
1227
+ "eval_steps_per_second": 0.4,
1228
+ "step": 2000
1229
+ },
1230
+ {
1231
+ "epoch": 2.1,
1232
+ "learning_rate": 3.949843260188088e-05,
1233
+ "loss": 0.0147,
1234
+ "step": 2010
1235
+ },
1236
+ {
1237
+ "epoch": 2.11,
1238
+ "learning_rate": 3.944618599791014e-05,
1239
+ "loss": 0.0126,
1240
+ "step": 2020
1241
+ },
1242
+ {
1243
+ "epoch": 2.12,
1244
+ "learning_rate": 3.939393939393939e-05,
1245
+ "loss": 0.0143,
1246
+ "step": 2030
1247
+ },
1248
+ {
1249
+ "epoch": 2.13,
1250
+ "learning_rate": 3.934169278996865e-05,
1251
+ "loss": 0.0147,
1252
+ "step": 2040
1253
+ },
1254
+ {
1255
+ "epoch": 2.14,
1256
+ "learning_rate": 3.9289446185997915e-05,
1257
+ "loss": 0.0134,
1258
+ "step": 2050
1259
+ },
1260
+ {
1261
+ "epoch": 2.15,
1262
+ "learning_rate": 3.9237199582027165e-05,
1263
+ "loss": 0.013,
1264
+ "step": 2060
1265
+ },
1266
+ {
1267
+ "epoch": 2.16,
1268
+ "learning_rate": 3.918495297805643e-05,
1269
+ "loss": 0.0217,
1270
+ "step": 2070
1271
+ },
1272
+ {
1273
+ "epoch": 2.17,
1274
+ "learning_rate": 3.9132706374085686e-05,
1275
+ "loss": 0.0174,
1276
+ "step": 2080
1277
+ },
1278
+ {
1279
+ "epoch": 2.18,
1280
+ "learning_rate": 3.908045977011495e-05,
1281
+ "loss": 0.0116,
1282
+ "step": 2090
1283
+ },
1284
+ {
1285
+ "epoch": 2.19,
1286
+ "learning_rate": 3.90282131661442e-05,
1287
+ "loss": 0.0169,
1288
+ "step": 2100
1289
+ },
1290
+ {
1291
+ "epoch": 2.2,
1292
+ "learning_rate": 3.897596656217346e-05,
1293
+ "loss": 0.0199,
1294
+ "step": 2110
1295
+ },
1296
+ {
1297
+ "epoch": 2.22,
1298
+ "learning_rate": 3.892371995820272e-05,
1299
+ "loss": 0.0124,
1300
+ "step": 2120
1301
+ },
1302
+ {
1303
+ "epoch": 2.23,
1304
+ "learning_rate": 3.887147335423197e-05,
1305
+ "loss": 0.0114,
1306
+ "step": 2130
1307
+ },
1308
+ {
1309
+ "epoch": 2.24,
1310
+ "learning_rate": 3.8819226750261236e-05,
1311
+ "loss": 0.0128,
1312
+ "step": 2140
1313
+ },
1314
+ {
1315
+ "epoch": 2.25,
1316
+ "learning_rate": 3.876698014629049e-05,
1317
+ "loss": 0.0097,
1318
+ "step": 2150
1319
+ },
1320
+ {
1321
+ "epoch": 2.26,
1322
+ "learning_rate": 3.871473354231975e-05,
1323
+ "loss": 0.0132,
1324
+ "step": 2160
1325
+ },
1326
+ {
1327
+ "epoch": 2.27,
1328
+ "learning_rate": 3.866248693834901e-05,
1329
+ "loss": 0.0133,
1330
+ "step": 2170
1331
+ },
1332
+ {
1333
+ "epoch": 2.28,
1334
+ "learning_rate": 3.8610240334378265e-05,
1335
+ "loss": 0.0094,
1336
+ "step": 2180
1337
+ },
1338
+ {
1339
+ "epoch": 2.29,
1340
+ "learning_rate": 3.855799373040753e-05,
1341
+ "loss": 0.0185,
1342
+ "step": 2190
1343
+ },
1344
+ {
1345
+ "epoch": 2.3,
1346
+ "learning_rate": 3.850574712643678e-05,
1347
+ "loss": 0.017,
1348
+ "step": 2200
1349
+ },
1350
+ {
1351
+ "epoch": 2.31,
1352
+ "learning_rate": 3.845350052246604e-05,
1353
+ "loss": 0.0128,
1354
+ "step": 2210
1355
+ },
1356
+ {
1357
+ "epoch": 2.32,
1358
+ "learning_rate": 3.84012539184953e-05,
1359
+ "loss": 0.0121,
1360
+ "step": 2220
1361
+ },
1362
+ {
1363
+ "epoch": 2.33,
1364
+ "learning_rate": 3.834900731452456e-05,
1365
+ "loss": 0.015,
1366
+ "step": 2230
1367
+ },
1368
+ {
1369
+ "epoch": 2.34,
1370
+ "learning_rate": 3.8296760710553815e-05,
1371
+ "loss": 0.0077,
1372
+ "step": 2240
1373
+ },
1374
+ {
1375
+ "epoch": 2.35,
1376
+ "learning_rate": 3.824451410658307e-05,
1377
+ "loss": 0.0118,
1378
+ "step": 2250
1379
+ },
1380
+ {
1381
+ "epoch": 2.36,
1382
+ "learning_rate": 3.8192267502612336e-05,
1383
+ "loss": 0.0133,
1384
+ "step": 2260
1385
+ },
1386
+ {
1387
+ "epoch": 2.37,
1388
+ "learning_rate": 3.8140020898641586e-05,
1389
+ "loss": 0.0156,
1390
+ "step": 2270
1391
+ },
1392
+ {
1393
+ "epoch": 2.38,
1394
+ "learning_rate": 3.808777429467085e-05,
1395
+ "loss": 0.0131,
1396
+ "step": 2280
1397
+ },
1398
+ {
1399
+ "epoch": 2.39,
1400
+ "learning_rate": 3.803552769070011e-05,
1401
+ "loss": 0.01,
1402
+ "step": 2290
1403
+ },
1404
+ {
1405
+ "epoch": 2.4,
1406
+ "learning_rate": 3.7983281086729364e-05,
1407
+ "loss": 0.0166,
1408
+ "step": 2300
1409
+ },
1410
+ {
1411
+ "epoch": 2.41,
1412
+ "learning_rate": 3.793103448275862e-05,
1413
+ "loss": 0.0125,
1414
+ "step": 2310
1415
+ },
1416
+ {
1417
+ "epoch": 2.42,
1418
+ "learning_rate": 3.787878787878788e-05,
1419
+ "loss": 0.0092,
1420
+ "step": 2320
1421
+ },
1422
+ {
1423
+ "epoch": 2.43,
1424
+ "learning_rate": 3.782654127481714e-05,
1425
+ "loss": 0.0105,
1426
+ "step": 2330
1427
+ },
1428
+ {
1429
+ "epoch": 2.45,
1430
+ "learning_rate": 3.777429467084639e-05,
1431
+ "loss": 0.0138,
1432
+ "step": 2340
1433
+ },
1434
+ {
1435
+ "epoch": 2.46,
1436
+ "learning_rate": 3.772204806687566e-05,
1437
+ "loss": 0.0136,
1438
+ "step": 2350
1439
+ },
1440
+ {
1441
+ "epoch": 2.47,
1442
+ "learning_rate": 3.7669801462904914e-05,
1443
+ "loss": 0.0153,
1444
+ "step": 2360
1445
+ },
1446
+ {
1447
+ "epoch": 2.48,
1448
+ "learning_rate": 3.761755485893417e-05,
1449
+ "loss": 0.0094,
1450
+ "step": 2370
1451
+ },
1452
+ {
1453
+ "epoch": 2.49,
1454
+ "learning_rate": 3.756530825496343e-05,
1455
+ "loss": 0.0081,
1456
+ "step": 2380
1457
+ },
1458
+ {
1459
+ "epoch": 2.5,
1460
+ "learning_rate": 3.7513061650992686e-05,
1461
+ "loss": 0.0169,
1462
+ "step": 2390
1463
+ },
1464
+ {
1465
+ "epoch": 2.51,
1466
+ "learning_rate": 3.746081504702195e-05,
1467
+ "loss": 0.0086,
1468
+ "step": 2400
1469
+ },
1470
+ {
1471
+ "epoch": 2.52,
1472
+ "learning_rate": 3.740856844305121e-05,
1473
+ "loss": 0.0124,
1474
+ "step": 2410
1475
+ },
1476
+ {
1477
+ "epoch": 2.53,
1478
+ "learning_rate": 3.735632183908046e-05,
1479
+ "loss": 0.0143,
1480
+ "step": 2420
1481
+ },
1482
+ {
1483
+ "epoch": 2.54,
1484
+ "learning_rate": 3.730407523510972e-05,
1485
+ "loss": 0.0092,
1486
+ "step": 2430
1487
+ },
1488
+ {
1489
+ "epoch": 2.55,
1490
+ "learning_rate": 3.725182863113898e-05,
1491
+ "loss": 0.0147,
1492
+ "step": 2440
1493
+ },
1494
+ {
1495
+ "epoch": 2.56,
1496
+ "learning_rate": 3.7199582027168236e-05,
1497
+ "loss": 0.009,
1498
+ "step": 2450
1499
+ },
1500
+ {
1501
+ "epoch": 2.57,
1502
+ "learning_rate": 3.714733542319749e-05,
1503
+ "loss": 0.0083,
1504
+ "step": 2460
1505
+ },
1506
+ {
1507
+ "epoch": 2.58,
1508
+ "learning_rate": 3.709508881922675e-05,
1509
+ "loss": 0.0146,
1510
+ "step": 2470
1511
+ },
1512
+ {
1513
+ "epoch": 2.59,
1514
+ "learning_rate": 3.7042842215256014e-05,
1515
+ "loss": 0.0142,
1516
+ "step": 2480
1517
+ },
1518
+ {
1519
+ "epoch": 2.6,
1520
+ "learning_rate": 3.6990595611285264e-05,
1521
+ "loss": 0.0097,
1522
+ "step": 2490
1523
+ },
1524
+ {
1525
+ "epoch": 2.61,
1526
+ "learning_rate": 3.693834900731453e-05,
1527
+ "loss": 0.0067,
1528
+ "step": 2500
1529
+ },
1530
+ {
1531
+ "epoch": 2.62,
1532
+ "learning_rate": 3.6886102403343786e-05,
1533
+ "loss": 0.0085,
1534
+ "step": 2510
1535
+ },
1536
+ {
1537
+ "epoch": 2.63,
1538
+ "learning_rate": 3.683385579937304e-05,
1539
+ "loss": 0.01,
1540
+ "step": 2520
1541
+ },
1542
+ {
1543
+ "epoch": 2.64,
1544
+ "learning_rate": 3.67816091954023e-05,
1545
+ "loss": 0.0128,
1546
+ "step": 2530
1547
+ },
1548
+ {
1549
+ "epoch": 2.65,
1550
+ "learning_rate": 3.672936259143156e-05,
1551
+ "loss": 0.0113,
1552
+ "step": 2540
1553
+ },
1554
+ {
1555
+ "epoch": 2.66,
1556
+ "learning_rate": 3.667711598746082e-05,
1557
+ "loss": 0.0098,
1558
+ "step": 2550
1559
+ },
1560
+ {
1561
+ "epoch": 2.68,
1562
+ "learning_rate": 3.662486938349007e-05,
1563
+ "loss": 0.0159,
1564
+ "step": 2560
1565
+ },
1566
+ {
1567
+ "epoch": 2.69,
1568
+ "learning_rate": 3.6572622779519335e-05,
1569
+ "loss": 0.0188,
1570
+ "step": 2570
1571
+ },
1572
+ {
1573
+ "epoch": 2.7,
1574
+ "learning_rate": 3.652037617554859e-05,
1575
+ "loss": 0.0149,
1576
+ "step": 2580
1577
+ },
1578
+ {
1579
+ "epoch": 2.71,
1580
+ "learning_rate": 3.646812957157785e-05,
1581
+ "loss": 0.0159,
1582
+ "step": 2590
1583
+ },
1584
+ {
1585
+ "epoch": 2.72,
1586
+ "learning_rate": 3.641588296760711e-05,
1587
+ "loss": 0.0093,
1588
+ "step": 2600
1589
+ },
1590
+ {
1591
+ "epoch": 2.73,
1592
+ "learning_rate": 3.6363636363636364e-05,
1593
+ "loss": 0.0115,
1594
+ "step": 2610
1595
+ },
1596
+ {
1597
+ "epoch": 2.74,
1598
+ "learning_rate": 3.631138975966563e-05,
1599
+ "loss": 0.0138,
1600
+ "step": 2620
1601
+ },
1602
+ {
1603
+ "epoch": 2.75,
1604
+ "learning_rate": 3.625914315569488e-05,
1605
+ "loss": 0.011,
1606
+ "step": 2630
1607
+ },
1608
+ {
1609
+ "epoch": 2.76,
1610
+ "learning_rate": 3.620689655172414e-05,
1611
+ "loss": 0.0144,
1612
+ "step": 2640
1613
+ },
1614
+ {
1615
+ "epoch": 2.77,
1616
+ "learning_rate": 3.61546499477534e-05,
1617
+ "loss": 0.0098,
1618
+ "step": 2650
1619
+ },
1620
+ {
1621
+ "epoch": 2.78,
1622
+ "learning_rate": 3.610240334378266e-05,
1623
+ "loss": 0.0132,
1624
+ "step": 2660
1625
+ },
1626
+ {
1627
+ "epoch": 2.79,
1628
+ "learning_rate": 3.6050156739811914e-05,
1629
+ "loss": 0.0125,
1630
+ "step": 2670
1631
+ },
1632
+ {
1633
+ "epoch": 2.8,
1634
+ "learning_rate": 3.599791013584117e-05,
1635
+ "loss": 0.0118,
1636
+ "step": 2680
1637
+ },
1638
+ {
1639
+ "epoch": 2.81,
1640
+ "learning_rate": 3.5945663531870435e-05,
1641
+ "loss": 0.0131,
1642
+ "step": 2690
1643
+ },
1644
+ {
1645
+ "epoch": 2.82,
1646
+ "learning_rate": 3.5893416927899686e-05,
1647
+ "loss": 0.0081,
1648
+ "step": 2700
1649
+ },
1650
+ {
1651
+ "epoch": 2.83,
1652
+ "learning_rate": 3.584117032392895e-05,
1653
+ "loss": 0.0115,
1654
+ "step": 2710
1655
+ },
1656
+ {
1657
+ "epoch": 2.84,
1658
+ "learning_rate": 3.578892371995821e-05,
1659
+ "loss": 0.0109,
1660
+ "step": 2720
1661
+ },
1662
+ {
1663
+ "epoch": 2.85,
1664
+ "learning_rate": 3.573667711598746e-05,
1665
+ "loss": 0.0149,
1666
+ "step": 2730
1667
+ },
1668
+ {
1669
+ "epoch": 2.86,
1670
+ "learning_rate": 3.568443051201672e-05,
1671
+ "loss": 0.0093,
1672
+ "step": 2740
1673
+ },
1674
+ {
1675
+ "epoch": 2.87,
1676
+ "learning_rate": 3.563218390804598e-05,
1677
+ "loss": 0.0124,
1678
+ "step": 2750
1679
+ },
1680
+ {
1681
+ "epoch": 2.88,
1682
+ "learning_rate": 3.557993730407524e-05,
1683
+ "loss": 0.0094,
1684
+ "step": 2760
1685
+ },
1686
+ {
1687
+ "epoch": 2.89,
1688
+ "learning_rate": 3.552769070010449e-05,
1689
+ "loss": 0.0105,
1690
+ "step": 2770
1691
+ },
1692
+ {
1693
+ "epoch": 2.9,
1694
+ "learning_rate": 3.547544409613375e-05,
1695
+ "loss": 0.0221,
1696
+ "step": 2780
1697
+ },
1698
+ {
1699
+ "epoch": 2.92,
1700
+ "learning_rate": 3.5423197492163014e-05,
1701
+ "loss": 0.0088,
1702
+ "step": 2790
1703
+ },
1704
+ {
1705
+ "epoch": 2.93,
1706
+ "learning_rate": 3.5370950888192264e-05,
1707
+ "loss": 0.0077,
1708
+ "step": 2800
1709
+ },
1710
+ {
1711
+ "epoch": 2.94,
1712
+ "learning_rate": 3.531870428422153e-05,
1713
+ "loss": 0.0106,
1714
+ "step": 2810
1715
+ },
1716
+ {
1717
+ "epoch": 2.95,
1718
+ "learning_rate": 3.5266457680250785e-05,
1719
+ "loss": 0.0113,
1720
+ "step": 2820
1721
+ },
1722
+ {
1723
+ "epoch": 2.96,
1724
+ "learning_rate": 3.521421107628005e-05,
1725
+ "loss": 0.012,
1726
+ "step": 2830
1727
+ },
1728
+ {
1729
+ "epoch": 2.97,
1730
+ "learning_rate": 3.51619644723093e-05,
1731
+ "loss": 0.0094,
1732
+ "step": 2840
1733
+ },
1734
+ {
1735
+ "epoch": 2.98,
1736
+ "learning_rate": 3.510971786833856e-05,
1737
+ "loss": 0.0088,
1738
+ "step": 2850
1739
+ },
1740
+ {
1741
+ "epoch": 2.99,
1742
+ "learning_rate": 3.505747126436782e-05,
1743
+ "loss": 0.0162,
1744
+ "step": 2860
1745
+ },
1746
+ {
1747
+ "epoch": 3.0,
1748
+ "learning_rate": 3.500522466039707e-05,
1749
+ "loss": 0.0104,
1750
+ "step": 2870
1751
+ },
1752
+ {
1753
+ "epoch": 3.01,
1754
+ "learning_rate": 3.4952978056426335e-05,
1755
+ "loss": 0.0045,
1756
+ "step": 2880
1757
+ },
1758
+ {
1759
+ "epoch": 3.02,
1760
+ "learning_rate": 3.490073145245559e-05,
1761
+ "loss": 0.0076,
1762
+ "step": 2890
1763
+ },
1764
+ {
1765
+ "epoch": 3.03,
1766
+ "learning_rate": 3.484848484848485e-05,
1767
+ "loss": 0.0092,
1768
+ "step": 2900
1769
+ },
1770
+ {
1771
+ "epoch": 3.04,
1772
+ "learning_rate": 3.479623824451411e-05,
1773
+ "loss": 0.0088,
1774
+ "step": 2910
1775
+ },
1776
+ {
1777
+ "epoch": 3.05,
1778
+ "learning_rate": 3.4743991640543364e-05,
1779
+ "loss": 0.0068,
1780
+ "step": 2920
1781
+ },
1782
+ {
1783
+ "epoch": 3.06,
1784
+ "learning_rate": 3.469174503657263e-05,
1785
+ "loss": 0.0037,
1786
+ "step": 2930
1787
+ },
1788
+ {
1789
+ "epoch": 3.07,
1790
+ "learning_rate": 3.463949843260188e-05,
1791
+ "loss": 0.008,
1792
+ "step": 2940
1793
+ },
1794
+ {
1795
+ "epoch": 3.08,
1796
+ "learning_rate": 3.458725182863114e-05,
1797
+ "loss": 0.0053,
1798
+ "step": 2950
1799
+ },
1800
+ {
1801
+ "epoch": 3.09,
1802
+ "learning_rate": 3.45350052246604e-05,
1803
+ "loss": 0.0085,
1804
+ "step": 2960
1805
+ },
1806
+ {
1807
+ "epoch": 3.1,
1808
+ "learning_rate": 3.4482758620689657e-05,
1809
+ "loss": 0.0122,
1810
+ "step": 2970
1811
+ },
1812
+ {
1813
+ "epoch": 3.11,
1814
+ "learning_rate": 3.4430512016718914e-05,
1815
+ "loss": 0.0096,
1816
+ "step": 2980
1817
+ },
1818
+ {
1819
+ "epoch": 3.12,
1820
+ "learning_rate": 3.437826541274817e-05,
1821
+ "loss": 0.0079,
1822
+ "step": 2990
1823
+ },
1824
+ {
1825
+ "epoch": 3.13,
1826
+ "learning_rate": 3.4326018808777435e-05,
1827
+ "loss": 0.01,
1828
+ "step": 3000
1829
+ },
1830
+ {
1831
+ "epoch": 3.13,
1832
+ "eval_acc": 0.7444168716019433,
1833
+ "eval_cer": 0.031845653184565316,
1834
+ "eval_loss": 0.017891723662614822,
1835
+ "eval_runtime": 127.5496,
1836
+ "eval_samples_per_second": 3.16,
1837
+ "eval_steps_per_second": 0.4,
1838
+ "step": 3000
1839
+ },
1840
+ {
1841
+ "epoch": 3.15,
1842
+ "learning_rate": 3.4273772204806685e-05,
1843
+ "loss": 0.0105,
1844
+ "step": 3010
1845
+ },
1846
+ {
1847
+ "epoch": 3.16,
1848
+ "learning_rate": 3.422152560083595e-05,
1849
+ "loss": 0.0091,
1850
+ "step": 3020
1851
+ },
1852
+ {
1853
+ "epoch": 3.17,
1854
+ "learning_rate": 3.4169278996865206e-05,
1855
+ "loss": 0.005,
1856
+ "step": 3030
1857
+ },
1858
+ {
1859
+ "epoch": 3.18,
1860
+ "learning_rate": 3.4117032392894464e-05,
1861
+ "loss": 0.0069,
1862
+ "step": 3040
1863
+ },
1864
+ {
1865
+ "epoch": 3.19,
1866
+ "learning_rate": 3.406478578892372e-05,
1867
+ "loss": 0.0066,
1868
+ "step": 3050
1869
+ },
1870
+ {
1871
+ "epoch": 3.2,
1872
+ "learning_rate": 3.401253918495298e-05,
1873
+ "loss": 0.0065,
1874
+ "step": 3060
1875
+ },
1876
+ {
1877
+ "epoch": 3.21,
1878
+ "learning_rate": 3.396029258098224e-05,
1879
+ "loss": 0.0098,
1880
+ "step": 3070
1881
+ },
1882
+ {
1883
+ "epoch": 3.22,
1884
+ "learning_rate": 3.390804597701149e-05,
1885
+ "loss": 0.0089,
1886
+ "step": 3080
1887
+ },
1888
+ {
1889
+ "epoch": 3.23,
1890
+ "learning_rate": 3.3855799373040756e-05,
1891
+ "loss": 0.0082,
1892
+ "step": 3090
1893
+ },
1894
+ {
1895
+ "epoch": 3.24,
1896
+ "learning_rate": 3.3803552769070014e-05,
1897
+ "loss": 0.0089,
1898
+ "step": 3100
1899
+ },
1900
+ {
1901
+ "epoch": 3.25,
1902
+ "learning_rate": 3.375130616509927e-05,
1903
+ "loss": 0.0097,
1904
+ "step": 3110
1905
+ },
1906
+ {
1907
+ "epoch": 3.26,
1908
+ "learning_rate": 3.369905956112853e-05,
1909
+ "loss": 0.0079,
1910
+ "step": 3120
1911
+ },
1912
+ {
1913
+ "epoch": 3.27,
1914
+ "learning_rate": 3.3646812957157785e-05,
1915
+ "loss": 0.0079,
1916
+ "step": 3130
1917
+ },
1918
+ {
1919
+ "epoch": 3.28,
1920
+ "learning_rate": 3.359456635318705e-05,
1921
+ "loss": 0.0062,
1922
+ "step": 3140
1923
+ },
1924
+ {
1925
+ "epoch": 3.29,
1926
+ "learning_rate": 3.35423197492163e-05,
1927
+ "loss": 0.0095,
1928
+ "step": 3150
1929
+ },
1930
+ {
1931
+ "epoch": 3.3,
1932
+ "learning_rate": 3.3490073145245557e-05,
1933
+ "loss": 0.0089,
1934
+ "step": 3160
1935
+ },
1936
+ {
1937
+ "epoch": 3.31,
1938
+ "learning_rate": 3.343782654127482e-05,
1939
+ "loss": 0.0107,
1940
+ "step": 3170
1941
+ },
1942
+ {
1943
+ "epoch": 3.32,
1944
+ "learning_rate": 3.338557993730408e-05,
1945
+ "loss": 0.0135,
1946
+ "step": 3180
1947
+ },
1948
+ {
1949
+ "epoch": 3.33,
1950
+ "learning_rate": 3.3333333333333335e-05,
1951
+ "loss": 0.006,
1952
+ "step": 3190
1953
+ },
1954
+ {
1955
+ "epoch": 3.34,
1956
+ "learning_rate": 3.328108672936259e-05,
1957
+ "loss": 0.0078,
1958
+ "step": 3200
1959
+ },
1960
+ {
1961
+ "epoch": 3.35,
1962
+ "learning_rate": 3.322884012539185e-05,
1963
+ "loss": 0.005,
1964
+ "step": 3210
1965
+ },
1966
+ {
1967
+ "epoch": 3.36,
1968
+ "learning_rate": 3.3176593521421106e-05,
1969
+ "loss": 0.0073,
1970
+ "step": 3220
1971
+ },
1972
+ {
1973
+ "epoch": 3.38,
1974
+ "learning_rate": 3.3124346917450364e-05,
1975
+ "loss": 0.0084,
1976
+ "step": 3230
1977
+ },
1978
+ {
1979
+ "epoch": 3.39,
1980
+ "learning_rate": 3.307210031347963e-05,
1981
+ "loss": 0.0077,
1982
+ "step": 3240
1983
+ },
1984
+ {
1985
+ "epoch": 3.4,
1986
+ "learning_rate": 3.3019853709508885e-05,
1987
+ "loss": 0.0065,
1988
+ "step": 3250
1989
+ },
1990
+ {
1991
+ "epoch": 3.41,
1992
+ "learning_rate": 3.296760710553814e-05,
1993
+ "loss": 0.0052,
1994
+ "step": 3260
1995
+ },
1996
+ {
1997
+ "epoch": 3.42,
1998
+ "learning_rate": 3.29153605015674e-05,
1999
+ "loss": 0.0078,
2000
+ "step": 3270
2001
+ },
2002
+ {
2003
+ "epoch": 3.43,
2004
+ "learning_rate": 3.2863113897596656e-05,
2005
+ "loss": 0.0056,
2006
+ "step": 3280
2007
+ },
2008
+ {
2009
+ "epoch": 3.44,
2010
+ "learning_rate": 3.2810867293625914e-05,
2011
+ "loss": 0.0084,
2012
+ "step": 3290
2013
+ },
2014
+ {
2015
+ "epoch": 3.45,
2016
+ "learning_rate": 3.275862068965517e-05,
2017
+ "loss": 0.0077,
2018
+ "step": 3300
2019
+ },
2020
+ {
2021
+ "epoch": 3.46,
2022
+ "learning_rate": 3.2706374085684435e-05,
2023
+ "loss": 0.0071,
2024
+ "step": 3310
2025
+ },
2026
+ {
2027
+ "epoch": 3.47,
2028
+ "learning_rate": 3.265412748171369e-05,
2029
+ "loss": 0.0084,
2030
+ "step": 3320
2031
+ },
2032
+ {
2033
+ "epoch": 3.48,
2034
+ "learning_rate": 3.260188087774295e-05,
2035
+ "loss": 0.0079,
2036
+ "step": 3330
2037
+ },
2038
+ {
2039
+ "epoch": 3.49,
2040
+ "learning_rate": 3.2549634273772206e-05,
2041
+ "loss": 0.0058,
2042
+ "step": 3340
2043
+ },
2044
+ {
2045
+ "epoch": 3.5,
2046
+ "learning_rate": 3.2497387669801463e-05,
2047
+ "loss": 0.0074,
2048
+ "step": 3350
2049
+ },
2050
+ {
2051
+ "epoch": 3.51,
2052
+ "learning_rate": 3.244514106583073e-05,
2053
+ "loss": 0.007,
2054
+ "step": 3360
2055
+ },
2056
+ {
2057
+ "epoch": 3.52,
2058
+ "learning_rate": 3.239289446185998e-05,
2059
+ "loss": 0.0089,
2060
+ "step": 3370
2061
+ },
2062
+ {
2063
+ "epoch": 3.53,
2064
+ "learning_rate": 3.234064785788924e-05,
2065
+ "loss": 0.0059,
2066
+ "step": 3380
2067
+ },
2068
+ {
2069
+ "epoch": 3.54,
2070
+ "learning_rate": 3.22884012539185e-05,
2071
+ "loss": 0.0082,
2072
+ "step": 3390
2073
+ },
2074
+ {
2075
+ "epoch": 3.55,
2076
+ "learning_rate": 3.2236154649947756e-05,
2077
+ "loss": 0.0107,
2078
+ "step": 3400
2079
+ },
2080
+ {
2081
+ "epoch": 3.56,
2082
+ "learning_rate": 3.218390804597701e-05,
2083
+ "loss": 0.0063,
2084
+ "step": 3410
2085
+ },
2086
+ {
2087
+ "epoch": 3.57,
2088
+ "learning_rate": 3.213166144200627e-05,
2089
+ "loss": 0.0059,
2090
+ "step": 3420
2091
+ },
2092
+ {
2093
+ "epoch": 3.58,
2094
+ "learning_rate": 3.2079414838035534e-05,
2095
+ "loss": 0.0074,
2096
+ "step": 3430
2097
+ },
2098
+ {
2099
+ "epoch": 3.59,
2100
+ "learning_rate": 3.2027168234064785e-05,
2101
+ "loss": 0.011,
2102
+ "step": 3440
2103
+ },
2104
+ {
2105
+ "epoch": 3.61,
2106
+ "learning_rate": 3.197492163009405e-05,
2107
+ "loss": 0.0107,
2108
+ "step": 3450
2109
+ },
2110
+ {
2111
+ "epoch": 3.62,
2112
+ "learning_rate": 3.1922675026123306e-05,
2113
+ "loss": 0.0071,
2114
+ "step": 3460
2115
+ },
2116
+ {
2117
+ "epoch": 3.63,
2118
+ "learning_rate": 3.1870428422152556e-05,
2119
+ "loss": 0.0073,
2120
+ "step": 3470
2121
+ },
2122
+ {
2123
+ "epoch": 3.64,
2124
+ "learning_rate": 3.181818181818182e-05,
2125
+ "loss": 0.0079,
2126
+ "step": 3480
2127
+ },
2128
+ {
2129
+ "epoch": 3.65,
2130
+ "learning_rate": 3.176593521421108e-05,
2131
+ "loss": 0.0068,
2132
+ "step": 3490
2133
+ },
2134
+ {
2135
+ "epoch": 3.66,
2136
+ "learning_rate": 3.171368861024034e-05,
2137
+ "loss": 0.0085,
2138
+ "step": 3500
2139
+ },
2140
+ {
2141
+ "epoch": 3.67,
2142
+ "learning_rate": 3.166144200626959e-05,
2143
+ "loss": 0.0087,
2144
+ "step": 3510
2145
+ },
2146
+ {
2147
+ "epoch": 3.68,
2148
+ "learning_rate": 3.160919540229885e-05,
2149
+ "loss": 0.005,
2150
+ "step": 3520
2151
+ },
2152
+ {
2153
+ "epoch": 3.69,
2154
+ "learning_rate": 3.155694879832811e-05,
2155
+ "loss": 0.0051,
2156
+ "step": 3530
2157
+ },
2158
+ {
2159
+ "epoch": 3.7,
2160
+ "learning_rate": 3.1504702194357363e-05,
2161
+ "loss": 0.0076,
2162
+ "step": 3540
2163
+ },
2164
+ {
2165
+ "epoch": 3.71,
2166
+ "learning_rate": 3.145245559038663e-05,
2167
+ "loss": 0.0054,
2168
+ "step": 3550
2169
+ },
2170
+ {
2171
+ "epoch": 3.72,
2172
+ "learning_rate": 3.1400208986415885e-05,
2173
+ "loss": 0.0083,
2174
+ "step": 3560
2175
+ },
2176
+ {
2177
+ "epoch": 3.73,
2178
+ "learning_rate": 3.134796238244515e-05,
2179
+ "loss": 0.0076,
2180
+ "step": 3570
2181
+ },
2182
+ {
2183
+ "epoch": 3.74,
2184
+ "learning_rate": 3.12957157784744e-05,
2185
+ "loss": 0.0094,
2186
+ "step": 3580
2187
+ },
2188
+ {
2189
+ "epoch": 3.75,
2190
+ "learning_rate": 3.1243469174503656e-05,
2191
+ "loss": 0.0083,
2192
+ "step": 3590
2193
+ },
2194
+ {
2195
+ "epoch": 3.76,
2196
+ "learning_rate": 3.119122257053292e-05,
2197
+ "loss": 0.0073,
2198
+ "step": 3600
2199
+ },
2200
+ {
2201
+ "epoch": 3.77,
2202
+ "learning_rate": 3.113897596656217e-05,
2203
+ "loss": 0.0063,
2204
+ "step": 3610
2205
+ },
2206
+ {
2207
+ "epoch": 3.78,
2208
+ "learning_rate": 3.1086729362591434e-05,
2209
+ "loss": 0.0086,
2210
+ "step": 3620
2211
+ },
2212
+ {
2213
+ "epoch": 3.79,
2214
+ "learning_rate": 3.103448275862069e-05,
2215
+ "loss": 0.0118,
2216
+ "step": 3630
2217
+ },
2218
+ {
2219
+ "epoch": 3.8,
2220
+ "learning_rate": 3.098223615464995e-05,
2221
+ "loss": 0.0088,
2222
+ "step": 3640
2223
+ },
2224
+ {
2225
+ "epoch": 3.81,
2226
+ "learning_rate": 3.0929989550679206e-05,
2227
+ "loss": 0.0064,
2228
+ "step": 3650
2229
+ },
2230
+ {
2231
+ "epoch": 3.82,
2232
+ "learning_rate": 3.087774294670846e-05,
2233
+ "loss": 0.0065,
2234
+ "step": 3660
2235
+ },
2236
+ {
2237
+ "epoch": 3.83,
2238
+ "learning_rate": 3.082549634273773e-05,
2239
+ "loss": 0.0057,
2240
+ "step": 3670
2241
+ },
2242
+ {
2243
+ "epoch": 3.85,
2244
+ "learning_rate": 3.077324973876698e-05,
2245
+ "loss": 0.0129,
2246
+ "step": 3680
2247
+ },
2248
+ {
2249
+ "epoch": 3.86,
2250
+ "learning_rate": 3.072100313479624e-05,
2251
+ "loss": 0.0119,
2252
+ "step": 3690
2253
+ },
2254
+ {
2255
+ "epoch": 3.87,
2256
+ "learning_rate": 3.06687565308255e-05,
2257
+ "loss": 0.0074,
2258
+ "step": 3700
2259
+ },
2260
+ {
2261
+ "epoch": 3.88,
2262
+ "learning_rate": 3.0616509926854756e-05,
2263
+ "loss": 0.0058,
2264
+ "step": 3710
2265
+ },
2266
+ {
2267
+ "epoch": 3.89,
2268
+ "learning_rate": 3.056426332288401e-05,
2269
+ "loss": 0.0049,
2270
+ "step": 3720
2271
+ },
2272
+ {
2273
+ "epoch": 3.9,
2274
+ "learning_rate": 3.0512016718913274e-05,
2275
+ "loss": 0.0075,
2276
+ "step": 3730
2277
+ },
2278
+ {
2279
+ "epoch": 3.91,
2280
+ "learning_rate": 3.045977011494253e-05,
2281
+ "loss": 0.0087,
2282
+ "step": 3740
2283
+ },
2284
+ {
2285
+ "epoch": 3.92,
2286
+ "learning_rate": 3.0407523510971785e-05,
2287
+ "loss": 0.0071,
2288
+ "step": 3750
2289
+ },
2290
+ {
2291
+ "epoch": 3.93,
2292
+ "learning_rate": 3.0355276907001045e-05,
2293
+ "loss": 0.0101,
2294
+ "step": 3760
2295
+ },
2296
+ {
2297
+ "epoch": 3.94,
2298
+ "learning_rate": 3.0303030303030306e-05,
2299
+ "loss": 0.008,
2300
+ "step": 3770
2301
+ },
2302
+ {
2303
+ "epoch": 3.95,
2304
+ "learning_rate": 3.0250783699059566e-05,
2305
+ "loss": 0.0077,
2306
+ "step": 3780
2307
+ },
2308
+ {
2309
+ "epoch": 3.96,
2310
+ "learning_rate": 3.019853709508882e-05,
2311
+ "loss": 0.0114,
2312
+ "step": 3790
2313
+ },
2314
+ {
2315
+ "epoch": 3.97,
2316
+ "learning_rate": 3.0146290491118077e-05,
2317
+ "loss": 0.0067,
2318
+ "step": 3800
2319
+ },
2320
+ {
2321
+ "epoch": 3.98,
2322
+ "learning_rate": 3.0094043887147338e-05,
2323
+ "loss": 0.0067,
2324
+ "step": 3810
2325
+ },
2326
+ {
2327
+ "epoch": 3.99,
2328
+ "learning_rate": 3.004179728317659e-05,
2329
+ "loss": 0.0099,
2330
+ "step": 3820
2331
+ },
2332
+ {
2333
+ "epoch": 4.0,
2334
+ "learning_rate": 2.9989550679205852e-05,
2335
+ "loss": 0.0078,
2336
+ "step": 3830
2337
+ },
2338
+ {
2339
+ "epoch": 4.01,
2340
+ "learning_rate": 2.9937304075235113e-05,
2341
+ "loss": 0.0042,
2342
+ "step": 3840
2343
+ },
2344
+ {
2345
+ "epoch": 4.02,
2346
+ "learning_rate": 2.988505747126437e-05,
2347
+ "loss": 0.0056,
2348
+ "step": 3850
2349
+ },
2350
+ {
2351
+ "epoch": 4.03,
2352
+ "learning_rate": 2.9832810867293627e-05,
2353
+ "loss": 0.0043,
2354
+ "step": 3860
2355
+ },
2356
+ {
2357
+ "epoch": 4.04,
2358
+ "learning_rate": 2.9780564263322884e-05,
2359
+ "loss": 0.0042,
2360
+ "step": 3870
2361
+ },
2362
+ {
2363
+ "epoch": 4.05,
2364
+ "learning_rate": 2.9728317659352145e-05,
2365
+ "loss": 0.0062,
2366
+ "step": 3880
2367
+ },
2368
+ {
2369
+ "epoch": 4.06,
2370
+ "learning_rate": 2.96760710553814e-05,
2371
+ "loss": 0.0057,
2372
+ "step": 3890
2373
+ },
2374
+ {
2375
+ "epoch": 4.08,
2376
+ "learning_rate": 2.962382445141066e-05,
2377
+ "loss": 0.0069,
2378
+ "step": 3900
2379
+ },
2380
+ {
2381
+ "epoch": 4.09,
2382
+ "learning_rate": 2.957157784743992e-05,
2383
+ "loss": 0.0072,
2384
+ "step": 3910
2385
+ },
2386
+ {
2387
+ "epoch": 4.1,
2388
+ "learning_rate": 2.9519331243469177e-05,
2389
+ "loss": 0.0072,
2390
+ "step": 3920
2391
+ },
2392
+ {
2393
+ "epoch": 4.11,
2394
+ "learning_rate": 2.946708463949843e-05,
2395
+ "loss": 0.0032,
2396
+ "step": 3930
2397
+ },
2398
+ {
2399
+ "epoch": 4.12,
2400
+ "learning_rate": 2.941483803552769e-05,
2401
+ "loss": 0.0089,
2402
+ "step": 3940
2403
+ },
2404
+ {
2405
+ "epoch": 4.13,
2406
+ "learning_rate": 2.9362591431556952e-05,
2407
+ "loss": 0.0043,
2408
+ "step": 3950
2409
+ },
2410
+ {
2411
+ "epoch": 4.14,
2412
+ "learning_rate": 2.9310344827586206e-05,
2413
+ "loss": 0.0063,
2414
+ "step": 3960
2415
+ },
2416
+ {
2417
+ "epoch": 4.15,
2418
+ "learning_rate": 2.9258098223615466e-05,
2419
+ "loss": 0.0045,
2420
+ "step": 3970
2421
+ },
2422
+ {
2423
+ "epoch": 4.16,
2424
+ "learning_rate": 2.9205851619644723e-05,
2425
+ "loss": 0.0058,
2426
+ "step": 3980
2427
+ },
2428
+ {
2429
+ "epoch": 4.17,
2430
+ "learning_rate": 2.9153605015673984e-05,
2431
+ "loss": 0.005,
2432
+ "step": 3990
2433
+ },
2434
+ {
2435
+ "epoch": 4.18,
2436
+ "learning_rate": 2.9101358411703238e-05,
2437
+ "loss": 0.0044,
2438
+ "step": 4000
2439
+ },
2440
+ {
2441
+ "epoch": 4.18,
2442
+ "eval_acc": 0.769230767322008,
2443
+ "eval_cer": 0.0299860529986053,
2444
+ "eval_loss": 0.01652107760310173,
2445
+ "eval_runtime": 126.6833,
2446
+ "eval_samples_per_second": 3.181,
2447
+ "eval_steps_per_second": 0.403,
2448
+ "step": 4000
2449
+ },
2450
+ {
2451
+ "epoch": 4.19,
2452
+ "learning_rate": 2.90491118077325e-05,
2453
+ "loss": 0.0068,
2454
+ "step": 4010
2455
+ },
2456
+ {
2457
+ "epoch": 4.2,
2458
+ "learning_rate": 2.899686520376176e-05,
2459
+ "loss": 0.0039,
2460
+ "step": 4020
2461
+ },
2462
+ {
2463
+ "epoch": 4.21,
2464
+ "learning_rate": 2.8944618599791013e-05,
2465
+ "loss": 0.01,
2466
+ "step": 4030
2467
+ },
2468
+ {
2469
+ "epoch": 4.22,
2470
+ "learning_rate": 2.8892371995820273e-05,
2471
+ "loss": 0.0046,
2472
+ "step": 4040
2473
+ },
2474
+ {
2475
+ "epoch": 4.23,
2476
+ "learning_rate": 2.884012539184953e-05,
2477
+ "loss": 0.0096,
2478
+ "step": 4050
2479
+ },
2480
+ {
2481
+ "epoch": 4.24,
2482
+ "learning_rate": 2.878787878787879e-05,
2483
+ "loss": 0.006,
2484
+ "step": 4060
2485
+ },
2486
+ {
2487
+ "epoch": 4.25,
2488
+ "learning_rate": 2.8735632183908045e-05,
2489
+ "loss": 0.0052,
2490
+ "step": 4070
2491
+ },
2492
+ {
2493
+ "epoch": 4.26,
2494
+ "learning_rate": 2.8683385579937305e-05,
2495
+ "loss": 0.005,
2496
+ "step": 4080
2497
+ },
2498
+ {
2499
+ "epoch": 4.27,
2500
+ "learning_rate": 2.8631138975966566e-05,
2501
+ "loss": 0.004,
2502
+ "step": 4090
2503
+ },
2504
+ {
2505
+ "epoch": 4.28,
2506
+ "learning_rate": 2.857889237199582e-05,
2507
+ "loss": 0.0045,
2508
+ "step": 4100
2509
+ },
2510
+ {
2511
+ "epoch": 4.29,
2512
+ "learning_rate": 2.8526645768025077e-05,
2513
+ "loss": 0.0037,
2514
+ "step": 4110
2515
+ },
2516
+ {
2517
+ "epoch": 4.31,
2518
+ "learning_rate": 2.8474399164054338e-05,
2519
+ "loss": 0.0125,
2520
+ "step": 4120
2521
+ },
2522
+ {
2523
+ "epoch": 4.32,
2524
+ "learning_rate": 2.8422152560083598e-05,
2525
+ "loss": 0.0057,
2526
+ "step": 4130
2527
+ },
2528
+ {
2529
+ "epoch": 4.33,
2530
+ "learning_rate": 2.8369905956112852e-05,
2531
+ "loss": 0.0057,
2532
+ "step": 4140
2533
+ },
2534
+ {
2535
+ "epoch": 4.34,
2536
+ "learning_rate": 2.8317659352142113e-05,
2537
+ "loss": 0.0062,
2538
+ "step": 4150
2539
+ },
2540
+ {
2541
+ "epoch": 4.35,
2542
+ "learning_rate": 2.8265412748171373e-05,
2543
+ "loss": 0.0074,
2544
+ "step": 4160
2545
+ },
2546
+ {
2547
+ "epoch": 4.36,
2548
+ "learning_rate": 2.8213166144200627e-05,
2549
+ "loss": 0.0034,
2550
+ "step": 4170
2551
+ },
2552
+ {
2553
+ "epoch": 4.37,
2554
+ "learning_rate": 2.8160919540229884e-05,
2555
+ "loss": 0.0055,
2556
+ "step": 4180
2557
+ },
2558
+ {
2559
+ "epoch": 4.38,
2560
+ "learning_rate": 2.8108672936259145e-05,
2561
+ "loss": 0.0055,
2562
+ "step": 4190
2563
+ },
2564
+ {
2565
+ "epoch": 4.39,
2566
+ "learning_rate": 2.8056426332288405e-05,
2567
+ "loss": 0.0041,
2568
+ "step": 4200
2569
+ },
2570
+ {
2571
+ "epoch": 4.4,
2572
+ "learning_rate": 2.800417972831766e-05,
2573
+ "loss": 0.0089,
2574
+ "step": 4210
2575
+ },
2576
+ {
2577
+ "epoch": 4.41,
2578
+ "learning_rate": 2.795193312434692e-05,
2579
+ "loss": 0.0045,
2580
+ "step": 4220
2581
+ },
2582
+ {
2583
+ "epoch": 4.42,
2584
+ "learning_rate": 2.7899686520376177e-05,
2585
+ "loss": 0.0056,
2586
+ "step": 4230
2587
+ },
2588
+ {
2589
+ "epoch": 4.43,
2590
+ "learning_rate": 2.7847439916405434e-05,
2591
+ "loss": 0.0036,
2592
+ "step": 4240
2593
+ },
2594
+ {
2595
+ "epoch": 4.44,
2596
+ "learning_rate": 2.779519331243469e-05,
2597
+ "loss": 0.0036,
2598
+ "step": 4250
2599
+ },
2600
+ {
2601
+ "epoch": 4.45,
2602
+ "learning_rate": 2.774294670846395e-05,
2603
+ "loss": 0.004,
2604
+ "step": 4260
2605
+ },
2606
+ {
2607
+ "epoch": 4.46,
2608
+ "learning_rate": 2.7690700104493212e-05,
2609
+ "loss": 0.0049,
2610
+ "step": 4270
2611
+ },
2612
+ {
2613
+ "epoch": 4.47,
2614
+ "learning_rate": 2.7638453500522466e-05,
2615
+ "loss": 0.0042,
2616
+ "step": 4280
2617
+ },
2618
+ {
2619
+ "epoch": 4.48,
2620
+ "learning_rate": 2.7586206896551727e-05,
2621
+ "loss": 0.0066,
2622
+ "step": 4290
2623
+ },
2624
+ {
2625
+ "epoch": 4.49,
2626
+ "learning_rate": 2.7533960292580984e-05,
2627
+ "loss": 0.0054,
2628
+ "step": 4300
2629
+ },
2630
+ {
2631
+ "epoch": 4.5,
2632
+ "learning_rate": 2.7481713688610244e-05,
2633
+ "loss": 0.0041,
2634
+ "step": 4310
2635
+ },
2636
+ {
2637
+ "epoch": 4.51,
2638
+ "learning_rate": 2.7429467084639498e-05,
2639
+ "loss": 0.0034,
2640
+ "step": 4320
2641
+ },
2642
+ {
2643
+ "epoch": 4.52,
2644
+ "learning_rate": 2.737722048066876e-05,
2645
+ "loss": 0.0065,
2646
+ "step": 4330
2647
+ },
2648
+ {
2649
+ "epoch": 4.54,
2650
+ "learning_rate": 2.732497387669802e-05,
2651
+ "loss": 0.0062,
2652
+ "step": 4340
2653
+ },
2654
+ {
2655
+ "epoch": 4.55,
2656
+ "learning_rate": 2.7272727272727273e-05,
2657
+ "loss": 0.0068,
2658
+ "step": 4350
2659
+ },
2660
+ {
2661
+ "epoch": 4.56,
2662
+ "learning_rate": 2.722048066875653e-05,
2663
+ "loss": 0.0045,
2664
+ "step": 4360
2665
+ },
2666
+ {
2667
+ "epoch": 4.57,
2668
+ "learning_rate": 2.716823406478579e-05,
2669
+ "loss": 0.0089,
2670
+ "step": 4370
2671
+ },
2672
+ {
2673
+ "epoch": 4.58,
2674
+ "learning_rate": 2.711598746081505e-05,
2675
+ "loss": 0.0046,
2676
+ "step": 4380
2677
+ },
2678
+ {
2679
+ "epoch": 4.59,
2680
+ "learning_rate": 2.7063740856844305e-05,
2681
+ "loss": 0.0061,
2682
+ "step": 4390
2683
+ },
2684
+ {
2685
+ "epoch": 4.6,
2686
+ "learning_rate": 2.7011494252873566e-05,
2687
+ "loss": 0.0036,
2688
+ "step": 4400
2689
+ },
2690
+ {
2691
+ "epoch": 4.61,
2692
+ "learning_rate": 2.6959247648902823e-05,
2693
+ "loss": 0.0048,
2694
+ "step": 4410
2695
+ },
2696
+ {
2697
+ "epoch": 4.62,
2698
+ "learning_rate": 2.690700104493208e-05,
2699
+ "loss": 0.0028,
2700
+ "step": 4420
2701
+ },
2702
+ {
2703
+ "epoch": 4.63,
2704
+ "learning_rate": 2.6854754440961337e-05,
2705
+ "loss": 0.0037,
2706
+ "step": 4430
2707
+ },
2708
+ {
2709
+ "epoch": 4.64,
2710
+ "learning_rate": 2.6802507836990598e-05,
2711
+ "loss": 0.0039,
2712
+ "step": 4440
2713
+ },
2714
+ {
2715
+ "epoch": 4.65,
2716
+ "learning_rate": 2.675026123301986e-05,
2717
+ "loss": 0.0077,
2718
+ "step": 4450
2719
+ },
2720
+ {
2721
+ "epoch": 4.66,
2722
+ "learning_rate": 2.6698014629049112e-05,
2723
+ "loss": 0.0035,
2724
+ "step": 4460
2725
+ },
2726
+ {
2727
+ "epoch": 4.67,
2728
+ "learning_rate": 2.6645768025078373e-05,
2729
+ "loss": 0.008,
2730
+ "step": 4470
2731
+ },
2732
+ {
2733
+ "epoch": 4.68,
2734
+ "learning_rate": 2.659352142110763e-05,
2735
+ "loss": 0.0053,
2736
+ "step": 4480
2737
+ },
2738
+ {
2739
+ "epoch": 4.69,
2740
+ "learning_rate": 2.6541274817136884e-05,
2741
+ "loss": 0.0031,
2742
+ "step": 4490
2743
+ },
2744
+ {
2745
+ "epoch": 4.7,
2746
+ "learning_rate": 2.6489028213166144e-05,
2747
+ "loss": 0.0038,
2748
+ "step": 4500
2749
+ },
2750
+ {
2751
+ "epoch": 4.71,
2752
+ "learning_rate": 2.6436781609195405e-05,
2753
+ "loss": 0.0042,
2754
+ "step": 4510
2755
+ },
2756
+ {
2757
+ "epoch": 4.72,
2758
+ "learning_rate": 2.6384535005224666e-05,
2759
+ "loss": 0.0073,
2760
+ "step": 4520
2761
+ },
2762
+ {
2763
+ "epoch": 4.73,
2764
+ "learning_rate": 2.633228840125392e-05,
2765
+ "loss": 0.0056,
2766
+ "step": 4530
2767
+ },
2768
+ {
2769
+ "epoch": 4.74,
2770
+ "learning_rate": 2.6280041797283177e-05,
2771
+ "loss": 0.0039,
2772
+ "step": 4540
2773
+ },
2774
+ {
2775
+ "epoch": 4.75,
2776
+ "learning_rate": 2.6227795193312437e-05,
2777
+ "loss": 0.0045,
2778
+ "step": 4550
2779
+ },
2780
+ {
2781
+ "epoch": 4.76,
2782
+ "learning_rate": 2.617554858934169e-05,
2783
+ "loss": 0.0065,
2784
+ "step": 4560
2785
+ },
2786
+ {
2787
+ "epoch": 4.78,
2788
+ "learning_rate": 2.612330198537095e-05,
2789
+ "loss": 0.0039,
2790
+ "step": 4570
2791
+ },
2792
+ {
2793
+ "epoch": 4.79,
2794
+ "learning_rate": 2.6071055381400212e-05,
2795
+ "loss": 0.0043,
2796
+ "step": 4580
2797
+ },
2798
+ {
2799
+ "epoch": 4.8,
2800
+ "learning_rate": 2.601880877742947e-05,
2801
+ "loss": 0.0033,
2802
+ "step": 4590
2803
+ },
2804
+ {
2805
+ "epoch": 4.81,
2806
+ "learning_rate": 2.5966562173458726e-05,
2807
+ "loss": 0.0035,
2808
+ "step": 4600
2809
+ },
2810
+ {
2811
+ "epoch": 4.82,
2812
+ "learning_rate": 2.5914315569487984e-05,
2813
+ "loss": 0.0046,
2814
+ "step": 4610
2815
+ },
2816
+ {
2817
+ "epoch": 4.83,
2818
+ "learning_rate": 2.5862068965517244e-05,
2819
+ "loss": 0.0026,
2820
+ "step": 4620
2821
+ },
2822
+ {
2823
+ "epoch": 4.84,
2824
+ "learning_rate": 2.5809822361546498e-05,
2825
+ "loss": 0.0031,
2826
+ "step": 4630
2827
+ },
2828
+ {
2829
+ "epoch": 4.85,
2830
+ "learning_rate": 2.575757575757576e-05,
2831
+ "loss": 0.0042,
2832
+ "step": 4640
2833
+ },
2834
+ {
2835
+ "epoch": 4.86,
2836
+ "learning_rate": 2.570532915360502e-05,
2837
+ "loss": 0.0037,
2838
+ "step": 4650
2839
+ },
2840
+ {
2841
+ "epoch": 4.87,
2842
+ "learning_rate": 2.5653082549634276e-05,
2843
+ "loss": 0.0034,
2844
+ "step": 4660
2845
+ },
2846
+ {
2847
+ "epoch": 4.88,
2848
+ "learning_rate": 2.560083594566353e-05,
2849
+ "loss": 0.006,
2850
+ "step": 4670
2851
+ },
2852
+ {
2853
+ "epoch": 4.89,
2854
+ "learning_rate": 2.554858934169279e-05,
2855
+ "loss": 0.005,
2856
+ "step": 4680
2857
+ },
2858
+ {
2859
+ "epoch": 4.9,
2860
+ "learning_rate": 2.549634273772205e-05,
2861
+ "loss": 0.0033,
2862
+ "step": 4690
2863
+ },
2864
+ {
2865
+ "epoch": 4.91,
2866
+ "learning_rate": 2.5444096133751305e-05,
2867
+ "loss": 0.0061,
2868
+ "step": 4700
2869
+ },
2870
+ {
2871
+ "epoch": 4.92,
2872
+ "learning_rate": 2.5391849529780566e-05,
2873
+ "loss": 0.0059,
2874
+ "step": 4710
2875
+ },
2876
+ {
2877
+ "epoch": 4.93,
2878
+ "learning_rate": 2.5339602925809823e-05,
2879
+ "loss": 0.0046,
2880
+ "step": 4720
2881
+ },
2882
+ {
2883
+ "epoch": 4.94,
2884
+ "learning_rate": 2.5287356321839083e-05,
2885
+ "loss": 0.0062,
2886
+ "step": 4730
2887
+ },
2888
+ {
2889
+ "epoch": 4.95,
2890
+ "learning_rate": 2.5235109717868337e-05,
2891
+ "loss": 0.0033,
2892
+ "step": 4740
2893
+ },
2894
+ {
2895
+ "epoch": 4.96,
2896
+ "learning_rate": 2.5182863113897598e-05,
2897
+ "loss": 0.0065,
2898
+ "step": 4750
2899
+ },
2900
+ {
2901
+ "epoch": 4.97,
2902
+ "learning_rate": 2.5130616509926858e-05,
2903
+ "loss": 0.0048,
2904
+ "step": 4760
2905
+ },
2906
+ {
2907
+ "epoch": 4.98,
2908
+ "learning_rate": 2.5078369905956112e-05,
2909
+ "loss": 0.0043,
2910
+ "step": 4770
2911
+ },
2912
+ {
2913
+ "epoch": 4.99,
2914
+ "learning_rate": 2.5026123301985373e-05,
2915
+ "loss": 0.0055,
2916
+ "step": 4780
2917
+ },
2918
+ {
2919
+ "epoch": 5.01,
2920
+ "learning_rate": 2.497387669801463e-05,
2921
+ "loss": 0.0039,
2922
+ "step": 4790
2923
+ },
2924
+ {
2925
+ "epoch": 5.02,
2926
+ "learning_rate": 2.4921630094043887e-05,
2927
+ "loss": 0.0025,
2928
+ "step": 4800
2929
+ },
2930
+ {
2931
+ "epoch": 5.03,
2932
+ "learning_rate": 2.4869383490073148e-05,
2933
+ "loss": 0.0039,
2934
+ "step": 4810
2935
+ },
2936
+ {
2937
+ "epoch": 5.04,
2938
+ "learning_rate": 2.4817136886102405e-05,
2939
+ "loss": 0.0048,
2940
+ "step": 4820
2941
+ },
2942
+ {
2943
+ "epoch": 5.05,
2944
+ "learning_rate": 2.4764890282131662e-05,
2945
+ "loss": 0.005,
2946
+ "step": 4830
2947
+ },
2948
+ {
2949
+ "epoch": 5.06,
2950
+ "learning_rate": 2.4712643678160922e-05,
2951
+ "loss": 0.0039,
2952
+ "step": 4840
2953
+ },
2954
+ {
2955
+ "epoch": 5.07,
2956
+ "learning_rate": 2.466039707419018e-05,
2957
+ "loss": 0.0096,
2958
+ "step": 4850
2959
+ },
2960
+ {
2961
+ "epoch": 5.08,
2962
+ "learning_rate": 2.4608150470219437e-05,
2963
+ "loss": 0.0046,
2964
+ "step": 4860
2965
+ },
2966
+ {
2967
+ "epoch": 5.09,
2968
+ "learning_rate": 2.4555903866248694e-05,
2969
+ "loss": 0.0026,
2970
+ "step": 4870
2971
+ },
2972
+ {
2973
+ "epoch": 5.1,
2974
+ "learning_rate": 2.4503657262277955e-05,
2975
+ "loss": 0.0043,
2976
+ "step": 4880
2977
+ },
2978
+ {
2979
+ "epoch": 5.11,
2980
+ "learning_rate": 2.4451410658307212e-05,
2981
+ "loss": 0.0031,
2982
+ "step": 4890
2983
+ },
2984
+ {
2985
+ "epoch": 5.12,
2986
+ "learning_rate": 2.439916405433647e-05,
2987
+ "loss": 0.005,
2988
+ "step": 4900
2989
+ },
2990
+ {
2991
+ "epoch": 5.13,
2992
+ "learning_rate": 2.4346917450365726e-05,
2993
+ "loss": 0.002,
2994
+ "step": 4910
2995
+ },
2996
+ {
2997
+ "epoch": 5.14,
2998
+ "learning_rate": 2.4294670846394983e-05,
2999
+ "loss": 0.0021,
3000
+ "step": 4920
3001
+ },
3002
+ {
3003
+ "epoch": 5.15,
3004
+ "learning_rate": 2.4242424242424244e-05,
3005
+ "loss": 0.0052,
3006
+ "step": 4930
3007
+ },
3008
+ {
3009
+ "epoch": 5.16,
3010
+ "learning_rate": 2.41901776384535e-05,
3011
+ "loss": 0.006,
3012
+ "step": 4940
3013
+ },
3014
+ {
3015
+ "epoch": 5.17,
3016
+ "learning_rate": 2.413793103448276e-05,
3017
+ "loss": 0.0028,
3018
+ "step": 4950
3019
+ },
3020
+ {
3021
+ "epoch": 5.18,
3022
+ "learning_rate": 2.408568443051202e-05,
3023
+ "loss": 0.0026,
3024
+ "step": 4960
3025
+ },
3026
+ {
3027
+ "epoch": 5.19,
3028
+ "learning_rate": 2.4033437826541276e-05,
3029
+ "loss": 0.0033,
3030
+ "step": 4970
3031
+ },
3032
+ {
3033
+ "epoch": 5.2,
3034
+ "learning_rate": 2.3981191222570533e-05,
3035
+ "loss": 0.003,
3036
+ "step": 4980
3037
+ },
3038
+ {
3039
+ "epoch": 5.21,
3040
+ "learning_rate": 2.392894461859979e-05,
3041
+ "loss": 0.0051,
3042
+ "step": 4990
3043
+ },
3044
+ {
3045
+ "epoch": 5.22,
3046
+ "learning_rate": 2.387669801462905e-05,
3047
+ "loss": 0.0062,
3048
+ "step": 5000
3049
+ },
3050
+ {
3051
+ "epoch": 5.22,
3052
+ "eval_acc": 0.7915632734700663,
3053
+ "eval_cer": 0.026731752673175267,
3054
+ "eval_loss": 0.01478270161896944,
3055
+ "eval_runtime": 128.2616,
3056
+ "eval_samples_per_second": 3.142,
3057
+ "eval_steps_per_second": 0.398,
3058
+ "step": 5000
3059
+ },
3060
+ {
3061
+ "epoch": 5.24,
3062
+ "learning_rate": 2.3824451410658308e-05,
3063
+ "loss": 0.0038,
3064
+ "step": 5010
3065
+ },
3066
+ {
3067
+ "epoch": 5.25,
3068
+ "learning_rate": 2.377220480668757e-05,
3069
+ "loss": 0.0061,
3070
+ "step": 5020
3071
+ },
3072
+ {
3073
+ "epoch": 5.26,
3074
+ "learning_rate": 2.3719958202716826e-05,
3075
+ "loss": 0.0035,
3076
+ "step": 5030
3077
+ },
3078
+ {
3079
+ "epoch": 5.27,
3080
+ "learning_rate": 2.3667711598746083e-05,
3081
+ "loss": 0.0034,
3082
+ "step": 5040
3083
+ },
3084
+ {
3085
+ "epoch": 5.28,
3086
+ "learning_rate": 2.361546499477534e-05,
3087
+ "loss": 0.0033,
3088
+ "step": 5050
3089
+ },
3090
+ {
3091
+ "epoch": 5.29,
3092
+ "learning_rate": 2.3563218390804597e-05,
3093
+ "loss": 0.0036,
3094
+ "step": 5060
3095
+ },
3096
+ {
3097
+ "epoch": 5.3,
3098
+ "learning_rate": 2.3510971786833858e-05,
3099
+ "loss": 0.004,
3100
+ "step": 5070
3101
+ },
3102
+ {
3103
+ "epoch": 5.31,
3104
+ "learning_rate": 2.3458725182863115e-05,
3105
+ "loss": 0.0023,
3106
+ "step": 5080
3107
+ },
3108
+ {
3109
+ "epoch": 5.32,
3110
+ "learning_rate": 2.3406478578892372e-05,
3111
+ "loss": 0.0024,
3112
+ "step": 5090
3113
+ },
3114
+ {
3115
+ "epoch": 5.33,
3116
+ "learning_rate": 2.335423197492163e-05,
3117
+ "loss": 0.0026,
3118
+ "step": 5100
3119
+ },
3120
+ {
3121
+ "epoch": 5.34,
3122
+ "learning_rate": 2.330198537095089e-05,
3123
+ "loss": 0.0031,
3124
+ "step": 5110
3125
+ },
3126
+ {
3127
+ "epoch": 5.35,
3128
+ "learning_rate": 2.3249738766980147e-05,
3129
+ "loss": 0.0032,
3130
+ "step": 5120
3131
+ },
3132
+ {
3133
+ "epoch": 5.36,
3134
+ "learning_rate": 2.3197492163009404e-05,
3135
+ "loss": 0.0032,
3136
+ "step": 5130
3137
+ },
3138
+ {
3139
+ "epoch": 5.37,
3140
+ "learning_rate": 2.3145245559038665e-05,
3141
+ "loss": 0.0043,
3142
+ "step": 5140
3143
+ },
3144
+ {
3145
+ "epoch": 5.38,
3146
+ "learning_rate": 2.3092998955067922e-05,
3147
+ "loss": 0.0039,
3148
+ "step": 5150
3149
+ },
3150
+ {
3151
+ "epoch": 5.39,
3152
+ "learning_rate": 2.304075235109718e-05,
3153
+ "loss": 0.0026,
3154
+ "step": 5160
3155
+ },
3156
+ {
3157
+ "epoch": 5.4,
3158
+ "learning_rate": 2.2988505747126437e-05,
3159
+ "loss": 0.0032,
3160
+ "step": 5170
3161
+ },
3162
+ {
3163
+ "epoch": 5.41,
3164
+ "learning_rate": 2.2936259143155697e-05,
3165
+ "loss": 0.0043,
3166
+ "step": 5180
3167
+ },
3168
+ {
3169
+ "epoch": 5.42,
3170
+ "learning_rate": 2.2884012539184954e-05,
3171
+ "loss": 0.0037,
3172
+ "step": 5190
3173
+ },
3174
+ {
3175
+ "epoch": 5.43,
3176
+ "learning_rate": 2.283176593521421e-05,
3177
+ "loss": 0.0038,
3178
+ "step": 5200
3179
+ },
3180
+ {
3181
+ "epoch": 5.44,
3182
+ "learning_rate": 2.2779519331243472e-05,
3183
+ "loss": 0.0038,
3184
+ "step": 5210
3185
+ },
3186
+ {
3187
+ "epoch": 5.45,
3188
+ "learning_rate": 2.272727272727273e-05,
3189
+ "loss": 0.004,
3190
+ "step": 5220
3191
+ },
3192
+ {
3193
+ "epoch": 5.46,
3194
+ "learning_rate": 2.2675026123301986e-05,
3195
+ "loss": 0.0025,
3196
+ "step": 5230
3197
+ },
3198
+ {
3199
+ "epoch": 5.48,
3200
+ "learning_rate": 2.2622779519331244e-05,
3201
+ "loss": 0.0023,
3202
+ "step": 5240
3203
+ },
3204
+ {
3205
+ "epoch": 5.49,
3206
+ "learning_rate": 2.2570532915360504e-05,
3207
+ "loss": 0.0024,
3208
+ "step": 5250
3209
+ },
3210
+ {
3211
+ "epoch": 5.5,
3212
+ "learning_rate": 2.251828631138976e-05,
3213
+ "loss": 0.0021,
3214
+ "step": 5260
3215
+ },
3216
+ {
3217
+ "epoch": 5.51,
3218
+ "learning_rate": 2.246603970741902e-05,
3219
+ "loss": 0.0045,
3220
+ "step": 5270
3221
+ },
3222
+ {
3223
+ "epoch": 5.52,
3224
+ "learning_rate": 2.2413793103448276e-05,
3225
+ "loss": 0.0026,
3226
+ "step": 5280
3227
+ },
3228
+ {
3229
+ "epoch": 5.53,
3230
+ "learning_rate": 2.2361546499477533e-05,
3231
+ "loss": 0.0044,
3232
+ "step": 5290
3233
+ },
3234
+ {
3235
+ "epoch": 5.54,
3236
+ "learning_rate": 2.2309299895506794e-05,
3237
+ "loss": 0.0046,
3238
+ "step": 5300
3239
+ },
3240
+ {
3241
+ "epoch": 5.55,
3242
+ "learning_rate": 2.225705329153605e-05,
3243
+ "loss": 0.0038,
3244
+ "step": 5310
3245
+ },
3246
+ {
3247
+ "epoch": 5.56,
3248
+ "learning_rate": 2.220480668756531e-05,
3249
+ "loss": 0.009,
3250
+ "step": 5320
3251
+ },
3252
+ {
3253
+ "epoch": 5.57,
3254
+ "learning_rate": 2.215256008359457e-05,
3255
+ "loss": 0.0048,
3256
+ "step": 5330
3257
+ },
3258
+ {
3259
+ "epoch": 5.58,
3260
+ "learning_rate": 2.2100313479623826e-05,
3261
+ "loss": 0.0033,
3262
+ "step": 5340
3263
+ },
3264
+ {
3265
+ "epoch": 5.59,
3266
+ "learning_rate": 2.2048066875653083e-05,
3267
+ "loss": 0.0024,
3268
+ "step": 5350
3269
+ },
3270
+ {
3271
+ "epoch": 5.6,
3272
+ "learning_rate": 2.199582027168234e-05,
3273
+ "loss": 0.002,
3274
+ "step": 5360
3275
+ },
3276
+ {
3277
+ "epoch": 5.61,
3278
+ "learning_rate": 2.19435736677116e-05,
3279
+ "loss": 0.0032,
3280
+ "step": 5370
3281
+ },
3282
+ {
3283
+ "epoch": 5.62,
3284
+ "learning_rate": 2.1891327063740858e-05,
3285
+ "loss": 0.0055,
3286
+ "step": 5380
3287
+ },
3288
+ {
3289
+ "epoch": 5.63,
3290
+ "learning_rate": 2.183908045977012e-05,
3291
+ "loss": 0.0059,
3292
+ "step": 5390
3293
+ },
3294
+ {
3295
+ "epoch": 5.64,
3296
+ "learning_rate": 2.1786833855799376e-05,
3297
+ "loss": 0.0024,
3298
+ "step": 5400
3299
+ },
3300
+ {
3301
+ "epoch": 5.65,
3302
+ "learning_rate": 2.173458725182863e-05,
3303
+ "loss": 0.0039,
3304
+ "step": 5410
3305
+ },
3306
+ {
3307
+ "epoch": 5.66,
3308
+ "learning_rate": 2.168234064785789e-05,
3309
+ "loss": 0.002,
3310
+ "step": 5420
3311
+ },
3312
+ {
3313
+ "epoch": 5.67,
3314
+ "learning_rate": 2.1630094043887147e-05,
3315
+ "loss": 0.0061,
3316
+ "step": 5430
3317
+ },
3318
+ {
3319
+ "epoch": 5.68,
3320
+ "learning_rate": 2.1577847439916408e-05,
3321
+ "loss": 0.0056,
3322
+ "step": 5440
3323
+ },
3324
+ {
3325
+ "epoch": 5.69,
3326
+ "learning_rate": 2.1525600835945665e-05,
3327
+ "loss": 0.0032,
3328
+ "step": 5450
3329
+ },
3330
+ {
3331
+ "epoch": 5.71,
3332
+ "learning_rate": 2.1473354231974925e-05,
3333
+ "loss": 0.0029,
3334
+ "step": 5460
3335
+ },
3336
+ {
3337
+ "epoch": 5.72,
3338
+ "learning_rate": 2.142110762800418e-05,
3339
+ "loss": 0.0017,
3340
+ "step": 5470
3341
+ },
3342
+ {
3343
+ "epoch": 5.73,
3344
+ "learning_rate": 2.1368861024033436e-05,
3345
+ "loss": 0.005,
3346
+ "step": 5480
3347
+ },
3348
+ {
3349
+ "epoch": 5.74,
3350
+ "learning_rate": 2.1316614420062697e-05,
3351
+ "loss": 0.0051,
3352
+ "step": 5490
3353
+ },
3354
+ {
3355
+ "epoch": 5.75,
3356
+ "learning_rate": 2.1264367816091954e-05,
3357
+ "loss": 0.0047,
3358
+ "step": 5500
3359
+ },
3360
+ {
3361
+ "epoch": 5.76,
3362
+ "learning_rate": 2.1212121212121215e-05,
3363
+ "loss": 0.0032,
3364
+ "step": 5510
3365
+ },
3366
+ {
3367
+ "epoch": 5.77,
3368
+ "learning_rate": 2.1159874608150472e-05,
3369
+ "loss": 0.004,
3370
+ "step": 5520
3371
+ },
3372
+ {
3373
+ "epoch": 5.78,
3374
+ "learning_rate": 2.110762800417973e-05,
3375
+ "loss": 0.0059,
3376
+ "step": 5530
3377
+ },
3378
+ {
3379
+ "epoch": 5.79,
3380
+ "learning_rate": 2.1055381400208986e-05,
3381
+ "loss": 0.0045,
3382
+ "step": 5540
3383
+ },
3384
+ {
3385
+ "epoch": 5.8,
3386
+ "learning_rate": 2.1003134796238247e-05,
3387
+ "loss": 0.0051,
3388
+ "step": 5550
3389
+ },
3390
+ {
3391
+ "epoch": 5.81,
3392
+ "learning_rate": 2.0950888192267504e-05,
3393
+ "loss": 0.005,
3394
+ "step": 5560
3395
+ },
3396
+ {
3397
+ "epoch": 5.82,
3398
+ "learning_rate": 2.089864158829676e-05,
3399
+ "loss": 0.0039,
3400
+ "step": 5570
3401
+ },
3402
+ {
3403
+ "epoch": 5.83,
3404
+ "learning_rate": 2.0846394984326022e-05,
3405
+ "loss": 0.0036,
3406
+ "step": 5580
3407
+ },
3408
+ {
3409
+ "epoch": 5.84,
3410
+ "learning_rate": 2.079414838035528e-05,
3411
+ "loss": 0.0041,
3412
+ "step": 5590
3413
+ },
3414
+ {
3415
+ "epoch": 5.85,
3416
+ "learning_rate": 2.0741901776384536e-05,
3417
+ "loss": 0.002,
3418
+ "step": 5600
3419
+ },
3420
+ {
3421
+ "epoch": 5.86,
3422
+ "learning_rate": 2.0689655172413793e-05,
3423
+ "loss": 0.0067,
3424
+ "step": 5610
3425
+ },
3426
+ {
3427
+ "epoch": 5.87,
3428
+ "learning_rate": 2.0637408568443054e-05,
3429
+ "loss": 0.0061,
3430
+ "step": 5620
3431
+ },
3432
+ {
3433
+ "epoch": 5.88,
3434
+ "learning_rate": 2.058516196447231e-05,
3435
+ "loss": 0.0041,
3436
+ "step": 5630
3437
+ },
3438
+ {
3439
+ "epoch": 5.89,
3440
+ "learning_rate": 2.0532915360501568e-05,
3441
+ "loss": 0.0063,
3442
+ "step": 5640
3443
+ },
3444
+ {
3445
+ "epoch": 5.9,
3446
+ "learning_rate": 2.0480668756530825e-05,
3447
+ "loss": 0.0036,
3448
+ "step": 5650
3449
+ },
3450
+ {
3451
+ "epoch": 5.91,
3452
+ "learning_rate": 2.0428422152560083e-05,
3453
+ "loss": 0.0034,
3454
+ "step": 5660
3455
+ },
3456
+ {
3457
+ "epoch": 5.92,
3458
+ "learning_rate": 2.0376175548589343e-05,
3459
+ "loss": 0.0036,
3460
+ "step": 5670
3461
+ },
3462
+ {
3463
+ "epoch": 5.94,
3464
+ "learning_rate": 2.03239289446186e-05,
3465
+ "loss": 0.0021,
3466
+ "step": 5680
3467
+ },
3468
+ {
3469
+ "epoch": 5.95,
3470
+ "learning_rate": 2.027168234064786e-05,
3471
+ "loss": 0.0027,
3472
+ "step": 5690
3473
+ },
3474
+ {
3475
+ "epoch": 5.96,
3476
+ "learning_rate": 2.0219435736677118e-05,
3477
+ "loss": 0.004,
3478
+ "step": 5700
3479
+ },
3480
+ {
3481
+ "epoch": 5.97,
3482
+ "learning_rate": 2.0167189132706375e-05,
3483
+ "loss": 0.0031,
3484
+ "step": 5710
3485
+ },
3486
+ {
3487
+ "epoch": 5.98,
3488
+ "learning_rate": 2.0114942528735632e-05,
3489
+ "loss": 0.0022,
3490
+ "step": 5720
3491
+ },
3492
+ {
3493
+ "epoch": 5.99,
3494
+ "learning_rate": 2.006269592476489e-05,
3495
+ "loss": 0.0029,
3496
+ "step": 5730
3497
+ },
3498
+ {
3499
+ "epoch": 6.0,
3500
+ "learning_rate": 2.001044932079415e-05,
3501
+ "loss": 0.0062,
3502
+ "step": 5740
3503
+ },
3504
+ {
3505
+ "epoch": 6.01,
3506
+ "learning_rate": 1.9958202716823407e-05,
3507
+ "loss": 0.0016,
3508
+ "step": 5750
3509
+ },
3510
+ {
3511
+ "epoch": 6.02,
3512
+ "learning_rate": 1.9905956112852668e-05,
3513
+ "loss": 0.0019,
3514
+ "step": 5760
3515
+ },
3516
+ {
3517
+ "epoch": 6.03,
3518
+ "learning_rate": 1.9853709508881925e-05,
3519
+ "loss": 0.003,
3520
+ "step": 5770
3521
+ },
3522
+ {
3523
+ "epoch": 6.04,
3524
+ "learning_rate": 1.980146290491118e-05,
3525
+ "loss": 0.002,
3526
+ "step": 5780
3527
+ },
3528
+ {
3529
+ "epoch": 6.05,
3530
+ "learning_rate": 1.974921630094044e-05,
3531
+ "loss": 0.003,
3532
+ "step": 5790
3533
+ },
3534
+ {
3535
+ "epoch": 6.06,
3536
+ "learning_rate": 1.9696969696969697e-05,
3537
+ "loss": 0.0033,
3538
+ "step": 5800
3539
+ },
3540
+ {
3541
+ "epoch": 6.07,
3542
+ "learning_rate": 1.9644723092998957e-05,
3543
+ "loss": 0.0016,
3544
+ "step": 5810
3545
+ },
3546
+ {
3547
+ "epoch": 6.08,
3548
+ "learning_rate": 1.9592476489028214e-05,
3549
+ "loss": 0.004,
3550
+ "step": 5820
3551
+ },
3552
+ {
3553
+ "epoch": 6.09,
3554
+ "learning_rate": 1.9540229885057475e-05,
3555
+ "loss": 0.0025,
3556
+ "step": 5830
3557
+ },
3558
+ {
3559
+ "epoch": 6.1,
3560
+ "learning_rate": 1.948798328108673e-05,
3561
+ "loss": 0.0025,
3562
+ "step": 5840
3563
+ },
3564
+ {
3565
+ "epoch": 6.11,
3566
+ "learning_rate": 1.9435736677115986e-05,
3567
+ "loss": 0.0031,
3568
+ "step": 5850
3569
+ },
3570
+ {
3571
+ "epoch": 6.12,
3572
+ "learning_rate": 1.9383490073145247e-05,
3573
+ "loss": 0.002,
3574
+ "step": 5860
3575
+ },
3576
+ {
3577
+ "epoch": 6.13,
3578
+ "learning_rate": 1.9331243469174504e-05,
3579
+ "loss": 0.0027,
3580
+ "step": 5870
3581
+ },
3582
+ {
3583
+ "epoch": 6.14,
3584
+ "learning_rate": 1.9278996865203764e-05,
3585
+ "loss": 0.0019,
3586
+ "step": 5880
3587
+ },
3588
+ {
3589
+ "epoch": 6.15,
3590
+ "learning_rate": 1.922675026123302e-05,
3591
+ "loss": 0.0033,
3592
+ "step": 5890
3593
+ },
3594
+ {
3595
+ "epoch": 6.17,
3596
+ "learning_rate": 1.917450365726228e-05,
3597
+ "loss": 0.0027,
3598
+ "step": 5900
3599
+ },
3600
+ {
3601
+ "epoch": 6.18,
3602
+ "learning_rate": 1.9122257053291536e-05,
3603
+ "loss": 0.0048,
3604
+ "step": 5910
3605
+ },
3606
+ {
3607
+ "epoch": 6.19,
3608
+ "learning_rate": 1.9070010449320793e-05,
3609
+ "loss": 0.0016,
3610
+ "step": 5920
3611
+ },
3612
+ {
3613
+ "epoch": 6.2,
3614
+ "learning_rate": 1.9017763845350054e-05,
3615
+ "loss": 0.002,
3616
+ "step": 5930
3617
+ },
3618
+ {
3619
+ "epoch": 6.21,
3620
+ "learning_rate": 1.896551724137931e-05,
3621
+ "loss": 0.0027,
3622
+ "step": 5940
3623
+ },
3624
+ {
3625
+ "epoch": 6.22,
3626
+ "learning_rate": 1.891327063740857e-05,
3627
+ "loss": 0.0021,
3628
+ "step": 5950
3629
+ },
3630
+ {
3631
+ "epoch": 6.23,
3632
+ "learning_rate": 1.886102403343783e-05,
3633
+ "loss": 0.0068,
3634
+ "step": 5960
3635
+ },
3636
+ {
3637
+ "epoch": 6.24,
3638
+ "learning_rate": 1.8808777429467086e-05,
3639
+ "loss": 0.0017,
3640
+ "step": 5970
3641
+ },
3642
+ {
3643
+ "epoch": 6.25,
3644
+ "learning_rate": 1.8756530825496343e-05,
3645
+ "loss": 0.0013,
3646
+ "step": 5980
3647
+ },
3648
+ {
3649
+ "epoch": 6.26,
3650
+ "learning_rate": 1.8704284221525603e-05,
3651
+ "loss": 0.0033,
3652
+ "step": 5990
3653
+ },
3654
+ {
3655
+ "epoch": 6.27,
3656
+ "learning_rate": 1.865203761755486e-05,
3657
+ "loss": 0.0024,
3658
+ "step": 6000
3659
+ },
3660
+ {
3661
+ "epoch": 6.27,
3662
+ "eval_acc": 0.8064516109021052,
3663
+ "eval_cer": 0.02487215248721525,
3664
+ "eval_loss": 0.014062131755053997,
3665
+ "eval_runtime": 122.3207,
3666
+ "eval_samples_per_second": 3.295,
3667
+ "eval_steps_per_second": 0.417,
3668
+ "step": 6000
3669
+ },
3670
+ {
3671
+ "epoch": 6.28,
3672
+ "learning_rate": 1.8599791013584118e-05,
3673
+ "loss": 0.006,
3674
+ "step": 6010
3675
+ },
3676
+ {
3677
+ "epoch": 6.29,
3678
+ "learning_rate": 1.8547544409613375e-05,
3679
+ "loss": 0.0014,
3680
+ "step": 6020
3681
+ },
3682
+ {
3683
+ "epoch": 6.3,
3684
+ "learning_rate": 1.8495297805642632e-05,
3685
+ "loss": 0.0016,
3686
+ "step": 6030
3687
+ },
3688
+ {
3689
+ "epoch": 6.31,
3690
+ "learning_rate": 1.8443051201671893e-05,
3691
+ "loss": 0.0038,
3692
+ "step": 6040
3693
+ },
3694
+ {
3695
+ "epoch": 6.32,
3696
+ "learning_rate": 1.839080459770115e-05,
3697
+ "loss": 0.0032,
3698
+ "step": 6050
3699
+ },
3700
+ {
3701
+ "epoch": 6.33,
3702
+ "learning_rate": 1.833855799373041e-05,
3703
+ "loss": 0.0036,
3704
+ "step": 6060
3705
+ },
3706
+ {
3707
+ "epoch": 6.34,
3708
+ "learning_rate": 1.8286311389759668e-05,
3709
+ "loss": 0.0016,
3710
+ "step": 6070
3711
+ },
3712
+ {
3713
+ "epoch": 6.35,
3714
+ "learning_rate": 1.8234064785788925e-05,
3715
+ "loss": 0.0038,
3716
+ "step": 6080
3717
+ },
3718
+ {
3719
+ "epoch": 6.36,
3720
+ "learning_rate": 1.8181818181818182e-05,
3721
+ "loss": 0.0042,
3722
+ "step": 6090
3723
+ },
3724
+ {
3725
+ "epoch": 6.37,
3726
+ "learning_rate": 1.812957157784744e-05,
3727
+ "loss": 0.003,
3728
+ "step": 6100
3729
+ },
3730
+ {
3731
+ "epoch": 6.38,
3732
+ "learning_rate": 1.80773249738767e-05,
3733
+ "loss": 0.0035,
3734
+ "step": 6110
3735
+ },
3736
+ {
3737
+ "epoch": 6.39,
3738
+ "learning_rate": 1.8025078369905957e-05,
3739
+ "loss": 0.0025,
3740
+ "step": 6120
3741
+ },
3742
+ {
3743
+ "epoch": 6.41,
3744
+ "learning_rate": 1.7972831765935218e-05,
3745
+ "loss": 0.0022,
3746
+ "step": 6130
3747
+ },
3748
+ {
3749
+ "epoch": 6.42,
3750
+ "learning_rate": 1.7920585161964475e-05,
3751
+ "loss": 0.0021,
3752
+ "step": 6140
3753
+ },
3754
+ {
3755
+ "epoch": 6.43,
3756
+ "learning_rate": 1.786833855799373e-05,
3757
+ "loss": 0.0037,
3758
+ "step": 6150
3759
+ },
3760
+ {
3761
+ "epoch": 6.44,
3762
+ "learning_rate": 1.781609195402299e-05,
3763
+ "loss": 0.0037,
3764
+ "step": 6160
3765
+ },
3766
+ {
3767
+ "epoch": 6.45,
3768
+ "learning_rate": 1.7763845350052246e-05,
3769
+ "loss": 0.0023,
3770
+ "step": 6170
3771
+ },
3772
+ {
3773
+ "epoch": 6.46,
3774
+ "learning_rate": 1.7711598746081507e-05,
3775
+ "loss": 0.0041,
3776
+ "step": 6180
3777
+ },
3778
+ {
3779
+ "epoch": 6.47,
3780
+ "learning_rate": 1.7659352142110764e-05,
3781
+ "loss": 0.0063,
3782
+ "step": 6190
3783
+ },
3784
+ {
3785
+ "epoch": 6.48,
3786
+ "learning_rate": 1.7607105538140025e-05,
3787
+ "loss": 0.0024,
3788
+ "step": 6200
3789
+ },
3790
+ {
3791
+ "epoch": 6.49,
3792
+ "learning_rate": 1.755485893416928e-05,
3793
+ "loss": 0.0033,
3794
+ "step": 6210
3795
+ },
3796
+ {
3797
+ "epoch": 6.5,
3798
+ "learning_rate": 1.7502612330198536e-05,
3799
+ "loss": 0.0025,
3800
+ "step": 6220
3801
+ },
3802
+ {
3803
+ "epoch": 6.51,
3804
+ "learning_rate": 1.7450365726227796e-05,
3805
+ "loss": 0.0019,
3806
+ "step": 6230
3807
+ },
3808
+ {
3809
+ "epoch": 6.52,
3810
+ "learning_rate": 1.7398119122257053e-05,
3811
+ "loss": 0.0015,
3812
+ "step": 6240
3813
+ },
3814
+ {
3815
+ "epoch": 6.53,
3816
+ "learning_rate": 1.7345872518286314e-05,
3817
+ "loss": 0.0026,
3818
+ "step": 6250
3819
+ },
3820
+ {
3821
+ "epoch": 6.54,
3822
+ "learning_rate": 1.729362591431557e-05,
3823
+ "loss": 0.0024,
3824
+ "step": 6260
3825
+ },
3826
+ {
3827
+ "epoch": 6.55,
3828
+ "learning_rate": 1.7241379310344828e-05,
3829
+ "loss": 0.0018,
3830
+ "step": 6270
3831
+ },
3832
+ {
3833
+ "epoch": 6.56,
3834
+ "learning_rate": 1.7189132706374085e-05,
3835
+ "loss": 0.0029,
3836
+ "step": 6280
3837
+ },
3838
+ {
3839
+ "epoch": 6.57,
3840
+ "learning_rate": 1.7136886102403343e-05,
3841
+ "loss": 0.0031,
3842
+ "step": 6290
3843
+ },
3844
+ {
3845
+ "epoch": 6.58,
3846
+ "learning_rate": 1.7084639498432603e-05,
3847
+ "loss": 0.0036,
3848
+ "step": 6300
3849
+ },
3850
+ {
3851
+ "epoch": 6.59,
3852
+ "learning_rate": 1.703239289446186e-05,
3853
+ "loss": 0.0028,
3854
+ "step": 6310
3855
+ },
3856
+ {
3857
+ "epoch": 6.6,
3858
+ "learning_rate": 1.698014629049112e-05,
3859
+ "loss": 0.0021,
3860
+ "step": 6320
3861
+ },
3862
+ {
3863
+ "epoch": 6.61,
3864
+ "learning_rate": 1.6927899686520378e-05,
3865
+ "loss": 0.0029,
3866
+ "step": 6330
3867
+ },
3868
+ {
3869
+ "epoch": 6.62,
3870
+ "learning_rate": 1.6875653082549635e-05,
3871
+ "loss": 0.0023,
3872
+ "step": 6340
3873
+ },
3874
+ {
3875
+ "epoch": 6.64,
3876
+ "learning_rate": 1.6823406478578893e-05,
3877
+ "loss": 0.0013,
3878
+ "step": 6350
3879
+ },
3880
+ {
3881
+ "epoch": 6.65,
3882
+ "learning_rate": 1.677115987460815e-05,
3883
+ "loss": 0.0023,
3884
+ "step": 6360
3885
+ },
3886
+ {
3887
+ "epoch": 6.66,
3888
+ "learning_rate": 1.671891327063741e-05,
3889
+ "loss": 0.0016,
3890
+ "step": 6370
3891
+ },
3892
+ {
3893
+ "epoch": 6.67,
3894
+ "learning_rate": 1.6666666666666667e-05,
3895
+ "loss": 0.0019,
3896
+ "step": 6380
3897
+ },
3898
+ {
3899
+ "epoch": 6.68,
3900
+ "learning_rate": 1.6614420062695925e-05,
3901
+ "loss": 0.003,
3902
+ "step": 6390
3903
+ },
3904
+ {
3905
+ "epoch": 6.69,
3906
+ "learning_rate": 1.6562173458725182e-05,
3907
+ "loss": 0.0018,
3908
+ "step": 6400
3909
+ },
3910
+ {
3911
+ "epoch": 6.7,
3912
+ "learning_rate": 1.6509926854754442e-05,
3913
+ "loss": 0.0029,
3914
+ "step": 6410
3915
+ },
3916
+ {
3917
+ "epoch": 6.71,
3918
+ "learning_rate": 1.64576802507837e-05,
3919
+ "loss": 0.0039,
3920
+ "step": 6420
3921
+ },
3922
+ {
3923
+ "epoch": 6.72,
3924
+ "learning_rate": 1.6405433646812957e-05,
3925
+ "loss": 0.0011,
3926
+ "step": 6430
3927
+ },
3928
+ {
3929
+ "epoch": 6.73,
3930
+ "learning_rate": 1.6353187042842217e-05,
3931
+ "loss": 0.0019,
3932
+ "step": 6440
3933
+ },
3934
+ {
3935
+ "epoch": 6.74,
3936
+ "learning_rate": 1.6300940438871475e-05,
3937
+ "loss": 0.0028,
3938
+ "step": 6450
3939
+ },
3940
+ {
3941
+ "epoch": 6.75,
3942
+ "learning_rate": 1.6248693834900732e-05,
3943
+ "loss": 0.004,
3944
+ "step": 6460
3945
+ },
3946
+ {
3947
+ "epoch": 6.76,
3948
+ "learning_rate": 1.619644723092999e-05,
3949
+ "loss": 0.0021,
3950
+ "step": 6470
3951
+ },
3952
+ {
3953
+ "epoch": 6.77,
3954
+ "learning_rate": 1.614420062695925e-05,
3955
+ "loss": 0.0026,
3956
+ "step": 6480
3957
+ },
3958
+ {
3959
+ "epoch": 6.78,
3960
+ "learning_rate": 1.6091954022988507e-05,
3961
+ "loss": 0.0024,
3962
+ "step": 6490
3963
+ },
3964
+ {
3965
+ "epoch": 6.79,
3966
+ "learning_rate": 1.6039707419017767e-05,
3967
+ "loss": 0.003,
3968
+ "step": 6500
3969
+ },
3970
+ {
3971
+ "epoch": 6.8,
3972
+ "learning_rate": 1.5987460815047024e-05,
3973
+ "loss": 0.0029,
3974
+ "step": 6510
3975
+ },
3976
+ {
3977
+ "epoch": 6.81,
3978
+ "learning_rate": 1.5935214211076278e-05,
3979
+ "loss": 0.0012,
3980
+ "step": 6520
3981
+ },
3982
+ {
3983
+ "epoch": 6.82,
3984
+ "learning_rate": 1.588296760710554e-05,
3985
+ "loss": 0.0023,
3986
+ "step": 6530
3987
+ },
3988
+ {
3989
+ "epoch": 6.83,
3990
+ "learning_rate": 1.5830721003134796e-05,
3991
+ "loss": 0.0023,
3992
+ "step": 6540
3993
+ },
3994
+ {
3995
+ "epoch": 6.84,
3996
+ "learning_rate": 1.5778474399164057e-05,
3997
+ "loss": 0.0025,
3998
+ "step": 6550
3999
+ },
4000
+ {
4001
+ "epoch": 6.85,
4002
+ "learning_rate": 1.5726227795193314e-05,
4003
+ "loss": 0.0029,
4004
+ "step": 6560
4005
+ },
4006
+ {
4007
+ "epoch": 6.87,
4008
+ "learning_rate": 1.5673981191222574e-05,
4009
+ "loss": 0.0036,
4010
+ "step": 6570
4011
+ },
4012
+ {
4013
+ "epoch": 6.88,
4014
+ "learning_rate": 1.5621734587251828e-05,
4015
+ "loss": 0.0033,
4016
+ "step": 6580
4017
+ },
4018
+ {
4019
+ "epoch": 6.89,
4020
+ "learning_rate": 1.5569487983281085e-05,
4021
+ "loss": 0.0028,
4022
+ "step": 6590
4023
+ },
4024
+ {
4025
+ "epoch": 6.9,
4026
+ "learning_rate": 1.5517241379310346e-05,
4027
+ "loss": 0.003,
4028
+ "step": 6600
4029
+ },
4030
+ {
4031
+ "epoch": 6.91,
4032
+ "learning_rate": 1.5464994775339603e-05,
4033
+ "loss": 0.0014,
4034
+ "step": 6610
4035
+ },
4036
+ {
4037
+ "epoch": 6.92,
4038
+ "learning_rate": 1.5412748171368864e-05,
4039
+ "loss": 0.0019,
4040
+ "step": 6620
4041
+ },
4042
+ {
4043
+ "epoch": 6.93,
4044
+ "learning_rate": 1.536050156739812e-05,
4045
+ "loss": 0.0031,
4046
+ "step": 6630
4047
+ },
4048
+ {
4049
+ "epoch": 6.94,
4050
+ "learning_rate": 1.5308254963427378e-05,
4051
+ "loss": 0.0025,
4052
+ "step": 6640
4053
+ },
4054
+ {
4055
+ "epoch": 6.95,
4056
+ "learning_rate": 1.5256008359456637e-05,
4057
+ "loss": 0.0016,
4058
+ "step": 6650
4059
+ },
4060
+ {
4061
+ "epoch": 6.96,
4062
+ "learning_rate": 1.5203761755485892e-05,
4063
+ "loss": 0.0028,
4064
+ "step": 6660
4065
+ },
4066
+ {
4067
+ "epoch": 6.97,
4068
+ "learning_rate": 1.5151515151515153e-05,
4069
+ "loss": 0.0034,
4070
+ "step": 6670
4071
+ },
4072
+ {
4073
+ "epoch": 6.98,
4074
+ "learning_rate": 1.509926854754441e-05,
4075
+ "loss": 0.0016,
4076
+ "step": 6680
4077
+ },
4078
+ {
4079
+ "epoch": 6.99,
4080
+ "learning_rate": 1.5047021943573669e-05,
4081
+ "loss": 0.0017,
4082
+ "step": 6690
4083
+ },
4084
+ {
4085
+ "epoch": 7.0,
4086
+ "learning_rate": 1.4994775339602926e-05,
4087
+ "loss": 0.0032,
4088
+ "step": 6700
4089
+ },
4090
+ {
4091
+ "epoch": 7.01,
4092
+ "learning_rate": 1.4942528735632185e-05,
4093
+ "loss": 0.0037,
4094
+ "step": 6710
4095
+ },
4096
+ {
4097
+ "epoch": 7.02,
4098
+ "learning_rate": 1.4890282131661442e-05,
4099
+ "loss": 0.0037,
4100
+ "step": 6720
4101
+ },
4102
+ {
4103
+ "epoch": 7.03,
4104
+ "learning_rate": 1.48380355276907e-05,
4105
+ "loss": 0.0025,
4106
+ "step": 6730
4107
+ },
4108
+ {
4109
+ "epoch": 7.04,
4110
+ "learning_rate": 1.478578892371996e-05,
4111
+ "loss": 0.0017,
4112
+ "step": 6740
4113
+ },
4114
+ {
4115
+ "epoch": 7.05,
4116
+ "learning_rate": 1.4733542319749215e-05,
4117
+ "loss": 0.003,
4118
+ "step": 6750
4119
+ },
4120
+ {
4121
+ "epoch": 7.06,
4122
+ "learning_rate": 1.4681295715778476e-05,
4123
+ "loss": 0.001,
4124
+ "step": 6760
4125
+ },
4126
+ {
4127
+ "epoch": 7.07,
4128
+ "learning_rate": 1.4629049111807733e-05,
4129
+ "loss": 0.0026,
4130
+ "step": 6770
4131
+ },
4132
+ {
4133
+ "epoch": 7.08,
4134
+ "learning_rate": 1.4576802507836992e-05,
4135
+ "loss": 0.003,
4136
+ "step": 6780
4137
+ },
4138
+ {
4139
+ "epoch": 7.1,
4140
+ "learning_rate": 1.452455590386625e-05,
4141
+ "loss": 0.0018,
4142
+ "step": 6790
4143
+ },
4144
+ {
4145
+ "epoch": 7.11,
4146
+ "learning_rate": 1.4472309299895506e-05,
4147
+ "loss": 0.0013,
4148
+ "step": 6800
4149
+ },
4150
+ {
4151
+ "epoch": 7.12,
4152
+ "learning_rate": 1.4420062695924765e-05,
4153
+ "loss": 0.002,
4154
+ "step": 6810
4155
+ },
4156
+ {
4157
+ "epoch": 7.13,
4158
+ "learning_rate": 1.4367816091954022e-05,
4159
+ "loss": 0.0011,
4160
+ "step": 6820
4161
+ },
4162
+ {
4163
+ "epoch": 7.14,
4164
+ "learning_rate": 1.4315569487983283e-05,
4165
+ "loss": 0.0015,
4166
+ "step": 6830
4167
+ },
4168
+ {
4169
+ "epoch": 7.15,
4170
+ "learning_rate": 1.4263322884012539e-05,
4171
+ "loss": 0.0013,
4172
+ "step": 6840
4173
+ },
4174
+ {
4175
+ "epoch": 7.16,
4176
+ "learning_rate": 1.4211076280041799e-05,
4177
+ "loss": 0.0035,
4178
+ "step": 6850
4179
+ },
4180
+ {
4181
+ "epoch": 7.17,
4182
+ "learning_rate": 1.4158829676071056e-05,
4183
+ "loss": 0.0022,
4184
+ "step": 6860
4185
+ },
4186
+ {
4187
+ "epoch": 7.18,
4188
+ "learning_rate": 1.4106583072100313e-05,
4189
+ "loss": 0.0021,
4190
+ "step": 6870
4191
+ },
4192
+ {
4193
+ "epoch": 7.19,
4194
+ "learning_rate": 1.4054336468129572e-05,
4195
+ "loss": 0.002,
4196
+ "step": 6880
4197
+ },
4198
+ {
4199
+ "epoch": 7.2,
4200
+ "learning_rate": 1.400208986415883e-05,
4201
+ "loss": 0.0012,
4202
+ "step": 6890
4203
+ },
4204
+ {
4205
+ "epoch": 7.21,
4206
+ "learning_rate": 1.3949843260188088e-05,
4207
+ "loss": 0.0018,
4208
+ "step": 6900
4209
+ },
4210
+ {
4211
+ "epoch": 7.22,
4212
+ "learning_rate": 1.3897596656217346e-05,
4213
+ "loss": 0.0026,
4214
+ "step": 6910
4215
+ },
4216
+ {
4217
+ "epoch": 7.23,
4218
+ "learning_rate": 1.3845350052246606e-05,
4219
+ "loss": 0.0016,
4220
+ "step": 6920
4221
+ },
4222
+ {
4223
+ "epoch": 7.24,
4224
+ "learning_rate": 1.3793103448275863e-05,
4225
+ "loss": 0.0022,
4226
+ "step": 6930
4227
+ },
4228
+ {
4229
+ "epoch": 7.25,
4230
+ "learning_rate": 1.3740856844305122e-05,
4231
+ "loss": 0.0018,
4232
+ "step": 6940
4233
+ },
4234
+ {
4235
+ "epoch": 7.26,
4236
+ "learning_rate": 1.368861024033438e-05,
4237
+ "loss": 0.001,
4238
+ "step": 6950
4239
+ },
4240
+ {
4241
+ "epoch": 7.27,
4242
+ "learning_rate": 1.3636363636363637e-05,
4243
+ "loss": 0.0019,
4244
+ "step": 6960
4245
+ },
4246
+ {
4247
+ "epoch": 7.28,
4248
+ "learning_rate": 1.3584117032392895e-05,
4249
+ "loss": 0.0014,
4250
+ "step": 6970
4251
+ },
4252
+ {
4253
+ "epoch": 7.29,
4254
+ "learning_rate": 1.3531870428422153e-05,
4255
+ "loss": 0.0019,
4256
+ "step": 6980
4257
+ },
4258
+ {
4259
+ "epoch": 7.3,
4260
+ "learning_rate": 1.3479623824451411e-05,
4261
+ "loss": 0.0032,
4262
+ "step": 6990
4263
+ },
4264
+ {
4265
+ "epoch": 7.31,
4266
+ "learning_rate": 1.3427377220480669e-05,
4267
+ "loss": 0.0017,
4268
+ "step": 7000
4269
+ },
4270
+ {
4271
+ "epoch": 7.31,
4272
+ "eval_acc": 0.821339948334144,
4273
+ "eval_cer": 0.021385402138540215,
4274
+ "eval_loss": 0.013037490658462048,
4275
+ "eval_runtime": 122.7957,
4276
+ "eval_samples_per_second": 3.282,
4277
+ "eval_steps_per_second": 0.415,
4278
+ "step": 7000
4279
+ }
4280
+ ],
4281
+ "max_steps": 9570,
4282
+ "num_train_epochs": 10,
4283
+ "total_flos": 4.3872550778956677e+18,
4284
+ "trial_name": null,
4285
+ "trial_params": null
4286
+ }
checkpoint/trocr-custdata-8000/checkpoint-7000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/checkpoint-8000/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/trocr-custdata-8000/checkpoint-8000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e8fc5824d11fae659bd8cc84e5a5b85293d52feed6feb4ddbae35178b2ca499
3
+ size 243062533
checkpoint/{trocr-custdata/checkpoint-4000 → trocr-custdata-8000/checkpoint-8000}/preprocessor_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/checkpoint-8000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56dad6884a338de70b07a843fc411c7ac1a5eaba6ab3f189e177c9c961020180
3
+ size 122142261
checkpoint/trocr-custdata-8000/checkpoint-8000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8227d407367919e234d93a77ff702ac1838ea162b87b0ae0688429bcd1c47928
3
+ size 627
checkpoint/trocr-custdata-8000/checkpoint-8000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint/trocr-custdata-8000/checkpoint-8000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/checkpoint-9000/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/trocr-custdata-8000/checkpoint-9000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a5bc50521d76202acbcb3db74432c504cb3883447cf071e1af13d78fd85736e
3
+ size 243062533
checkpoint/trocr-custdata-8000/checkpoint-9000/preprocessor_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_rescale": true,
4
+ "do_resize": true,
5
+ "image_mean": [
6
+ 0.5,
7
+ 0.5,
8
+ 0.5
9
+ ],
10
+ "image_processor_type": "ViTImageProcessor",
11
+ "image_std": [
12
+ 0.5,
13
+ 0.5,
14
+ 0.5
15
+ ],
16
+ "resample": 2,
17
+ "rescale_factor": 0.00392156862745098,
18
+ "size": {
19
+ "height": 384,
20
+ "width": 384
21
+ }
22
+ }
checkpoint/trocr-custdata-8000/checkpoint-9000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21a8c0e318035ca8cbedaf2b4cf6f003a3dbed918acfe437e4eba0830b474d30
3
+ size 122142261
checkpoint/trocr-custdata-8000/checkpoint-9000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3e2391b91f8e2fc30a46edf444ba9d8794e155351f540a427cd83bf25775816
3
+ size 627
checkpoint/trocr-custdata-8000/checkpoint-9000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint/trocr-custdata-8000/checkpoint-9000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/last/config.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "_name_or_path": "./cust-data/weights",
4
+ "architectures": [
5
+ "VisionEncoderDecoderModel"
6
+ ],
7
+ "decoder": {
8
+ "_name_or_path": "",
9
+ "activation_dropout": 0.0,
10
+ "activation_function": "relu",
11
+ "add_cross_attention": true,
12
+ "architectures": null,
13
+ "attention_dropout": 0.0,
14
+ "bad_words_ids": null,
15
+ "begin_suppress_tokens": null,
16
+ "bos_token_id": 0,
17
+ "chunk_size_feed_forward": 0,
18
+ "classifier_dropout": 0.0,
19
+ "cross_attention_hidden_size": 384,
20
+ "d_model": 256,
21
+ "decoder_attention_heads": 8,
22
+ "decoder_ffn_dim": 1024,
23
+ "decoder_layerdrop": 0.0,
24
+ "decoder_layers": 6,
25
+ "decoder_start_token_id": 2,
26
+ "diversity_penalty": 0.0,
27
+ "do_sample": false,
28
+ "dropout": 0.1,
29
+ "early_stopping": false,
30
+ "encoder_no_repeat_ngram_size": 0,
31
+ "eos_token_id": 2,
32
+ "exponential_decay_length_penalty": null,
33
+ "finetuning_task": null,
34
+ "forced_bos_token_id": null,
35
+ "forced_eos_token_id": null,
36
+ "id2label": {
37
+ "0": "LABEL_0",
38
+ "1": "LABEL_1"
39
+ },
40
+ "init_std": 0.02,
41
+ "is_decoder": true,
42
+ "is_encoder_decoder": false,
43
+ "label2id": {
44
+ "LABEL_0": 0,
45
+ "LABEL_1": 1
46
+ },
47
+ "layernorm_embedding": true,
48
+ "length_penalty": 1.0,
49
+ "max_length": 20,
50
+ "max_position_embeddings": 512,
51
+ "min_length": 0,
52
+ "model_type": "trocr",
53
+ "no_repeat_ngram_size": 0,
54
+ "num_beam_groups": 1,
55
+ "num_beams": 1,
56
+ "num_return_sequences": 1,
57
+ "output_attentions": false,
58
+ "output_hidden_states": false,
59
+ "output_scores": false,
60
+ "pad_token_id": 1,
61
+ "prefix": null,
62
+ "problem_type": null,
63
+ "pruned_heads": {},
64
+ "remove_invalid_values": false,
65
+ "repetition_penalty": 1.0,
66
+ "return_dict": true,
67
+ "return_dict_in_generate": false,
68
+ "scale_embedding": true,
69
+ "sep_token_id": null,
70
+ "suppress_tokens": null,
71
+ "task_specific_params": null,
72
+ "temperature": 1.0,
73
+ "tf_legacy_loss": false,
74
+ "tie_encoder_decoder": false,
75
+ "tie_word_embeddings": false,
76
+ "tokenizer_class": null,
77
+ "top_k": 50,
78
+ "top_p": 1.0,
79
+ "torch_dtype": null,
80
+ "torchscript": false,
81
+ "transformers_version": "4.26.0.dev0",
82
+ "typical_p": 1.0,
83
+ "use_bfloat16": false,
84
+ "use_cache": false,
85
+ "use_learned_position_embeddings": true,
86
+ "vocab_size": 3319
87
+ },
88
+ "decoder_start_token_id": 0,
89
+ "early_stopping": true,
90
+ "encoder": {
91
+ "_name_or_path": "",
92
+ "add_cross_attention": false,
93
+ "architectures": null,
94
+ "attention_probs_dropout_prob": 0.0,
95
+ "bad_words_ids": null,
96
+ "begin_suppress_tokens": null,
97
+ "bos_token_id": null,
98
+ "chunk_size_feed_forward": 0,
99
+ "cross_attention_hidden_size": null,
100
+ "decoder_start_token_id": null,
101
+ "diversity_penalty": 0.0,
102
+ "do_sample": false,
103
+ "early_stopping": false,
104
+ "encoder_no_repeat_ngram_size": 0,
105
+ "encoder_stride": 16,
106
+ "eos_token_id": null,
107
+ "exponential_decay_length_penalty": null,
108
+ "finetuning_task": null,
109
+ "forced_bos_token_id": null,
110
+ "forced_eos_token_id": null,
111
+ "hidden_act": "gelu",
112
+ "hidden_dropout_prob": 0.0,
113
+ "hidden_size": 384,
114
+ "id2label": {
115
+ "0": "LABEL_0",
116
+ "1": "LABEL_1"
117
+ },
118
+ "image_size": 384,
119
+ "initializer_range": 0.02,
120
+ "intermediate_size": 1536,
121
+ "is_decoder": false,
122
+ "is_encoder_decoder": false,
123
+ "label2id": {
124
+ "LABEL_0": 0,
125
+ "LABEL_1": 1
126
+ },
127
+ "layer_norm_eps": 1e-12,
128
+ "length_penalty": 1.0,
129
+ "max_length": 20,
130
+ "min_length": 0,
131
+ "model_type": "deit",
132
+ "no_repeat_ngram_size": 0,
133
+ "num_attention_heads": 6,
134
+ "num_beam_groups": 1,
135
+ "num_beams": 1,
136
+ "num_channels": 3,
137
+ "num_hidden_layers": 12,
138
+ "num_return_sequences": 1,
139
+ "output_attentions": false,
140
+ "output_hidden_states": false,
141
+ "output_scores": false,
142
+ "pad_token_id": null,
143
+ "patch_size": 16,
144
+ "prefix": null,
145
+ "problem_type": null,
146
+ "pruned_heads": {},
147
+ "qkv_bias": true,
148
+ "remove_invalid_values": false,
149
+ "repetition_penalty": 1.0,
150
+ "return_dict": true,
151
+ "return_dict_in_generate": false,
152
+ "sep_token_id": null,
153
+ "suppress_tokens": null,
154
+ "task_specific_params": null,
155
+ "temperature": 1.0,
156
+ "tf_legacy_loss": false,
157
+ "tie_encoder_decoder": false,
158
+ "tie_word_embeddings": true,
159
+ "tokenizer_class": null,
160
+ "top_k": 50,
161
+ "top_p": 1.0,
162
+ "torch_dtype": null,
163
+ "torchscript": false,
164
+ "transformers_version": "4.26.0.dev0",
165
+ "typical_p": 1.0,
166
+ "use_bfloat16": false
167
+ },
168
+ "eos_token_id": 2,
169
+ "is_encoder_decoder": true,
170
+ "length_penalty": 2.0,
171
+ "max_length": 256,
172
+ "model_type": "vision-encoder-decoder",
173
+ "no_repeat_ngram_size": 3,
174
+ "num_beams": 4,
175
+ "pad_token_id": 1,
176
+ "tie_word_embeddings": false,
177
+ "torch_dtype": "float32",
178
+ "transformers_version": null,
179
+ "vocab_size": 3319
180
+ }
checkpoint/{trocr-custdata → trocr-custdata-8000}/last/merges.txt RENAMED
File without changes
checkpoint/{trocr-custdata → trocr-custdata-8000}/last/preprocessor_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/last/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8737fad887eff5e78b0c672e9aede35436c1a4eb9dcbe99c2a0874cc811a8256
3
+ size 122142261
checkpoint/{trocr-custdata → trocr-custdata-8000}/last/special_tokens_map.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/last/tokenizer.json ADDED
@@ -0,0 +1,3408 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "<s>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": true,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "<pad>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": true,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "</s>",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": true,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "<unk>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": true,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "<mask>",
45
+ "single_word": false,
46
+ "lstrip": true,
47
+ "rstrip": false,
48
+ "normalized": true,
49
+ "special": true
50
+ }
51
+ ],
52
+ "normalizer": null,
53
+ "pre_tokenizer": {
54
+ "type": "ByteLevel",
55
+ "add_prefix_space": false,
56
+ "trim_offsets": true,
57
+ "use_regex": true
58
+ },
59
+ "post_processor": {
60
+ "type": "RobertaProcessing",
61
+ "sep": [
62
+ "</s>",
63
+ 2
64
+ ],
65
+ "cls": [
66
+ "<s>",
67
+ 0
68
+ ],
69
+ "trim_offsets": true,
70
+ "add_prefix_space": false
71
+ },
72
+ "decoder": {
73
+ "type": "ByteLevel",
74
+ "add_prefix_space": true,
75
+ "trim_offsets": true,
76
+ "use_regex": true
77
+ },
78
+ "model": {
79
+ "type": "BPE",
80
+ "dropout": null,
81
+ "unk_token": null,
82
+ "continuing_subword_prefix": "",
83
+ "end_of_word_suffix": "",
84
+ "fuse_unk": false,
85
+ "vocab": {
86
+ "<s>": 0,
87
+ "<pad>": 1,
88
+ "</s>": 2,
89
+ "<unk>": 3,
90
+ "<mask>": 4,
91
+ "奏": 5,
92
+ "音": 6,
93
+ "照": 7,
94
+ "把": 8,
95
+ "不": 9,
96
+ "尹": 10,
97
+ "伙": 11,
98
+ "蜘": 12,
99
+ "僭": 13,
100
+ "使": 14,
101
+ "啡": 15,
102
+ "柄": 16,
103
+ "羹": 17,
104
+ "嘩": 18,
105
+ "龍": 19,
106
+ "歷": 20,
107
+ "藍": 21,
108
+ "唞": 22,
109
+ "猶": 23,
110
+ "嘉": 24,
111
+ "衝": 25,
112
+ "寨": 26,
113
+ "炮": 27,
114
+ "銜": 28,
115
+ "喬": 29,
116
+ "睛": 30,
117
+ "庸": 31,
118
+ "耶": 32,
119
+ "伺": 33,
120
+ "片": 34,
121
+ "艘": 35,
122
+ "俐": 36,
123
+ "恨": 37,
124
+ "索": 38,
125
+ "哩": 39,
126
+ "哨": 40,
127
+ "銅": 41,
128
+ "璃": 42,
129
+ "貴": 43,
130
+ "斷": 44,
131
+ "達": 45,
132
+ "癩": 46,
133
+ "斜": 47,
134
+ "駛": 48,
135
+ "煲": 49,
136
+ "殘": 50,
137
+ "翻": 51,
138
+ "墨": 52,
139
+ "與": 53,
140
+ "更": 54,
141
+ "多": 55,
142
+ "博": 56,
143
+ "巨": 57,
144
+ "啓": 58,
145
+ "號": 59,
146
+ "扣": 60,
147
+ "等": 61,
148
+ "於": 62,
149
+ "勉": 63,
150
+ "脷": 64,
151
+ "盒": 65,
152
+ "扶": 66,
153
+ "濫": 67,
154
+ "式": 68,
155
+ "球": 69,
156
+ "嚿": 70,
157
+ "靴": 71,
158
+ "柯": 72,
159
+ "嘢": 73,
160
+ "后": 74,
161
+ "楊": 75,
162
+ "典": 76,
163
+ "燙": 77,
164
+ "鋪": 78,
165
+ "瀨": 79,
166
+ "肆": 80,
167
+ "赴": 81,
168
+ "牧": 82,
169
+ "紋": 83,
170
+ "況": 84,
171
+ "覆": 85,
172
+ "蜀": 86,
173
+ "省": 87,
174
+ "罩": 88,
175
+ "纜": 89,
176
+ "時": 90,
177
+ "擪": 91,
178
+ "騰": 92,
179
+ "腐": 93,
180
+ "配": 94,
181
+ "頗": 95,
182
+ "吉": 96,
183
+ "歐": 97,
184
+ "8": 98,
185
+ "卸": 99,
186
+ "醃": 100,
187
+ "鏡": 101,
188
+ "佢": 102,
189
+ "坊": 103,
190
+ "搖": 104,
191
+ "城": 105,
192
+ "戶": 106,
193
+ "基": 107,
194
+ "敷": 108,
195
+ "已": 109,
196
+ "對": 110,
197
+ "幻": 111,
198
+ "旅": 112,
199
+ "俘": 113,
200
+ "健": 114,
201
+ "悔": 115,
202
+ "飼": 116,
203
+ "宅": 117,
204
+ "燉": 118,
205
+ "櫃": 119,
206
+ "扳": 120,
207
+ "3": 121,
208
+ "h": 122,
209
+ "扮": 123,
210
+ "啋": 124,
211
+ "裁": 125,
212
+ "攀": 126,
213
+ "鵑": 127,
214
+ "有": 128,
215
+ "喻": 129,
216
+ "忟": 130,
217
+ "整": 131,
218
+ "賠": 132,
219
+ "守": 133,
220
+ "扑": 134,
221
+ "瑞": 135,
222
+ "善": 136,
223
+ "o": 137,
224
+ "假": 138,
225
+ "短": 139,
226
+ "念": 140,
227
+ "濛": 141,
228
+ "問": 142,
229
+ "盈": 143,
230
+ "擸": 144,
231
+ "旋": 145,
232
+ "漾": 146,
233
+ "蝦": 147,
234
+ "荒": 148,
235
+ "沼": 149,
236
+ "樽": 150,
237
+ "迴": 151,
238
+ "甕": 152,
239
+ "又": 153,
240
+ "嶺": 154,
241
+ "播": 155,
242
+ "下": 156,
243
+ "誰": 157,
244
+ "簿": 158,
245
+ "恙": 159,
246
+ "蛋": 160,
247
+ "V": 161,
248
+ "陣": 162,
249
+ "縫": 163,
250
+ "忌": 164,
251
+ "鬚": 165,
252
+ "字": 166,
253
+ "搬": 167,
254
+ "蚊": 168,
255
+ "方": 169,
256
+ "昧": 170,
257
+ "彈": 171,
258
+ "梗": 172,
259
+ "捽": 173,
260
+ "遇": 174,
261
+ "攻": 175,
262
+ "徑": 176,
263
+ "特": 177,
264
+ "末": 178,
265
+ "水": 179,
266
+ "t": 180,
267
+ "府": 181,
268
+ "鑽": 182,
269
+ "組": 183,
270
+ "恆": 184,
271
+ "喀": 185,
272
+ "肯": 186,
273
+ "輻": 187,
274
+ "蕾": 188,
275
+ "蝶": 189,
276
+ "拎": 190,
277
+ "藥": 191,
278
+ "址": 192,
279
+ "椅": 193,
280
+ "既": 194,
281
+ "蒸": 195,
282
+ "哂": 196,
283
+ "薯": 197,
284
+ "攣": 198,
285
+ "譽": 199,
286
+ "失": 200,
287
+ "淡": 201,
288
+ "六": 202,
289
+ "陰": 203,
290
+ "丁": 204,
291
+ "逗": 205,
292
+ "玻": 206,
293
+ "宙": 207,
294
+ "晚": 208,
295
+ "擇": 209,
296
+ "胭": 210,
297
+ "鍋": 211,
298
+ "剝": 212,
299
+ "遍": 213,
300
+ "鋭": 214,
301
+ "鮮": 215,
302
+ "伍": 216,
303
+ "漆": 217,
304
+ "停": 218,
305
+ "靜": 219,
306
+ "譚": 220,
307
+ "慧": 221,
308
+ "際": 222,
309
+ "症": 223,
310
+ "商": 224,
311
+ "各": 225,
312
+ "囚": 226,
313
+ "騎": 227,
314
+ "壩": 228,
315
+ "蹊": 229,
316
+ "偷": 230,
317
+ "漲": 231,
318
+ "恢": 232,
319
+ "嫁": 233,
320
+ "閏": 234,
321
+ "懶": 235,
322
+ "遠": 236,
323
+ "掌": 237,
324
+ "廳": 238,
325
+ "措": 239,
326
+ "老": 240,
327
+ "寺": 241,
328
+ "亂": 242,
329
+ "醇": 243,
330
+ "至": 244,
331
+ "綱": 245,
332
+ "欖": 246,
333
+ "屙": 247,
334
+ "睥": 248,
335
+ "承": 249,
336
+ "喂": 250,
337
+ "屢": 251,
338
+ "綽": 252,
339
+ "鼎": 253,
340
+ "虐": 254,
341
+ "煞": 255,
342
+ "玉": 256,
343
+ "瀏": 257,
344
+ "揪": 258,
345
+ "拃": 259,
346
+ "奪": 260,
347
+ "祭": 261,
348
+ "司": 262,
349
+ "刀": 263,
350
+ "縱": 264,
351
+ "形": 265,
352
+ "像": 266,
353
+ "哲": 267,
354
+ "相": 268,
355
+ "控": 269,
356
+ "休": 270,
357
+ "碌": 271,
358
+ "遴": 272,
359
+ "診": 273,
360
+ "泊": 274,
361
+ "牌": 275,
362
+ "最": 276,
363
+ "穫": 277,
364
+ "屬": 278,
365
+ "皆": 279,
366
+ "爭": 280,
367
+ "繁": 281,
368
+ "冰": 282,
369
+ "秒": 283,
370
+ "飾": 284,
371
+ "沒": 285,
372
+ "艱": 286,
373
+ "皈": 287,
374
+ "賭": 288,
375
+ "喎": 289,
376
+ "腔": 290,
377
+ "憶": 291,
378
+ "巷": 292,
379
+ "綁": 293,
380
+ "薰": 294,
381
+ "抓": 295,
382
+ "颱": 296,
383
+ "焚": 297,
384
+ "明": 298,
385
+ "嚐": 299,
386
+ "操": 300,
387
+ "油": 301,
388
+ "巡": 302,
389
+ "囑": 303,
390
+ "悠": 304,
391
+ "罐": 305,
392
+ "稿": 306,
393
+ "條": 307,
394
+ "敵": 308,
395
+ "險": 309,
396
+ "埋": 310,
397
+ "佛": 311,
398
+ "瀾": 312,
399
+ "膚": 313,
400
+ "閒": 314,
401
+ "趺": 315,
402
+ "何": 316,
403
+ "茸": 317,
404
+ "超": 318,
405
+ "亞": 319,
406
+ "鏹": 320,
407
+ "爸": 321,
408
+ "納": 322,
409
+ "甲": 323,
410
+ "耽": 324,
411
+ "危": 325,
412
+ "鏈": 326,
413
+ "針": 327,
414
+ "賞": 328,
415
+ "堤": 329,
416
+ "創": 330,
417
+ "澳": 331,
418
+ "飪": 332,
419
+ "押": 333,
420
+ "蕉": 334,
421
+ "心": 335,
422
+ "鑣": 336,
423
+ "耷": 337,
424
+ "西": 338,
425
+ "吖": 339,
426
+ "耕": 340,
427
+ "噬": 341,
428
+ "睇": 342,
429
+ "q": 343,
430
+ "法": 344,
431
+ "嘅": 345,
432
+ "隻": 346,
433
+ "禮": 347,
434
+ "我": 348,
435
+ "改": 349,
436
+ "啱": 350,
437
+ "此": 351,
438
+ "娘": 352,
439
+ "曳": 353,
440
+ "深": 354,
441
+ "寥": 355,
442
+ "奧": 356,
443
+ "酌": 357,
444
+ "瞞": 358,
445
+ "澄": 359,
446
+ "伏": 360,
447
+ "宿": 361,
448
+ "代": 362,
449
+ "葱": 363,
450
+ "唐": 364,
451
+ "濟": 365,
452
+ "刨": 366,
453
+ "追": 367,
454
+ "怪": 368,
455
+ "攝": 369,
456
+ "稜": 370,
457
+ "搭": 371,
458
+ "網": 372,
459
+ "遣": 373,
460
+ "沊": 374,
461
+ "淆": 375,
462
+ "迅": 376,
463
+ "螺": 377,
464
+ "憐": 378,
465
+ "符": 379,
466
+ "吓": 380,
467
+ "蜆": 381,
468
+ "O": 382,
469
+ "得": 383,
470
+ "勒": 384,
471
+ "俱": 385,
472
+ "無": 386,
473
+ "搵": 387,
474
+ "嘥": 388,
475
+ "馴": 389,
476
+ "鬼": 390,
477
+ "功": 391,
478
+ "願": 392,
479
+ "錢": 393,
480
+ "修": 394,
481
+ "胞": 395,
482
+ "趨": 396,
483
+ "虧": 397,
484
+ "邦": 398,
485
+ "杜": 399,
486
+ "理": 400,
487
+ "起": 401,
488
+ "路": 402,
489
+ "扇": 403,
490
+ "月": 404,
491
+ "釀": 405,
492
+ "b": 406,
493
+ "塊": 407,
494
+ "着": 408,
495
+ "隧": 409,
496
+ "篤": 410,
497
+ "螢": 411,
498
+ "冷": 412,
499
+ "瘤": 413,
500
+ "掩": 414,
501
+ "升": 415,
502
+ "蝕": 416,
503
+ "群": 417,
504
+ "義": 418,
505
+ "勾": 419,
506
+ "投": 420,
507
+ "粹": 421,
508
+ "泰": 422,
509
+ "食": 423,
510
+ "磅": 424,
511
+ "野": 425,
512
+ "撚": 426,
513
+ "殺": 427,
514
+ "屎": 428,
515
+ "爪": 429,
516
+ "啫": 430,
517
+ "永": 431,
518
+ "汀": 432,
519
+ "紥": 433,
520
+ "安": 434,
521
+ "梁": 435,
522
+ "還": 436,
523
+ "積": 437,
524
+ "佻": 438,
525
+ "伊": 439,
526
+ "滑": 440,
527
+ "劉": 441,
528
+ "歇": 442,
529
+ "坐": 443,
530
+ "板": 444,
531
+ "棋": 445,
532
+ "誘": 446,
533
+ "壺": 447,
534
+ "編": 448,
535
+ "似": 449,
536
+ "悶": 450,
537
+ "鹹": 451,
538
+ "罕": 452,
539
+ "難": 453,
540
+ "蠢": 454,
541
+ "郵": 455,
542
+ "唉": 456,
543
+ "狂": 457,
544
+ "羊": 458,
545
+ "上": 459,
546
+ "入": 460,
547
+ "樑": 461,
548
+ "學": 462,
549
+ "模": 463,
550
+ "耘": 464,
551
+ "潺": 465,
552
+ "練": 466,
553
+ "驅": 467,
554
+ "吾": 468,
555
+ "凍": 469,
556
+ "咿": 470,
557
+ "抬": 471,
558
+ "卦": 472,
559
+ "攜": 473,
560
+ "噪": 474,
561
+ "偏": 475,
562
+ "蠟": 476,
563
+ "麼": 477,
564
+ "霎": 478,
565
+ "誓": 479,
566
+ "故": 480,
567
+ "姓": 481,
568
+ "欠": 482,
569
+ "令": 483,
570
+ "梅": 484,
571
+ "」": 485,
572
+ "窄": 486,
573
+ "布": 487,
574
+ "豁": 488,
575
+ "妖": 489,
576
+ "喝": 490,
577
+ "項": 491,
578
+ "嘗": 492,
579
+ "職": 493,
580
+ "卡": 494,
581
+ "朦": 495,
582
+ "俾": 496,
583
+ "拉": 497,
584
+ "指": 498,
585
+ "釣": 499,
586
+ "套": 500,
587
+ "悲": 501,
588
+ "仃": 502,
589
+ "旺": 503,
590
+ "恭": 504,
591
+ "拖": 505,
592
+ "痠": 506,
593
+ "遙": 507,
594
+ "象": 508,
595
+ "界": 509,
596
+ "珊": 510,
597
+ "瘦": 511,
598
+ "裨": 512,
599
+ "別": 513,
600
+ "衩": 514,
601
+ "皮": 515,
602
+ "師": 516,
603
+ "牙": 517,
604
+ "揚": 518,
605
+ "寶": 519,
606
+ "欄": 520,
607
+ "刪": 521,
608
+ "溝": 522,
609
+ "重": 523,
610
+ "泡": 524,
611
+ "賓": 525,
612
+ "樹": 526,
613
+ "稽": 527,
614
+ "胡": 528,
615
+ "芥": 529,
616
+ "戀": 530,
617
+ "桌": 531,
618
+ "翼": 532,
619
+ "崩": 533,
620
+ "泳": 534,
621
+ "昇": 535,
622
+ "勵": 536,
623
+ "枝": 537,
624
+ "J": 538,
625
+ "績": 539,
626
+ "胎": 540,
627
+ "蔔": 541,
628
+ "思": 542,
629
+ "凶": 543,
630
+ "搏": 544,
631
+ "孕": 545,
632
+ "甫": 546,
633
+ "嚴": 547,
634
+ "2": 548,
635
+ "貶": 549,
636
+ "圓": 550,
637
+ "察": 551,
638
+ "血": 552,
639
+ "賽": 553,
640
+ "聲": 554,
641
+ "0": 555,
642
+ "婉": 556,
643
+ "濕": 557,
644
+ "裂": 558,
645
+ "!": 559,
646
+ "琢": 560,
647
+ "參": 561,
648
+ "析": 562,
649
+ "黨": 563,
650
+ "森": 564,
651
+ "惡": 565,
652
+ "委": 566,
653
+ "捉": 567,
654
+ "僧": 568,
655
+ "飛": 569,
656
+ "愛": 570,
657
+ "儉": 571,
658
+ "仲": 572,
659
+ "憲": 573,
660
+ "喳": 574,
661
+ "貪": 575,
662
+ "挽": 576,
663
+ "爬": 577,
664
+ "跨": 578,
665
+ "%": 579,
666
+ "盛": 580,
667
+ "鯨": 581,
668
+ "咗": 582,
669
+ "禾": 583,
670
+ "童": 584,
671
+ "瘡": 585,
672
+ "蘿": 586,
673
+ "麒": 587,
674
+ "攞": 588,
675
+ "膜": 589,
676
+ "軭": 590,
677
+ "𧕴": 591,
678
+ "插": 592,
679
+ "妃": 593,
680
+ "進": 594,
681
+ "呈": 595,
682
+ "拇": 596,
683
+ "標": 597,
684
+ "唔": 598,
685
+ "隍": 599,
686
+ "擘": 600,
687
+ "扎": 601,
688
+ "障": 602,
689
+ "捨": 603,
690
+ "靠": 604,
691
+ "刻": 605,
692
+ "磷": 606,
693
+ "麻": 607,
694
+ "腫": 608,
695
+ "玲": 609,
696
+ "梘": 610,
697
+ "蹤": 611,
698
+ "蒙": 612,
699
+ "杰": 613,
700
+ "蝴": 614,
701
+ "喱": 615,
702
+ "𠸎": 616,
703
+ "查": 617,
704
+ "預": 618,
705
+ "熟": 619,
706
+ "足": 620,
707
+ "瓜": 621,
708
+ "忙": 622,
709
+ "臉": 623,
710
+ "咄": 624,
711
+ "圍": 625,
712
+ "僱": 626,
713
+ "爽": 627,
714
+ "廹": 628,
715
+ "移": 629,
716
+ "炬": 630,
717
+ "息": 631,
718
+ "坡": 632,
719
+ "爐": 633,
720
+ "場": 634,
721
+ "臣": 635,
722
+ "丫": 636,
723
+ "勢": 637,
724
+ "覺": 638,
725
+ "威": 639,
726
+ "妳": 640,
727
+ "左": 641,
728
+ "牛": 642,
729
+ "寞": 643,
730
+ "𠵇": 644,
731
+ "腳": 645,
732
+ "冊": 646,
733
+ "塌": 647,
734
+ "潭": 648,
735
+ "研": 649,
736
+ "拐": 650,
737
+ "糕": 651,
738
+ "迎": 652,
739
+ "翹": 653,
740
+ "笨": 654,
741
+ "蔗": 655,
742
+ "穢": 656,
743
+ "震": 657,
744
+ "畏": 658,
745
+ "婚": 659,
746
+ "良": 660,
747
+ "惠": 661,
748
+ "遁": 662,
749
+ "臌": 663,
750
+ "姐": 664,
751
+ "䂻": 665,
752
+ "悦": 666,
753
+ "頁": 667,
754
+ "𡁻": 668,
755
+ "蟹": 669,
756
+ "菌": 670,
757
+ "敬": 671,
758
+ "茂": 672,
759
+ "邊": 673,
760
+ "凌": 674,
761
+ "關": 675,
762
+ "英": 676,
763
+ "曬": 677,
764
+ "啦": 678,
765
+ "餅": 679,
766
+ "審": 680,
767
+ "需": 681,
768
+ "歡": 682,
769
+ "負": 683,
770
+ "鳳": 684,
771
+ "枳": 685,
772
+ "芭": 686,
773
+ "鬧": 687,
774
+ "冠": 688,
775
+ "謀": 689,
776
+ "零": 690,
777
+ "衞": 691,
778
+ "悟": 692,
779
+ "狗": 693,
780
+ "細": 694,
781
+ "舒": 695,
782
+ "招": 696,
783
+ "罹": 697,
784
+ "擠": 698,
785
+ "蘑": 699,
786
+ "秦": 700,
787
+ "樂": 701,
788
+ "腦": 702,
789
+ "響": 703,
790
+ "蟲": 704,
791
+ "一": 705,
792
+ "徊": 706,
793
+ "膉": 707,
794
+ "供": 708,
795
+ "l": 709,
796
+ "戥": 710,
797
+ "詳": 711,
798
+ "遮": 712,
799
+ "蒜": 713,
800
+ "硯": 714,
801
+ "尖": 715,
802
+ "挑": 716,
803
+ "企": 717,
804
+ "屌": 718,
805
+ "公": 719,
806
+ "完": 720,
807
+ "吟": 721,
808
+ "觸": 722,
809
+ "棉": 723,
810
+ "���": 724,
811
+ "文": 725,
812
+ "爆": 726,
813
+ "毀": 727,
814
+ "跪": 728,
815
+ "榴": 729,
816
+ "麖": 730,
817
+ "藐": 731,
818
+ "趣": 732,
819
+ "百": 733,
820
+ "拗": 734,
821
+ "蹈": 735,
822
+ "餸": 736,
823
+ "鐳": 737,
824
+ "菇": 738,
825
+ "權": 739,
826
+ "極": 740,
827
+ "詭": 741,
828
+ "八": 742,
829
+ "夾": 743,
830
+ "佈": 744,
831
+ "契": 745,
832
+ "攊": 746,
833
+ "朝": 747,
834
+ "漠": 748,
835
+ "彌": 749,
836
+ "匿": 750,
837
+ "葺": 751,
838
+ "眛": 752,
839
+ "應": 753,
840
+ "豬": 754,
841
+ "植": 755,
842
+ "両": 756,
843
+ "渧": 757,
844
+ "覽": 758,
845
+ "聊": 759,
846
+ "咁": 760,
847
+ "採": 761,
848
+ "尼": 762,
849
+ "謬": 763,
850
+ "農": 764,
851
+ "彼": 765,
852
+ "買": 766,
853
+ "皇": 767,
854
+ "欲": 768,
855
+ "污": 769,
856
+ "魚": 770,
857
+ "慷": 771,
858
+ "顯": 772,
859
+ "瓶": 773,
860
+ "白": 774,
861
+ "乍": 775,
862
+ "王": 776,
863
+ " ": 777,
864
+ "稱": 778,
865
+ "孫": 779,
866
+ "滲": 780,
867
+ "面": 781,
868
+ "辯": 782,
869
+ "秤": 783,
870
+ "腩": 784,
871
+ "桔": 785,
872
+ "財": 786,
873
+ "毆": 787,
874
+ "遭": 788,
875
+ "掉": 789,
876
+ "Y": 790,
877
+ "氧": 791,
878
+ "垢": 792,
879
+ "黃": 793,
880
+ "傘": 794,
881
+ "裇": 795,
882
+ "倪": 796,
883
+ "嘍": 797,
884
+ "較": 798,
885
+ "弊": 799,
886
+ "世": 800,
887
+ "篇": 801,
888
+ "萃": 802,
889
+ "鬟": 803,
890
+ "櫳": 804,
891
+ "嬰": 805,
892
+ "懼": 806,
893
+ "噴": 807,
894
+ "噶": 808,
895
+ "貫": 809,
896
+ "波": 810,
897
+ "奢": 811,
898
+ "救": 812,
899
+ "謊": 813,
900
+ "墜": 814,
901
+ "疊": 815,
902
+ "眾": 816,
903
+ "據": 817,
904
+ "效": 818,
905
+ "容": 819,
906
+ "掂": 820,
907
+ "欺": 821,
908
+ "凸": 822,
909
+ "靡": 823,
910
+ "鑑": 824,
911
+ "妨": 825,
912
+ "二": 826,
913
+ "跳": 827,
914
+ "腍": 828,
915
+ "瓣": 829,
916
+ "割": 830,
917
+ "奉": 831,
918
+ "新": 832,
919
+ "丑": 833,
920
+ "椗": 834,
921
+ "贖": 835,
922
+ "頇": 836,
923
+ "厘": 837,
924
+ "碳": 838,
925
+ "阻": 839,
926
+ "郎": 840,
927
+ "誡": 841,
928
+ "行": 842,
929
+ "4": 843,
930
+ "畀": 844,
931
+ "絕": 845,
932
+ "咳": 846,
933
+ "i": 847,
934
+ "戈": 848,
935
+ "簧": 849,
936
+ "卵": 850,
937
+ "書": 851,
938
+ "蹲": 852,
939
+ "訂": 853,
940
+ "輔": 854,
941
+ "箍": 855,
942
+ "惕": 856,
943
+ "尾": 857,
944
+ "鬙": 858,
945
+ "荷": 859,
946
+ "臘": 860,
947
+ "肝": 861,
948
+ "要": 862,
949
+ "俏": 863,
950
+ "雪": 864,
951
+ "擲": 865,
952
+ "漪": 866,
953
+ "違": 867,
954
+ "苛": 868,
955
+ "豆": 869,
956
+ "道": 870,
957
+ "鑿": 871,
958
+ "教": 872,
959
+ "愁": 873,
960
+ "局": 874,
961
+ "服": 875,
962
+ "苦": 876,
963
+ "蓮": 877,
964
+ "瞌": 878,
965
+ "銷": 879,
966
+ "祇": 880,
967
+ "但": 881,
968
+ "作": 882,
969
+ "蛙": 883,
970
+ "擊": 884,
971
+ "掣": 885,
972
+ "軸": 886,
973
+ "噚": 887,
974
+ "脅": 888,
975
+ "刷": 889,
976
+ "期": 890,
977
+ "紙": 891,
978
+ "解": 892,
979
+ "剩": 893,
980
+ "賤": 894,
981
+ "印": 895,
982
+ "急": 896,
983
+ "湃": 897,
984
+ "賬": 898,
985
+ "構": 899,
986
+ "器": 900,
987
+ "他": 901,
988
+ "勸": 902,
989
+ "票": 903,
990
+ "襯": 904,
991
+ "先": 905,
992
+ "屏": 906,
993
+ "烘": 907,
994
+ "涉": 908,
995
+ "粵": 909,
996
+ "胥": 910,
997
+ "堆": 911,
998
+ "摧": 912,
999
+ "母": 913,
1000
+ "攤": 914,
1001
+ "仆": 915,
1002
+ "贈": 916,
1003
+ "怖": 917,
1004
+ "共": 918,
1005
+ "乓": 919,
1006
+ "逾": 920,
1007
+ "半": 921,
1008
+ "a": 922,
1009
+ "草": 923,
1010
+ "述": 924,
1011
+ "肩": 925,
1012
+ "登": 926,
1013
+ "頹": 927,
1014
+ "俄": 928,
1015
+ "惶": 929,
1016
+ "晝": 930,
1017
+ "抗": 931,
1018
+ "侈": 932,
1019
+ "填": 933,
1020
+ "譬": 934,
1021
+ "鋒": 935,
1022
+ "搶": 936,
1023
+ "實": 937,
1024
+ "律": 938,
1025
+ "繹": 939,
1026
+ "碧": 940,
1027
+ "添": 941,
1028
+ "交": 942,
1029
+ "望": 943,
1030
+ "㬹": 944,
1031
+ "榜": 945,
1032
+ "暢": 946,
1033
+ "框": 947,
1034
+ "胸": 948,
1035
+ "囉": 949,
1036
+ "親": 950,
1037
+ "潔": 951,
1038
+ "信": 952,
1039
+ "韓": 953,
1040
+ "知": 954,
1041
+ "𠺘": 955,
1042
+ "初": 956,
1043
+ "掁": 957,
1044
+ "露": 958,
1045
+ "官": 959,
1046
+ "堅": 960,
1047
+ "許": 961,
1048
+ "暉": 962,
1049
+ "揀": 963,
1050
+ "氹": 964,
1051
+ "單": 965,
1052
+ "迦": 966,
1053
+ "谷": 967,
1054
+ "褪": 968,
1055
+ "撒": 969,
1056
+ "廣": 970,
1057
+ "輪": 971,
1058
+ "結": 972,
1059
+ "渦": 973,
1060
+ "邨": 974,
1061
+ "高": 975,
1062
+ "井": 976,
1063
+ "奔": 977,
1064
+ "魏": 978,
1065
+ "腰": 979,
1066
+ "。": 980,
1067
+ "芽": 981,
1068
+ "軚": 982,
1069
+ "F": 983,
1070
+ "密": 984,
1071
+ "雹": 985,
1072
+ "鐸": 986,
1073
+ "齣": 987,
1074
+ "牢": 988,
1075
+ "通": 989,
1076
+ "麵": 990,
1077
+ "鴉": 991,
1078
+ "眷": 992,
1079
+ "廟": 993,
1080
+ "��": 994,
1081
+ "墓": 995,
1082
+ "氏": 996,
1083
+ "瀕": 997,
1084
+ "擬": 998,
1085
+ "鷹": 999,
1086
+ "談": 1000,
1087
+ "蓋": 1001,
1088
+ "匯": 1002,
1089
+ "言": 1003,
1090
+ "塗": 1004,
1091
+ "藝": 1005,
1092
+ "壞": 1006,
1093
+ "肺": 1007,
1094
+ "動": 1008,
1095
+ "萬": 1009,
1096
+ "娃": 1010,
1097
+ "底": 1011,
1098
+ "灰": 1012,
1099
+ "夕": 1013,
1100
+ "姦": 1014,
1101
+ "丸": 1015,
1102
+ "G": 1016,
1103
+ "刁": 1017,
1104
+ "奄": 1018,
1105
+ "槤": 1019,
1106
+ "自": 1020,
1107
+ "殃": 1021,
1108
+ "消": 1022,
1109
+ "啹": 1023,
1110
+ "夜": 1024,
1111
+ "態": 1025,
1112
+ "扯": 1026,
1113
+ "滂": 1027,
1114
+ "垂": 1028,
1115
+ "給": 1029,
1116
+ "勿": 1030,
1117
+ "右": 1031,
1118
+ "分": 1032,
1119
+ "裸": 1033,
1120
+ "能": 1034,
1121
+ "稚": 1035,
1122
+ "業": 1036,
1123
+ "喉": 1037,
1124
+ "制": 1038,
1125
+ "摺": 1039,
1126
+ "糧": 1040,
1127
+ "勤": 1041,
1128
+ "潤": 1042,
1129
+ "絨": 1043,
1130
+ "成": 1044,
1131
+ "撐": 1045,
1132
+ "臍": 1046,
1133
+ "棍": 1047,
1134
+ "嘆": 1048,
1135
+ "訪": 1049,
1136
+ "做": 1050,
1137
+ "裝": 1051,
1138
+ "傢": 1052,
1139
+ "窒": 1053,
1140
+ "丶": 1054,
1141
+ "累": 1055,
1142
+ "惑": 1056,
1143
+ "悼": 1057,
1144
+ "遂": 1058,
1145
+ "擺": 1059,
1146
+ "跋": 1060,
1147
+ "築": 1061,
1148
+ "劈": 1062,
1149
+ "誤": 1063,
1150
+ "湧": 1064,
1151
+ "御": 1065,
1152
+ "廈": 1066,
1153
+ "淺": 1067,
1154
+ "淫": 1068,
1155
+ "A": 1069,
1156
+ "揼": 1070,
1157
+ "陪": 1071,
1158
+ "匹": 1072,
1159
+ "呎": 1073,
1160
+ "部": 1074,
1161
+ "中": 1075,
1162
+ "固": 1076,
1163
+ "奮": 1077,
1164
+ "糊": 1078,
1165
+ "稀": 1079,
1166
+ "逕": 1080,
1167
+ "詢": 1081,
1168
+ "洋": 1082,
1169
+ "枚": 1083,
1170
+ "檸": 1084,
1171
+ "顫": 1085,
1172
+ "風": 1086,
1173
+ "駐": 1087,
1174
+ "責": 1088,
1175
+ "埲": 1089,
1176
+ "c": 1090,
1177
+ "孩": 1091,
1178
+ "階": 1092,
1179
+ "回": 1093,
1180
+ "洪": 1094,
1181
+ "束": 1095,
1182
+ "孺": 1096,
1183
+ "訊": 1097,
1184
+ "芻": 1098,
1185
+ "藉": 1099,
1186
+ "嫡": 1100,
1187
+ "晾": 1101,
1188
+ "庭": 1102,
1189
+ "士": 1103,
1190
+ "懦": 1104,
1191
+ "刑": 1105,
1192
+ "如": 1106,
1193
+ "支": 1107,
1194
+ "店": 1108,
1195
+ "R": 1109,
1196
+ "咋": 1110,
1197
+ "膊": 1111,
1198
+ "熨": 1112,
1199
+ "妝": 1113,
1200
+ "藏": 1114,
1201
+ "興": 1115,
1202
+ "犀": 1116,
1203
+ "馳": 1117,
1204
+ "嚒": 1118,
1205
+ "掃": 1119,
1206
+ "境": 1120,
1207
+ "利": 1121,
1208
+ "𠺌": 1122,
1209
+ "父": 1123,
1210
+ "汁": 1124,
1211
+ "殉": 1125,
1212
+ "捋": 1126,
1213
+ "緋": 1127,
1214
+ "窩": 1128,
1215
+ "伯": 1129,
1216
+ "匆": 1130,
1217
+ "討": 1131,
1218
+ "旬": 1132,
1219
+ "鎖": 1133,
1220
+ "逃": 1134,
1221
+ "股": 1135,
1222
+ "菠": 1136,
1223
+ "紀": 1137,
1224
+ "定": 1138,
1225
+ "5": 1139,
1226
+ "貸": 1140,
1227
+ "慶": 1141,
1228
+ "建": 1142,
1229
+ "昏": 1143,
1230
+ "順": 1144,
1231
+ "歸": 1145,
1232
+ "君": 1146,
1233
+ "魔": 1147,
1234
+ "𠱸": 1148,
1235
+ "怨": 1149,
1236
+ "醫": 1150,
1237
+ "盲": 1151,
1238
+ "焰": 1152,
1239
+ "犯": 1153,
1240
+ "嶄": 1154,
1241
+ "括": 1155,
1242
+ "㾓": 1156,
1243
+ "注": 1157,
1244
+ "渠": 1158,
1245
+ "寫": 1159,
1246
+ "尊": 1160,
1247
+ "訴": 1161,
1248
+ "乞": 1162,
1249
+ "須": 1163,
1250
+ "垃": 1164,
1251
+ "沛": 1165,
1252
+ "富": 1166,
1253
+ "空": 1167,
1254
+ "格": 1168,
1255
+ "謹": 1169,
1256
+ "獵": 1170,
1257
+ "岌": 1171,
1258
+ "曆": 1172,
1259
+ "愧": 1173,
1260
+ "埗": 1174,
1261
+ "題": 1175,
1262
+ "擔": 1176,
1263
+ "曝": 1177,
1264
+ "鱗": 1178,
1265
+ "芯": 1179,
1266
+ "快": 1180,
1267
+ "拮": 1181,
1268
+ "提": 1182,
1269
+ "》": 1183,
1270
+ "傾": 1184,
1271
+ "賴": 1185,
1272
+ "鍾": 1186,
1273
+ "妙": 1187,
1274
+ "笪": 1188,
1275
+ "采": 1189,
1276
+ "絀": 1190,
1277
+ "跌": 1191,
1278
+ "鈔": 1192,
1279
+ "墅": 1193,
1280
+ "罷": 1194,
1281
+ "刊": 1195,
1282
+ "托": 1196,
1283
+ "幫": 1197,
1284
+ "撇": 1198,
1285
+ "依": 1199,
1286
+ "忠": 1200,
1287
+ "衷": 1201,
1288
+ "頂": 1202,
1289
+ "擂": 1203,
1290
+ "嘞": 1204,
1291
+ "乳": 1205,
1292
+ "沽": 1206,
1293
+ "卅": 1207,
1294
+ "哦": 1208,
1295
+ "識": 1209,
1296
+ "漢": 1210,
1297
+ "份": 1211,
1298
+ "封": 1212,
1299
+ "膏": 1213,
1300
+ "陸": 1214,
1301
+ "粒": 1215,
1302
+ "搞": 1216,
1303
+ "哺": 1217,
1304
+ "擳": 1218,
1305
+ "捍": 1219,
1306
+ "欣": 1220,
1307
+ "趁": 1221,
1308
+ "渡": 1222,
1309
+ "厚": 1223,
1310
+ "貧": 1224,
1311
+ "噁": 1225,
1312
+ "磨": 1226,
1313
+ "攰": 1227,
1314
+ "統": 1228,
1315
+ "P": 1229,
1316
+ "撤": 1230,
1317
+ "篷": 1231,
1318
+ "髮": 1232,
1319
+ "啞": 1233,
1320
+ "吻": 1234,
1321
+ "汐": 1235,
1322
+ "過": 1236,
1323
+ "摵": 1237,
1324
+ "飽": 1238,
1325
+ "緣": 1239,
1326
+ "具": 1240,
1327
+ "呀": 1241,
1328
+ "捕": 1242,
1329
+ "准": 1243,
1330
+ "（": 1244,
1331
+ "社": 1245,
1332
+ "曖": 1246,
1333
+ "臟": 1247,
1334
+ "齒": 1248,
1335
+ "剛": 1249,
1336
+ "工": 1250,
1337
+ "延": 1251,
1338
+ "及": 1252,
1339
+ "罅": 1253,
1340
+ "牒": 1254,
1341
+ "欽": 1255,
1342
+ "卧": 1256,
1343
+ "迄": 1257,
1344
+ "端": 1258,
1345
+ "夷": 1259,
1346
+ "恐": 1260,
1347
+ "易": 1261,
1348
+ "島": 1262,
1349
+ "松": 1263,
1350
+ "鹿": 1264,
1351
+ "陽": 1265,
1352
+ "介": 1266,
1353
+ "間": 1267,
1354
+ "掹": 1268,
1355
+ "免": 1269,
1356
+ "腥": 1270,
1357
+ "殼": 1271,
1358
+ "朱": 1272,
1359
+ "吳": 1273,
1360
+ "所": 1274,
1361
+ "纖": 1275,
1362
+ "療": 1276,
1363
+ "貼": 1277,
1364
+ "肉": 1278,
1365
+ "誇": 1279,
1366
+ "斑": 1280,
1367
+ "帥": 1281,
1368
+ "稔": 1282,
1369
+ "咧": 1283,
1370
+ "就": 1284,
1371
+ "評": 1285,
1372
+ "策": 1286,
1373
+ "𡃁": 1287,
1374
+ "偶": 1288,
1375
+ "婆": 1289,
1376
+ "聚": 1290,
1377
+ "茲": 1291,
1378
+ "只": 1292,
1379
+ "躝": 1293,
1380
+ "混": 1294,
1381
+ "么": 1295,
1382
+ "專": 1296,
1383
+ "陋": 1297,
1384
+ "／": 1298,
1385
+ "粗": 1299,
1386
+ "科": 1300,
1387
+ "類": 1301,
1388
+ "煮": 1302,
1389
+ "晃": 1303,
1390
+ "漣": 1304,
1391
+ "昔": 1305,
1392
+ "雅": 1306,
1393
+ "銘": 1307,
1394
+ "載": 1308,
1395
+ "獸": 1309,
1396
+ "拋": 1310,
1397
+ "牀": 1311,
1398
+ "傻": 1312,
1399
+ "掯": 1313,
1400
+ "鋸": 1314,
1401
+ "晰": 1315,
1402
+ "嬲": 1316,
1403
+ "奇": 1317,
1404
+ "疏": 1318,
1405
+ "嘈": 1319,
1406
+ "渴": 1320,
1407
+ "濾": 1321,
1408
+ "啲": 1322,
1409
+ "綜": 1323,
1410
+ "梳": 1324,
1411
+ "秋": 1325,
1412
+ "戚": 1326,
1413
+ "署": 1327,
1414
+ "煉": 1328,
1415
+ "孔": 1329,
1416
+ "龕": 1330,
1417
+ "兇": 1331,
1418
+ "麗": 1332,
1419
+ "撳": 1333,
1420
+ "費": 1334,
1421
+ "Q": 1335,
1422
+ "居": 1336,
1423
+ "嗤": 1337,
1424
+ "劑": 1338,
1425
+ "漚": 1339,
1426
+ "阿": 1340,
1427
+ "羽": 1341,
1428
+ "舌": 1342,
1429
+ "赦": 1343,
1430
+ "援": 1344,
1431
+ "並": 1345,
1432
+ "體": 1346,
1433
+ "麪": 1347,
1434
+ "拆": 1348,
1435
+ "仕": 1349,
1436
+ "放": 1350,
1437
+ "微": 1351,
1438
+ "施": 1352,
1439
+ "戇": 1353,
1440
+ "困": 1354,
1441
+ "鶴": 1355,
1442
+ "(": 1356,
1443
+ "側": 1357,
1444
+ "縮": 1358,
1445
+ "蛇": 1359,
1446
+ "五": 1360,
1447
+ "槍": 1361,
1448
+ "取": 1362,
1449
+ "離": 1363,
1450
+ "剔": 1364,
1451
+ "咸": 1365,
1452
+ "探": 1366,
1453
+ "袱": 1367,
1454
+ "隔": 1368,
1455
+ "燒": 1369,
1456
+ "犧": 1370,
1457
+ "酬": 1371,
1458
+ "逢": 1372,
1459
+ "死": 1373,
1460
+ "哎": 1374,
1461
+ "埔": 1375,
1462
+ "角": 1376,
1463
+ "購": 1377,
1464
+ "曼": 1378,
1465
+ "踏": 1379,
1466
+ "慣": 1380,
1467
+ "餘": 1381,
1468
+ "紛": 1382,
1469
+ "蹺": 1383,
1470
+ "金": 1384,
1471
+ "仔": 1385,
1472
+ "畢": 1386,
1473
+ "爾": 1387,
1474
+ "競": 1388,
1475
+ "遲": 1389,
1476
+ "蓬": 1390,
1477
+ "瀝": 1391,
1478
+ "精": 1392,
1479
+ "芳": 1393,
1480
+ "券": 1394,
1481
+ "殖": 1395,
1482
+ "胃": 1396,
1483
+ "液": 1397,
1484
+ "巢": 1398,
1485
+ "撰": 1399,
1486
+ "：": 1400,
1487
+ "非": 1401,
1488
+ "誠": 1402,
1489
+ "唓": 1403,
1490
+ "亦": 1404,
1491
+ "舍": 1405,
1492
+ "型": 1406,
1493
+ "彩": 1407,
1494
+ "滯": 1408,
1495
+ "碗": 1409,
1496
+ "婪": 1410,
1497
+ "䟕": 1411,
1498
+ "妥": 1412,
1499
+ "佗": 1413,
1500
+ "玫": 1414,
1501
+ "之": 1415,
1502
+ "慮": 1416,
1503
+ "剷": 1417,
1504
+ "疫": 1418,
1505
+ "尿": 1419,
1506
+ "敦": 1420,
1507
+ "笠": 1421,
1508
+ "般": 1422,
1509
+ "員": 1423,
1510
+ "名": 1424,
1511
+ "唧": 1425,
1512
+ "厭": 1426,
1513
+ "啖": 1427,
1514
+ "走": 1428,
1515
+ "丈": 1429,
1516
+ "弗": 1430,
1517
+ "儡": 1431,
1518
+ "福": 1432,
1519
+ "黐": 1433,
1520
+ "醬": 1434,
1521
+ "情": 1435,
1522
+ "姻": 1436,
1523
+ "綿": 1437,
1524
+ "y": 1438,
1525
+ "撈": 1439,
1526
+ "務": 1440,
1527
+ "澎": 1441,
1528
+ "頸": 1442,
1529
+ "𠺫": 1443,
1530
+ "疹": 1444,
1531
+ "卜": 1445,
1532
+ "！": 1446,
1533
+ "峽": 1447,
1534
+ "傳": 1448,
1535
+ "剁": 1449,
1536
+ "嚇": 1450,
1537
+ "旨": 1451,
1538
+ "孝": 1452,
1539
+ "求": 1453,
1540
+ "驟": 1454,
1541
+ "躉": 1455,
1542
+ "前": 1456,
1543
+ "任": 1457,
1544
+ "增": 1458,
1545
+ "卯": 1459,
1546
+ "痕": 1460,
1547
+ "孖": 1461,
1548
+ "逸": 1462,
1549
+ "髹": 1463,
1550
+ "氣": 1464,
1551
+ "握": 1465,
1552
+ "豪": 1466,
1553
+ "抖": 1467,
1554
+ "錯": 1468,
1555
+ "顧": 1469,
1556
+ "俗": 1470,
1557
+ "斃": 1471,
1558
+ "霸": 1472,
1559
+ "資": 1473,
1560
+ "瘟": 1474,
1561
+ "n": 1475,
1562
+ "甩": 1476,
1563
+ "藕": 1477,
1564
+ "鳴": 1478,
1565
+ "架": 1479,
1566
+ "湯": 1480,
1567
+ "憑": 1481,
1568
+ "浮": 1482,
1569
+ "開": 1483,
1570
+ "盞": 1484,
1571
+ "帳": 1485,
1572
+ "往": 1486,
1573
+ "避": 1487,
1574
+ "勻": 1488,
1575
+ "軌": 1489,
1576
+ "裏": 1490,
1577
+ "仇": 1491,
1578
+ "康": 1492,
1579
+ "續": 1493,
1580
+ "借": 1494,
1581
+ "蹄": 1495,
1582
+ "同": 1496,
1583
+ "幹": 1497,
1584
+ "鬱": 1498,
1585
+ "？": 1499,
1586
+ "嚨": 1500,
1587
+ "煤": 1501,
1588
+ "橫": 1502,
1589
+ "彿": 1503,
1590
+ "燥": 1504,
1591
+ "C": 1505,
1592
+ "侄": 1506,
1593
+ "祿": 1507,
1594
+ "痺": 1508,
1595
+ "鈴": 1509,
1596
+ "熊": 1510,
1597
+ "s": 1511,
1598
+ "蟎": 1512,
1599
+ "班": 1513,
1600
+ "子": 1514,
1601
+ "造": 1515,
1602
+ "矢": 1516,
1603
+ "愾": 1517,
1604
+ "族": 1518,
1605
+ "𠱁": 1519,
1606
+ "啜": 1520,
1607
+ "副": 1521,
1608
+ "蓆": 1522,
1609
+ "墟": 1523,
1610
+ "答": 1524,
1611
+ "抄": 1525,
1612
+ "駕": 1526,
1613
+ "鬍": 1527,
1614
+ "凜": 1528,
1615
+ "治": 1529,
1616
+ "崖": 1530,
1617
+ "系": 1531,
1618
+ "愈": 1532,
1619
+ "繩": 1533,
1620
+ "補": 1534,
1621
+ "蒲": 1535,
1622
+ "引": 1536,
1623
+ "併": 1537,
1624
+ "g": 1538,
1625
+ "火": 1539,
1626
+ "橡": 1540,
1627
+ "咓": 1541,
1628
+ "踿": 1542,
1629
+ "禁": 1543,
1630
+ "甜": 1544,
1631
+ "冥": 1545,
1632
+ "嶼": 1546,
1633
+ "十": 1547,
1634
+ "太": 1548,
1635
+ "發": 1549,
1636
+ "賣": 1550,
1637
+ "切": 1551,
1638
+ "徘": 1552,
1639
+ "蓉": 1553,
1640
+ "都": 1554,
1641
+ "耗": 1555,
1642
+ "叭": 1556,
1643
+ "揸": 1557,
1644
+ "旭": 1558,
1645
+ "當": 1559,
1646
+ "x": 1560,
1647
+ "狀": 1561,
1648
+ "竅": 1562,
1649
+ "劍": 1563,
1650
+ "巧": 1564,
1651
+ "崇": 1565,
1652
+ "事": 1566,
1653
+ "揉": 1567,
1654
+ "穿": 1568,
1655
+ "痰": 1569,
1656
+ "清": 1570,
1657
+ "漬": 1571,
1658
+ "嚏": 1572,
1659
+ "為": 1573,
1660
+ "客": 1574,
1661
+ "吊": 1575,
1662
+ "瞬": 1576,
1663
+ "窈": 1577,
1664
+ "臨": 1578,
1665
+ "筒": 1579,
1666
+ "噉": 1580,
1667
+ "勇": 1581,
1668
+ "峯": 1582,
1669
+ "棲": 1583,
1670
+ "麂": 1584,
1671
+ "炭": 1585,
1672
+ "鼻": 1586,
1673
+ "廂": 1587,
1674
+ "大": 1588,
1675
+ "咫": 1589,
1676
+ "鉛": 1590,
1677
+ "蹟": 1591,
1678
+ "嘴": 1592,
1679
+ "少": 1593,
1680
+ "猜": 1594,
1681
+ "洛": 1595,
1682
+ "災": 1596,
1683
+ "遊": 1597,
1684
+ "啤": 1598,
1685
+ "朋": 1599,
1686
+ "幣": 1600,
1687
+ "斗": 1601,
1688
+ "現": 1602,
1689
+ "剪": 1603,
1690
+ "聰": 1604,
1691
+ "昆": 1605,
1692
+ "台": 1606,
1693
+ "辭": 1607,
1694
+ "剽": 1608,
1695
+ "融": 1609,
1696
+ "囇": 1610,
1697
+ "位": 1611,
1698
+ "潦": 1612,
1699
+ "暫": 1613,
1700
+ "係": 1614,
1701
+ "雜": 1615,
1702
+ "盆": 1616,
1703
+ "屆": 1617,
1704
+ "呼": 1618,
1705
+ "罵": 1619,
1706
+ "爺": 1620,
1707
+ "柴": 1621,
1708
+ "耿": 1622,
1709
+ "表": 1623,
1710
+ "菖": 1624,
1711
+ "包": 1625,
1712
+ "壽": 1626,
1713
+ "用": 1627,
1714
+ "鯁": 1628,
1715
+ "毗": 1629,
1716
+ "值": 1630,
1717
+ "仍": 1631,
1718
+ "監": 1632,
1719
+ "疤": 1633,
1720
+ "估": 1634,
1721
+ "廊": 1635,
1722
+ "衍": 1636,
1723
+ "蕊": 1637,
1724
+ "溪": 1638,
1725
+ "視": 1639,
1726
+ "芝": 1640,
1727
+ "缸": 1641,
1728
+ "雀": 1642,
1729
+ "兩": 1643,
1730
+ "歌": 1644,
1731
+ "碰": 1645,
1732
+ "醜": 1646,
1733
+ "京": 1647,
1734
+ "咬": 1648,
1735
+ "砂": 1649,
1736
+ "霉": 1650,
1737
+ "脾": 1651,
1738
+ "惹": 1652,
1739
+ "翔": 1653,
1740
+ "誨": 1654,
1741
+ "總": 1655,
1742
+ "議": 1656,
1743
+ "乘": 1657,
1744
+ "掗": 1658,
1745
+ "鍛": 1659,
1746
+ "忍": 1660,
1747
+ "𨀤": 1661,
1748
+ "日": 1662,
1749
+ "靈": 1663,
1750
+ "弟": 1664,
1751
+ "浴": 1665,
1752
+ "氼": 1666,
1753
+ "考": 1667,
1754
+ "倔": 1668,
1755
+ "澀": 1669,
1756
+ "盜": 1670,
1757
+ "祢": 1671,
1758
+ "復": 1672,
1759
+ "層": 1673,
1760
+ "凰": 1674,
1761
+ "賜": 1675,
1762
+ "車": 1676,
1763
+ "辨": 1677,
1764
+ "是": 1678,
1765
+ "魅": 1679,
1766
+ "罔": 1680,
1767
+ "鎮": 1681,
1768
+ "嫲": 1682,
1769
+ "確": 1683,
1770
+ "斯": 1684,
1771
+ "禿": 1685,
1772
+ "孭": 1686,
1773
+ "誼": 1687,
1774
+ "古": 1688,
1775
+ "破": 1689,
1776
+ "仁": 1690,
1777
+ "劣": 1691,
1778
+ "喪": 1692,
1779
+ "便": 1693,
1780
+ "黎": 1694,
1781
+ "率": 1695,
1782
+ "序": 1696,
1783
+ "售": 1697,
1784
+ "普": 1698,
1785
+ "薑": 1699,
1786
+ "曉": 1700,
1787
+ "蛛": 1701,
1788
+ "久": 1702,
1789
+ "顛": 1703,
1790
+ "瑕": 1704,
1791
+ "蒐": 1705,
1792
+ "劊": 1706,
1793
+ "菜": 1707,
1794
+ "騙": 1708,
1795
+ "己": 1709,
1796
+ "寸": 1710,
1797
+ "晒": 1711,
1798
+ "倒": 1712,
1799
+ "滴": 1713,
1800
+ "喼": 1714,
1801
+ "扼": 1715,
1802
+ "緩": 1716,
1803
+ "製": 1717,
1804
+ "朗": 1718,
1805
+ "揈": 1719,
1806
+ "讓": 1720,
1807
+ "屋": 1721,
1808
+ "伐": 1722,
1809
+ "輛": 1723,
1810
+ "閲": 1724,
1811
+ "嗎": 1725,
1812
+ "洱": 1726,
1813
+ "檢": 1727,
1814
+ "四": 1728,
1815
+ "甘": 1729,
1816
+ "寡": 1730,
1817
+ "鬥": 1731,
1818
+ "紳": 1732,
1819
+ "z": 1733,
1820
+ "𠴕": 1734,
1821
+ "好": 1735,
1822
+ "徬": 1736,
1823
+ "屈": 1737,
1824
+ "侯": 1738,
1825
+ "桂": 1739,
1826
+ "轉": 1740,
1827
+ "睬": 1741,
1828
+ "p": 1742,
1829
+ "翁": 1743,
1830
+ "儒": 1744,
1831
+ "梵": 1745,
1832
+ "卿": 1746,
1833
+ "碟": 1747,
1834
+ "叔": 1748,
1835
+ "嚕": 1749,
1836
+ "竹": 1750,
1837
+ "番": 1751,
1838
+ "苓": 1752,
1839
+ "院": 1753,
1840
+ "泄": 1754,
1841
+ "晶": 1755,
1842
+ "按": 1756,
1843
+ "質": 1757,
1844
+ "座": 1758,
1845
+ "芒": 1759,
1846
+ "果": 1760,
1847
+ "希": 1761,
1848
+ "錫": 1762,
1849
+ "茶": 1763,
1850
+ "土": 1764,
1851
+ "醉": 1765,
1852
+ "物": 1766,
1853
+ "…": 1767,
1854
+ "虞": 1768,
1855
+ "囗": 1769,
1856
+ "贏": 1770,
1857
+ "軀": 1771,
1858
+ "莊": 1772,
1859
+ "鯊": 1773,
1860
+ "錶": 1774,
1861
+ "譜": 1775,
1862
+ "池": 1776,
1863
+ "糞": 1777,
1864
+ "崗": 1778,
1865
+ "挫": 1779,
1866
+ "X": 1780,
1867
+ "檔": 1781,
1868
+ "攪": 1782,
1869
+ "充": 1783,
1870
+ "碼": 1784,
1871
+ "喺": 1785,
1872
+ "綴": 1786,
1873
+ "領": 1787,
1874
+ "珍": 1788,
1875
+ "逐": 1789,
1876
+ "袍": 1790,
1877
+ "巾": 1791,
1878
+ "蘇": 1792,
1879
+ "夏": 1793,
1880
+ "綫": 1794,
1881
+ "几": 1795,
1882
+ "噏": 1796,
1883
+ "運": 1797,
1884
+ "樸": 1798,
1885
+ "繞": 1799,
1886
+ "搣": 1800,
1887
+ "T": 1801,
1888
+ "睞": 1802,
1889
+ "敍": 1803,
1890
+ "勁": 1804,
1891
+ "恥": 1805,
1892
+ "到": 1806,
1893
+ "告": 1807,
1894
+ "悉": 1808,
1895
+ "喘": 1809,
1896
+ "罨": 1810,
1897
+ "挈": 1811,
1898
+ "挨": 1812,
1899
+ "遢": 1813,
1900
+ "𢲡": 1814,
1901
+ "巴": 1815,
1902
+ "殊": 1816,
1903
+ "柚": 1817,
1904
+ "輩": 1818,
1905
+ "猛": 1819,
1906
+ "贊": 1820,
1907
+ "坪": 1821,
1908
+ "妻": 1822,
1909
+ "癦": 1823,
1910
+ "炒": 1824,
1911
+ "狼": 1825,
1912
+ "曱": 1826,
1913
+ "辰": 1827,
1914
+ "冇": 1828,
1915
+ "囡": 1829,
1916
+ "窟": 1830,
1917
+ "氈": 1831,
1918
+ "氛": 1832,
1919
+ "鋼": 1833,
1920
+ "活": 1834,
1921
+ "速": 1835,
1922
+ "哽": 1836,
1923
+ "木": 1837,
1924
+ "外": 1838,
1925
+ "靖": 1839,
1926
+ "助": 1840,
1927
+ "規": 1841,
1928
+ "旱": 1842,
1929
+ "豚": 1843,
1930
+ "吋": 1844,
1931
+ "轆": 1845,
1932
+ "臺": 1846,
1933
+ "墩": 1847,
1934
+ "頻": 1848,
1935
+ "樖": 1849,
1936
+ "飯": 1850,
1937
+ "砌": 1851,
1938
+ "點": 1852,
1939
+ "洗": 1853,
1940
+ "忳": 1854,
1941
+ "楷": 1855,
1942
+ "鄰": 1856,
1943
+ "矯": 1857,
1944
+ "想": 1858,
1945
+ "品": 1859,
1946
+ "馬": 1860,
1947
+ "早": 1861,
1948
+ "罪": 1862,
1949
+ "㗎": 1863,
1950
+ "酸": 1864,
1951
+ "呢": 1865,
1952
+ "坑": 1866,
1953
+ "侶": 1867,
1954
+ "團": 1868,
1955
+ "跡": 1869,
1956
+ "千": 1870,
1957
+ "錦": 1871,
1958
+ "嘎": 1872,
1959
+ "嚡": 1873,
1960
+ "覓": 1874,
1961
+ "山": 1875,
1962
+ "腹": 1876,
1963
+ "畸": 1877,
1964
+ "丟": 1878,
1965
+ "歲": 1879,
1966
+ "和": 1880,
1967
+ "泓": 1881,
1968
+ "慳": 1882,
1969
+ "園": 1883,
1970
+ "謠": 1884,
1971
+ "祖": 1885,
1972
+ "廠": 1886,
1973
+ "扱": 1887,
1974
+ "暑": 1888,
1975
+ "貌": 1889,
1976
+ "妄": 1890,
1977
+ "巫": 1891,
1978
+ "赤": 1892,
1979
+ "麥": 1893,
1980
+ "丹": 1894,
1981
+ "辦": 1895,
1982
+ "穎": 1896,
1983
+ "竊": 1897,
1984
+ "幾": 1898,
1985
+ "埞": 1899,
1986
+ "獎": 1900,
1987
+ "獅": 1901,
1988
+ "E": 1902,
1989
+ "侍": 1903,
1990
+ "橢": 1904,
1991
+ "濃": 1905,
1992
+ "夫": 1906,
1993
+ "憤": 1907,
1994
+ "闢": 1908,
1995
+ "尺": 1909,
1996
+ "胖": 1910,
1997
+ "吽": 1911,
1998
+ "洶": 1912,
1999
+ "織": 1913,
2000
+ "㔆": 1914,
2001
+ "試": 1915,
2002
+ "脆": 1916,
2003
+ "域": 1917,
2004
+ "案": 1918,
2005
+ "雄": 1919,
2006
+ "否": 1920,
2007
+ "滋": 1921,
2008
+ "席": 1922,
2009
+ "受": 1923,
2010
+ "旗": 1924,
2011
+ "吞": 1925,
2012
+ "徹": 1926,
2013
+ "仄": 1927,
2014
+ "謙": 1928,
2015
+ "以": 1929,
2016
+ "甚": 1930,
2017
+ "搽": 1931,
2018
+ "扁": 1932,
2019
+ "檬": 1933,
2020
+ "天": 1934,
2021
+ "測": 1935,
2022
+ "曹": 1936,
2023
+ "帖": 1937,
2024
+ "敢": 1938,
2025
+ "𠺝": 1939,
2026
+ "吧": 1940,
2027
+ "娶": 1941,
2028
+ "奚": 1942,
2029
+ "𨋢": 1943,
2030
+ "批": 1944,
2031
+ "將": 1945,
2032
+ "反": 1946,
2033
+ "娜": 1947,
2034
+ "膝": 1948,
2035
+ "披": 1949,
2036
+ "背": 1950,
2037
+ "薪": 1951,
2038
+ "迭": 1952,
2039
+ "葛": 1953,
2040
+ "姑": 1954,
2041
+ "款": 1955,
2042
+ "庫": 1956,
2043
+ "掅": 1957,
2044
+ "溫": 1958,
2045
+ "區": 1959,
2046
+ "南": 1960,
2047
+ "市": 1961,
2048
+ "枕": 1962,
2049
+ "恤": 1963,
2050
+ "#": 1964,
2051
+ "核": 1965,
2052
+ "K": 1966,
2053
+ "由": 1967,
2054
+ "被": 1968,
2055
+ "因": 1969,
2056
+ "坦": 1970,
2057
+ "們": 1971,
2058
+ "兜": 1972,
2059
+ "磋": 1973,
2060
+ "聖": 1974,
2061
+ "煩": 1975,
2062
+ "革": 1976,
2063
+ "授": 1977,
2064
+ "柒": 1978,
2065
+ "聯": 1979,
2066
+ "攏": 1980,
2067
+ "魂": 1981,
2068
+ "憎": 1982,
2069
+ "聆": 1983,
2070
+ "摘": 1984,
2071
+ "薦": 1985,
2072
+ "鈕": 1986,
2073
+ "色": 1987,
2074
+ "遵": 1988,
2075
+ "演": 1989,
2076
+ "待": 1990,
2077
+ "廉": 1991,
2078
+ "I": 1992,
2079
+ "掠": 1993,
2080
+ "兢": 1994,
2081
+ "夢": 1995,
2082
+ "𣲷": 1996,
2083
+ "潰": 1997,
2084
+ "味": 1998,
2085
+ "髖": 1999,
2086
+ "撼": 2000,
2087
+ "男": 2001,
2088
+ "寒": 2002,
2089
+ "豉": 2003,
2090
+ "折": 2004,
2091
+ "f": 2005,
2092
+ "吵": 2006,
2093
+ "數": 2007,
2094
+ "推": 2008,
2095
+ "毅": 2009,
2096
+ "帶": 2010,
2097
+ "街": 2011,
2098
+ "劖": 2012,
2099
+ "桃": 2013,
2100
+ "叉": 2014,
2101
+ "優": 2015,
2102
+ "智": 2016,
2103
+ "伴": 2017,
2104
+ "葬": 2018,
2105
+ "鼓": 2019,
2106
+ "刮": 2020,
2107
+ "傅": 2021,
2108
+ "諸": 2022,
2109
+ "疵": 2023,
2110
+ "忘": 2024,
2111
+ "爛": 2025,
2112
+ "隨": 2026,
2113
+ "低": 2027,
2114
+ "霖": 2028,
2115
+ "塔": 2029,
2116
+ "鍵": 2030,
2117
+ "瞰": 2031,
2118
+ "含": 2032,
2119
+ "刺": 2033,
2120
+ "臭": 2034,
2121
+ "鈎": 2035,
2122
+ "咒": 2036,
2123
+ "防": 2037,
2124
+ "瓊": 2038,
2125
+ "志": 2039,
2126
+ "女": 2040,
2127
+ "躬": 2041,
2128
+ "魷": 2042,
2129
+ "讚": 2043,
2130
+ "㷫": 2044,
2131
+ "冬": 2045,
2132
+ "俬": 2046,
2133
+ "筆": 2047,
2134
+ "潑": 2048,
2135
+ "慈": 2049,
2136
+ "平": 2050,
2137
+ "琵": 2051,
2138
+ "拼": 2052,
2139
+ "腺": 2053,
2140
+ "輕": 2054,
2141
+ "咀": 2055,
2142
+ "碇": 2056,
2143
+ "偉": 2057,
2144
+ "撻": 2058,
2145
+ "沱": 2059,
2146
+ "恒": 2060,
2147
+ "損": 2061,
2148
+ "付": 2062,
2149
+ "佑": 2063,
2150
+ "警": 2064,
2151
+ "闊": 2065,
2152
+ "首": 2066,
2153
+ "櫈": 2067,
2154
+ "偈": 2068,
2155
+ "腕": 2069,
2156
+ "姣": 2070,
2157
+ "箭": 2071,
2158
+ "蛹": 2072,
2159
+ "素": 2073,
2160
+ "襟": 2074,
2161
+ "絡": 2075,
2162
+ "誦": 2076,
2163
+ "計": 2077,
2164
+ "染": 2078,
2165
+ "擤": 2079,
2166
+ "步": 2080,
2167
+ "涕": 2081,
2168
+ "拾": 2082,
2169
+ "沐": 2083,
2170
+ "港": 2084,
2171
+ "盟": 2085,
2172
+ "乎": 2086,
2173
+ "產": 2087,
2174
+ "腸": 2088,
2175
+ "榮": 2089,
2176
+ "斬": 2090,
2177
+ "頓": 2091,
2178
+ "仿": 2092,
2179
+ "抌": 2093,
2180
+ "送": 2094,
2181
+ "賀": 2095,
2182
+ "諾": 2096,
2183
+ "簡": 2097,
2184
+ "註": 2098,
2185
+ "概": 2099,
2186
+ "揦": 2100,
2187
+ "碑": 2101,
2188
+ "漏": 2102,
2189
+ "霍": 2103,
2190
+ "即": 2104,
2191
+ "踩": 2105,
2192
+ "勝": 2106,
2193
+ "餓": 2107,
2194
+ "設": 2108,
2195
+ "抽": 2109,
2196
+ "脈": 2110,
2197
+ "証": 2111,
2198
+ "透": 2112,
2199
+ "次": 2113,
2200
+ "慘": 2114,
2201
+ "鑊": 2115,
2202
+ "友": 2116,
2203
+ "羅": 2117,
2204
+ "綠": 2118,
2205
+ "流": 2119,
2206
+ "聾": 2120,
2207
+ "豫": 2121,
2208
+ "艙": 2122,
2209
+ "襲": 2123,
2210
+ "鐵": 2124,
2211
+ "川": 2125,
2212
+ "役": 2126,
2213
+ "噃": 2127,
2214
+ "泌": 2128,
2215
+ "逼": 2129,
2216
+ "華": 2130,
2217
+ "偵": 2131,
2218
+ "誕": 2132,
2219
+ "菩": 2133,
2220
+ "傷": 2134,
2221
+ "雲": 2135,
2222
+ "踢": 2136,
2223
+ "凹": 2137,
2224
+ "環": 2138,
2225
+ "拔": 2139,
2226
+ "盪": 2140,
2227
+ "來": 2141,
2228
+ "﹐": 2142,
2229
+ "捅": 2143,
2230
+ "向": 2144,
2231
+ "衰": 2145,
2232
+ "晏": 2146,
2233
+ "閃": 2147,
2234
+ "械": 2148,
2235
+ "嗰": 2149,
2236
+ "媒": 2150,
2237
+ "著": 2151,
2238
+ "立": 2152,
2239
+ "澤": 2153,
2240
+ "催": 2154,
2241
+ "的": 2155,
2242
+ "獲": 2156,
2243
+ "呻": 2157,
2244
+ "埠": 2158,
2245
+ "棒": 2159,
2246
+ "毛": 2160,
2247
+ "、": 2161,
2248
+ "薩": 2162,
2249
+ "戒": 2163,
2250
+ "樓": 2164,
2251
+ "燃": 2165,
2252
+ "哣": 2166,
2253
+ "擁": 2167,
2254
+ "隆": 2168,
2255
+ "拘": 2169,
2256
+ "征": 2170,
2257
+ "骨": 2171,
2258
+ "神": 2172,
2259
+ "啊": 2173,
2260
+ "貞": 2174,
2261
+ "葵": 2175,
2262
+ "祈": 2176,
2263
+ "撞": 2177,
2264
+ "焗": 2178,
2265
+ "u": 2179,
2266
+ "辮": 2180,
2267
+ "瞓": 2181,
2268
+ "狐": 2182,
2269
+ "竟": 2183,
2270
+ "倍": 2184,
2271
+ "唸": 2185,
2272
+ "飄": 2186,
2273
+ "蘭": 2187,
2274
+ "撩": 2188,
2275
+ "𨈇": 2189,
2276
+ "秘": 2190,
2277
+ "例": 2191,
2278
+ "栽": 2192,
2279
+ "啝": 2193,
2280
+ "打": 2194,
2281
+ "午": 2195,
2282
+ "哈": 2196,
2283
+ "懈": 2197,
2284
+ "腮": 2198,
2285
+ "落": 2199,
2286
+ "籬": 2200,
2287
+ "鵝": 2201,
2288
+ "碩": 2202,
2289
+ "蝸": 2203,
2290
+ "倖": 2204,
2291
+ "育": 2205,
2292
+ "吹": 2206,
2293
+ "乜": 2207,
2294
+ "疲": 2208,
2295
+ "戴": 2209,
2296
+ "畫": 2210,
2297
+ "棟": 2211,
2298
+ "春": 2212,
2299
+ "盤": 2213,
2300
+ "禽": 2214,
2301
+ "冧": 2215,
2302
+ "淘": 2216,
2303
+ "矮": 2217,
2304
+ "米": 2218,
2305
+ "喇": 2219,
2306
+ "者": 2220,
2307
+ "壓": 2221,
2308
+ "翰": 2222,
2309
+ "褲": 2223,
2310
+ "閂": 2224,
2311
+ "葉": 2225,
2312
+ "横": 2226,
2313
+ "搪": 2227,
2314
+ "遞": 2228,
2315
+ "徒": 2229,
2316
+ "嗌": 2230,
2317
+ "版": 2231,
2318
+ "龐": 2232,
2319
+ "7": 2233,
2320
+ "辣": 2234,
2321
+ "亡": 2235,
2322
+ "簾": 2236,
2323
+ "熔": 2237,
2324
+ "虹": 2238,
2325
+ "兄": 2239,
2326
+ "彎": 2240,
2327
+ "租": 2241,
2328
+ "辜": 2242,
2329
+ "矇": 2243,
2330
+ "咖": 2244,
2331
+ "笑": 2245,
2332
+ "墮": 2246,
2333
+ "繼": 2247,
2334
+ "佣": 2248,
2335
+ "略": 2249,
2336
+ "j": 2250,
2337
+ "頌": 2251,
2338
+ "腑": 2252,
2339
+ "鞍": 2253,
2340
+ "艮": 2254,
2341
+ "或": 2255,
2342
+ "匪": 2256,
2343
+ "懵": 2257,
2344
+ "頒": 2258,
2345
+ "適": 2259,
2346
+ "郁": 2260,
2347
+ "閪": 2261,
2348
+ "獄": 2262,
2349
+ "脹": 2263,
2350
+ "6": 2264,
2351
+ "粉": 2265,
2352
+ "歪": 2266,
2353
+ "叮": 2267,
2354
+ "漫": 2268,
2355
+ "站": 2269,
2356
+ "椒": 2270,
2357
+ "揭": 2271,
2358
+ "協": 2272,
2359
+ "陶": 2273,
2360
+ "懇": 2274,
2361
+ "央": 2275,
2362
+ "癡": 2276,
2363
+ "徨": 2277,
2364
+ "耐": 2278,
2365
+ "S": 2279,
2366
+ "鉗": 2280,
2367
+ "沾": 2281,
2368
+ "鵡": 2282,
2369
+ "㖭": 2283,
2370
+ "邀": 2284,
2371
+ "撥": 2285,
2372
+ "存": 2286,
2373
+ "叫": 2287,
2374
+ "温": 2288,
2375
+ "塢": 2289,
2376
+ "隊": 2290,
2377
+ "閘": 2291,
2378
+ "弱": 2292,
2379
+ "挺": 2293,
2380
+ "乒": 2294,
2381
+ "始": 2295,
2382
+ "泛": 2296,
2383
+ "技": 2297,
2384
+ "麟": 2298,
2385
+ "約": 2299,
2386
+ "寢": 2300,
2387
+ "賊": 2301,
2388
+ "簽": 2302,
2389
+ "董": 2303,
2390
+ "仰": 2304,
2391
+ "𥄫": 2305,
2392
+ "絲": 2306,
2393
+ "氫": 2307,
2394
+ "督": 2308,
2395
+ "嚥": 2309,
2396
+ "伸": 2310,
2397
+ "𦧲": 2311,
2398
+ "持": 2312,
2399
+ "錄": 2313,
2400
+ "茄": 2314,
2401
+ "收": 2315,
2402
+ "觀": 2316,
2403
+ "第": 2317,
2404
+ "砸": 2318,
2405
+ "睹": 2319,
2406
+ "種": 2320,
2407
+ "勃": 2321,
2408
+ "獨": 2322,
2409
+ "枉": 2323,
2410
+ "描": 2324,
2411
+ "週": 2325,
2412
+ "棄": 2326,
2413
+ "矩": 2327,
2414
+ "訓": 2328,
2415
+ "朵": 2329,
2416
+ "琦": 2330,
2417
+ "僑": 2331,
2418
+ "厲": 2332,
2419
+ "9": 2333,
2420
+ "玩": 2334,
2421
+ "則": 2335,
2422
+ "箇": 2336,
2423
+ "陛": 2337,
2424
+ "豐": 2338,
2425
+ "庚": 2339,
2426
+ "會": 2340,
2427
+ "嚮": 2341,
2428
+ "傭": 2342,
2429
+ "缺": 2343,
2430
+ "海": 2344,
2431
+ "鹽": 2345,
2432
+ "憂": 2346,
2433
+ "～": 2347,
2434
+ "鄉": 2348,
2435
+ "州": 2349,
2436
+ "晉": 2350,
2437
+ "搜": 2351,
2438
+ "斤": 2352,
2439
+ "搗": 2353,
2440
+ "擴": 2354,
2441
+ "鴨": 2355,
2442
+ "手": 2356,
2443
+ "筷": 2357,
2444
+ "盎": 2358,
2445
+ "擗": 2359,
2446
+ "忤": 2360,
2447
+ "某": 2361,
2448
+ "蘋": 2362,
2449
+ "嫌": 2363,
2450
+ "駁": 2364,
2451
+ "熠": 2365,
2452
+ "擱": 2366,
2453
+ "w": 2367,
2454
+ "溦": 2368,
2455
+ "虔": 2369,
2456
+ "紮": 2370,
2457
+ "害": 2371,
2458
+ "嘔": 2372,
2459
+ "懷": 2373,
2460
+ "娩": 2374,
2461
+ "喫": 2375,
2462
+ "盼": 2376,
2463
+ "全": 2377,
2464
+ "再": 2378,
2465
+ "冤": 2379,
2466
+ "裙": 2380,
2467
+ "倉": 2381,
2468
+ "命": 2382,
2469
+ "邏": 2383,
2470
+ "降": 2384,
2471
+ "慾": 2385,
2472
+ "鳥": 2386,
2473
+ "煎": 2387,
2474
+ "穌": 2388,
2475
+ "養": 2389,
2476
+ "湊": 2390,
2477
+ "嬉": 2391,
2478
+ "搥": 2392,
2479
+ "捻": 2393,
2480
+ "賦": 2394,
2481
+ "緒": 2395,
2482
+ "亨": 2396,
2483
+ "騷": 2397,
2484
+ "讀": 2398,
2485
+ "昂": 2399,
2486
+ "串": 2400,
2487
+ "呆": 2401,
2488
+ "鬆": 2402,
2489
+ "喜": 2403,
2490
+ "珠": 2404,
2491
+ "骸": 2405,
2492
+ "鵲": 2406,
2493
+ "了": 2407,
2494
+ "烏": 2408,
2495
+ "力": 2409,
2496
+ "丼": 2410,
2497
+ "美": 2411,
2498
+ "糟": 2412,
2499
+ "桶": 2413,
2500
+ "在": 2414,
2501
+ "䟴": 2415,
2502
+ "出": 2416,
2503
+ "材": 2417,
2504
+ "乸": 2418,
2505
+ "泥": 2419,
2506
+ "齊": 2420,
2507
+ "該": 2421,
2508
+ "里": 2422,
2509
+ "俠": 2423,
2510
+ "輯": 2424,
2511
+ "湖": 2425,
2512
+ "癖": 2426,
2513
+ "棚": 2427,
2514
+ "柔": 2428,
2515
+ "洲": 2429,
2516
+ "慰": 2430,
2517
+ "卼": 2431,
2518
+ "婷": 2432,
2519
+ "閣": 2433,
2520
+ "經": 2434,
2521
+ "唾": 2435,
2522
+ "沿": 2436,
2523
+ "闔": 2437,
2524
+ "蒼": 2438,
2525
+ "減": 2439,
2526
+ "吸": 2440,
2527
+ "掘": 2441,
2528
+ "^": 2442,
2529
+ "突": 2443,
2530
+ "陌": 2444,
2531
+ "乖": 2445,
2532
+ "灘": 2446,
2533
+ "𢳂": 2447,
2534
+ "琶": 2448,
2535
+ "越": 2449,
2536
+ "詐": 2450,
2537
+ "佐": 2451,
2538
+ "課": 2452,
2539
+ "賺": 2453,
2540
+ "囂": 2454,
2541
+ "貿": 2455,
2542
+ "內": 2456,
2543
+ "踎": 2457,
2544
+ "U": 2458,
2545
+ "膠": 2459,
2546
+ "宰": 2460,
2547
+ "幕": 2461,
2548
+ "懸": 2462,
2549
+ "；": 2463,
2550
+ "校": 2464,
2551
+ "嗯": 2465,
2552
+ "唥": 2466,
2553
+ "料": 2467,
2554
+ "姆": 2468,
2555
+ "狡": 2469,
2556
+ "予": 2470,
2557
+ "眼": 2471,
2558
+ "癌": 2472,
2559
+ "奀": 2473,
2560
+ "唱": 2474,
2561
+ "蕩": 2475,
2562
+ "昌": 2476,
2563
+ "暴": 2477,
2564
+ "泉": 2478,
2565
+ "田": 2479,
2566
+ "彙": 2480,
2567
+ "慎": 2481,
2568
+ "燕": 2482,
2569
+ "晨": 2483,
2570
+ "戲": 2484,
2571
+ "凡": 2485,
2572
+ "繃": 2486,
2573
+ "霜": 2487,
2574
+ "喚": 2488,
2575
+ "熄": 2489,
2576
+ "鼠": 2490,
2577
+ "窗": 2491,
2578
+ "乾": 2492,
2579
+ "肇": 2493,
2580
+ "屍": 2494,
2581
+ "𢆡": 2495,
2582
+ "囈": 2496,
2583
+ "r": 2497,
2584
+ "七": 2498,
2585
+ "牲": 2499,
2586
+ "騮": 2500,
2587
+ "盃": 2501,
2588
+ "漸": 2502,
2589
+ "橋": 2503,
2590
+ "隱": 2504,
2591
+ "飲": 2505,
2592
+ "濁": 2506,
2593
+ "疾": 2507,
2594
+ "斲": 2508,
2595
+ "庇": 2509,
2596
+ "髀": 2510,
2597
+ "齋": 2511,
2598
+ "吱": 2512,
2599
+ "青": 2513,
2600
+ "婀": 2514,
2601
+ "乙": 2515,
2602
+ "額": 2516,
2603
+ "餉": 2517,
2604
+ "匡": 2518,
2605
+ "諒": 2519,
2606
+ "秀": 2520,
2607
+ "互": 2521,
2608
+ "哄": 2522,
2609
+ "煽": 2523,
2610
+ "雷": 2524,
2611
+ "嗲": 2525,
2612
+ "掟": 2526,
2613
+ "鐘": 2527,
2614
+ "抰": 2528,
2615
+ "託": 2529,
2616
+ "截": 2530,
2617
+ "咯": 2531,
2618
+ "翱": 2532,
2619
+ "《": 2533,
2620
+ "咭": 2534,
2621
+ "拓": 2535,
2622
+ "範": 2536,
2623
+ "膩": 2537,
2624
+ "磚": 2538,
2625
+ "霧": 2539,
2626
+ "益": 2540,
2627
+ "貨": 2541,
2628
+ "舞": 2542,
2629
+ "夠": 2543,
2630
+ "認": 2544,
2631
+ "'": 2545,
2632
+ "涌": 2546,
2633
+ "忽": 2547,
2634
+ "捐": 2548,
2635
+ "叻": 2549,
2636
+ "削": 2550,
2637
+ "捵": 2551,
2638
+ "暗": 2552,
2639
+ "愕": 2553,
2640
+ "揩": 2554,
2641
+ "灣": 2555,
2642
+ "術": 2556,
2643
+ "祉": 2557,
2644
+ "喊": 2558,
2645
+ "倦": 2559,
2646
+ "羔": 2560,
2647
+ "鏽": 2561,
2648
+ "涵": 2562,
2649
+ "億": 2563,
2650
+ "舟": 2564,
2651
+ "甥": 2565,
2652
+ "級": 2566,
2653
+ "候": 2567,
2654
+ "蜂": 2568,
2655
+ "焫": 2569,
2656
+ "連": 2570,
2657
+ "算": 2571,
2658
+ "雙": 2572,
2659
+ "囊": 2573,
2660
+ "摟": 2574,
2661
+ "樣": 2575,
2662
+ "緝": 2576,
2663
+ "餞": 2577,
2664
+ "劃": 2578,
2665
+ "嗚": 2579,
2666
+ "唳": 2580,
2667
+ "聽": 2581,
2668
+ "汽": 2582,
2669
+ "圖": 2583,
2670
+ "意": 2584,
2671
+ "處": 2585,
2672
+ "懾": 2586,
2673
+ "仝": 2587,
2674
+ "身": 2588,
2675
+ "茅": 2589,
2676
+ "輝": 2590,
2677
+ "汗": 2591,
2678
+ "劫": 2592,
2679
+ "性": 2593,
2680
+ "廢": 2594,
2681
+ "準": 2595,
2682
+ "塵": 2596,
2683
+ "陷": 2597,
2684
+ "祠": 2598,
2685
+ "附": 2599,
2686
+ "營": 2600,
2687
+ "迷": 2601,
2688
+ "淵": 2602,
2689
+ "澈": 2603,
2690
+ "炎": 2604,
2691
+ "瀆": 2605,
2692
+ "迪": 2606,
2693
+ "國": 2607,
2694
+ "孤": 2608,
2695
+ "論": 2609,
2696
+ "機": 2610,
2697
+ "叢": 2611,
2698
+ "褦": 2612,
2699
+ "籲": 2613,
2700
+ "召": 2614,
2701
+ "撫": 2615,
2702
+ "燈": 2616,
2703
+ "決": 2617,
2704
+ "怒": 2618,
2705
+ "循": 2619,
2706
+ "嚟": 2620,
2707
+ "瀉": 2621,
2708
+ "惟": 2622,
2709
+ "逆": 2623,
2710
+ "句": 2624,
2711
+ "儘": 2625,
2712
+ "鯉": 2626,
2713
+ "蕎": 2627,
2714
+ "礦": 2628,
2715
+ "盅": 2629,
2716
+ "謂": 2630,
2717
+ "堡": 2631,
2718
+ "圳": 2632,
2719
+ "亮": 2633,
2720
+ "選": 2634,
2721
+ "宵": 2635,
2722
+ "稟": 2636,
2723
+ "亭": 2637,
2724
+ "拳": 2638,
2725
+ "菲": 2639,
2726
+ "杞": 2640,
2727
+ "彷": 2641,
2728
+ "儀": 2642,
2729
+ "v": 2643,
2730
+ "税": 2644,
2731
+ "嘭": 2645,
2732
+ "然": 2646,
2733
+ "餐": 2647,
2734
+ "侵": 2648,
2735
+ "硬": 2649,
2736
+ "靚": 2650,
2737
+ "星": 2651,
2738
+ "什": 2652,
2739
+ "肌": 2653,
2740
+ "籌": 2654,
2741
+ "艇": 2655,
2742
+ "兵": 2656,
2743
+ "梯": 2657,
2744
+ "夥": 2658,
2745
+ "申": 2659,
2746
+ "限": 2660,
2747
+ "跟": 2661,
2748
+ "獻": 2662,
2749
+ "伶": 2663,
2750
+ "梨": 2664,
2751
+ "N": 2665,
2752
+ "贓": 2666,
2753
+ "李": 2667,
2754
+ "壁": 2668,
2755
+ "執": 2669,
2756
+ "近": 2670,
2757
+ "鎚": 2671,
2758
+ "炙": 2672,
2759
+ "寂": 2673,
2760
+ "醒": 2674,
2761
+ "釘": 2675,
2762
+ "乏": 2676,
2763
+ "異": 2677,
2764
+ "本": 2678,
2765
+ "脂": 2679,
2766
+ "奴": 2680,
2767
+ "捶": 2681,
2768
+ "狸": 2682,
2769
+ "e": 2683,
2770
+ "促": 2684,
2771
+ "真": 2685,
2772
+ "怕": 2686,
2773
+ "鞋": 2687,
2774
+ "主": 2688,
2775
+ "今": 2689,
2776
+ "佩": 2690,
2777
+ "漂": 2691,
2778
+ "峻": 2692,
2779
+ "燭": 2693,
2780
+ "癮": 2694,
2781
+ "氓": 2695,
2782
+ "長": 2696,
2783
+ "焦": 2697,
2784
+ "淪": 2698,
2785
+ "兼": 2699,
2786
+ "鍊": 2700,
2787
+ "唯": 2701,
2788
+ "竽": 2702,
2789
+ "虛": 2703,
2790
+ "趕": 2704,
2791
+ "趾": 2705,
2792
+ "江": 2706,
2793
+ "餵": 2707,
2794
+ "潮": 2708,
2795
+ "殿": 2709,
2796
+ "民": 2710,
2797
+ "鋤": 2711,
2798
+ "肚": 2712,
2799
+ "龜": 2713,
2800
+ "煙": 2714,
2801
+ "秉": 2715,
2802
+ "躍": 2716,
2803
+ "鮑": 2717,
2804
+ "仙": 2718,
2805
+ "扚": 2719,
2806
+ "替": 2720,
2807
+ "習": 2721,
2808
+ "鸚": 2722,
2809
+ "痞": 2723,
2810
+ "肥": 2724,
2811
+ "判": 2725,
2812
+ "林": 2726,
2813
+ "畜": 2727,
2814
+ "武": 2728,
2815
+ "痾": 2729,
2816
+ "村": 2730,
2817
+ "合": 2731,
2818
+ "粟": 2732,
2819
+ "紂": 2733,
2820
+ "拒": 2734,
2821
+ "溏": 2735,
2822
+ "撑": 2736,
2823
+ "謝": 2737,
2824
+ "剎": 2738,
2825
+ "焉": 2739,
2826
+ "摳": 2740,
2827
+ "扤": 2741,
2828
+ "季": 2742,
2829
+ "涯": 2743,
2830
+ "鮓": 2744,
2831
+ "敲": 2745,
2832
+ "袖": 2746,
2833
+ "狹": 2747,
2834
+ "迺": 2748,
2835
+ "倫": 2749,
2836
+ "噒": 2750,
2837
+ "𠵼": 2751,
2838
+ "紓": 2752,
2839
+ "兒": 2753,
2840
+ "煥": 2754,
2841
+ "政": 2755,
2842
+ "「": 2756,
2843
+ "狠": 2757,
2844
+ "礙": 2758,
2845
+ "眈": 2759,
2846
+ "慌": 2760,
2847
+ "莉": 2761,
2848
+ "涼": 2762,
2849
+ "蟬": 2763,
2850
+ "縣": 2764,
2851
+ "唆": 2765,
2852
+ "紅": 2766,
2853
+ "妹": 2767,
2854
+ "踐": 2768,
2855
+ "瑤": 2769,
2856
+ "敗": 2770,
2857
+ "腿": 2771,
2858
+ "帽": 2772,
2859
+ "粥": 2773,
2860
+ "蓄": 2774,
2861
+ "擅": 2775,
2862
+ "頑": 2776,
2863
+ "勛": 2777,
2864
+ "灌": 2778,
2865
+ "榃": 2779,
2866
+ "宋": 2780,
2867
+ "堪": 2781,
2868
+ "幅": 2782,
2869
+ "穩": 2783,
2870
+ "維": 2784,
2871
+ "從": 2785,
2872
+ "懂": 2786,
2873
+ "k": 2787,
2874
+ "沖": 2788,
2875
+ "鞭": 2789,
2876
+ "鎝": 2790,
2877
+ "岸": 2791,
2878
+ "拿": 2792,
2879
+ "酒": 2793,
2880
+ "射": 2794,
2881
+ "雞": 2795,
2882
+ "睡": 2796,
2883
+ "牡": 2797,
2884
+ "癢": 2798,
2885
+ "量": 2799,
2886
+ "籮": 2800,
2887
+ "節": 2801,
2888
+ "譯": 2802,
2889
+ "簷": 2803,
2890
+ "醖": 2804,
2891
+ "砭": 2805,
2892
+ "凱": 2806,
2893
+ "宇": 2807,
2894
+ "東": 2808,
2895
+ "擦": 2809,
2896
+ "比": 2810,
2897
+ "箱": 2811,
2898
+ "浸": 2812,
2899
+ "終": 2813,
2900
+ "懲": 2814,
2901
+ "窿": 2815,
2902
+ "剿": 2816,
2903
+ "拍": 2817,
2904
+ "小": 2818,
2905
+ "牽": 2819,
2906
+ "媽": 2820,
2907
+ "辛": 2821,
2908
+ "宜": 2822,
2909
+ "怯": 2823,
2910
+ "穴": 2824,
2911
+ "蠻": 2825,
2912
+ "闖": 2826,
2913
+ "琴": 2827,
2914
+ "溯": 2828,
2915
+ "，": 2829,
2916
+ "見": 2830,
2917
+ "莞": 2831,
2918
+ "赫": 2832,
2919
+ "詩": 2833,
2920
+ "宗": 2834,
2921
+ "哋": 2835,
2922
+ "筋": 2836,
2923
+ "轟": 2837,
2924
+ "&": 2838,
2925
+ "嚹": 2839,
2926
+ "證": 2840,
2927
+ "銀": 2841,
2928
+ "幸": 2842,
2929
+ "辱": 2843,
2930
+ "哥": 2844,
2931
+ "橙": 2845,
2932
+ "舊": 2846,
2933
+ "淋": 2847,
2934
+ "黑": 2848,
2935
+ "罰": 2849,
2936
+ "尋": 2850,
2937
+ "疑": 2851,
2938
+ "拜": 2852,
2939
+ "航": 2853,
2940
+ "吐": 2854,
2941
+ "熱": 2855,
2942
+ "哇": 2856,
2943
+ "咕": 2857,
2944
+ "船": 2858,
2945
+ "壅": 2859,
2946
+ "搓": 2860,
2947
+ "）": 2861,
2948
+ "凳": 2862,
2949
+ "置": 2863,
2950
+ "儲": 2864,
2951
+ "詞": 2865,
2952
+ "門": 2866,
2953
+ "培": 2867,
2954
+ "旁": 2868,
2955
+ "媳": 2869,
2956
+ "口": 2870,
2957
+ "嘟": 2871,
2958
+ "掛": 2872,
2959
+ "痴": 2873,
2960
+ "諗": 2874,
2961
+ "鯭": 2875,
2962
+ "廿": 2876,
2963
+ "閉": 2877,
2964
+ "瘍": 2878,
2965
+ "塘": 2879,
2966
+ "禦": 2880,
2967
+ "曾": 2881,
2968
+ "繫": 2882,
2969
+ "櫻": 2883,
2970
+ "咦": 2884,
2971
+ "榕": 2885,
2972
+ "棕": 2886,
2973
+ "可": 2887,
2974
+ "囖": 2888,
2975
+ "槓": 2889,
2976
+ "侖": 2890,
2977
+ "寄": 2891,
2978
+ "驚": 2892,
2979
+ "根": 2893,
2980
+ "姪": 2894,
2981
+ "頃": 2895,
2982
+ "晴": 2896,
2983
+ "姨": 2897,
2984
+ "努": 2898,
2985
+ "圾": 2899,
2986
+ "強": 2900,
2987
+ "寓": 2901,
2988
+ "挪": 2902,
2989
+ "戰": 2903,
2990
+ "貓": 2904,
2991
+ "塞": 2905,
2992
+ "跑": 2906,
2993
+ "漿": 2907,
2994
+ "孵": 2908,
2995
+ "肅": 2909,
2996
+ "履": 2910,
2997
+ "呔": 2911,
2998
+ "掙": 2912,
2999
+ "芬": 2913,
3000
+ "電": 2914,
3001
+ "病": 2915,
3002
+ "-": 2916,
3003
+ "楚": 2917,
3004
+ "纏": 2918,
3005
+ "途": 2919,
3006
+ "毋": 2920,
3007
+ "價": 2921,
3008
+ "人": 2922,
3009
+ "光": 2923,
3010
+ "每": 2924,
3011
+ "釋": 2925,
3012
+ "酷": 2926,
3013
+ "𠝹": 2927,
3014
+ "咇": 2928,
3015
+ "菢": 2929,
3016
+ "干": 2930,
3017
+ "炊": 2931,
3018
+ "説": 2932,
3019
+ "篩": 2933,
3020
+ "地": 2934,
3021
+ "苗": 2935,
3022
+ "𢳆": 2936,
3023
+ "糾": 2937,
3024
+ "薄": 2938,
3025
+ "年": 2939,
3026
+ "滿": 2940,
3027
+ "報": 2941,
3028
+ "杯": 2942,
3029
+ "揗": 2943,
3030
+ "淥": 2944,
3031
+ "常": 2945,
3032
+ "脊": 2946,
3033
+ "⋯": 2947,
3034
+ "唪": 2948,
3035
+ "化": 2949,
3036
+ "弦": 2950,
3037
+ "瓦": 2951,
3038
+ "漁": 2952,
3039
+ "餃": 2953,
3040
+ "堂": 2954,
3041
+ "寧": 2955,
3042
+ "樁": 2956,
3043
+ "純": 2957,
3044
+ "輸": 2958,
3045
+ "章": 2959,
3046
+ "潛": 2960,
3047
+ "寵": 2961,
3048
+ "磁": 2962,
3049
+ "哮": 2963,
3050
+ "柳": 2964,
3051
+ "歧": 2965,
3052
+ "宣": 2966,
3053
+ "卒": 2967,
3054
+ "籠": 2968,
3055
+ "紐": 2969,
3056
+ "擋": 2970,
3057
+ "眨": 2971,
3058
+ "耳": 2972,
3059
+ "複": 2973,
3060
+ "宴": 2974,
3061
+ "@": 2975,
3062
+ "室": 2976,
3063
+ "咪": 2977,
3064
+ "啩": 2978,
3065
+ "韻": 2979,
3066
+ "締": 2980,
3067
+ "朔": 2981,
3068
+ "舉": 2982,
3069
+ "繳": 2983,
3070
+ "闆": 2984,
3071
+ "卷": 2985,
3072
+ "仗": 2986,
3073
+ "游": 2987,
3074
+ "幽": 2988,
3075
+ "件": 2989,
3076
+ "舖": 2990,
3077
+ "渺": 2991,
3078
+ "其": 2992,
3079
+ "滾": 2993,
3080
+ "竭": 2994,
3081
+ "滓": 2995,
3082
+ "炸": 2996,
3083
+ "籍": 2997,
3084
+ "甴": 2998,
3085
+ "去": 2999,
3086
+ "挖": 3000,
3087
+ "嘛": 3001,
3088
+ "淨": 3002,
3089
+ "溶": 3003,
3090
+ "棺": 3004,
3091
+ "吮": 3005,
3092
+ "虎": 3006,
3093
+ "枱": 3007,
3094
+ "雌": 3008,
3095
+ "三": 3009,
3096
+ "酗": 3010,
3097
+ "抑": 3011,
3098
+ "線": 3012,
3099
+ "緊": 3013,
3100
+ "患": 3014,
3101
+ "攬": 3015,
3102
+ "/": 3016,
3103
+ "廁": 3017,
3104
+ "郊": 3018,
3105
+ "汰": 3019,
3106
+ "淒": 3020,
3107
+ "謎": 3021,
3108
+ "癲": 3022,
3109
+ "唂": 3023,
3110
+ "佔": 3024,
3111
+ "萌": 3025,
3112
+ "扭": 3026,
3113
+ "擒": 3027,
3114
+ "渣": 3028,
3115
+ "𢃇": 3029,
3116
+ "彳": 3030,
3117
+ "趷": 3031,
3118
+ "烈": 3032,
3119
+ "徵": 3033,
3120
+ "均": 3034,
3121
+ "秩": 3035,
3122
+ "奸": 3036,
3123
+ "跣": 3037,
3124
+ "生": 3038,
3125
+ "哼": 3039,
3126
+ "記": 3040,
3127
+ "唇": 3041,
3128
+ "派": 3042,
3129
+ "默": 3043,
3130
+ "備": 3044,
3131
+ "曲": 3045,
3132
+ "鈍": 3046,
3133
+ "嘯": 3047,
3134
+ "摩": 3048,
3135
+ "籤": 3049,
3136
+ "廚": 3050,
3137
+ "沉": 3051,
3138
+ "未": 3052,
3139
+ "看": 3053,
3140
+ "抱": 3054,
3141
+ "糯": 3055,
3142
+ "勞": 3056,
3143
+ "驍": 3057,
3144
+ "庶": 3058,
3145
+ ".": 3059,
3146
+ "必": 3060,
3147
+ "糍": 3061,
3148
+ "館": 3062,
3149
+ "正": 3063,
3150
+ "寬": 3064,
3151
+ "捱": 3065,
3152
+ "影": 3066,
3153
+ "塑": 3067,
3154
+ "誌": 3068,
3155
+ "史": 3069,
3156
+ "驗": 3070,
3157
+ "鳩": 3071,
3158
+ "德": 3072,
3159
+ "劇": 3073,
3160
+ "慨": 3074,
3161
+ "肪": 3075,
3162
+ "姿": 3076,
3163
+ "禍": 3077,
3164
+ "毒": 3078,
3165
+ "烹": 3079,
3166
+ "恕": 3080,
3167
+ "示": 3081,
3168
+ "調": 3082,
3169
+ "歎": 3083,
3170
+ "恩": 3084,
3171
+ "輾": 3085,
3172
+ "斥": 3086,
3173
+ "貢": 3087,
3174
+ "且": 3088,
3175
+ "呃": 3089,
3176
+ "稍": 3090,
3177
+ "吔": 3091,
3178
+ "花": 3092,
3179
+ "趙": 3093,
3180
+ "享": 3094,
3181
+ "傍": 3095,
3182
+ "另": 3096,
3183
+ "滄": 3097,
3184
+ "慕": 3098,
3185
+ "講": 3099,
3186
+ "艦": 3100,
3187
+ "津": 3101,
3188
+ "曦": 3102,
3189
+ "住": 3103,
3190
+ "趴": 3104,
3191
+ "護": 3105,
3192
+ "佳": 3106,
3193
+ "禧": 3107,
3194
+ "揾": 3108,
3195
+ "帛": 3109,
3196
+ "管": 3110,
3197
+ "些": 3111,
3198
+ "蠱": 3112,
3199
+ "鎅": 3113,
3200
+ "僅": 3114,
3201
+ "抆": 3115,
3202
+ "個": 3116,
3203
+ "變": 3117,
3204
+ "吝": 3118,
3205
+ "壯": 3119,
3206
+ "傀": 3120,
3207
+ "嚱": 3121,
3208
+ "雨": 3122,
3209
+ "究": 3123,
3210
+ "牆": 3124,
3211
+ "貝": 3125,
3212
+ "痛": 3126,
3213
+ "扒": 3127,
3214
+ "羨": 3128,
3215
+ "雖": 3129,
3216
+ "浪": 3130,
3217
+ "遜": 3131,
3218
+ "顏": 3132,
3219
+ "棵": 3133,
3220
+ "衡": 3134,
3221
+ "祝": 3135,
3222
+ "褸": 3136,
3223
+ "接": 3137,
3224
+ "燶": 3138,
3225
+ "段": 3139,
3226
+ "窮": 3140,
3227
+ "感": 3141,
3228
+ "匙": 3142,
3229
+ "峙": 3143,
3230
+ "香": 3144,
3231
+ "圈": 3145,
3232
+ "宮": 3146,
3233
+ "致": 3147,
3234
+ "排": 3148,
3235
+ "話": 3149,
3236
+ "邋": 3150,
3237
+ "屁": 3151,
3238
+ "蔴": 3152,
3239
+ "D": 3153,
3240
+ "M": 3154,
3241
+ "加": 3155,
3242
+ "緻": 3156,
3243
+ "軟": 3157,
3244
+ "滅": 3158,
3245
+ "景": 3159,
3246
+ "捷": 3160,
3247
+ "挹": 3161,
3248
+ "散": 3162,
3249
+ "凝": 3163,
3250
+ "耍": 3164,
3251
+ "找": 3165,
3252
+ "列": 3166,
3253
+ "衣": 3167,
3254
+ "陳": 3168,
3255
+ "沙": 3169,
3256
+ "傑": 3170,
3257
+ "直": 3171,
3258
+ "窕": 3172,
3259
+ "語": 3173,
3260
+ "辟": 3174,
3261
+ "裔": 3175,
3262
+ "袋": 3176,
3263
+ "北": 3177,
3264
+ "踭": 3178,
3265
+ "才": 3179,
3266
+ "集": 3180,
3267
+ "展": 3181,
3268
+ "河": 3182,
3269
+ "𠿪": 3183,
3270
+ "齡": 3184,
3271
+ "蕭": 3185,
3272
+ "擾": 3186,
3273
+ "惜": 3187,
3274
+ "d": 3188,
3275
+ "竇": 3189,
3276
+ "彤": 3190,
3277
+ "源": 3191,
3278
+ "盡": 3192,
3279
+ "檯": 3193,
3280
+ "留": 3194,
3281
+ "豔": 3195,
3282
+ "目": 3196,
3283
+ "斟": 3197,
3284
+ "烊": 3198,
3285
+ "鉸": 3199,
3286
+ "冚": 3200,
3287
+ "帝": 3201,
3288
+ "私": 3202,
3289
+ "揮": 3203,
3290
+ "𠻹": 3204,
3291
+ "聞": 3205,
3292
+ "原": 3206,
3293
+ "户": 3207,
3294
+ "嗱": 3208,
3295
+ "祥": 3209,
3296
+ "後": 3210,
3297
+ "膽": 3211,
3298
+ "嘖": 3212,
3299
+ "莫": 3213,
3300
+ "撕": 3214,
3301
+ "L": 3215,
3302
+ "蔭": 3216,
3303
+ "張": 3217,
3304
+ "吠": 3218,
3305
+ "距": 3219,
3306
+ "沈": 3220,
3307
+ "吼": 3221,
3308
+ "哀": 3222,
3309
+ "蜜": 3223,
3310
+ "若": 3224,
3311
+ "尚": 3225,
3312
+ "1": 3226,
3313
+ "紗": 3227,
3314
+ "邪": 3228,
3315
+ "斂": 3229,
3316
+ "瑰": 3230,
3317
+ "淚": 3231,
3318
+ "房": 3232,
3319
+ "紹": 3233,
3320
+ "瑜": 3234,
3321
+ "叛": 3235,
3322
+ "毫": 3236,
3323
+ "婦": 3237,
3324
+ "脛": 3238,
3325
+ "寰": 3239,
3326
+ "娛": 3240,
3327
+ "克": 3241,
3328
+ "剖": 3242,
3329
+ "遺": 3243,
3330
+ "m": 3244,
3331
+ "衫": 3245,
3332
+ "亍": 3246,
3333
+ "保": 3247,
3334
+ "迫": 3248,
3335
+ "剋": 3249,
3336
+ "翌": 3250,
3337
+ "家": 3251,
3338
+ "除": 3252,
3339
+ "拯": 3253,
3340
+ "恰": 3254,
3341
+ "返": 3255,
3342
+ "酥": 3256,
3343
+ "薇": 3257,
3344
+ "糖": 3258,
3345
+ ")": 3259,
3346
+ "導": 3260,
3347
+ "疇": 3261,
3348
+ "奶": 3262,
3349
+ "冒": 3263,
3350
+ "阱": 3264,
3351
+ "敏": 3265,
3352
+ "石": 3266,
3353
+ "憾": 3267,
3354
+ "軍": 3268,
3355
+ "暈": 3269,
3356
+ "差": 3270,
3357
+ "幼": 3271,
3358
+ "你": 3272,
3359
+ "啪": 3273,
3360
+ "慢": 3274,
3361
+ "頭": 3275,
3362
+ "摸": 3276,
3363
+ "程": 3277,
3364
+ "周": 3278,
3365
+ "九": 3279,
3366
+ "度": 3280,
3367
+ "止": 3281,
3368
+ "換": 3282,
3369
+ "纊": 3283,
3370
+ "韞": 3284,
3371
+ "壇": 3285,
3372
+ "舫": 3286,
3373
+ "B": 3287,
3374
+ "退": 3288,
3375
+ "韌": 3289,
3376
+ "逝": 3290,
3377
+ "元": 3291,
3378
+ "僻": 3292,
3379
+ "激": 3293,
3380
+ "姊": 3294,
3381
+ "佬": 3295,
3382
+ "鴿": 3296,
3383
+ "黏": 3297,
3384
+ "碎": 3298,
3385
+ "販": 3299,
3386
+ "羞": 3300,
3387
+ "侮": 3301,
3388
+ "請": 3302,
3389
+ "飢": 3303,
3390
+ "旦": 3304,
3391
+ "咩": 3305,
3392
+ "蠶": 3306,
3393
+ "W": 3307,
3394
+ "耀": 3308,
3395
+ "榨": 3309,
3396
+ "而": 3310,
3397
+ "抵": 3311,
3398
+ "襪": 3312,
3399
+ "*": 3313,
3400
+ "脱": 3314,
3401
+ "偽": 3315,
3402
+ "羈": 3316,
3403
+ "愚": 3317,
3404
+ "暖": 3318
3405
+ },
3406
+ "merges": []
3407
+ }
3408
+ }
checkpoint/{trocr-custdata → trocr-custdata-8000}/last/tokenizer_config.json RENAMED
File without changes
checkpoint/trocr-custdata-8000/last/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f0d357a4f97e7e0f70e94c6c65a78ce92aecb2097c32fbd6fd785d8556c4412
3
+ size 3579
checkpoint/trocr-custdata-8000/last/vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<s>":0,"<pad>":1,"</s>":2,"<unk>":3,"<mask>":4,"奏":5,"音":6,"照":7,"把":8,"不":9,"尹":10,"伙":11,"蜘":12,"僭":13,"使":14,"啡":15,"柄":16,"羹":17,"嘩":18,"龍":19,"歷":20,"藍":21,"唞":22,"猶":23,"嘉":24,"衝":25,"寨":26,"炮":27,"銜":28,"喬":29,"睛":30,"庸":31,"耶":32,"伺":33,"片":34,"艘":35,"俐":36,"恨":37,"索":38,"哩":39,"哨":40,"銅":41,"璃":42,"貴":43,"斷":44,"達":45,"癩":46,"斜":47,"駛":48,"煲":49,"殘":50,"翻":51,"墨":52,"與":53,"更":54,"多":55,"博":56,"巨":57,"啓":58,"號":59,"扣":60,"等":61,"於":62,"勉":63,"脷":64,"盒":65,"扶":66,"濫":67,"式":68,"球":69,"嚿":70,"靴":71,"柯":72,"嘢":73,"后":74,"楊":75,"典":76,"燙":77,"鋪":78,"瀨":79,"肆":80,"赴":81,"牧":82,"紋":83,"況":84,"覆":85,"蜀":86,"省":87,"罩":88,"纜":89,"時":90,"擪":91,"騰":92,"腐":93,"配":94,"頗":95,"吉":96,"歐":97,"8":98,"卸":99,"醃":100,"鏡":101,"佢":102,"坊":103,"搖":104,"城":105,"戶":106,"基":107,"敷":108,"已":109,"對":110,"幻":111,"旅":112,"俘":113,"健":114,"悔":115,"飼":116,"宅":117,"燉":118,"櫃":119,"扳":120,"3":121,"h":122,"扮":123,"啋":124,"裁":125,"攀":126,"鵑":127,"有":128,"喻":129,"忟":130,"整":131,"賠":132,"守":133,"扑":134,"瑞":135,"善":136,"o":137,"假":138,"短":139,"念":140,"濛":141,"問":142,"盈":143,"擸":144,"旋":145,"漾":146,"蝦":147,"荒":148,"沼":149,"樽":150,"迴":151,"甕":152,"又":153,"嶺":154,"播":155,"下":156,"誰":157,"簿":158,"恙":159,"蛋":160,"V":161,"陣":162,"縫":163,"忌":164,"鬚":165,"字":166,"搬":167,"蚊":168,"方":169,"昧":170,"彈":171,"梗":172,"捽":173,"遇":174,"攻":175,"徑":176,"特":177,"末":178,"水":179,"t":180,"府":181,"鑽":182,"組":183,"恆":184,"喀":185,"肯":186,"輻":187,"蕾":188,"蝶":189,"拎":190,"藥":191,"址":192,"椅":193,"既":194,"蒸":195,"哂":196,"薯":197,"攣":198,"譽":199,"失":200,"淡":201,"六":202,"陰":203,"丁":204,"逗":205,"玻":206,"宙":207,"晚":208,"擇":209,"胭":210,"鍋":211,"剝":212,"遍":213,"鋭":214,"鮮":215,"伍":216,"漆":217,"停":218,"靜":219,"譚":220,"慧":221,"際":222,"症":223,"商":224,"各":225,"囚":226,"騎":227,"壩":228,"蹊":229,"偷":230,"漲":231,"恢":232,"嫁":233,"閏":234,"懶":235,"遠":236,"掌":237,"廳":238,"措":239,"老":240,"寺":241,"亂":242,"醇":243,"至":244,"綱":245,"欖":246,"屙":247,"睥":248,"承":249,"喂":250,"屢":251,"綽":252,"鼎":253,"虐":254,"煞":255,"玉":256,"瀏":257,"揪":258,"拃":259,"奪":260,"祭":261,"司":262,"刀":263,"縱":264,"形":265,"像":266,"哲":267,"相":268,"控":269,"休":270,"碌":271,"遴":272,"診":273,"泊":274,"牌":275,"最":276,"穫":277,"屬":278,"皆":279,"爭":280,"繁":281,"冰":282,"秒":283,"飾":284,"沒":285,"艱":286,"皈":287,"賭":288,"喎":289,"腔":290,"憶":291,"巷":292,"綁":293,"薰":294,"抓":295,"颱":296,"焚":297,"明":298,"嚐":299,"操":300,"油":301,"巡":302,"囑":303,"悠":304,"罐":305,"稿":306,"條":307,"敵":308,"險":309,"埋":310,"佛":311,"瀾":312,"膚":313,"閒":314,"趺":315,"何":316,"茸":317,"超":318,"亞":319,"鏹":320,"爸":321,"納":322,"甲":323,"耽":324,"危":325,"鏈":326,"針":327,"賞":328,"堤":329,"創":330,"澳":331,"飪":332,"押":333,"蕉":334,"心":335,"鑣":336,"耷":337,"西":338,"吖":339,"耕":340,"噬":341,"睇":342,"q":343,"法":344,"嘅":345,"隻":346,"禮":347,"我":348,"改":349,"啱":350,"此":351,"娘":352,"曳":353,"深":354,"寥":355,"奧":356,"酌":357,"瞞":358,"澄":359,"伏":360,"宿":361,"代":362,"葱":363,"唐":364,"濟":365,"刨":366,"追":367,"怪":368,"攝":369,"稜":370,"搭":371,"網":372,"遣":373,"沊":374,"淆":375,"迅":376,"螺":377,"憐":378,"符":379,"吓":380,"蜆":381,"O":382,"得":383,"勒":384,"俱":385,"無":386,"搵":387,"嘥":388,"馴":389,"鬼":390,"功":391,"願":392,"錢":393,"修":394,"胞":395,"趨":396,"虧":397,"邦":398,"杜":399,"理":400,"起":401,"路":402,"扇":403,"月":404,"釀":405,"b":406,"塊":407,"着":408,"隧":409,"篤":410,"螢":411,"冷":412,"瘤":413,"掩":414,"升":415,"蝕":416,"群":417,"義":418,"勾":419,"投":420,"粹":421,"泰":422,"食":423,"磅":424,"野":425,"撚":426,"殺":427,"屎":428,"爪":429,"啫":430,"永":431,"汀":432,"紥":433,"安":434,"梁":435,"還":436,"積":437,"佻":438,"伊":439,"滑":440,"劉":441,"歇":442,"坐":443,"板":444,"棋":445,"誘":446,"壺":447,"編":448,"似":449,"悶":450,"鹹":451,"罕":452,"難":453,"蠢":454,"郵":455,"唉":456,"狂":457,"羊":458,"上":459,"入":460,"樑":461,"學":462,"模":463,"耘":464,"潺":465,"練":466,"驅":467,"吾":468,"凍":469,"咿":470,"抬":471,"卦":472,"攜":473,"噪":474,"偏":475,"蠟":476,"麼":477,"霎":478,"誓":479,"故":480,"姓":481,"欠":482,"令":483,"梅":484,"」":485,"窄":486,"布":487,"豁":488,"妖":489,"喝":490,"項":491,"嘗":492,"職":493,"卡":494,"朦":495,"俾":496,"拉":497,"指":498,"釣":499,"套":500,"悲":501,"仃":502,"旺":503,"恭":504,"拖":505,"痠":506,"遙":507,"象":508,"界":509,"珊":510,"瘦":511,"裨":512,"別":513,"衩":514,"皮":515,"師":516,"牙":517,"揚":518,"寶":519,"欄":520,"刪":521,"溝":522,"重":523,"泡":524,"賓":525,"樹":526,"稽":527,"胡":528,"芥":529,"戀":530,"桌":531,"翼":532,"崩":533,"泳":534,"昇":535,"勵":536,"枝":537,"J":538,"績":539,"胎":540,"蔔":541,"思":542,"凶":543,"搏":544,"孕":545,"甫":546,"嚴":547,"2":548,"貶":549,"圓":550,"察":551,"血":552,"賽":553,"聲":554,"0":555,"婉":556,"濕":557,"裂":558,"!":559,"琢":560,"參":561,"析":562,"黨":563,"森":564,"惡":565,"委":566,"捉":567,"僧":568,"飛":569,"愛":570,"儉":571,"仲":572,"憲":573,"喳":574,"貪":575,"挽":576,"爬":577,"跨":578,"%":579,"盛":580,"鯨":581,"咗":582,"禾":583,"童":584,"瘡":585,"蘿":586,"麒":587,"攞":588,"膜":589,"軭":590,"𧕴":591,"插":592,"妃":593,"進":594,"呈":595,"拇":596,"標":597,"唔":598,"隍":599,"擘":600,"扎":601,"障":602,"捨":603,"靠":604,"刻":605,"磷":606,"麻":607,"腫":608,"玲":609,"梘":610,"蹤":611,"蒙":612,"杰":613,"蝴":614,"喱":615,"𠸎":616,"查":617,"預":618,"熟":619,"足":620,"瓜":621,"忙":622,"臉":623,"咄":624,"圍":625,"僱":626,"爽":627,"廹":628,"移":629,"炬":630,"息":631,"坡":632,"爐":633,"場":634,"臣":635,"丫":636,"勢":637,"覺":638,"威":639,"妳":640,"左":641,"牛":642,"寞":643,"𠵇":644,"腳":645,"冊":646,"塌":647,"潭":648,"研":649,"拐":650,"糕":651,"迎":652,"翹":653,"笨":654,"蔗":655,"穢":656,"震":657,"畏":658,"婚":659,"良":660,"惠":661,"遁":662,"臌":663,"姐":664,"䂻":665,"悦":666,"頁":667,"𡁻":668,"蟹":669,"菌":670,"敬":671,"茂":672,"邊":673,"凌":674,"關":675,"英":676,"曬":677,"啦":678,"餅":679,"審":680,"需":681,"歡":682,"負":683,"鳳":684,"枳":685,"芭":686,"鬧":687,"冠":688,"謀":689,"零":690,"衞":691,"悟":692,"狗":693,"細":694,"舒":695,"招":696,"罹":697,"擠":698,"蘑":699,"秦":700,"樂":701,"腦":702,"響":703,"蟲":704,"一":705,"徊":706,"膉":707,"供":708,"l":709,"戥":710,"詳":711,"遮":712,"蒜":713,"硯":714,"尖":715,"挑":716,"企":717,"屌":718,"公":719,"完":720,"吟":721,"觸":722,"棉":723,"債":724,"文":725,"爆":726,"毀":727,"跪":728,"榴":729,"麖":730,"藐":731,"趣":732,"百":733,"拗":734,"蹈":735,"餸":736,"鐳":737,"菇":738,"權":739,"極":740,"詭":741,"八":742,"夾":743,"佈":744,"契":745,"攊":746,"朝":747,"漠":748,"彌":749,"匿":750,"葺":751,"眛":752,"應":753,"豬":754,"植":755,"両":756,"渧":757,"覽":758,"聊":759,"咁":760,"採":761,"尼":762,"謬":763,"農":764,"彼":765,"買":766,"皇":767,"欲":768,"污":769,"魚":770,"慷":771,"顯":772,"瓶":773,"白":774,"乍":775,"王":776," ":777,"稱":778,"孫":779,"滲":780,"面":781,"辯":782,"秤":783,"腩":784,"桔":785,"財":786,"毆":787,"遭":788,"掉":789,"Y":790,"氧":791,"垢":792,"黃":793,"傘":794,"裇":795,"倪":796,"嘍":797,"較":798,"弊":799,"世":800,"篇":801,"萃":802,"鬟":803,"櫳":804,"嬰":805,"懼":806,"噴":807,"噶":808,"貫":809,"波":810,"奢":811,"救":812,"謊":813,"墜":814,"疊":815,"眾":816,"據":817,"效":818,"容":819,"掂":820,"欺":821,"凸":822,"靡":823,"鑑":824,"妨":825,"二":826,"跳":827,"腍":828,"瓣":829,"割":830,"奉":831,"新":832,"丑":833,"椗":834,"贖":835,"頇":836,"厘":837,"碳":838,"阻":839,"郎":840,"誡":841,"行":842,"4":843,"畀":844,"絕":845,"咳":846,"i":847,"戈":848,"簧":849,"卵":850,"書":851,"蹲":852,"訂":853,"輔":854,"箍":855,"惕":856,"尾":857,"鬙":858,"荷":859,"臘":860,"肝":861,"要":862,"俏":863,"雪":864,"擲":865,"漪":866,"違":867,"苛":868,"豆":869,"道":870,"鑿":871,"教":872,"愁":873,"局":874,"服":875,"苦":876,"蓮":877,"瞌":878,"銷":879,"祇":880,"但":881,"作":882,"蛙":883,"擊":884,"掣":885,"軸":886,"噚":887,"脅":888,"刷":889,"期":890,"紙":891,"解":892,"剩":893,"賤":894,"印":895,"急":896,"湃":897,"賬":898,"構":899,"器":900,"他":901,"勸":902,"票":903,"襯":904,"先":905,"屏":906,"烘":907,"涉":908,"粵":909,"胥":910,"堆":911,"摧":912,"母":913,"攤":914,"仆":915,"贈":916,"怖":917,"共":918,"乓":919,"逾":920,"半":921,"a":922,"草":923,"述":924,"肩":925,"登":926,"頹":927,"俄":928,"惶":929,"晝":930,"抗":931,"侈":932,"填":933,"譬":934,"鋒":935,"搶":936,"實":937,"律":938,"繹":939,"碧":940,"添":941,"交":942,"望":943,"㬹":944,"榜":945,"暢":946,"框":947,"胸":948,"囉":949,"親":950,"潔":951,"信":952,"韓":953,"知":954,"𠺘":955,"初":956,"掁":957,"露":958,"官":959,"堅":960,"許":961,"暉":962,"揀":963,"氹":964,"單":965,"迦":966,"谷":967,"褪":968,"撒":969,"廣":970,"輪":971,"結":972,"渦":973,"邨":974,"高":975,"井":976,"奔":977,"魏":978,"腰":979,"。":980,"芽":981,"軚":982,"F":983,"密":984,"雹":985,"鐸":986,"齣":987,"牢":988,"通":989,"麵":990,"鴉":991,"眷":992,"廟":993,"揞":994,"墓":995,"氏":996,"瀕":997,"擬":998,"鷹":999,"談":1000,"蓋":1001,"匯":1002,"言":1003,"塗":1004,"藝":1005,"壞":1006,"肺":1007,"動":1008,"萬":1009,"娃":1010,"底":1011,"灰":1012,"夕":1013,"姦":1014,"丸":1015,"G":1016,"刁":1017,"奄":1018,"槤":1019,"自":1020,"殃":1021,"消":1022,"啹":1023,"夜":1024,"態":1025,"扯":1026,"滂":1027,"垂":1028,"給":1029,"勿":1030,"右":1031,"分":1032,"���":1033,"能":1034,"稚":1035,"業":1036,"喉":1037,"制":1038,"摺":1039,"糧":1040,"勤":1041,"潤":1042,"絨":1043,"成":1044,"撐":1045,"臍":1046,"棍":1047,"嘆":1048,"訪":1049,"做":1050,"裝":1051,"傢":1052,"窒":1053,"丶":1054,"累":1055,"惑":1056,"悼":1057,"遂":1058,"擺":1059,"跋":1060,"築":1061,"劈":1062,"誤":1063,"湧":1064,"御":1065,"廈":1066,"淺":1067,"淫":1068,"A":1069,"揼":1070,"陪":1071,"匹":1072,"呎":1073,"部":1074,"中":1075,"固":1076,"奮":1077,"糊":1078,"稀":1079,"逕":1080,"詢":1081,"洋":1082,"枚":1083,"檸":1084,"顫":1085,"風":1086,"駐":1087,"責":1088,"埲":1089,"c":1090,"孩":1091,"階":1092,"回":1093,"洪":1094,"束":1095,"孺":1096,"訊":1097,"芻":1098,"藉":1099,"嫡":1100,"晾":1101,"庭":1102,"士":1103,"懦":1104,"刑":1105,"如":1106,"支":1107,"店":1108,"R":1109,"咋":1110,"膊":1111,"熨":1112,"妝":1113,"藏":1114,"興":1115,"犀":1116,"馳":1117,"嚒":1118,"掃":1119,"境":1120,"利":1121,"𠺌":1122,"父":1123,"汁":1124,"殉":1125,"捋":1126,"緋":1127,"窩":1128,"伯":1129,"匆":1130,"討":1131,"旬":1132,"鎖":1133,"逃":1134,"股":1135,"菠":1136,"紀":1137,"定":1138,"5":1139,"貸":1140,"慶":1141,"建":1142,"昏":1143,"順":1144,"歸":1145,"君":1146,"魔":1147,"𠱸":1148,"怨":1149,"醫":1150,"盲":1151,"焰":1152,"犯":1153,"嶄":1154,"括":1155,"㾓":1156,"注":1157,"渠":1158,"寫":1159,"尊":1160,"訴":1161,"乞":1162,"須":1163,"垃":1164,"沛":1165,"富":1166,"空":1167,"格":1168,"謹":1169,"獵":1170,"岌":1171,"曆":1172,"愧":1173,"埗":1174,"題":1175,"擔":1176,"曝":1177,"鱗":1178,"芯":1179,"快":1180,"拮":1181,"提":1182,"》":1183,"傾":1184,"賴":1185,"鍾":1186,"妙":1187,"笪":1188,"采":1189,"絀":1190,"跌":1191,"鈔":1192,"墅":1193,"罷":1194,"刊":1195,"托":1196,"幫":1197,"撇":1198,"依":1199,"忠":1200,"衷":1201,"頂":1202,"擂":1203,"嘞":1204,"乳":1205,"沽":1206,"卅":1207,"哦":1208,"識":1209,"漢":1210,"份":1211,"封":1212,"膏":1213,"陸":1214,"粒":1215,"搞":1216,"哺":1217,"擳":1218,"捍":1219,"欣":1220,"趁":1221,"渡":1222,"厚":1223,"貧":1224,"噁":1225,"磨":1226,"攰":1227,"統":1228,"P":1229,"撤":1230,"篷":1231,"髮":1232,"啞":1233,"吻":1234,"汐":1235,"過":1236,"摵":1237,"飽":1238,"緣":1239,"具":1240,"呀":1241,"捕":1242,"准":1243,"（":1244,"社":1245,"曖":1246,"臟":1247,"齒":1248,"剛":1249,"工":1250,"延":1251,"及":1252,"罅":1253,"牒":1254,"欽":1255,"卧":1256,"迄":1257,"端":1258,"夷":1259,"恐":1260,"易":1261,"島":1262,"松":1263,"鹿":1264,"陽":1265,"介":1266,"間":1267,"掹":1268,"免":1269,"腥":1270,"殼":1271,"朱":1272,"吳":1273,"所":1274,"纖":1275,"療":1276,"貼":1277,"肉":1278,"誇":1279,"斑":1280,"帥":1281,"稔":1282,"咧":1283,"就":1284,"評":1285,"策":1286,"𡃁":1287,"偶":1288,"婆":1289,"聚":1290,"茲":1291,"只":1292,"躝":1293,"混":1294,"么":1295,"專":1296,"陋":1297,"／":1298,"粗":1299,"科":1300,"類":1301,"煮":1302,"晃":1303,"漣":1304,"昔":1305,"雅":1306,"銘":1307,"載":1308,"獸":1309,"拋":1310,"牀":1311,"傻":1312,"掯":1313,"鋸":1314,"晰":1315,"嬲":1316,"奇":1317,"疏":1318,"嘈":1319,"渴":1320,"濾":1321,"啲":1322,"綜":1323,"梳":1324,"秋":1325,"戚":1326,"署":1327,"煉":1328,"孔":1329,"龕":1330,"兇":1331,"麗":1332,"撳":1333,"費":1334,"Q":1335,"居":1336,"嗤":1337,"劑":1338,"漚":1339,"阿":1340,"羽":1341,"舌":1342,"赦":1343,"援":1344,"並":1345,"體":1346,"麪":1347,"拆":1348,"仕":1349,"放":1350,"微":1351,"施":1352,"戇":1353,"困":1354,"鶴":1355,"(":1356,"側":1357,"縮":1358,"蛇":1359,"五":1360,"槍":1361,"取":1362,"離":1363,"剔":1364,"咸":1365,"探":1366,"袱":1367,"隔":1368,"燒":1369,"犧":1370,"酬":1371,"逢":1372,"死":1373,"哎":1374,"埔":1375,"角":1376,"購":1377,"曼":1378,"踏":1379,"慣":1380,"餘":1381,"紛":1382,"蹺":1383,"金":1384,"仔":1385,"畢":1386,"爾":1387,"競":1388,"遲":1389,"蓬":1390,"瀝":1391,"精":1392,"芳":1393,"券":1394,"殖":1395,"胃":1396,"液":1397,"巢":1398,"撰":1399,"：":1400,"非":1401,"誠":1402,"唓":1403,"亦":1404,"舍":1405,"型":1406,"彩":1407,"滯":1408,"碗":1409,"婪":1410,"䟕":1411,"妥":1412,"佗":1413,"玫":1414,"之":1415,"慮":1416,"剷":1417,"疫":1418,"尿":1419,"敦":1420,"笠":1421,"般":1422,"員":1423,"名":1424,"唧":1425,"厭":1426,"啖":1427,"走":1428,"丈":1429,"弗":1430,"儡":1431,"福":1432,"黐":1433,"醬":1434,"情":1435,"姻":1436,"綿":1437,"y":1438,"撈":1439,"務":1440,"澎":1441,"頸":1442,"𠺫":1443,"疹":1444,"卜":1445,"！":1446,"峽":1447,"傳":1448,"剁":1449,"嚇":1450,"旨":1451,"孝":1452,"求":1453,"驟":1454,"躉":1455,"前":1456,"任":1457,"增":1458,"卯":1459,"痕":1460,"孖":1461,"逸":1462,"髹":1463,"氣":1464,"握":1465,"豪":1466,"抖":1467,"錯":1468,"顧":1469,"俗":1470,"斃":1471,"霸":1472,"資":1473,"瘟":1474,"n":1475,"甩":1476,"藕":1477,"鳴":1478,"架":1479,"湯":1480,"憑":1481,"浮":1482,"開":1483,"盞":1484,"帳":1485,"往":1486,"避":1487,"勻":1488,"軌":1489,"裏":1490,"仇":1491,"康":1492,"續":1493,"借":1494,"蹄":1495,"同":1496,"幹":1497,"鬱":1498,"？":1499,"嚨":1500,"煤":1501,"橫":1502,"彿":1503,"燥":1504,"C":1505,"侄":1506,"祿":1507,"痺":1508,"鈴":1509,"熊":1510,"s":1511,"蟎":1512,"班":1513,"子":1514,"造":1515,"矢":1516,"愾":1517,"族":1518,"𠱁":1519,"啜":1520,"副":1521,"蓆":1522,"墟":1523,"答":1524,"抄":1525,"駕":1526,"鬍":1527,"凜":1528,"治":1529,"崖":1530,"系":1531,"愈":1532,"繩":1533,"補":1534,"蒲":1535,"引":1536,"併":1537,"g":1538,"火":1539,"橡":1540,"咓":1541,"踿":1542,"禁":1543,"甜":1544,"冥":1545,"嶼":1546,"十":1547,"太":1548,"發":1549,"賣":1550,"切":1551,"徘":1552,"蓉":1553,"都":1554,"耗":1555,"叭":1556,"揸":1557,"旭":1558,"當":1559,"x":1560,"狀":1561,"竅":1562,"劍":1563,"巧":1564,"崇":1565,"事":1566,"揉":1567,"穿":1568,"痰":1569,"清":1570,"漬":1571,"嚏":1572,"為":1573,"客":1574,"吊":1575,"瞬":1576,"窈":1577,"臨":1578,"筒":1579,"噉":1580,"勇":1581,"峯":1582,"棲":1583,"麂":1584,"炭":1585,"鼻":1586,"廂":1587,"大":1588,"咫":1589,"鉛":1590,"蹟":1591,"嘴":1592,"少":1593,"猜":1594,"洛":1595,"災":1596,"遊":1597,"啤":1598,"朋":1599,"幣":1600,"斗":1601,"現":1602,"剪":1603,"聰":1604,"昆":1605,"台":1606,"辭":1607,"剽":1608,"融":1609,"囇":1610,"位":1611,"潦":1612,"暫":1613,"係":1614,"雜":1615,"盆":1616,"屆":1617,"呼":1618,"罵":1619,"爺":1620,"柴":1621,"耿":1622,"表":1623,"菖":1624,"包":1625,"壽":1626,"用":1627,"鯁":1628,"毗":1629,"值":1630,"仍":1631,"監":1632,"疤":1633,"估":1634,"廊":1635,"衍":1636,"蕊":1637,"溪":1638,"視":1639,"芝":1640,"缸":1641,"雀":1642,"兩":1643,"歌":1644,"碰":1645,"醜":1646,"京":1647,"咬":1648,"砂":1649,"霉":1650,"脾":1651,"惹":1652,"翔":1653,"誨":1654,"總":1655,"議":1656,"乘":1657,"掗":1658,"鍛":1659,"忍":1660,"𨀤":1661,"日":1662,"靈":1663,"弟":1664,"浴":1665,"氼":1666,"考":1667,"倔":1668,"澀":1669,"盜":1670,"祢":1671,"復":1672,"層":1673,"凰":1674,"賜":1675,"車":1676,"辨":1677,"是":1678,"魅":1679,"罔":1680,"鎮":1681,"嫲":1682,"確":1683,"斯":1684,"禿":1685,"孭":1686,"誼":1687,"古":1688,"破":1689,"仁":1690,"劣":1691,"喪":1692,"便":1693,"黎":1694,"率":1695,"序":1696,"售":1697,"普":1698,"薑":1699,"曉":1700,"蛛":1701,"久":1702,"顛":1703,"瑕":1704,"蒐":1705,"劊":1706,"菜":1707,"騙":1708,"己":1709,"寸":1710,"晒":1711,"倒":1712,"滴":1713,"喼":1714,"扼":1715,"緩":1716,"製":1717,"朗":1718,"揈":1719,"讓":1720,"屋":1721,"伐":1722,"輛":1723,"閲":1724,"嗎":1725,"洱":1726,"檢":1727,"四":1728,"甘":1729,"寡":1730,"鬥":1731,"紳":1732,"z":1733,"𠴕":1734,"好":1735,"徬":1736,"屈":1737,"侯":1738,"桂":1739,"轉":1740,"睬":1741,"p":1742,"翁":1743,"儒":1744,"梵":1745,"卿":1746,"碟":1747,"叔":1748,"嚕":1749,"竹":1750,"番":1751,"苓":1752,"院":1753,"泄":1754,"晶":1755,"按":1756,"質":1757,"座":1758,"芒":1759,"果":1760,"希":1761,"錫":1762,"茶":1763,"土":1764,"醉":1765,"物":1766,"…":1767,"虞":1768,"囗":1769,"贏":1770,"軀":1771,"莊":1772,"鯊":1773,"錶":1774,"譜":1775,"池":1776,"糞":1777,"崗":1778,"挫":1779,"X":1780,"檔":1781,"攪":1782,"充":1783,"碼":1784,"喺":1785,"綴":1786,"領":1787,"珍":1788,"逐":1789,"袍":1790,"巾":1791,"蘇":1792,"夏":1793,"綫":1794,"几":1795,"噏":1796,"運":1797,"樸":1798,"繞":1799,"搣":1800,"T":1801,"睞":1802,"敍":1803,"勁":1804,"恥":1805,"到":1806,"告":1807,"悉":1808,"喘":1809,"罨":1810,"挈":1811,"挨":1812,"遢":1813,"𢲡":1814,"巴":1815,"殊":1816,"柚":1817,"輩":1818,"猛":1819,"贊":1820,"坪":1821,"妻":1822,"癦":1823,"炒":1824,"狼":1825,"曱":1826,"辰":1827,"冇":1828,"囡":1829,"窟":1830,"氈":1831,"氛":1832,"鋼":1833,"活":1834,"速":1835,"哽":1836,"木":1837,"外":1838,"靖":1839,"助":1840,"規":1841,"旱":1842,"豚":1843,"吋":1844,"轆":1845,"臺":1846,"墩":1847,"頻":1848,"樖":1849,"飯":1850,"砌":1851,"點":1852,"洗":1853,"忳":1854,"楷":1855,"鄰":1856,"矯":1857,"想":1858,"品":1859,"馬":1860,"早":1861,"罪":1862,"㗎":1863,"酸":1864,"呢":1865,"坑":1866,"侶":1867,"團":1868,"跡":1869,"千":1870,"錦":1871,"嘎":1872,"嚡":1873,"覓":1874,"山":1875,"腹":1876,"畸":1877,"丟":1878,"歲":1879,"和":1880,"泓":1881,"慳":1882,"園":1883,"謠":1884,"祖":1885,"廠":1886,"扱":1887,"暑":1888,"貌":1889,"妄":1890,"巫":1891,"赤":1892,"麥":1893,"丹":1894,"辦":1895,"穎":1896,"竊":1897,"幾":1898,"埞":1899,"獎":1900,"獅":1901,"E":1902,"侍":1903,"橢":1904,"濃":1905,"夫":1906,"憤":1907,"闢":1908,"尺":1909,"胖":1910,"吽":1911,"洶":1912,"織":1913,"㔆":1914,"試":1915,"脆":1916,"域":1917,"案":1918,"雄":1919,"否":1920,"滋":1921,"席":1922,"受":1923,"旗":1924,"吞":1925,"徹":1926,"仄":1927,"謙":1928,"以":1929,"甚":1930,"搽":1931,"扁":1932,"檬":1933,"天":1934,"測":1935,"曹":1936,"帖":1937,"敢":1938,"𠺝":1939,"吧":1940,"娶":1941,"奚":1942,"𨋢":1943,"批":1944,"將":1945,"反":1946,"娜":1947,"膝":1948,"披":1949,"背":1950,"薪":1951,"迭":1952,"葛":1953,"姑":1954,"款":1955,"庫":1956,"掅":1957,"溫":1958,"區":1959,"南":1960,"市":1961,"枕":1962,"恤":1963,"#":1964,"核":1965,"K":1966,"由":1967,"被":1968,"因":1969,"坦":1970,"們":1971,"兜":1972,"磋":1973,"聖":1974,"煩":1975,"革":1976,"授":1977,"柒":1978,"聯":1979,"攏":1980,"魂":1981,"憎":1982,"聆":1983,"摘":1984,"薦":1985,"鈕":1986,"色":1987,"遵":1988,"演":1989,"待":1990,"廉":1991,"I":1992,"掠":1993,"兢":1994,"夢":1995,"𣲷":1996,"潰":1997,"味":1998,"髖":1999,"撼":2000,"男":2001,"寒":2002,"豉":2003,"折":2004,"f":2005,"吵":2006,"數":2007,"推":2008,"毅":2009,"帶":2010,"街":2011,"劖":2012,"桃":2013,"叉":2014,"優":2015,"智":2016,"伴":2017,"葬":2018,"鼓":2019,"刮":2020,"傅":2021,"諸":2022,"疵":2023,"忘":2024,"爛":2025,"隨":2026,"低":2027,"霖":2028,"塔":2029,"鍵":2030,"瞰":2031,"含":2032,"刺":2033,"臭":2034,"鈎":2035,"咒":2036,"防":2037,"瓊":2038,"志":2039,"女":2040,"躬":2041,"魷":2042,"讚":2043,"㷫":2044,"冬":2045,"俬":2046,"筆":2047,"潑":2048,"慈":2049,"平":2050,"琵":2051,"拼":2052,"腺":2053,"輕":2054,"咀":2055,"碇":2056,"偉":2057,"撻":2058,"沱":2059,"恒":2060,"損":2061,"付":2062,"佑":2063,"警":2064,"闊":2065,"首":2066,"櫈":2067,"偈":2068,"腕":2069,"姣":2070,"箭":2071,"蛹":2072,"素":2073,"襟":2074,"絡":2075,"誦":2076,"計":2077,"染":2078,"擤":2079,"步":2080,"涕":2081,"拾":2082,"沐":2083,"港":2084,"盟":2085,"乎":2086,"產":2087,"腸":2088,"榮":2089,"斬":2090,"頓":2091,"仿":2092,"抌":2093,"送":2094,"賀":2095,"諾":2096,"簡":2097,"註":2098,"概":2099,"揦":2100,"碑":2101,"漏":2102,"霍":2103,"即":2104,"踩":2105,"勝":2106,"餓":2107,"設":2108,"抽":2109,"脈":2110,"証":2111,"透":2112,"次":2113,"慘":2114,"鑊":2115,"友":2116,"羅":2117,"綠":2118,"流":2119,"聾":2120,"豫":2121,"艙":2122,"襲":2123,"鐵":2124,"川":2125,"役":2126,"噃":2127,"泌":2128,"逼":2129,"華":2130,"偵":2131,"誕":2132,"菩":2133,"傷":2134,"雲":2135,"踢":2136,"凹":2137,"環":2138,"拔":2139,"盪":2140,"來":2141,"﹐":2142,"捅":2143,"向":2144,"衰":2145,"晏":2146,"閃":2147,"械":2148,"嗰":2149,"媒":2150,"著":2151,"立":2152,"澤":2153,"催":2154,"的":2155,"獲":2156,"呻":2157,"埠":2158,"棒":2159,"毛":2160,"、":2161,"薩":2162,"戒":2163,"樓":2164,"燃":2165,"哣":2166,"擁":2167,"隆":2168,"拘":2169,"征":2170,"骨":2171,"神":2172,"啊":2173,"貞":2174,"葵":2175,"祈":2176,"撞":2177,"焗":2178,"u":2179,"辮":2180,"瞓":2181,"狐":2182,"竟":2183,"倍":2184,"唸":2185,"飄":2186,"蘭":2187,"撩":2188,"𨈇":2189,"秘":2190,"例":2191,"栽":2192,"啝":2193,"打":2194,"午":2195,"哈":2196,"懈":2197,"腮":2198,"落":2199,"籬":2200,"鵝":2201,"碩":2202,"蝸":2203,"倖":2204,"育":2205,"吹":2206,"乜":2207,"疲":2208,"戴":2209,"畫":2210,"棟":2211,"春":2212,"盤":2213,"禽":2214,"冧":2215,"淘":2216,"矮":2217,"米":2218,"喇":2219,"者":2220,"壓":2221,"翰":2222,"褲":2223,"閂":2224,"葉":2225,"横":2226,"搪":2227,"遞":2228,"徒":2229,"嗌":2230,"版":2231,"龐":2232,"7":2233,"辣":2234,"亡":2235,"簾":2236,"熔":2237,"虹":2238,"兄":2239,"彎":2240,"租":2241,"辜":2242,"矇":2243,"咖":2244,"笑":2245,"墮":2246,"繼":2247,"佣":2248,"略":2249,"j":2250,"頌":2251,"腑":2252,"鞍":2253,"艮":2254,"或":2255,"匪":2256,"懵":2257,"頒":2258,"適":2259,"郁":2260,"閪":2261,"獄":2262,"脹":2263,"6":2264,"粉":2265,"歪":2266,"叮":2267,"漫":2268,"站":2269,"椒":2270,"揭":2271,"協":2272,"陶":2273,"懇":2274,"央":2275,"癡":2276,"徨":2277,"耐":2278,"S":2279,"鉗":2280,"沾":2281,"鵡":2282,"㖭":2283,"邀":2284,"撥":2285,"存":2286,"叫":2287,"温":2288,"塢":2289,"隊":2290,"閘":2291,"弱":2292,"挺":2293,"乒":2294,"始":2295,"泛":2296,"技":2297,"麟":2298,"約":2299,"寢":2300,"賊":2301,"簽":2302,"董":2303,"仰":2304,"𥄫":2305,"絲":2306,"氫":2307,"督":2308,"嚥":2309,"伸":2310,"𦧲":2311,"持":2312,"錄":2313,"茄":2314,"收":2315,"觀":2316,"第":2317,"砸":2318,"睹":2319,"種":2320,"勃":2321,"獨":2322,"枉":2323,"描":2324,"週":2325,"棄":2326,"矩":2327,"訓":2328,"朵":2329,"琦":2330,"僑":2331,"厲":2332,"9":2333,"玩":2334,"則":2335,"箇":2336,"陛":2337,"豐":2338,"庚":2339,"會":2340,"嚮":2341,"傭":2342,"缺":2343,"海":2344,"鹽":2345,"憂":2346,"～":2347,"鄉":2348,"州":2349,"晉":2350,"搜":2351,"斤":2352,"搗":2353,"擴":2354,"鴨":2355,"手":2356,"筷":2357,"盎":2358,"擗":2359,"忤":2360,"某":2361,"蘋":2362,"嫌":2363,"駁":2364,"熠":2365,"擱":2366,"w":2367,"溦":2368,"虔":2369,"紮":2370,"害":2371,"嘔":2372,"懷":2373,"娩":2374,"喫":2375,"盼":2376,"全":2377,"再":2378,"冤":2379,"裙":2380,"倉":2381,"命":2382,"邏":2383,"降":2384,"慾":2385,"鳥":2386,"煎":2387,"穌":2388,"養":2389,"湊":2390,"嬉":2391,"搥":2392,"捻":2393,"賦":2394,"緒":2395,"亨":2396,"騷":2397,"讀":2398,"昂":2399,"串":2400,"呆":2401,"鬆":2402,"喜":2403,"珠":2404,"骸":2405,"鵲":2406,"了":2407,"烏":2408,"力":2409,"丼":2410,"美":2411,"糟":2412,"桶":2413,"在":2414,"䟴":2415,"出":2416,"材":2417,"乸":2418,"泥":2419,"齊":2420,"該":2421,"里":2422,"俠":2423,"輯":2424,"湖":2425,"癖":2426,"棚":2427,"柔":2428,"洲":2429,"慰":2430,"卼":2431,"婷":2432,"閣":2433,"經":2434,"唾":2435,"沿":2436,"闔":2437,"蒼":2438,"減":2439,"吸":2440,"掘":2441,"^":2442,"突":2443,"陌":2444,"乖":2445,"灘":2446,"𢳂":2447,"琶":2448,"越":2449,"詐":2450,"佐":2451,"課":2452,"賺":2453,"囂":2454,"貿":2455,"內":2456,"踎":2457,"U":2458,"膠":2459,"宰":2460,"幕":2461,"懸":2462,"；":2463,"校":2464,"嗯":2465,"唥":2466,"料":2467,"姆":2468,"狡":2469,"予":2470,"眼":2471,"癌":2472,"奀":2473,"唱":2474,"蕩":2475,"昌":2476,"暴":2477,"泉":2478,"田":2479,"彙":2480,"慎":2481,"燕":2482,"晨":2483,"戲":2484,"凡":2485,"繃":2486,"霜":2487,"喚":2488,"熄":2489,"鼠":2490,"窗":2491,"乾":2492,"肇":2493,"屍":2494,"𢆡":2495,"囈":2496,"r":2497,"七":2498,"牲":2499,"騮":2500,"盃":2501,"漸":2502,"橋":2503,"隱":2504,"飲":2505,"濁":2506,"疾":2507,"斲":2508,"庇":2509,"髀":2510,"齋":2511,"吱":2512,"青":2513,"婀":2514,"乙":2515,"額":2516,"餉":2517,"匡":2518,"諒":2519,"秀":2520,"互":2521,"哄":2522,"煽":2523,"雷":2524,"嗲":2525,"掟":2526,"鐘":2527,"抰":2528,"託":2529,"截":2530,"咯":2531,"翱":2532,"《":2533,"咭":2534,"拓":2535,"範":2536,"膩":2537,"磚":2538,"霧":2539,"益":2540,"貨":2541,"舞":2542,"夠":2543,"認":2544,"'":2545,"涌":2546,"忽":2547,"捐":2548,"叻":2549,"削":2550,"捵":2551,"暗":2552,"愕":2553,"揩":2554,"灣":2555,"術":2556,"祉":2557,"喊":2558,"倦":2559,"羔":2560,"鏽":2561,"涵":2562,"億":2563,"舟":2564,"甥":2565,"級":2566,"候":2567,"蜂":2568,"焫":2569,"連":2570,"算":2571,"雙":2572,"囊":2573,"摟":2574,"樣":2575,"緝":2576,"餞":2577,"劃":2578,"嗚":2579,"唳":2580,"聽":2581,"汽":2582,"圖":2583,"意":2584,"處":2585,"懾":2586,"仝":2587,"身":2588,"茅":2589,"輝":2590,"汗":2591,"劫":2592,"性":2593,"廢":2594,"準":2595,"塵":2596,"陷":2597,"祠":2598,"附":2599,"營":2600,"迷":2601,"淵":2602,"澈":2603,"炎":2604,"瀆":2605,"迪":2606,"國":2607,"孤":2608,"論":2609,"機":2610,"叢":2611,"褦":2612,"籲":2613,"召":2614,"撫":2615,"燈":2616,"決":2617,"怒":2618,"循":2619,"嚟":2620,"瀉":2621,"惟":2622,"逆":2623,"句":2624,"儘":2625,"鯉":2626,"蕎":2627,"礦":2628,"盅":2629,"謂":2630,"堡":2631,"圳":2632,"亮":2633,"選":2634,"宵":2635,"稟":2636,"亭":2637,"拳":2638,"菲":2639,"杞":2640,"彷":2641,"儀":2642,"v":2643,"税":2644,"嘭":2645,"然":2646,"餐":2647,"侵":2648,"硬":2649,"靚":2650,"星":2651,"什":2652,"肌":2653,"籌":2654,"艇":2655,"兵":2656,"梯":2657,"夥":2658,"申":2659,"限":2660,"跟":2661,"獻":2662,"伶":2663,"梨":2664,"N":2665,"贓":2666,"李":2667,"壁":2668,"執":2669,"近":2670,"鎚":2671,"炙":2672,"寂":2673,"醒":2674,"釘":2675,"乏":2676,"異":2677,"本":2678,"脂":2679,"奴":2680,"捶":2681,"狸":2682,"e":2683,"促":2684,"真":2685,"怕":2686,"鞋":2687,"主":2688,"今":2689,"佩":2690,"漂":2691,"峻":2692,"燭":2693,"癮":2694,"氓":2695,"長":2696,"焦":2697,"淪":2698,"兼":2699,"鍊":2700,"唯":2701,"竽":2702,"虛":2703,"趕":2704,"趾":2705,"江":2706,"餵":2707,"潮":2708,"殿":2709,"民":2710,"鋤":2711,"肚":2712,"龜":2713,"煙":2714,"秉":2715,"躍":2716,"鮑":2717,"仙":2718,"扚":2719,"替":2720,"習":2721,"鸚":2722,"痞":2723,"肥":2724,"判":2725,"林":2726,"畜":2727,"武":2728,"痾":2729,"村":2730,"合":2731,"粟":2732,"紂":2733,"拒":2734,"溏":2735,"撑":2736,"謝":2737,"剎":2738,"焉":2739,"摳":2740,"扤":2741,"季":2742,"涯":2743,"鮓":2744,"敲":2745,"袖":2746,"狹":2747,"迺":2748,"倫":2749,"噒":2750,"𠵼":2751,"紓":2752,"兒":2753,"煥":2754,"政":2755,"「":2756,"狠":2757,"礙":2758,"眈":2759,"慌":2760,"莉":2761,"涼":2762,"蟬":2763,"縣":2764,"唆":2765,"紅":2766,"妹":2767,"踐":2768,"瑤":2769,"敗":2770,"腿":2771,"帽":2772,"粥":2773,"蓄":2774,"擅":2775,"頑":2776,"勛":2777,"灌":2778,"榃":2779,"宋":2780,"堪":2781,"幅":2782,"穩":2783,"維":2784,"從":2785,"懂":2786,"k":2787,"沖":2788,"鞭":2789,"鎝":2790,"岸":2791,"拿":2792,"酒":2793,"射":2794,"雞":2795,"睡":2796,"牡":2797,"癢":2798,"量":2799,"籮":2800,"節":2801,"譯":2802,"簷":2803,"醖":2804,"砭":2805,"凱":2806,"宇":2807,"東":2808,"擦":2809,"比":2810,"箱":2811,"浸":2812,"終":2813,"懲":2814,"窿":2815,"剿":2816,"拍":2817,"小":2818,"牽":2819,"媽":2820,"辛":2821,"宜":2822,"怯":2823,"穴":2824,"蠻":2825,"闖":2826,"琴":2827,"溯":2828,"，":2829,"見":2830,"莞":2831,"赫":2832,"詩":2833,"宗":2834,"哋":2835,"筋":2836,"轟":2837,"&":2838,"嚹":2839,"證":2840,"銀":2841,"幸":2842,"辱":2843,"哥":2844,"橙":2845,"舊":2846,"淋":2847,"黑":2848,"罰":2849,"尋":2850,"疑":2851,"拜":2852,"航":2853,"吐":2854,"熱":2855,"哇":2856,"咕":2857,"船":2858,"壅":2859,"搓":2860,"）":2861,"凳":2862,"置":2863,"儲":2864,"詞":2865,"門":2866,"培":2867,"旁":2868,"媳":2869,"口":2870,"嘟":2871,"掛":2872,"痴":2873,"諗":2874,"鯭":2875,"廿":2876,"閉":2877,"瘍":2878,"塘":2879,"禦":2880,"曾":2881,"繫":2882,"櫻":2883,"咦":2884,"榕":2885,"棕":2886,"可":2887,"囖":2888,"槓":2889,"侖":2890,"寄":2891,"驚":2892,"根":2893,"姪":2894,"頃":2895,"晴":2896,"姨":2897,"努":2898,"圾":2899,"強":2900,"寓":2901,"挪":2902,"戰":2903,"貓":2904,"塞":2905,"跑":2906,"漿":2907,"孵":2908,"肅":2909,"履":2910,"呔":2911,"掙":2912,"芬":2913,"電":2914,"病":2915,"-":2916,"楚":2917,"纏":2918,"途":2919,"毋":2920,"價":2921,"人":2922,"光":2923,"每":2924,"釋":2925,"酷":2926,"𠝹":2927,"咇":2928,"菢":2929,"干":2930,"炊":2931,"説":2932,"篩":2933,"地":2934,"苗":2935,"𢳆":2936,"糾":2937,"薄":2938,"年":2939,"滿":2940,"報":2941,"杯":2942,"揗":2943,"淥":2944,"常":2945,"脊":2946,"⋯":2947,"唪":2948,"化":2949,"弦":2950,"瓦":2951,"漁":2952,"餃":2953,"堂":2954,"寧":2955,"樁":2956,"純":2957,"輸":2958,"章":2959,"潛":2960,"寵":2961,"磁":2962,"哮":2963,"柳":2964,"歧":2965,"宣":2966,"卒":2967,"籠":2968,"紐":2969,"擋":2970,"眨":2971,"耳":2972,"複":2973,"宴":2974,"@":2975,"室":2976,"咪":2977,"啩":2978,"韻":2979,"締":2980,"朔":2981,"舉":2982,"繳":2983,"闆":2984,"卷":2985,"仗":2986,"游":2987,"幽":2988,"件":2989,"舖":2990,"渺":2991,"其":2992,"滾":2993,"竭":2994,"滓":2995,"炸":2996,"籍":2997,"甴":2998,"去":2999,"挖":3000,"嘛":3001,"淨":3002,"溶":3003,"棺":3004,"吮":3005,"虎":3006,"枱":3007,"雌":3008,"三":3009,"酗":3010,"抑":3011,"線":3012,"緊":3013,"患":3014,"攬":3015,"/":3016,"廁":3017,"郊":3018,"汰":3019,"淒":3020,"謎":3021,"癲":3022,"唂":3023,"佔":3024,"萌":3025,"扭":3026,"擒":3027,"渣":3028,"𢃇":3029,"彳":3030,"趷":3031,"烈":3032,"徵":3033,"均":3034,"秩":3035,"奸":3036,"跣":3037,"生":3038,"哼":3039,"記":3040,"唇":3041,"派":3042,"默":3043,"備":3044,"曲":3045,"鈍":3046,"嘯":3047,"摩":3048,"籤":3049,"廚":3050,"沉":3051,"未":3052,"看":3053,"抱":3054,"糯":3055,"勞":3056,"驍":3057,"庶":3058,".":3059,"必":3060,"糍":3061,"館":3062,"正":3063,"寬":3064,"捱":3065,"影":3066,"塑":3067,"誌":3068,"史":3069,"驗":3070,"鳩":3071,"德":3072,"劇":3073,"慨":3074,"肪":3075,"姿":3076,"禍":3077,"毒":3078,"烹":3079,"恕":3080,"示":3081,"調":3082,"歎":3083,"恩":3084,"輾":3085,"斥":3086,"貢":3087,"且":3088,"呃":3089,"稍":3090,"吔":3091,"花":3092,"趙":3093,"享":3094,"傍":3095,"另":3096,"滄":3097,"慕":3098,"講":3099,"艦":3100,"津":3101,"曦":3102,"住":3103,"趴":3104,"護":3105,"佳":3106,"禧":3107,"揾":3108,"帛":3109,"管":3110,"些":3111,"蠱":3112,"鎅":3113,"僅":3114,"抆":3115,"個":3116,"變":3117,"吝":3118,"壯":3119,"傀":3120,"嚱":3121,"雨":3122,"究":3123,"牆":3124,"貝":3125,"痛":3126,"扒":3127,"羨":3128,"雖":3129,"浪":3130,"遜":3131,"顏":3132,"棵":3133,"衡":3134,"祝":3135,"褸":3136,"接":3137,"燶":3138,"段":3139,"窮":3140,"感":3141,"匙":3142,"峙":3143,"香":3144,"圈":3145,"宮":3146,"致":3147,"排":3148,"話":3149,"邋":3150,"屁":3151,"蔴":3152,"D":3153,"M":3154,"加":3155,"緻":3156,"軟":3157,"滅":3158,"景":3159,"捷":3160,"挹":3161,"散":3162,"凝":3163,"耍":3164,"找":3165,"列":3166,"衣":3167,"陳":3168,"沙":3169,"傑":3170,"直":3171,"窕":3172,"語":3173,"辟":3174,"裔":3175,"袋":3176,"北":3177,"踭":3178,"才":3179,"集":3180,"展":3181,"河":3182,"𠿪":3183,"齡":3184,"蕭":3185,"擾":3186,"惜":3187,"d":3188,"竇":3189,"彤":3190,"源":3191,"盡":3192,"檯":3193,"留":3194,"豔":3195,"目":3196,"斟":3197,"烊":3198,"鉸":3199,"冚":3200,"帝":3201,"私":3202,"揮":3203,"𠻹":3204,"聞":3205,"原":3206,"户":3207,"嗱":3208,"祥":3209,"後":3210,"膽":3211,"嘖":3212,"莫":3213,"撕":3214,"L":3215,"蔭":3216,"張":3217,"吠":3218,"距":3219,"沈":3220,"吼":3221,"哀":3222,"蜜":3223,"若":3224,"尚":3225,"1":3226,"紗":3227,"邪":3228,"斂":3229,"瑰":3230,"淚":3231,"房":3232,"紹":3233,"瑜":3234,"叛":3235,"毫":3236,"婦":3237,"脛":3238,"寰":3239,"娛":3240,"克":3241,"剖":3242,"遺":3243,"m":3244,"衫":3245,"亍":3246,"保":3247,"迫":3248,"剋":3249,"翌":3250,"家":3251,"除":3252,"拯":3253,"恰":3254,"返":3255,"酥":3256,"薇":3257,"糖":3258,")":3259,"導":3260,"疇":3261,"奶":3262,"冒":3263,"阱":3264,"敏":3265,"石":3266,"憾":3267,"軍":3268,"暈":3269,"差":3270,"幼":3271,"你":3272,"啪":3273,"慢":3274,"頭":3275,"摸":3276,"程":3277,"周":3278,"九":3279,"度":3280,"止":3281,"換":3282,"纊":3283,"韞":3284,"壇":3285,"舫":3286,"B":3287,"退":3288,"韌":3289,"逝":3290,"元":3291,"僻":3292,"激":3293,"姊":3294,"佬":3295,"鴿":3296,"黏":3297,"碎":3298,"販":3299,"羞":3300,"侮":3301,"請":3302,"飢":3303,"旦":3304,"咩":3305,"蠶":3306,"W":3307,"耀":3308,"榨":3309,"而":3310,"抵":3311,"襪":3312,"*":3313,"脱":3314,"偽":3315,"羈":3316,"愚":3317,"暖":3318}
checkpoint/trocr-custdata/checkpoint-1000/config.json DELETED
@@ -1,180 +0,0 @@
1
- {
2
- "_commit_hash": null,
3
- "_name_or_path": "./cust-data/weights",
4
- "architectures": [
5
- "VisionEncoderDecoderModel"
6
- ],
7
- "decoder": {
8
- "_name_or_path": "",
9
- "activation_dropout": 0.0,
10
- "activation_function": "relu",
11
- "add_cross_attention": true,
12
- "architectures": null,
13
- "attention_dropout": 0.0,
14
- "bad_words_ids": null,
15
- "begin_suppress_tokens": null,
16
- "bos_token_id": 0,
17
- "chunk_size_feed_forward": 0,
18
- "classifier_dropout": 0.0,
19
- "cross_attention_hidden_size": 384,
20
- "d_model": 256,
21
- "decoder_attention_heads": 8,
22
- "decoder_ffn_dim": 1024,
23
- "decoder_layerdrop": 0.0,
24
- "decoder_layers": 6,
25
- "decoder_start_token_id": 2,
26
- "diversity_penalty": 0.0,
27
- "do_sample": false,
28
- "dropout": 0.1,
29
- "early_stopping": false,
30
- "encoder_no_repeat_ngram_size": 0,
31
- "eos_token_id": 2,
32
- "exponential_decay_length_penalty": null,
33
- "finetuning_task": null,
34
- "forced_bos_token_id": null,
35
- "forced_eos_token_id": null,
36
- "id2label": {
37
- "0": "LABEL_0",
38
- "1": "LABEL_1"
39
- },
40
- "init_std": 0.02,
41
- "is_decoder": true,
42
- "is_encoder_decoder": false,
43
- "label2id": {
44
- "LABEL_0": 0,
45
- "LABEL_1": 1
46
- },
47
- "layernorm_embedding": true,
48
- "length_penalty": 1.0,
49
- "max_length": 20,
50
- "max_position_embeddings": 512,
51
- "min_length": 0,
52
- "model_type": "trocr",
53
- "no_repeat_ngram_size": 0,
54
- "num_beam_groups": 1,
55
- "num_beams": 1,
56
- "num_return_sequences": 1,
57
- "output_attentions": false,
58
- "output_hidden_states": false,
59
- "output_scores": false,
60
- "pad_token_id": 1,
61
- "prefix": null,
62
- "problem_type": null,
63
- "pruned_heads": {},
64
- "remove_invalid_values": false,
65
- "repetition_penalty": 1.0,
66
- "return_dict": true,
67
- "return_dict_in_generate": false,
68
- "scale_embedding": true,
69
- "sep_token_id": null,
70
- "suppress_tokens": null,
71
- "task_specific_params": null,
72
- "temperature": 1.0,
73
- "tf_legacy_loss": false,
74
- "tie_encoder_decoder": false,
75
- "tie_word_embeddings": false,
76
- "tokenizer_class": null,
77
- "top_k": 50,
78
- "top_p": 1.0,
79
- "torch_dtype": null,
80
- "torchscript": false,
81
- "transformers_version": "4.26.0.dev0",
82
- "typical_p": 1.0,
83
- "use_bfloat16": false,
84
- "use_cache": false,
85
- "use_learned_position_embeddings": true,
86
- "vocab_size": 2794
87
- },
88
- "decoder_start_token_id": 0,
89
- "early_stopping": true,
90
- "encoder": {
91
- "_name_or_path": "",
92
- "add_cross_attention": false,
93
- "architectures": null,
94
- "attention_probs_dropout_prob": 0.0,
95
- "bad_words_ids": null,
96
- "begin_suppress_tokens": null,
97
- "bos_token_id": null,
98
- "chunk_size_feed_forward": 0,
99
- "cross_attention_hidden_size": null,
100
- "decoder_start_token_id": null,
101
- "diversity_penalty": 0.0,
102
- "do_sample": false,
103
- "early_stopping": false,
104
- "encoder_no_repeat_ngram_size": 0,
105
- "encoder_stride": 16,
106
- "eos_token_id": null,
107
- "exponential_decay_length_penalty": null,
108
- "finetuning_task": null,
109
- "forced_bos_token_id": null,
110
- "forced_eos_token_id": null,
111
- "hidden_act": "gelu",
112
- "hidden_dropout_prob": 0.0,
113
- "hidden_size": 384,
114
- "id2label": {
115
- "0": "LABEL_0",
116
- "1": "LABEL_1"
117
- },
118
- "image_size": 384,
119
- "initializer_range": 0.02,
120
- "intermediate_size": 1536,
121
- "is_decoder": false,
122
- "is_encoder_decoder": false,
123
- "label2id": {
124
- "LABEL_0": 0,
125
- "LABEL_1": 1
126
- },
127
- "layer_norm_eps": 1e-12,
128
- "length_penalty": 1.0,
129
- "max_length": 20,
130
- "min_length": 0,
131
- "model_type": "deit",
132
- "no_repeat_ngram_size": 0,
133
- "num_attention_heads": 6,
134
- "num_beam_groups": 1,
135
- "num_beams": 1,
136
- "num_channels": 3,
137
- "num_hidden_layers": 12,
138
- "num_return_sequences": 1,
139
- "output_attentions": false,
140
- "output_hidden_states": false,
141
- "output_scores": false,
142
- "pad_token_id": null,
143
- "patch_size": 16,
144
- "prefix": null,
145
- "problem_type": null,
146
- "pruned_heads": {},
147
- "qkv_bias": true,
148
- "remove_invalid_values": false,
149
- "repetition_penalty": 1.0,
150
- "return_dict": true,
151
- "return_dict_in_generate": false,
152
- "sep_token_id": null,
153
- "suppress_tokens": null,
154
- "task_specific_params": null,
155
- "temperature": 1.0,
156
- "tf_legacy_loss": false,
157
- "tie_encoder_decoder": false,
158
- "tie_word_embeddings": true,
159
- "tokenizer_class": null,
160
- "top_k": 50,
161
- "top_p": 1.0,
162
- "torch_dtype": null,
163
- "torchscript": false,
164
- "transformers_version": "4.26.0.dev0",
165
- "typical_p": 1.0,
166
- "use_bfloat16": false
167
- },
168
- "eos_token_id": 2,
169
- "is_encoder_decoder": true,
170
- "length_penalty": 2.0,
171
- "max_length": 256,
172
- "model_type": "vision-encoder-decoder",
173
- "no_repeat_ngram_size": 3,
174
- "num_beams": 4,
175
- "pad_token_id": 1,
176
- "tie_word_embeddings": false,
177
- "torch_dtype": "float32",
178
- "transformers_version": null,
179
- "vocab_size": 2794
180
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
checkpoint/trocr-custdata/checkpoint-1000/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6489104d59cb14b7745d321ea4eac1789eefeb1c4a4d1871c60599141cacdf2e
3
- size 240912133
 
 
 
 
checkpoint/trocr-custdata/checkpoint-1000/pytorch_model.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:11c93666c1d0992537c885ebe8fd0dd5e5976a1a702c441b99c9b82e0d972eeb
3
- size 121067061
 
 
 
 
checkpoint/trocr-custdata/checkpoint-1000/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6ab06bdf28d4a500661240bdf19a0ef6a204e49e6f33c542f0aed44e28681e46
3
- size 627
 
 
 
 
checkpoint/trocr-custdata/checkpoint-1000/trainer_state.json DELETED
@@ -1,626 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 2.092050209205021,
5
- "global_step": 1000,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.02,
12
- "learning_rate": 4.989539748953975e-05,
13
- "loss": 0.3889,
14
- "step": 10
15
- },
16
- {
17
- "epoch": 0.04,
18
- "learning_rate": 4.97907949790795e-05,
19
- "loss": 0.245,
20
- "step": 20
21
- },
22
- {
23
- "epoch": 0.06,
24
- "learning_rate": 4.968619246861925e-05,
25
- "loss": 0.1722,
26
- "step": 30
27
- },
28
- {
29
- "epoch": 0.08,
30
- "learning_rate": 4.9581589958159e-05,
31
- "loss": 0.1545,
32
- "step": 40
33
- },
34
- {
35
- "epoch": 0.1,
36
- "learning_rate": 4.947698744769875e-05,
37
- "loss": 0.1614,
38
- "step": 50
39
- },
40
- {
41
- "epoch": 0.13,
42
- "learning_rate": 4.93723849372385e-05,
43
- "loss": 0.1332,
44
- "step": 60
45
- },
46
- {
47
- "epoch": 0.15,
48
- "learning_rate": 4.926778242677825e-05,
49
- "loss": 0.1098,
50
- "step": 70
51
- },
52
- {
53
- "epoch": 0.17,
54
- "learning_rate": 4.916317991631799e-05,
55
- "loss": 0.1298,
56
- "step": 80
57
- },
58
- {
59
- "epoch": 0.19,
60
- "learning_rate": 4.905857740585774e-05,
61
- "loss": 0.117,
62
- "step": 90
63
- },
64
- {
65
- "epoch": 0.21,
66
- "learning_rate": 4.895397489539749e-05,
67
- "loss": 0.114,
68
- "step": 100
69
- },
70
- {
71
- "epoch": 0.23,
72
- "learning_rate": 4.8849372384937244e-05,
73
- "loss": 0.1137,
74
- "step": 110
75
- },
76
- {
77
- "epoch": 0.25,
78
- "learning_rate": 4.8744769874476994e-05,
79
- "loss": 0.077,
80
- "step": 120
81
- },
82
- {
83
- "epoch": 0.27,
84
- "learning_rate": 4.864016736401674e-05,
85
- "loss": 0.0907,
86
- "step": 130
87
- },
88
- {
89
- "epoch": 0.29,
90
- "learning_rate": 4.853556485355649e-05,
91
- "loss": 0.0863,
92
- "step": 140
93
- },
94
- {
95
- "epoch": 0.31,
96
- "learning_rate": 4.843096234309624e-05,
97
- "loss": 0.0909,
98
- "step": 150
99
- },
100
- {
101
- "epoch": 0.33,
102
- "learning_rate": 4.832635983263599e-05,
103
- "loss": 0.0923,
104
- "step": 160
105
- },
106
- {
107
- "epoch": 0.36,
108
- "learning_rate": 4.822175732217574e-05,
109
- "loss": 0.0898,
110
- "step": 170
111
- },
112
- {
113
- "epoch": 0.38,
114
- "learning_rate": 4.811715481171548e-05,
115
- "loss": 0.0852,
116
- "step": 180
117
- },
118
- {
119
- "epoch": 0.4,
120
- "learning_rate": 4.8012552301255234e-05,
121
- "loss": 0.0739,
122
- "step": 190
123
- },
124
- {
125
- "epoch": 0.42,
126
- "learning_rate": 4.7907949790794984e-05,
127
- "loss": 0.0801,
128
- "step": 200
129
- },
130
- {
131
- "epoch": 0.44,
132
- "learning_rate": 4.7803347280334734e-05,
133
- "loss": 0.076,
134
- "step": 210
135
- },
136
- {
137
- "epoch": 0.46,
138
- "learning_rate": 4.7698744769874485e-05,
139
- "loss": 0.0772,
140
- "step": 220
141
- },
142
- {
143
- "epoch": 0.48,
144
- "learning_rate": 4.759414225941423e-05,
145
- "loss": 0.0595,
146
- "step": 230
147
- },
148
- {
149
- "epoch": 0.5,
150
- "learning_rate": 4.748953974895398e-05,
151
- "loss": 0.0612,
152
- "step": 240
153
- },
154
- {
155
- "epoch": 0.52,
156
- "learning_rate": 4.738493723849373e-05,
157
- "loss": 0.0513,
158
- "step": 250
159
- },
160
- {
161
- "epoch": 0.54,
162
- "learning_rate": 4.728033472803347e-05,
163
- "loss": 0.0555,
164
- "step": 260
165
- },
166
- {
167
- "epoch": 0.56,
168
- "learning_rate": 4.7175732217573224e-05,
169
- "loss": 0.0731,
170
- "step": 270
171
- },
172
- {
173
- "epoch": 0.59,
174
- "learning_rate": 4.7071129707112974e-05,
175
- "loss": 0.0677,
176
- "step": 280
177
- },
178
- {
179
- "epoch": 0.61,
180
- "learning_rate": 4.696652719665272e-05,
181
- "loss": 0.0673,
182
- "step": 290
183
- },
184
- {
185
- "epoch": 0.63,
186
- "learning_rate": 4.686192468619247e-05,
187
- "loss": 0.0552,
188
- "step": 300
189
- },
190
- {
191
- "epoch": 0.65,
192
- "learning_rate": 4.675732217573222e-05,
193
- "loss": 0.0646,
194
- "step": 310
195
- },
196
- {
197
- "epoch": 0.67,
198
- "learning_rate": 4.665271966527197e-05,
199
- "loss": 0.0681,
200
- "step": 320
201
- },
202
- {
203
- "epoch": 0.69,
204
- "learning_rate": 4.654811715481171e-05,
205
- "loss": 0.05,
206
- "step": 330
207
- },
208
- {
209
- "epoch": 0.71,
210
- "learning_rate": 4.644351464435146e-05,
211
- "loss": 0.0621,
212
- "step": 340
213
- },
214
- {
215
- "epoch": 0.73,
216
- "learning_rate": 4.6338912133891213e-05,
217
- "loss": 0.0557,
218
- "step": 350
219
- },
220
- {
221
- "epoch": 0.75,
222
- "learning_rate": 4.6234309623430964e-05,
223
- "loss": 0.0682,
224
- "step": 360
225
- },
226
- {
227
- "epoch": 0.77,
228
- "learning_rate": 4.6129707112970714e-05,
229
- "loss": 0.0543,
230
- "step": 370
231
- },
232
- {
233
- "epoch": 0.79,
234
- "learning_rate": 4.602510460251046e-05,
235
- "loss": 0.0576,
236
- "step": 380
237
- },
238
- {
239
- "epoch": 0.82,
240
- "learning_rate": 4.592050209205021e-05,
241
- "loss": 0.0551,
242
- "step": 390
243
- },
244
- {
245
- "epoch": 0.84,
246
- "learning_rate": 4.581589958158996e-05,
247
- "loss": 0.0627,
248
- "step": 400
249
- },
250
- {
251
- "epoch": 0.86,
252
- "learning_rate": 4.571129707112971e-05,
253
- "loss": 0.0708,
254
- "step": 410
255
- },
256
- {
257
- "epoch": 0.88,
258
- "learning_rate": 4.560669456066946e-05,
259
- "loss": 0.0616,
260
- "step": 420
261
- },
262
- {
263
- "epoch": 0.9,
264
- "learning_rate": 4.5502092050209203e-05,
265
- "loss": 0.0496,
266
- "step": 430
267
- },
268
- {
269
- "epoch": 0.92,
270
- "learning_rate": 4.5397489539748954e-05,
271
- "loss": 0.0444,
272
- "step": 440
273
- },
274
- {
275
- "epoch": 0.94,
276
- "learning_rate": 4.5292887029288704e-05,
277
- "loss": 0.0538,
278
- "step": 450
279
- },
280
- {
281
- "epoch": 0.96,
282
- "learning_rate": 4.5188284518828455e-05,
283
- "loss": 0.0553,
284
- "step": 460
285
- },
286
- {
287
- "epoch": 0.98,
288
- "learning_rate": 4.5083682008368205e-05,
289
- "loss": 0.0499,
290
- "step": 470
291
- },
292
- {
293
- "epoch": 1.0,
294
- "learning_rate": 4.497907949790795e-05,
295
- "loss": 0.059,
296
- "step": 480
297
- },
298
- {
299
- "epoch": 1.03,
300
- "learning_rate": 4.48744769874477e-05,
301
- "loss": 0.0433,
302
- "step": 490
303
- },
304
- {
305
- "epoch": 1.05,
306
- "learning_rate": 4.476987447698745e-05,
307
- "loss": 0.0408,
308
- "step": 500
309
- },
310
- {
311
- "epoch": 1.07,
312
- "learning_rate": 4.46652719665272e-05,
313
- "loss": 0.0469,
314
- "step": 510
315
- },
316
- {
317
- "epoch": 1.09,
318
- "learning_rate": 4.456066945606695e-05,
319
- "loss": 0.0397,
320
- "step": 520
321
- },
322
- {
323
- "epoch": 1.11,
324
- "learning_rate": 4.4456066945606694e-05,
325
- "loss": 0.0329,
326
- "step": 530
327
- },
328
- {
329
- "epoch": 1.13,
330
- "learning_rate": 4.4351464435146445e-05,
331
- "loss": 0.0421,
332
- "step": 540
333
- },
334
- {
335
- "epoch": 1.15,
336
- "learning_rate": 4.4246861924686195e-05,
337
- "loss": 0.0357,
338
- "step": 550
339
- },
340
- {
341
- "epoch": 1.17,
342
- "learning_rate": 4.4142259414225946e-05,
343
- "loss": 0.0383,
344
- "step": 560
345
- },
346
- {
347
- "epoch": 1.19,
348
- "learning_rate": 4.4037656903765696e-05,
349
- "loss": 0.0364,
350
- "step": 570
351
- },
352
- {
353
- "epoch": 1.21,
354
- "learning_rate": 4.393305439330544e-05,
355
- "loss": 0.0338,
356
- "step": 580
357
- },
358
- {
359
- "epoch": 1.23,
360
- "learning_rate": 4.382845188284519e-05,
361
- "loss": 0.0415,
362
- "step": 590
363
- },
364
- {
365
- "epoch": 1.26,
366
- "learning_rate": 4.372384937238494e-05,
367
- "loss": 0.026,
368
- "step": 600
369
- },
370
- {
371
- "epoch": 1.28,
372
- "learning_rate": 4.361924686192469e-05,
373
- "loss": 0.0366,
374
- "step": 610
375
- },
376
- {
377
- "epoch": 1.3,
378
- "learning_rate": 4.351464435146444e-05,
379
- "loss": 0.03,
380
- "step": 620
381
- },
382
- {
383
- "epoch": 1.32,
384
- "learning_rate": 4.3410041841004185e-05,
385
- "loss": 0.0313,
386
- "step": 630
387
- },
388
- {
389
- "epoch": 1.34,
390
- "learning_rate": 4.3305439330543936e-05,
391
- "loss": 0.0277,
392
- "step": 640
393
- },
394
- {
395
- "epoch": 1.36,
396
- "learning_rate": 4.3200836820083686e-05,
397
- "loss": 0.0351,
398
- "step": 650
399
- },
400
- {
401
- "epoch": 1.38,
402
- "learning_rate": 4.3096234309623436e-05,
403
- "loss": 0.0277,
404
- "step": 660
405
- },
406
- {
407
- "epoch": 1.4,
408
- "learning_rate": 4.299163179916319e-05,
409
- "loss": 0.0334,
410
- "step": 670
411
- },
412
- {
413
- "epoch": 1.42,
414
- "learning_rate": 4.288702928870293e-05,
415
- "loss": 0.0254,
416
- "step": 680
417
- },
418
- {
419
- "epoch": 1.44,
420
- "learning_rate": 4.278242677824268e-05,
421
- "loss": 0.0327,
422
- "step": 690
423
- },
424
- {
425
- "epoch": 1.46,
426
- "learning_rate": 4.267782426778243e-05,
427
- "loss": 0.0298,
428
- "step": 700
429
- },
430
- {
431
- "epoch": 1.49,
432
- "learning_rate": 4.257322175732218e-05,
433
- "loss": 0.0328,
434
- "step": 710
435
- },
436
- {
437
- "epoch": 1.51,
438
- "learning_rate": 4.246861924686193e-05,
439
- "loss": 0.0299,
440
- "step": 720
441
- },
442
- {
443
- "epoch": 1.53,
444
- "learning_rate": 4.2364016736401676e-05,
445
- "loss": 0.0371,
446
- "step": 730
447
- },
448
- {
449
- "epoch": 1.55,
450
- "learning_rate": 4.2259414225941426e-05,
451
- "loss": 0.0298,
452
- "step": 740
453
- },
454
- {
455
- "epoch": 1.57,
456
- "learning_rate": 4.215481171548118e-05,
457
- "loss": 0.034,
458
- "step": 750
459
- },
460
- {
461
- "epoch": 1.59,
462
- "learning_rate": 4.205020920502093e-05,
463
- "loss": 0.0273,
464
- "step": 760
465
- },
466
- {
467
- "epoch": 1.61,
468
- "learning_rate": 4.194560669456067e-05,
469
- "loss": 0.0323,
470
- "step": 770
471
- },
472
- {
473
- "epoch": 1.63,
474
- "learning_rate": 4.184100418410042e-05,
475
- "loss": 0.0293,
476
- "step": 780
477
- },
478
- {
479
- "epoch": 1.65,
480
- "learning_rate": 4.173640167364017e-05,
481
- "loss": 0.0196,
482
- "step": 790
483
- },
484
- {
485
- "epoch": 1.67,
486
- "learning_rate": 4.1631799163179915e-05,
487
- "loss": 0.036,
488
- "step": 800
489
- },
490
- {
491
- "epoch": 1.69,
492
- "learning_rate": 4.1527196652719666e-05,
493
- "loss": 0.0368,
494
- "step": 810
495
- },
496
- {
497
- "epoch": 1.72,
498
- "learning_rate": 4.1422594142259416e-05,
499
- "loss": 0.032,
500
- "step": 820
501
- },
502
- {
503
- "epoch": 1.74,
504
- "learning_rate": 4.131799163179916e-05,
505
- "loss": 0.0242,
506
- "step": 830
507
- },
508
- {
509
- "epoch": 1.76,
510
- "learning_rate": 4.121338912133891e-05,
511
- "loss": 0.0303,
512
- "step": 840
513
- },
514
- {
515
- "epoch": 1.78,
516
- "learning_rate": 4.110878661087866e-05,
517
- "loss": 0.036,
518
- "step": 850
519
- },
520
- {
521
- "epoch": 1.8,
522
- "learning_rate": 4.100418410041841e-05,
523
- "loss": 0.0379,
524
- "step": 860
525
- },
526
- {
527
- "epoch": 1.82,
528
- "learning_rate": 4.089958158995816e-05,
529
- "loss": 0.0321,
530
- "step": 870
531
- },
532
- {
533
- "epoch": 1.84,
534
- "learning_rate": 4.0794979079497905e-05,
535
- "loss": 0.0259,
536
- "step": 880
537
- },
538
- {
539
- "epoch": 1.86,
540
- "learning_rate": 4.0690376569037656e-05,
541
- "loss": 0.031,
542
- "step": 890
543
- },
544
- {
545
- "epoch": 1.88,
546
- "learning_rate": 4.0585774058577406e-05,
547
- "loss": 0.0339,
548
- "step": 900
549
- },
550
- {
551
- "epoch": 1.9,
552
- "learning_rate": 4.048117154811716e-05,
553
- "loss": 0.0326,
554
- "step": 910
555
- },
556
- {
557
- "epoch": 1.92,
558
- "learning_rate": 4.037656903765691e-05,
559
- "loss": 0.0309,
560
- "step": 920
561
- },
562
- {
563
- "epoch": 1.95,
564
- "learning_rate": 4.027196652719665e-05,
565
- "loss": 0.0269,
566
- "step": 930
567
- },
568
- {
569
- "epoch": 1.97,
570
- "learning_rate": 4.01673640167364e-05,
571
- "loss": 0.0289,
572
- "step": 940
573
- },
574
- {
575
- "epoch": 1.99,
576
- "learning_rate": 4.006276150627615e-05,
577
- "loss": 0.0348,
578
- "step": 950
579
- },
580
- {
581
- "epoch": 2.01,
582
- "learning_rate": 3.99581589958159e-05,
583
- "loss": 0.0343,
584
- "step": 960
585
- },
586
- {
587
- "epoch": 2.03,
588
- "learning_rate": 3.985355648535565e-05,
589
- "loss": 0.03,
590
- "step": 970
591
- },
592
- {
593
- "epoch": 2.05,
594
- "learning_rate": 3.9748953974895396e-05,
595
- "loss": 0.0178,
596
- "step": 980
597
- },
598
- {
599
- "epoch": 2.07,
600
- "learning_rate": 3.964435146443515e-05,
601
- "loss": 0.0285,
602
- "step": 990
603
- },
604
- {
605
- "epoch": 2.09,
606
- "learning_rate": 3.95397489539749e-05,
607
- "loss": 0.0312,
608
- "step": 1000
609
- },
610
- {
611
- "epoch": 2.09,
612
- "eval_acc": 0.5445544527497305,
613
- "eval_cer": 0.06267806267806268,
614
- "eval_loss": 0.029834048822522163,
615
- "eval_runtime": 75.6807,
616
- "eval_samples_per_second": 2.669,
617
- "eval_steps_per_second": 0.344,
618
- "step": 1000
619
- }
620
- ],
621
- "max_steps": 4780,
622
- "num_train_epochs": 10,
623
- "total_flos": 6.232735807753421e+17,
624
- "trial_name": null,
625
- "trial_params": null
626
- }