Hugging Face's logo

rttl-ai
/
BIOptimus

Fill-Mask
Transformers
PyTorch
English
bert
biology
medical
Inference Endpoints
Model card Files Community
2
rttl commited on
Commit
3d960bb
•
1 Parent(s): e67babf

Upload 11 files

Browse files
Files changed (11) hide show
  1. config.json +27 -0
  2. optimizer.pt +3 -0
  3. pytorch_model.bin +3 -0
  4. rng_state.pth +3 -0
  5. scaler.pt +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +7 -0
  8. tokenizer_config.json +16 -0
  9. trainer_state.json +3736 -0
  10. training_args.bin +3 -0
  11. vocab.txt +0 -0
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/home/vera/Desktop/pretrain_biomedical/18th-epoch-bioptimus03-wwm-02-pubmed22-resumed-all-mask/checkpoint-32670",
3
+ "architectures": [
4
+ "BertForMaskedLM"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "classifier_dropout": null,
8
+ "gradient_checkpointing": false,
9
+ "hidden_act": "gelu",
10
+ "hidden_dropout_prob": 0.1,
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 3072,
14
+ "layer_norm_eps": 1e-12,
15
+ "max_position_embeddings": 512,
16
+ "model_type": "bert",
17
+ "num_attention_heads": 12,
18
+ "num_hidden_layers": 12,
19
+ "output_hidden_states": true,
20
+ "pad_token_id": 0,
21
+ "position_embedding_type": "absolute",
22
+ "torch_dtype": "float32",
23
+ "transformers_version": "4.22.0.dev0",
24
+ "type_vocab_size": 2,
25
+ "use_cache": true,
26
+ "vocab_size": 30522
27
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9de261be27ec34144542579680bf892c282aa7f987ac0f4998e644a4bc0cb74a
3
+ size 876231833
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e3cf53e3afcd53815d7f0aed7fee9e8a3d7490d1aa33f3237365df1223c853eb
3
+ size 438128811
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c6957be144fb392ec6368443571dc91d1f00c3baac863a8707ee3f3aad67960
3
+ size 15459
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:656cc7f253be8c2ee0e441bfb5b5ab6b163626e4e887b1373239994f03deaf8e
3
+ size 559
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c63331045453745ef07d53139dab98fc965152ba9fb16b7b156219d73a4a06de
3
+ size 623
special_tokens_map.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "pad_token": "[PAD]",
5
+ "sep_token": "[SEP]",
6
+ "unk_token": "[UNK]"
7
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "do_basic_tokenize": true,
4
+ "do_lower_case": true,
5
+ "mask_token": "[MASK]",
6
+ "model_max_length": 1000000000000000019884624838656,
7
+ "name_or_path": "/home/vera/Desktop/pretrain_biomedical/18th-epoch-bioptimus03-wwm-02-pubmed22-resumed-all-mask/checkpoint-32670",
8
+ "never_split": null,
9
+ "pad_token": "[PAD]",
10
+ "sep_token": "[SEP]",
11
+ "special_tokens_map_file": null,
12
+ "strip_accents": null,
13
+ "tokenize_chinese_chars": true,
14
+ "tokenizer_class": "BertTokenizer",
15
+ "unk_token": "[UNK]"
16
+ }
trainer_state.json ADDED
@@ -0,0 +1,3736 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 5.930971148695186,
5
+ "global_step": 310000,
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": 9.98e-07,
13
+ "loss": 1.6418,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 1.998e-06,
19
+ "loss": 1.6387,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 2.9980000000000003e-06,
25
+ "loss": 1.6386,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 3.9980000000000005e-06,
31
+ "loss": 1.6357,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 4.998e-06,
37
+ "loss": 1.633,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 5.996000000000001e-06,
43
+ "loss": 1.6367,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 6.9960000000000004e-06,
49
+ "loss": 1.6346,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 7.996000000000001e-06,
55
+ "loss": 1.6368,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 8.996e-06,
61
+ "loss": 1.6311,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 9.994000000000001e-06,
67
+ "loss": 1.6388,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.11,
72
+ "learning_rate": 9.990399474578892e-06,
73
+ "loss": 1.6331,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 9.98074099830011e-06,
79
+ "loss": 1.6372,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 9.971082522021327e-06,
85
+ "loss": 1.6372,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.13,
90
+ "learning_rate": 9.961424045742545e-06,
91
+ "loss": 1.6315,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.14,
96
+ "learning_rate": 9.95178488641632e-06,
97
+ "loss": 1.6344,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.15,
102
+ "learning_rate": 9.942126410137538e-06,
103
+ "loss": 1.638,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.16,
108
+ "learning_rate": 9.932467933858756e-06,
109
+ "loss": 1.6345,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.17,
114
+ "learning_rate": 9.922809457579972e-06,
115
+ "loss": 1.6324,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.18,
120
+ "learning_rate": 9.913170298253748e-06,
121
+ "loss": 1.6336,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.19,
126
+ "learning_rate": 9.903511821974966e-06,
127
+ "loss": 1.6334,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.2,
132
+ "learning_rate": 9.893853345696184e-06,
133
+ "loss": 1.6336,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.21,
138
+ "learning_rate": 9.884194869417402e-06,
139
+ "loss": 1.6323,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.22,
144
+ "learning_rate": 9.874555710091176e-06,
145
+ "loss": 1.6342,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.23,
150
+ "learning_rate": 9.864897233812394e-06,
151
+ "loss": 1.6329,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.24,
156
+ "learning_rate": 9.855238757533612e-06,
157
+ "loss": 1.6333,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.25,
162
+ "learning_rate": 9.84558028125483e-06,
163
+ "loss": 1.634,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.26,
168
+ "learning_rate": 9.835921804976048e-06,
169
+ "loss": 1.6347,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 0.27,
174
+ "learning_rate": 9.826263328697266e-06,
175
+ "loss": 1.6276,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 0.28,
180
+ "learning_rate": 9.81662416937104e-06,
181
+ "loss": 1.632,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 0.29,
186
+ "learning_rate": 9.806965693092258e-06,
187
+ "loss": 1.6322,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 0.3,
192
+ "learning_rate": 9.797307216813476e-06,
193
+ "loss": 1.6318,
194
+ "step": 15500
195
+ },
196
+ {
197
+ "epoch": 0.31,
198
+ "learning_rate": 9.787648740534694e-06,
199
+ "loss": 1.6286,
200
+ "step": 16000
201
+ },
202
+ {
203
+ "epoch": 0.32,
204
+ "learning_rate": 9.77800958120847e-06,
205
+ "loss": 1.6347,
206
+ "step": 16500
207
+ },
208
+ {
209
+ "epoch": 0.33,
210
+ "learning_rate": 9.768351104929688e-06,
211
+ "loss": 1.6333,
212
+ "step": 17000
213
+ },
214
+ {
215
+ "epoch": 0.33,
216
+ "learning_rate": 9.758692628650906e-06,
217
+ "loss": 1.6345,
218
+ "step": 17500
219
+ },
220
+ {
221
+ "epoch": 0.34,
222
+ "learning_rate": 9.749034152372122e-06,
223
+ "loss": 1.6311,
224
+ "step": 18000
225
+ },
226
+ {
227
+ "epoch": 0.35,
228
+ "learning_rate": 9.73937567609334e-06,
229
+ "loss": 1.6308,
230
+ "step": 18500
231
+ },
232
+ {
233
+ "epoch": 0.36,
234
+ "learning_rate": 9.729736516767115e-06,
235
+ "loss": 1.6299,
236
+ "step": 19000
237
+ },
238
+ {
239
+ "epoch": 0.37,
240
+ "learning_rate": 9.720078040488333e-06,
241
+ "loss": 1.6284,
242
+ "step": 19500
243
+ },
244
+ {
245
+ "epoch": 0.38,
246
+ "learning_rate": 9.71041956420955e-06,
247
+ "loss": 1.6301,
248
+ "step": 20000
249
+ },
250
+ {
251
+ "epoch": 0.39,
252
+ "learning_rate": 9.700761087930768e-06,
253
+ "loss": 1.6305,
254
+ "step": 20500
255
+ },
256
+ {
257
+ "epoch": 0.4,
258
+ "learning_rate": 9.691121928604545e-06,
259
+ "loss": 1.6309,
260
+ "step": 21000
261
+ },
262
+ {
263
+ "epoch": 0.41,
264
+ "learning_rate": 9.681463452325762e-06,
265
+ "loss": 1.6328,
266
+ "step": 21500
267
+ },
268
+ {
269
+ "epoch": 0.42,
270
+ "learning_rate": 9.67180497604698e-06,
271
+ "loss": 1.6292,
272
+ "step": 22000
273
+ },
274
+ {
275
+ "epoch": 0.43,
276
+ "learning_rate": 9.662146499768197e-06,
277
+ "loss": 1.6292,
278
+ "step": 22500
279
+ },
280
+ {
281
+ "epoch": 0.44,
282
+ "learning_rate": 9.652488023489415e-06,
283
+ "loss": 1.6295,
284
+ "step": 23000
285
+ },
286
+ {
287
+ "epoch": 0.45,
288
+ "learning_rate": 9.64284886416319e-06,
289
+ "loss": 1.627,
290
+ "step": 23500
291
+ },
292
+ {
293
+ "epoch": 0.46,
294
+ "learning_rate": 9.633190387884409e-06,
295
+ "loss": 1.6281,
296
+ "step": 24000
297
+ },
298
+ {
299
+ "epoch": 0.47,
300
+ "learning_rate": 9.623531911605627e-06,
301
+ "loss": 1.6273,
302
+ "step": 24500
303
+ },
304
+ {
305
+ "epoch": 0.48,
306
+ "learning_rate": 9.613873435326844e-06,
307
+ "loss": 1.6305,
308
+ "step": 25000
309
+ },
310
+ {
311
+ "epoch": 0.49,
312
+ "learning_rate": 9.604214959048062e-06,
313
+ "loss": 1.6286,
314
+ "step": 25500
315
+ },
316
+ {
317
+ "epoch": 0.5,
318
+ "learning_rate": 9.594575799721837e-06,
319
+ "loss": 1.628,
320
+ "step": 26000
321
+ },
322
+ {
323
+ "epoch": 0.51,
324
+ "learning_rate": 9.584917323443055e-06,
325
+ "loss": 1.6279,
326
+ "step": 26500
327
+ },
328
+ {
329
+ "epoch": 0.52,
330
+ "learning_rate": 9.575258847164273e-06,
331
+ "loss": 1.6248,
332
+ "step": 27000
333
+ },
334
+ {
335
+ "epoch": 0.53,
336
+ "learning_rate": 9.565600370885489e-06,
337
+ "loss": 1.6244,
338
+ "step": 27500
339
+ },
340
+ {
341
+ "epoch": 0.54,
342
+ "learning_rate": 9.555961211559265e-06,
343
+ "loss": 1.6279,
344
+ "step": 28000
345
+ },
346
+ {
347
+ "epoch": 0.55,
348
+ "learning_rate": 9.546302735280483e-06,
349
+ "loss": 1.6269,
350
+ "step": 28500
351
+ },
352
+ {
353
+ "epoch": 0.55,
354
+ "learning_rate": 9.536644259001701e-06,
355
+ "loss": 1.6275,
356
+ "step": 29000
357
+ },
358
+ {
359
+ "epoch": 0.56,
360
+ "learning_rate": 9.526985782722919e-06,
361
+ "loss": 1.6281,
362
+ "step": 29500
363
+ },
364
+ {
365
+ "epoch": 0.57,
366
+ "learning_rate": 9.517327306444137e-06,
367
+ "loss": 1.6248,
368
+ "step": 30000
369
+ },
370
+ {
371
+ "epoch": 0.58,
372
+ "learning_rate": 9.507668830165355e-06,
373
+ "loss": 1.6251,
374
+ "step": 30500
375
+ },
376
+ {
377
+ "epoch": 0.59,
378
+ "learning_rate": 9.498010353886573e-06,
379
+ "loss": 1.6255,
380
+ "step": 31000
381
+ },
382
+ {
383
+ "epoch": 0.6,
384
+ "learning_rate": 9.488371194560347e-06,
385
+ "loss": 1.6276,
386
+ "step": 31500
387
+ },
388
+ {
389
+ "epoch": 0.61,
390
+ "learning_rate": 9.478712718281565e-06,
391
+ "loss": 1.6276,
392
+ "step": 32000
393
+ },
394
+ {
395
+ "epoch": 0.62,
396
+ "learning_rate": 9.469054242002781e-06,
397
+ "loss": 1.6269,
398
+ "step": 32500
399
+ },
400
+ {
401
+ "epoch": 0.63,
402
+ "learning_rate": 9.459395765723999e-06,
403
+ "loss": 1.6281,
404
+ "step": 33000
405
+ },
406
+ {
407
+ "epoch": 0.64,
408
+ "learning_rate": 9.449756606397775e-06,
409
+ "loss": 1.6231,
410
+ "step": 33500
411
+ },
412
+ {
413
+ "epoch": 0.65,
414
+ "learning_rate": 9.440098130118993e-06,
415
+ "loss": 1.625,
416
+ "step": 34000
417
+ },
418
+ {
419
+ "epoch": 0.66,
420
+ "learning_rate": 9.430439653840211e-06,
421
+ "loss": 1.6261,
422
+ "step": 34500
423
+ },
424
+ {
425
+ "epoch": 0.67,
426
+ "learning_rate": 9.420781177561429e-06,
427
+ "loss": 1.6254,
428
+ "step": 35000
429
+ },
430
+ {
431
+ "epoch": 0.68,
432
+ "learning_rate": 9.411122701282647e-06,
433
+ "loss": 1.6253,
434
+ "step": 35500
435
+ },
436
+ {
437
+ "epoch": 0.69,
438
+ "learning_rate": 9.401483541956421e-06,
439
+ "loss": 1.6272,
440
+ "step": 36000
441
+ },
442
+ {
443
+ "epoch": 0.7,
444
+ "learning_rate": 9.39182506567764e-06,
445
+ "loss": 1.6277,
446
+ "step": 36500
447
+ },
448
+ {
449
+ "epoch": 0.71,
450
+ "learning_rate": 9.382166589398857e-06,
451
+ "loss": 1.6239,
452
+ "step": 37000
453
+ },
454
+ {
455
+ "epoch": 0.72,
456
+ "learning_rate": 9.372508113120075e-06,
457
+ "loss": 1.6266,
458
+ "step": 37500
459
+ },
460
+ {
461
+ "epoch": 0.73,
462
+ "learning_rate": 9.362849636841293e-06,
463
+ "loss": 1.6233,
464
+ "step": 38000
465
+ },
466
+ {
467
+ "epoch": 0.74,
468
+ "learning_rate": 9.353210477515067e-06,
469
+ "loss": 1.6268,
470
+ "step": 38500
471
+ },
472
+ {
473
+ "epoch": 0.75,
474
+ "learning_rate": 9.343552001236285e-06,
475
+ "loss": 1.6226,
476
+ "step": 39000
477
+ },
478
+ {
479
+ "epoch": 0.76,
480
+ "learning_rate": 9.333893524957503e-06,
481
+ "loss": 1.6237,
482
+ "step": 39500
483
+ },
484
+ {
485
+ "epoch": 0.77,
486
+ "learning_rate": 9.324235048678721e-06,
487
+ "loss": 1.6261,
488
+ "step": 40000
489
+ },
490
+ {
491
+ "epoch": 0.77,
492
+ "learning_rate": 9.31457657239994e-06,
493
+ "loss": 1.6256,
494
+ "step": 40500
495
+ },
496
+ {
497
+ "epoch": 0.78,
498
+ "learning_rate": 9.304937413073715e-06,
499
+ "loss": 1.6223,
500
+ "step": 41000
501
+ },
502
+ {
503
+ "epoch": 0.79,
504
+ "learning_rate": 9.295278936794933e-06,
505
+ "loss": 1.6245,
506
+ "step": 41500
507
+ },
508
+ {
509
+ "epoch": 0.8,
510
+ "learning_rate": 9.28562046051615e-06,
511
+ "loss": 1.6265,
512
+ "step": 42000
513
+ },
514
+ {
515
+ "epoch": 0.81,
516
+ "learning_rate": 9.275961984237367e-06,
517
+ "loss": 1.6237,
518
+ "step": 42500
519
+ },
520
+ {
521
+ "epoch": 0.82,
522
+ "learning_rate": 9.266322824911142e-06,
523
+ "loss": 1.6212,
524
+ "step": 43000
525
+ },
526
+ {
527
+ "epoch": 0.83,
528
+ "learning_rate": 9.25666434863236e-06,
529
+ "loss": 1.6212,
530
+ "step": 43500
531
+ },
532
+ {
533
+ "epoch": 0.84,
534
+ "learning_rate": 9.247005872353578e-06,
535
+ "loss": 1.6216,
536
+ "step": 44000
537
+ },
538
+ {
539
+ "epoch": 0.85,
540
+ "learning_rate": 9.237347396074796e-06,
541
+ "loss": 1.6276,
542
+ "step": 44500
543
+ },
544
+ {
545
+ "epoch": 0.86,
546
+ "learning_rate": 9.227688919796014e-06,
547
+ "loss": 1.6307,
548
+ "step": 45000
549
+ },
550
+ {
551
+ "epoch": 0.87,
552
+ "learning_rate": 9.21804976046979e-06,
553
+ "loss": 1.6272,
554
+ "step": 45500
555
+ },
556
+ {
557
+ "epoch": 0.88,
558
+ "learning_rate": 9.208391284191008e-06,
559
+ "loss": 1.6252,
560
+ "step": 46000
561
+ },
562
+ {
563
+ "epoch": 0.89,
564
+ "learning_rate": 9.198732807912224e-06,
565
+ "loss": 1.6256,
566
+ "step": 46500
567
+ },
568
+ {
569
+ "epoch": 0.9,
570
+ "learning_rate": 9.189074331633442e-06,
571
+ "loss": 1.6262,
572
+ "step": 47000
573
+ },
574
+ {
575
+ "epoch": 0.91,
576
+ "learning_rate": 9.17941585535466e-06,
577
+ "loss": 1.6246,
578
+ "step": 47500
579
+ },
580
+ {
581
+ "epoch": 0.92,
582
+ "learning_rate": 9.169776696028436e-06,
583
+ "loss": 1.6261,
584
+ "step": 48000
585
+ },
586
+ {
587
+ "epoch": 0.93,
588
+ "learning_rate": 9.160118219749654e-06,
589
+ "loss": 1.6241,
590
+ "step": 48500
591
+ },
592
+ {
593
+ "epoch": 0.94,
594
+ "learning_rate": 9.150459743470872e-06,
595
+ "loss": 1.624,
596
+ "step": 49000
597
+ },
598
+ {
599
+ "epoch": 0.95,
600
+ "learning_rate": 9.14080126719209e-06,
601
+ "loss": 1.6239,
602
+ "step": 49500
603
+ },
604
+ {
605
+ "epoch": 0.96,
606
+ "learning_rate": 9.131162107865864e-06,
607
+ "loss": 1.6216,
608
+ "step": 50000
609
+ },
610
+ {
611
+ "epoch": 0.97,
612
+ "learning_rate": 9.121503631587082e-06,
613
+ "loss": 1.6294,
614
+ "step": 50500
615
+ },
616
+ {
617
+ "epoch": 0.98,
618
+ "learning_rate": 9.1118451553083e-06,
619
+ "loss": 1.6241,
620
+ "step": 51000
621
+ },
622
+ {
623
+ "epoch": 0.99,
624
+ "learning_rate": 9.102186679029516e-06,
625
+ "loss": 1.6193,
626
+ "step": 51500
627
+ },
628
+ {
629
+ "epoch": 0.99,
630
+ "learning_rate": 9.092547519703292e-06,
631
+ "loss": 1.6234,
632
+ "step": 52000
633
+ },
634
+ {
635
+ "epoch": 1.0,
636
+ "learning_rate": 9.08288904342451e-06,
637
+ "loss": 1.6231,
638
+ "step": 52500
639
+ },
640
+ {
641
+ "epoch": 1.01,
642
+ "learning_rate": 9.073230567145728e-06,
643
+ "loss": 1.6248,
644
+ "step": 53000
645
+ },
646
+ {
647
+ "epoch": 1.02,
648
+ "learning_rate": 9.063572090866946e-06,
649
+ "loss": 1.6273,
650
+ "step": 53500
651
+ },
652
+ {
653
+ "epoch": 1.03,
654
+ "learning_rate": 9.053913614588164e-06,
655
+ "loss": 1.6226,
656
+ "step": 54000
657
+ },
658
+ {
659
+ "epoch": 1.04,
660
+ "learning_rate": 9.044274455261938e-06,
661
+ "loss": 1.619,
662
+ "step": 54500
663
+ },
664
+ {
665
+ "epoch": 1.05,
666
+ "learning_rate": 9.034615978983156e-06,
667
+ "loss": 1.6266,
668
+ "step": 55000
669
+ },
670
+ {
671
+ "epoch": 1.06,
672
+ "learning_rate": 9.024957502704374e-06,
673
+ "loss": 1.6216,
674
+ "step": 55500
675
+ },
676
+ {
677
+ "epoch": 1.07,
678
+ "learning_rate": 9.015299026425592e-06,
679
+ "loss": 1.6224,
680
+ "step": 56000
681
+ },
682
+ {
683
+ "epoch": 1.08,
684
+ "learning_rate": 9.005640550146808e-06,
685
+ "loss": 1.6238,
686
+ "step": 56500
687
+ },
688
+ {
689
+ "epoch": 1.09,
690
+ "learning_rate": 8.996001390820584e-06,
691
+ "loss": 1.6201,
692
+ "step": 57000
693
+ },
694
+ {
695
+ "epoch": 1.1,
696
+ "learning_rate": 8.986342914541802e-06,
697
+ "loss": 1.6222,
698
+ "step": 57500
699
+ },
700
+ {
701
+ "epoch": 1.11,
702
+ "learning_rate": 8.97668443826302e-06,
703
+ "loss": 1.6233,
704
+ "step": 58000
705
+ },
706
+ {
707
+ "epoch": 1.12,
708
+ "learning_rate": 8.967025961984238e-06,
709
+ "loss": 1.6259,
710
+ "step": 58500
711
+ },
712
+ {
713
+ "epoch": 1.13,
714
+ "learning_rate": 8.957386802658014e-06,
715
+ "loss": 1.6218,
716
+ "step": 59000
717
+ },
718
+ {
719
+ "epoch": 1.14,
720
+ "learning_rate": 8.94772832637923e-06,
721
+ "loss": 1.6247,
722
+ "step": 59500
723
+ },
724
+ {
725
+ "epoch": 1.15,
726
+ "learning_rate": 8.938069850100449e-06,
727
+ "loss": 1.6239,
728
+ "step": 60000
729
+ },
730
+ {
731
+ "epoch": 1.16,
732
+ "learning_rate": 8.928411373821666e-06,
733
+ "loss": 1.621,
734
+ "step": 60500
735
+ },
736
+ {
737
+ "epoch": 1.17,
738
+ "learning_rate": 8.918752897542884e-06,
739
+ "loss": 1.6251,
740
+ "step": 61000
741
+ },
742
+ {
743
+ "epoch": 1.18,
744
+ "learning_rate": 8.909094421264102e-06,
745
+ "loss": 1.6237,
746
+ "step": 61500
747
+ },
748
+ {
749
+ "epoch": 1.19,
750
+ "learning_rate": 8.899455261937877e-06,
751
+ "loss": 1.6236,
752
+ "step": 62000
753
+ },
754
+ {
755
+ "epoch": 1.2,
756
+ "learning_rate": 8.889796785659095e-06,
757
+ "loss": 1.6223,
758
+ "step": 62500
759
+ },
760
+ {
761
+ "epoch": 1.21,
762
+ "learning_rate": 8.880138309380313e-06,
763
+ "loss": 1.6218,
764
+ "step": 63000
765
+ },
766
+ {
767
+ "epoch": 1.21,
768
+ "learning_rate": 8.87047983310153e-06,
769
+ "loss": 1.6249,
770
+ "step": 63500
771
+ },
772
+ {
773
+ "epoch": 1.22,
774
+ "learning_rate": 8.860821356822748e-06,
775
+ "loss": 1.6212,
776
+ "step": 64000
777
+ },
778
+ {
779
+ "epoch": 1.23,
780
+ "learning_rate": 8.851162880543966e-06,
781
+ "loss": 1.6219,
782
+ "step": 64500
783
+ },
784
+ {
785
+ "epoch": 1.24,
786
+ "learning_rate": 8.841523721217742e-06,
787
+ "loss": 1.6229,
788
+ "step": 65000
789
+ },
790
+ {
791
+ "epoch": 1.25,
792
+ "learning_rate": 8.83186524493896e-06,
793
+ "loss": 1.6214,
794
+ "step": 65500
795
+ },
796
+ {
797
+ "epoch": 1.26,
798
+ "learning_rate": 8.822206768660177e-06,
799
+ "loss": 1.6218,
800
+ "step": 66000
801
+ },
802
+ {
803
+ "epoch": 1.27,
804
+ "learning_rate": 8.812548292381395e-06,
805
+ "loss": 1.6271,
806
+ "step": 66500
807
+ },
808
+ {
809
+ "epoch": 1.28,
810
+ "learning_rate": 8.802909133055169e-06,
811
+ "loss": 1.6239,
812
+ "step": 67000
813
+ },
814
+ {
815
+ "epoch": 1.29,
816
+ "learning_rate": 8.793250656776387e-06,
817
+ "loss": 1.6238,
818
+ "step": 67500
819
+ },
820
+ {
821
+ "epoch": 1.3,
822
+ "learning_rate": 8.783592180497605e-06,
823
+ "loss": 1.62,
824
+ "step": 68000
825
+ },
826
+ {
827
+ "epoch": 1.31,
828
+ "learning_rate": 8.773933704218823e-06,
829
+ "loss": 1.6235,
830
+ "step": 68500
831
+ },
832
+ {
833
+ "epoch": 1.32,
834
+ "learning_rate": 8.764294544892599e-06,
835
+ "loss": 1.6227,
836
+ "step": 69000
837
+ },
838
+ {
839
+ "epoch": 1.33,
840
+ "learning_rate": 8.754636068613817e-06,
841
+ "loss": 1.6222,
842
+ "step": 69500
843
+ },
844
+ {
845
+ "epoch": 1.34,
846
+ "learning_rate": 8.744977592335035e-06,
847
+ "loss": 1.6215,
848
+ "step": 70000
849
+ },
850
+ {
851
+ "epoch": 1.35,
852
+ "learning_rate": 8.735319116056251e-06,
853
+ "loss": 1.624,
854
+ "step": 70500
855
+ },
856
+ {
857
+ "epoch": 1.36,
858
+ "learning_rate": 8.725660639777469e-06,
859
+ "loss": 1.6271,
860
+ "step": 71000
861
+ },
862
+ {
863
+ "epoch": 1.37,
864
+ "learning_rate": 8.716002163498687e-06,
865
+ "loss": 1.6218,
866
+ "step": 71500
867
+ },
868
+ {
869
+ "epoch": 1.38,
870
+ "learning_rate": 8.706363004172463e-06,
871
+ "loss": 1.6211,
872
+ "step": 72000
873
+ },
874
+ {
875
+ "epoch": 1.39,
876
+ "learning_rate": 8.696704527893681e-06,
877
+ "loss": 1.6274,
878
+ "step": 72500
879
+ },
880
+ {
881
+ "epoch": 1.4,
882
+ "learning_rate": 8.687046051614899e-06,
883
+ "loss": 1.624,
884
+ "step": 73000
885
+ },
886
+ {
887
+ "epoch": 1.41,
888
+ "learning_rate": 8.677387575336117e-06,
889
+ "loss": 1.624,
890
+ "step": 73500
891
+ },
892
+ {
893
+ "epoch": 1.42,
894
+ "learning_rate": 8.667729099057333e-06,
895
+ "loss": 1.6234,
896
+ "step": 74000
897
+ },
898
+ {
899
+ "epoch": 1.43,
900
+ "learning_rate": 8.658089939731109e-06,
901
+ "loss": 1.6189,
902
+ "step": 74500
903
+ },
904
+ {
905
+ "epoch": 1.43,
906
+ "learning_rate": 8.648431463452327e-06,
907
+ "loss": 1.6221,
908
+ "step": 75000
909
+ },
910
+ {
911
+ "epoch": 1.44,
912
+ "learning_rate": 8.638772987173543e-06,
913
+ "loss": 1.6224,
914
+ "step": 75500
915
+ },
916
+ {
917
+ "epoch": 1.45,
918
+ "learning_rate": 8.629114510894761e-06,
919
+ "loss": 1.6212,
920
+ "step": 76000
921
+ },
922
+ {
923
+ "epoch": 1.46,
924
+ "learning_rate": 8.619456034615979e-06,
925
+ "loss": 1.6189,
926
+ "step": 76500
927
+ },
928
+ {
929
+ "epoch": 1.47,
930
+ "learning_rate": 8.609816875289755e-06,
931
+ "loss": 1.6169,
932
+ "step": 77000
933
+ },
934
+ {
935
+ "epoch": 1.48,
936
+ "learning_rate": 8.600158399010973e-06,
937
+ "loss": 1.6216,
938
+ "step": 77500
939
+ },
940
+ {
941
+ "epoch": 1.49,
942
+ "learning_rate": 8.590499922732191e-06,
943
+ "loss": 1.6229,
944
+ "step": 78000
945
+ },
946
+ {
947
+ "epoch": 1.5,
948
+ "learning_rate": 8.580841446453409e-06,
949
+ "loss": 1.6219,
950
+ "step": 78500
951
+ },
952
+ {
953
+ "epoch": 1.51,
954
+ "learning_rate": 8.571202287127183e-06,
955
+ "loss": 1.6243,
956
+ "step": 79000
957
+ },
958
+ {
959
+ "epoch": 1.52,
960
+ "learning_rate": 8.561543810848401e-06,
961
+ "loss": 1.6197,
962
+ "step": 79500
963
+ },
964
+ {
965
+ "epoch": 1.53,
966
+ "learning_rate": 8.55188533456962e-06,
967
+ "loss": 1.6199,
968
+ "step": 80000
969
+ },
970
+ {
971
+ "epoch": 1.54,
972
+ "learning_rate": 8.542226858290835e-06,
973
+ "loss": 1.6204,
974
+ "step": 80500
975
+ },
976
+ {
977
+ "epoch": 1.55,
978
+ "learning_rate": 8.532568382012053e-06,
979
+ "loss": 1.6242,
980
+ "step": 81000
981
+ },
982
+ {
983
+ "epoch": 1.56,
984
+ "learning_rate": 8.522909905733271e-06,
985
+ "loss": 1.6201,
986
+ "step": 81500
987
+ },
988
+ {
989
+ "epoch": 1.57,
990
+ "learning_rate": 8.513270746407047e-06,
991
+ "loss": 1.6237,
992
+ "step": 82000
993
+ },
994
+ {
995
+ "epoch": 1.58,
996
+ "learning_rate": 8.503612270128265e-06,
997
+ "loss": 1.6191,
998
+ "step": 82500
999
+ },
1000
+ {
1001
+ "epoch": 1.59,
1002
+ "learning_rate": 8.493953793849483e-06,
1003
+ "loss": 1.6215,
1004
+ "step": 83000
1005
+ },
1006
+ {
1007
+ "epoch": 1.6,
1008
+ "learning_rate": 8.484295317570701e-06,
1009
+ "loss": 1.6233,
1010
+ "step": 83500
1011
+ },
1012
+ {
1013
+ "epoch": 1.61,
1014
+ "learning_rate": 8.474656158244476e-06,
1015
+ "loss": 1.6219,
1016
+ "step": 84000
1017
+ },
1018
+ {
1019
+ "epoch": 1.62,
1020
+ "learning_rate": 8.464997681965694e-06,
1021
+ "loss": 1.6202,
1022
+ "step": 84500
1023
+ },
1024
+ {
1025
+ "epoch": 1.63,
1026
+ "learning_rate": 8.455339205686912e-06,
1027
+ "loss": 1.6207,
1028
+ "step": 85000
1029
+ },
1030
+ {
1031
+ "epoch": 1.64,
1032
+ "learning_rate": 8.44568072940813e-06,
1033
+ "loss": 1.6219,
1034
+ "step": 85500
1035
+ },
1036
+ {
1037
+ "epoch": 1.65,
1038
+ "learning_rate": 8.436022253129347e-06,
1039
+ "loss": 1.6233,
1040
+ "step": 86000
1041
+ },
1042
+ {
1043
+ "epoch": 1.65,
1044
+ "learning_rate": 8.426383093803122e-06,
1045
+ "loss": 1.6221,
1046
+ "step": 86500
1047
+ },
1048
+ {
1049
+ "epoch": 1.66,
1050
+ "learning_rate": 8.41672461752434e-06,
1051
+ "loss": 1.6177,
1052
+ "step": 87000
1053
+ },
1054
+ {
1055
+ "epoch": 1.67,
1056
+ "learning_rate": 8.407066141245558e-06,
1057
+ "loss": 1.6232,
1058
+ "step": 87500
1059
+ },
1060
+ {
1061
+ "epoch": 1.68,
1062
+ "learning_rate": 8.397407664966776e-06,
1063
+ "loss": 1.6236,
1064
+ "step": 88000
1065
+ },
1066
+ {
1067
+ "epoch": 1.69,
1068
+ "learning_rate": 8.387749188687994e-06,
1069
+ "loss": 1.6204,
1070
+ "step": 88500
1071
+ },
1072
+ {
1073
+ "epoch": 1.7,
1074
+ "learning_rate": 8.378090712409211e-06,
1075
+ "loss": 1.6189,
1076
+ "step": 89000
1077
+ },
1078
+ {
1079
+ "epoch": 1.71,
1080
+ "learning_rate": 8.368451553082988e-06,
1081
+ "loss": 1.6251,
1082
+ "step": 89500
1083
+ },
1084
+ {
1085
+ "epoch": 1.72,
1086
+ "learning_rate": 8.358793076804204e-06,
1087
+ "loss": 1.6237,
1088
+ "step": 90000
1089
+ },
1090
+ {
1091
+ "epoch": 1.73,
1092
+ "learning_rate": 8.349134600525422e-06,
1093
+ "loss": 1.6214,
1094
+ "step": 90500
1095
+ },
1096
+ {
1097
+ "epoch": 1.74,
1098
+ "learning_rate": 8.33947612424664e-06,
1099
+ "loss": 1.6232,
1100
+ "step": 91000
1101
+ },
1102
+ {
1103
+ "epoch": 1.75,
1104
+ "learning_rate": 8.329817647967858e-06,
1105
+ "loss": 1.6204,
1106
+ "step": 91500
1107
+ },
1108
+ {
1109
+ "epoch": 1.76,
1110
+ "learning_rate": 8.320159171689075e-06,
1111
+ "loss": 1.6224,
1112
+ "step": 92000
1113
+ },
1114
+ {
1115
+ "epoch": 1.77,
1116
+ "learning_rate": 8.31052001236285e-06,
1117
+ "loss": 1.6234,
1118
+ "step": 92500
1119
+ },
1120
+ {
1121
+ "epoch": 1.78,
1122
+ "learning_rate": 8.300861536084068e-06,
1123
+ "loss": 1.6222,
1124
+ "step": 93000
1125
+ },
1126
+ {
1127
+ "epoch": 1.79,
1128
+ "learning_rate": 8.291203059805286e-06,
1129
+ "loss": 1.6242,
1130
+ "step": 93500
1131
+ },
1132
+ {
1133
+ "epoch": 1.8,
1134
+ "learning_rate": 8.281544583526504e-06,
1135
+ "loss": 1.6213,
1136
+ "step": 94000
1137
+ },
1138
+ {
1139
+ "epoch": 1.81,
1140
+ "learning_rate": 8.271905424200278e-06,
1141
+ "loss": 1.6185,
1142
+ "step": 94500
1143
+ },
1144
+ {
1145
+ "epoch": 1.82,
1146
+ "learning_rate": 8.262246947921496e-06,
1147
+ "loss": 1.6207,
1148
+ "step": 95000
1149
+ },
1150
+ {
1151
+ "epoch": 1.83,
1152
+ "learning_rate": 8.252588471642714e-06,
1153
+ "loss": 1.6218,
1154
+ "step": 95500
1155
+ },
1156
+ {
1157
+ "epoch": 1.84,
1158
+ "learning_rate": 8.242929995363932e-06,
1159
+ "loss": 1.6242,
1160
+ "step": 96000
1161
+ },
1162
+ {
1163
+ "epoch": 1.85,
1164
+ "learning_rate": 8.23327151908515e-06,
1165
+ "loss": 1.6207,
1166
+ "step": 96500
1167
+ },
1168
+ {
1169
+ "epoch": 1.86,
1170
+ "learning_rate": 8.223632359758926e-06,
1171
+ "loss": 1.6217,
1172
+ "step": 97000
1173
+ },
1174
+ {
1175
+ "epoch": 1.87,
1176
+ "learning_rate": 8.213973883480142e-06,
1177
+ "loss": 1.6192,
1178
+ "step": 97500
1179
+ },
1180
+ {
1181
+ "epoch": 1.87,
1182
+ "learning_rate": 8.20431540720136e-06,
1183
+ "loss": 1.6162,
1184
+ "step": 98000
1185
+ },
1186
+ {
1187
+ "epoch": 1.88,
1188
+ "learning_rate": 8.194656930922578e-06,
1189
+ "loss": 1.6201,
1190
+ "step": 98500
1191
+ },
1192
+ {
1193
+ "epoch": 1.89,
1194
+ "learning_rate": 8.185017771596354e-06,
1195
+ "loss": 1.621,
1196
+ "step": 99000
1197
+ },
1198
+ {
1199
+ "epoch": 1.9,
1200
+ "learning_rate": 8.17535929531757e-06,
1201
+ "loss": 1.6224,
1202
+ "step": 99500
1203
+ },
1204
+ {
1205
+ "epoch": 1.91,
1206
+ "learning_rate": 8.165700819038788e-06,
1207
+ "loss": 1.6233,
1208
+ "step": 100000
1209
+ },
1210
+ {
1211
+ "epoch": 1.92,
1212
+ "learning_rate": 8.156042342760006e-06,
1213
+ "loss": 1.6191,
1214
+ "step": 100500
1215
+ },
1216
+ {
1217
+ "epoch": 1.93,
1218
+ "learning_rate": 8.146383866481224e-06,
1219
+ "loss": 1.6213,
1220
+ "step": 101000
1221
+ },
1222
+ {
1223
+ "epoch": 1.94,
1224
+ "learning_rate": 8.136725390202442e-06,
1225
+ "loss": 1.6252,
1226
+ "step": 101500
1227
+ },
1228
+ {
1229
+ "epoch": 1.95,
1230
+ "learning_rate": 8.127086230876218e-06,
1231
+ "loss": 1.6195,
1232
+ "step": 102000
1233
+ },
1234
+ {
1235
+ "epoch": 1.96,
1236
+ "learning_rate": 8.117427754597436e-06,
1237
+ "loss": 1.6176,
1238
+ "step": 102500
1239
+ },
1240
+ {
1241
+ "epoch": 1.97,
1242
+ "learning_rate": 8.107769278318654e-06,
1243
+ "loss": 1.6205,
1244
+ "step": 103000
1245
+ },
1246
+ {
1247
+ "epoch": 1.98,
1248
+ "learning_rate": 8.098110802039872e-06,
1249
+ "loss": 1.6205,
1250
+ "step": 103500
1251
+ },
1252
+ {
1253
+ "epoch": 1.99,
1254
+ "learning_rate": 8.088452325761088e-06,
1255
+ "loss": 1.6224,
1256
+ "step": 104000
1257
+ },
1258
+ {
1259
+ "epoch": 2.0,
1260
+ "learning_rate": 8.078813166434863e-06,
1261
+ "loss": 1.6204,
1262
+ "step": 104500
1263
+ },
1264
+ {
1265
+ "epoch": 2.01,
1266
+ "learning_rate": 8.06915469015608e-06,
1267
+ "loss": 1.6198,
1268
+ "step": 105000
1269
+ },
1270
+ {
1271
+ "epoch": 2.02,
1272
+ "learning_rate": 8.059496213877298e-06,
1273
+ "loss": 1.6217,
1274
+ "step": 105500
1275
+ },
1276
+ {
1277
+ "epoch": 2.03,
1278
+ "learning_rate": 8.049837737598516e-06,
1279
+ "loss": 1.6196,
1280
+ "step": 106000
1281
+ },
1282
+ {
1283
+ "epoch": 2.04,
1284
+ "learning_rate": 8.040179261319734e-06,
1285
+ "loss": 1.6195,
1286
+ "step": 106500
1287
+ },
1288
+ {
1289
+ "epoch": 2.05,
1290
+ "learning_rate": 8.03054010199351e-06,
1291
+ "loss": 1.6219,
1292
+ "step": 107000
1293
+ },
1294
+ {
1295
+ "epoch": 2.06,
1296
+ "learning_rate": 8.020881625714728e-06,
1297
+ "loss": 1.6221,
1298
+ "step": 107500
1299
+ },
1300
+ {
1301
+ "epoch": 2.07,
1302
+ "learning_rate": 8.011223149435946e-06,
1303
+ "loss": 1.6193,
1304
+ "step": 108000
1305
+ },
1306
+ {
1307
+ "epoch": 2.08,
1308
+ "learning_rate": 8.001564673157164e-06,
1309
+ "loss": 1.6199,
1310
+ "step": 108500
1311
+ },
1312
+ {
1313
+ "epoch": 2.09,
1314
+ "learning_rate": 7.99190619687838e-06,
1315
+ "loss": 1.622,
1316
+ "step": 109000
1317
+ },
1318
+ {
1319
+ "epoch": 2.09,
1320
+ "learning_rate": 7.982267037552157e-06,
1321
+ "loss": 1.6222,
1322
+ "step": 109500
1323
+ },
1324
+ {
1325
+ "epoch": 2.1,
1326
+ "learning_rate": 7.972608561273375e-06,
1327
+ "loss": 1.6199,
1328
+ "step": 110000
1329
+ },
1330
+ {
1331
+ "epoch": 2.11,
1332
+ "learning_rate": 7.962950084994592e-06,
1333
+ "loss": 1.6212,
1334
+ "step": 110500
1335
+ },
1336
+ {
1337
+ "epoch": 2.12,
1338
+ "learning_rate": 7.95329160871581e-06,
1339
+ "loss": 1.6177,
1340
+ "step": 111000
1341
+ },
1342
+ {
1343
+ "epoch": 2.13,
1344
+ "learning_rate": 7.943633132437028e-06,
1345
+ "loss": 1.6214,
1346
+ "step": 111500
1347
+ },
1348
+ {
1349
+ "epoch": 2.14,
1350
+ "learning_rate": 7.933974656158245e-06,
1351
+ "loss": 1.6179,
1352
+ "step": 112000
1353
+ },
1354
+ {
1355
+ "epoch": 2.15,
1356
+ "learning_rate": 7.92433549683202e-06,
1357
+ "loss": 1.6199,
1358
+ "step": 112500
1359
+ },
1360
+ {
1361
+ "epoch": 2.16,
1362
+ "learning_rate": 7.914677020553239e-06,
1363
+ "loss": 1.6212,
1364
+ "step": 113000
1365
+ },
1366
+ {
1367
+ "epoch": 2.17,
1368
+ "learning_rate": 7.905018544274455e-06,
1369
+ "loss": 1.6208,
1370
+ "step": 113500
1371
+ },
1372
+ {
1373
+ "epoch": 2.18,
1374
+ "learning_rate": 7.895360067995673e-06,
1375
+ "loss": 1.6182,
1376
+ "step": 114000
1377
+ },
1378
+ {
1379
+ "epoch": 2.19,
1380
+ "learning_rate": 7.885720908669449e-06,
1381
+ "loss": 1.6159,
1382
+ "step": 114500
1383
+ },
1384
+ {
1385
+ "epoch": 2.2,
1386
+ "learning_rate": 7.876062432390667e-06,
1387
+ "loss": 1.6184,
1388
+ "step": 115000
1389
+ },
1390
+ {
1391
+ "epoch": 2.21,
1392
+ "learning_rate": 7.866403956111885e-06,
1393
+ "loss": 1.6225,
1394
+ "step": 115500
1395
+ },
1396
+ {
1397
+ "epoch": 2.22,
1398
+ "learning_rate": 7.856745479833103e-06,
1399
+ "loss": 1.6181,
1400
+ "step": 116000
1401
+ },
1402
+ {
1403
+ "epoch": 2.23,
1404
+ "learning_rate": 7.84708700355432e-06,
1405
+ "loss": 1.6179,
1406
+ "step": 116500
1407
+ },
1408
+ {
1409
+ "epoch": 2.24,
1410
+ "learning_rate": 7.837428527275539e-06,
1411
+ "loss": 1.6224,
1412
+ "step": 117000
1413
+ },
1414
+ {
1415
+ "epoch": 2.25,
1416
+ "learning_rate": 7.827789367949313e-06,
1417
+ "loss": 1.617,
1418
+ "step": 117500
1419
+ },
1420
+ {
1421
+ "epoch": 2.26,
1422
+ "learning_rate": 7.818130891670531e-06,
1423
+ "loss": 1.6193,
1424
+ "step": 118000
1425
+ },
1426
+ {
1427
+ "epoch": 2.27,
1428
+ "learning_rate": 7.808472415391749e-06,
1429
+ "loss": 1.617,
1430
+ "step": 118500
1431
+ },
1432
+ {
1433
+ "epoch": 2.28,
1434
+ "learning_rate": 7.798813939112965e-06,
1435
+ "loss": 1.618,
1436
+ "step": 119000
1437
+ },
1438
+ {
1439
+ "epoch": 2.29,
1440
+ "learning_rate": 7.789174779786741e-06,
1441
+ "loss": 1.6211,
1442
+ "step": 119500
1443
+ },
1444
+ {
1445
+ "epoch": 2.3,
1446
+ "learning_rate": 7.779516303507959e-06,
1447
+ "loss": 1.6171,
1448
+ "step": 120000
1449
+ },
1450
+ {
1451
+ "epoch": 2.31,
1452
+ "learning_rate": 7.769857827229177e-06,
1453
+ "loss": 1.619,
1454
+ "step": 120500
1455
+ },
1456
+ {
1457
+ "epoch": 2.31,
1458
+ "learning_rate": 7.760199350950395e-06,
1459
+ "loss": 1.6201,
1460
+ "step": 121000
1461
+ },
1462
+ {
1463
+ "epoch": 2.32,
1464
+ "learning_rate": 7.75056019162417e-06,
1465
+ "loss": 1.6184,
1466
+ "step": 121500
1467
+ },
1468
+ {
1469
+ "epoch": 2.33,
1470
+ "learning_rate": 7.740901715345387e-06,
1471
+ "loss": 1.6187,
1472
+ "step": 122000
1473
+ },
1474
+ {
1475
+ "epoch": 2.34,
1476
+ "learning_rate": 7.731243239066605e-06,
1477
+ "loss": 1.6174,
1478
+ "step": 122500
1479
+ },
1480
+ {
1481
+ "epoch": 2.35,
1482
+ "learning_rate": 7.721584762787823e-06,
1483
+ "loss": 1.6184,
1484
+ "step": 123000
1485
+ },
1486
+ {
1487
+ "epoch": 2.36,
1488
+ "learning_rate": 7.711926286509041e-06,
1489
+ "loss": 1.617,
1490
+ "step": 123500
1491
+ },
1492
+ {
1493
+ "epoch": 2.37,
1494
+ "learning_rate": 7.702287127182815e-06,
1495
+ "loss": 1.6202,
1496
+ "step": 124000
1497
+ },
1498
+ {
1499
+ "epoch": 2.38,
1500
+ "learning_rate": 7.692628650904033e-06,
1501
+ "loss": 1.6212,
1502
+ "step": 124500
1503
+ },
1504
+ {
1505
+ "epoch": 2.39,
1506
+ "learning_rate": 7.682970174625251e-06,
1507
+ "loss": 1.6175,
1508
+ "step": 125000
1509
+ },
1510
+ {
1511
+ "epoch": 2.4,
1512
+ "learning_rate": 7.67331169834647e-06,
1513
+ "loss": 1.6202,
1514
+ "step": 125500
1515
+ },
1516
+ {
1517
+ "epoch": 2.41,
1518
+ "learning_rate": 7.663653222067687e-06,
1519
+ "loss": 1.6202,
1520
+ "step": 126000
1521
+ },
1522
+ {
1523
+ "epoch": 2.42,
1524
+ "learning_rate": 7.654014062741463e-06,
1525
+ "loss": 1.6184,
1526
+ "step": 126500
1527
+ },
1528
+ {
1529
+ "epoch": 2.43,
1530
+ "learning_rate": 7.644355586462681e-06,
1531
+ "loss": 1.617,
1532
+ "step": 127000
1533
+ },
1534
+ {
1535
+ "epoch": 2.44,
1536
+ "learning_rate": 7.634697110183899e-06,
1537
+ "loss": 1.6186,
1538
+ "step": 127500
1539
+ },
1540
+ {
1541
+ "epoch": 2.45,
1542
+ "learning_rate": 7.625038633905115e-06,
1543
+ "loss": 1.6184,
1544
+ "step": 128000
1545
+ },
1546
+ {
1547
+ "epoch": 2.46,
1548
+ "learning_rate": 7.615399474578891e-06,
1549
+ "loss": 1.6176,
1550
+ "step": 128500
1551
+ },
1552
+ {
1553
+ "epoch": 2.47,
1554
+ "learning_rate": 7.605740998300109e-06,
1555
+ "loss": 1.6201,
1556
+ "step": 129000
1557
+ },
1558
+ {
1559
+ "epoch": 2.48,
1560
+ "learning_rate": 7.5960825220213265e-06,
1561
+ "loss": 1.619,
1562
+ "step": 129500
1563
+ },
1564
+ {
1565
+ "epoch": 2.49,
1566
+ "learning_rate": 7.5864240457425444e-06,
1567
+ "loss": 1.6237,
1568
+ "step": 130000
1569
+ },
1570
+ {
1571
+ "epoch": 2.5,
1572
+ "learning_rate": 7.576765569463762e-06,
1573
+ "loss": 1.6201,
1574
+ "step": 130500
1575
+ },
1576
+ {
1577
+ "epoch": 2.51,
1578
+ "learning_rate": 7.56710709318498e-06,
1579
+ "loss": 1.6199,
1580
+ "step": 131000
1581
+ },
1582
+ {
1583
+ "epoch": 2.52,
1584
+ "learning_rate": 7.5574679338587556e-06,
1585
+ "loss": 1.6186,
1586
+ "step": 131500
1587
+ },
1588
+ {
1589
+ "epoch": 2.53,
1590
+ "learning_rate": 7.5478094575799735e-06,
1591
+ "loss": 1.6196,
1592
+ "step": 132000
1593
+ },
1594
+ {
1595
+ "epoch": 2.54,
1596
+ "learning_rate": 7.538150981301191e-06,
1597
+ "loss": 1.6209,
1598
+ "step": 132500
1599
+ },
1600
+ {
1601
+ "epoch": 2.54,
1602
+ "learning_rate": 7.528492505022408e-06,
1603
+ "loss": 1.618,
1604
+ "step": 133000
1605
+ },
1606
+ {
1607
+ "epoch": 2.55,
1608
+ "learning_rate": 7.5188340287436256e-06,
1609
+ "loss": 1.616,
1610
+ "step": 133500
1611
+ },
1612
+ {
1613
+ "epoch": 2.56,
1614
+ "learning_rate": 7.5091755524648435e-06,
1615
+ "loss": 1.6182,
1616
+ "step": 134000
1617
+ },
1618
+ {
1619
+ "epoch": 2.57,
1620
+ "learning_rate": 7.499536393138619e-06,
1621
+ "loss": 1.6191,
1622
+ "step": 134500
1623
+ },
1624
+ {
1625
+ "epoch": 2.58,
1626
+ "learning_rate": 7.489877916859837e-06,
1627
+ "loss": 1.6191,
1628
+ "step": 135000
1629
+ },
1630
+ {
1631
+ "epoch": 2.59,
1632
+ "learning_rate": 7.480219440581055e-06,
1633
+ "loss": 1.6152,
1634
+ "step": 135500
1635
+ },
1636
+ {
1637
+ "epoch": 2.6,
1638
+ "learning_rate": 7.4705609643022726e-06,
1639
+ "loss": 1.6199,
1640
+ "step": 136000
1641
+ },
1642
+ {
1643
+ "epoch": 2.61,
1644
+ "learning_rate": 7.460921804976048e-06,
1645
+ "loss": 1.6196,
1646
+ "step": 136500
1647
+ },
1648
+ {
1649
+ "epoch": 2.62,
1650
+ "learning_rate": 7.451263328697266e-06,
1651
+ "loss": 1.619,
1652
+ "step": 137000
1653
+ },
1654
+ {
1655
+ "epoch": 2.63,
1656
+ "learning_rate": 7.441604852418483e-06,
1657
+ "loss": 1.6162,
1658
+ "step": 137500
1659
+ },
1660
+ {
1661
+ "epoch": 2.64,
1662
+ "learning_rate": 7.431946376139701e-06,
1663
+ "loss": 1.617,
1664
+ "step": 138000
1665
+ },
1666
+ {
1667
+ "epoch": 2.65,
1668
+ "learning_rate": 7.422287899860918e-06,
1669
+ "loss": 1.6192,
1670
+ "step": 138500
1671
+ },
1672
+ {
1673
+ "epoch": 2.66,
1674
+ "learning_rate": 7.412648740534694e-06,
1675
+ "loss": 1.6184,
1676
+ "step": 139000
1677
+ },
1678
+ {
1679
+ "epoch": 2.67,
1680
+ "learning_rate": 7.402990264255912e-06,
1681
+ "loss": 1.6161,
1682
+ "step": 139500
1683
+ },
1684
+ {
1685
+ "epoch": 2.68,
1686
+ "learning_rate": 7.393331787977129e-06,
1687
+ "loss": 1.6179,
1688
+ "step": 140000
1689
+ },
1690
+ {
1691
+ "epoch": 2.69,
1692
+ "learning_rate": 7.383673311698347e-06,
1693
+ "loss": 1.616,
1694
+ "step": 140500
1695
+ },
1696
+ {
1697
+ "epoch": 2.7,
1698
+ "learning_rate": 7.374034152372123e-06,
1699
+ "loss": 1.6199,
1700
+ "step": 141000
1701
+ },
1702
+ {
1703
+ "epoch": 2.71,
1704
+ "learning_rate": 7.36437567609334e-06,
1705
+ "loss": 1.621,
1706
+ "step": 141500
1707
+ },
1708
+ {
1709
+ "epoch": 2.72,
1710
+ "learning_rate": 7.354717199814558e-06,
1711
+ "loss": 1.6177,
1712
+ "step": 142000
1713
+ },
1714
+ {
1715
+ "epoch": 2.73,
1716
+ "learning_rate": 7.345058723535775e-06,
1717
+ "loss": 1.619,
1718
+ "step": 142500
1719
+ },
1720
+ {
1721
+ "epoch": 2.74,
1722
+ "learning_rate": 7.33541956420955e-06,
1723
+ "loss": 1.6187,
1724
+ "step": 143000
1725
+ },
1726
+ {
1727
+ "epoch": 2.75,
1728
+ "learning_rate": 7.325761087930768e-06,
1729
+ "loss": 1.6187,
1730
+ "step": 143500
1731
+ },
1732
+ {
1733
+ "epoch": 2.76,
1734
+ "learning_rate": 7.316102611651986e-06,
1735
+ "loss": 1.619,
1736
+ "step": 144000
1737
+ },
1738
+ {
1739
+ "epoch": 2.76,
1740
+ "learning_rate": 7.306444135373204e-06,
1741
+ "loss": 1.6207,
1742
+ "step": 144500
1743
+ },
1744
+ {
1745
+ "epoch": 2.77,
1746
+ "learning_rate": 7.296785659094422e-06,
1747
+ "loss": 1.6177,
1748
+ "step": 145000
1749
+ },
1750
+ {
1751
+ "epoch": 2.78,
1752
+ "learning_rate": 7.287146499768197e-06,
1753
+ "loss": 1.6186,
1754
+ "step": 145500
1755
+ },
1756
+ {
1757
+ "epoch": 2.79,
1758
+ "learning_rate": 7.277488023489415e-06,
1759
+ "loss": 1.6175,
1760
+ "step": 146000
1761
+ },
1762
+ {
1763
+ "epoch": 2.8,
1764
+ "learning_rate": 7.267829547210633e-06,
1765
+ "loss": 1.6151,
1766
+ "step": 146500
1767
+ },
1768
+ {
1769
+ "epoch": 2.81,
1770
+ "learning_rate": 7.258171070931851e-06,
1771
+ "loss": 1.6145,
1772
+ "step": 147000
1773
+ },
1774
+ {
1775
+ "epoch": 2.82,
1776
+ "learning_rate": 7.2485319116056256e-06,
1777
+ "loss": 1.6206,
1778
+ "step": 147500
1779
+ },
1780
+ {
1781
+ "epoch": 2.83,
1782
+ "learning_rate": 7.2388734353268435e-06,
1783
+ "loss": 1.6189,
1784
+ "step": 148000
1785
+ },
1786
+ {
1787
+ "epoch": 2.84,
1788
+ "learning_rate": 7.2292149590480606e-06,
1789
+ "loss": 1.6163,
1790
+ "step": 148500
1791
+ },
1792
+ {
1793
+ "epoch": 2.85,
1794
+ "learning_rate": 7.2195564827692785e-06,
1795
+ "loss": 1.617,
1796
+ "step": 149000
1797
+ },
1798
+ {
1799
+ "epoch": 2.86,
1800
+ "learning_rate": 7.209917323443054e-06,
1801
+ "loss": 1.6202,
1802
+ "step": 149500
1803
+ },
1804
+ {
1805
+ "epoch": 2.87,
1806
+ "learning_rate": 7.200258847164272e-06,
1807
+ "loss": 1.6177,
1808
+ "step": 150000
1809
+ },
1810
+ {
1811
+ "epoch": 2.88,
1812
+ "learning_rate": 7.19060037088549e-06,
1813
+ "loss": 1.6173,
1814
+ "step": 150500
1815
+ },
1816
+ {
1817
+ "epoch": 2.89,
1818
+ "learning_rate": 7.1809418946067075e-06,
1819
+ "loss": 1.6186,
1820
+ "step": 151000
1821
+ },
1822
+ {
1823
+ "epoch": 2.9,
1824
+ "learning_rate": 7.1712834183279255e-06,
1825
+ "loss": 1.619,
1826
+ "step": 151500
1827
+ },
1828
+ {
1829
+ "epoch": 2.91,
1830
+ "learning_rate": 7.161644259001701e-06,
1831
+ "loss": 1.6189,
1832
+ "step": 152000
1833
+ },
1834
+ {
1835
+ "epoch": 2.92,
1836
+ "learning_rate": 7.151985782722918e-06,
1837
+ "loss": 1.6184,
1838
+ "step": 152500
1839
+ },
1840
+ {
1841
+ "epoch": 2.93,
1842
+ "learning_rate": 7.142327306444136e-06,
1843
+ "loss": 1.6178,
1844
+ "step": 153000
1845
+ },
1846
+ {
1847
+ "epoch": 2.94,
1848
+ "learning_rate": 7.132668830165354e-06,
1849
+ "loss": 1.6174,
1850
+ "step": 153500
1851
+ },
1852
+ {
1853
+ "epoch": 2.95,
1854
+ "learning_rate": 7.123029670839129e-06,
1855
+ "loss": 1.6175,
1856
+ "step": 154000
1857
+ },
1858
+ {
1859
+ "epoch": 2.96,
1860
+ "learning_rate": 7.113371194560347e-06,
1861
+ "loss": 1.6187,
1862
+ "step": 154500
1863
+ },
1864
+ {
1865
+ "epoch": 2.97,
1866
+ "learning_rate": 7.103712718281565e-06,
1867
+ "loss": 1.6213,
1868
+ "step": 155000
1869
+ },
1870
+ {
1871
+ "epoch": 2.98,
1872
+ "learning_rate": 7.094054242002783e-06,
1873
+ "loss": 1.6151,
1874
+ "step": 155500
1875
+ },
1876
+ {
1877
+ "epoch": 2.98,
1878
+ "learning_rate": 7.084415082676558e-06,
1879
+ "loss": 1.6154,
1880
+ "step": 156000
1881
+ },
1882
+ {
1883
+ "epoch": 2.99,
1884
+ "learning_rate": 7.074756606397776e-06,
1885
+ "loss": 1.6157,
1886
+ "step": 156500
1887
+ },
1888
+ {
1889
+ "epoch": 3.0,
1890
+ "learning_rate": 7.065098130118992e-06,
1891
+ "loss": 1.6197,
1892
+ "step": 157000
1893
+ },
1894
+ {
1895
+ "epoch": 3.01,
1896
+ "learning_rate": 7.05543965384021e-06,
1897
+ "loss": 1.6173,
1898
+ "step": 157500
1899
+ },
1900
+ {
1901
+ "epoch": 3.02,
1902
+ "learning_rate": 7.045781177561428e-06,
1903
+ "loss": 1.6182,
1904
+ "step": 158000
1905
+ },
1906
+ {
1907
+ "epoch": 3.03,
1908
+ "learning_rate": 7.036122701282646e-06,
1909
+ "loss": 1.6156,
1910
+ "step": 158500
1911
+ },
1912
+ {
1913
+ "epoch": 3.04,
1914
+ "learning_rate": 7.026483541956421e-06,
1915
+ "loss": 1.6168,
1916
+ "step": 159000
1917
+ },
1918
+ {
1919
+ "epoch": 3.05,
1920
+ "learning_rate": 7.016825065677639e-06,
1921
+ "loss": 1.6189,
1922
+ "step": 159500
1923
+ },
1924
+ {
1925
+ "epoch": 3.06,
1926
+ "learning_rate": 7.007166589398857e-06,
1927
+ "loss": 1.6204,
1928
+ "step": 160000
1929
+ },
1930
+ {
1931
+ "epoch": 3.07,
1932
+ "learning_rate": 6.997508113120075e-06,
1933
+ "loss": 1.6139,
1934
+ "step": 160500
1935
+ },
1936
+ {
1937
+ "epoch": 3.08,
1938
+ "learning_rate": 6.98786895379385e-06,
1939
+ "loss": 1.6178,
1940
+ "step": 161000
1941
+ },
1942
+ {
1943
+ "epoch": 3.09,
1944
+ "learning_rate": 6.978210477515068e-06,
1945
+ "loss": 1.6171,
1946
+ "step": 161500
1947
+ },
1948
+ {
1949
+ "epoch": 3.1,
1950
+ "learning_rate": 6.968552001236285e-06,
1951
+ "loss": 1.6155,
1952
+ "step": 162000
1953
+ },
1954
+ {
1955
+ "epoch": 3.11,
1956
+ "learning_rate": 6.958893524957503e-06,
1957
+ "loss": 1.618,
1958
+ "step": 162500
1959
+ },
1960
+ {
1961
+ "epoch": 3.12,
1962
+ "learning_rate": 6.9492543656312785e-06,
1963
+ "loss": 1.6159,
1964
+ "step": 163000
1965
+ },
1966
+ {
1967
+ "epoch": 3.13,
1968
+ "learning_rate": 6.939595889352496e-06,
1969
+ "loss": 1.614,
1970
+ "step": 163500
1971
+ },
1972
+ {
1973
+ "epoch": 3.14,
1974
+ "learning_rate": 6.929937413073714e-06,
1975
+ "loss": 1.617,
1976
+ "step": 164000
1977
+ },
1978
+ {
1979
+ "epoch": 3.15,
1980
+ "learning_rate": 6.920278936794932e-06,
1981
+ "loss": 1.6172,
1982
+ "step": 164500
1983
+ },
1984
+ {
1985
+ "epoch": 3.16,
1986
+ "learning_rate": 6.9106397774687075e-06,
1987
+ "loss": 1.6129,
1988
+ "step": 165000
1989
+ },
1990
+ {
1991
+ "epoch": 3.17,
1992
+ "learning_rate": 6.9009813011899254e-06,
1993
+ "loss": 1.6201,
1994
+ "step": 165500
1995
+ },
1996
+ {
1997
+ "epoch": 3.18,
1998
+ "learning_rate": 6.891322824911143e-06,
1999
+ "loss": 1.6163,
2000
+ "step": 166000
2001
+ },
2002
+ {
2003
+ "epoch": 3.19,
2004
+ "learning_rate": 6.88166434863236e-06,
2005
+ "loss": 1.6196,
2006
+ "step": 166500
2007
+ },
2008
+ {
2009
+ "epoch": 3.2,
2010
+ "learning_rate": 6.8720058723535775e-06,
2011
+ "loss": 1.6174,
2012
+ "step": 167000
2013
+ },
2014
+ {
2015
+ "epoch": 3.2,
2016
+ "learning_rate": 6.862366713027353e-06,
2017
+ "loss": 1.6116,
2018
+ "step": 167500
2019
+ },
2020
+ {
2021
+ "epoch": 3.21,
2022
+ "learning_rate": 6.852708236748571e-06,
2023
+ "loss": 1.61,
2024
+ "step": 168000
2025
+ },
2026
+ {
2027
+ "epoch": 3.22,
2028
+ "learning_rate": 6.843049760469789e-06,
2029
+ "loss": 1.6156,
2030
+ "step": 168500
2031
+ },
2032
+ {
2033
+ "epoch": 3.23,
2034
+ "learning_rate": 6.833391284191007e-06,
2035
+ "loss": 1.6167,
2036
+ "step": 169000
2037
+ },
2038
+ {
2039
+ "epoch": 3.24,
2040
+ "learning_rate": 6.8237328079122245e-06,
2041
+ "loss": 1.6116,
2042
+ "step": 169500
2043
+ },
2044
+ {
2045
+ "epoch": 3.25,
2046
+ "learning_rate": 6.814093648586e-06,
2047
+ "loss": 1.617,
2048
+ "step": 170000
2049
+ },
2050
+ {
2051
+ "epoch": 3.26,
2052
+ "learning_rate": 6.804435172307218e-06,
2053
+ "loss": 1.6188,
2054
+ "step": 170500
2055
+ },
2056
+ {
2057
+ "epoch": 3.27,
2058
+ "learning_rate": 6.794776696028436e-06,
2059
+ "loss": 1.6116,
2060
+ "step": 171000
2061
+ },
2062
+ {
2063
+ "epoch": 3.28,
2064
+ "learning_rate": 6.785118219749653e-06,
2065
+ "loss": 1.6157,
2066
+ "step": 171500
2067
+ },
2068
+ {
2069
+ "epoch": 3.29,
2070
+ "learning_rate": 6.775479060423428e-06,
2071
+ "loss": 1.6157,
2072
+ "step": 172000
2073
+ },
2074
+ {
2075
+ "epoch": 3.3,
2076
+ "learning_rate": 6.765820584144646e-06,
2077
+ "loss": 1.616,
2078
+ "step": 172500
2079
+ },
2080
+ {
2081
+ "epoch": 3.31,
2082
+ "learning_rate": 6.756162107865864e-06,
2083
+ "loss": 1.6168,
2084
+ "step": 173000
2085
+ },
2086
+ {
2087
+ "epoch": 3.32,
2088
+ "learning_rate": 6.746503631587081e-06,
2089
+ "loss": 1.6169,
2090
+ "step": 173500
2091
+ },
2092
+ {
2093
+ "epoch": 3.33,
2094
+ "learning_rate": 6.736845155308299e-06,
2095
+ "loss": 1.6156,
2096
+ "step": 174000
2097
+ },
2098
+ {
2099
+ "epoch": 3.34,
2100
+ "learning_rate": 6.727186679029517e-06,
2101
+ "loss": 1.6176,
2102
+ "step": 174500
2103
+ },
2104
+ {
2105
+ "epoch": 3.35,
2106
+ "learning_rate": 6.717547519703292e-06,
2107
+ "loss": 1.6182,
2108
+ "step": 175000
2109
+ },
2110
+ {
2111
+ "epoch": 3.36,
2112
+ "learning_rate": 6.70788904342451e-06,
2113
+ "loss": 1.6145,
2114
+ "step": 175500
2115
+ },
2116
+ {
2117
+ "epoch": 3.37,
2118
+ "learning_rate": 6.698230567145728e-06,
2119
+ "loss": 1.6162,
2120
+ "step": 176000
2121
+ },
2122
+ {
2123
+ "epoch": 3.38,
2124
+ "learning_rate": 6.688572090866945e-06,
2125
+ "loss": 1.6162,
2126
+ "step": 176500
2127
+ },
2128
+ {
2129
+ "epoch": 3.39,
2130
+ "learning_rate": 6.678913614588163e-06,
2131
+ "loss": 1.6166,
2132
+ "step": 177000
2133
+ },
2134
+ {
2135
+ "epoch": 3.4,
2136
+ "learning_rate": 6.669274455261938e-06,
2137
+ "loss": 1.6141,
2138
+ "step": 177500
2139
+ },
2140
+ {
2141
+ "epoch": 3.41,
2142
+ "learning_rate": 6.659615978983156e-06,
2143
+ "loss": 1.6158,
2144
+ "step": 178000
2145
+ },
2146
+ {
2147
+ "epoch": 3.42,
2148
+ "learning_rate": 6.649957502704374e-06,
2149
+ "loss": 1.6186,
2150
+ "step": 178500
2151
+ },
2152
+ {
2153
+ "epoch": 3.42,
2154
+ "learning_rate": 6.640299026425592e-06,
2155
+ "loss": 1.6137,
2156
+ "step": 179000
2157
+ },
2158
+ {
2159
+ "epoch": 3.43,
2160
+ "learning_rate": 6.63064055014681e-06,
2161
+ "loss": 1.6178,
2162
+ "step": 179500
2163
+ },
2164
+ {
2165
+ "epoch": 3.44,
2166
+ "learning_rate": 6.620982073868028e-06,
2167
+ "loss": 1.62,
2168
+ "step": 180000
2169
+ },
2170
+ {
2171
+ "epoch": 3.45,
2172
+ "learning_rate": 6.611342914541803e-06,
2173
+ "loss": 1.6134,
2174
+ "step": 180500
2175
+ },
2176
+ {
2177
+ "epoch": 3.46,
2178
+ "learning_rate": 6.601684438263019e-06,
2179
+ "loss": 1.6139,
2180
+ "step": 181000
2181
+ },
2182
+ {
2183
+ "epoch": 3.47,
2184
+ "learning_rate": 6.592025961984237e-06,
2185
+ "loss": 1.6166,
2186
+ "step": 181500
2187
+ },
2188
+ {
2189
+ "epoch": 3.48,
2190
+ "learning_rate": 6.582367485705455e-06,
2191
+ "loss": 1.616,
2192
+ "step": 182000
2193
+ },
2194
+ {
2195
+ "epoch": 3.49,
2196
+ "learning_rate": 6.5727283263792304e-06,
2197
+ "loss": 1.6191,
2198
+ "step": 182500
2199
+ },
2200
+ {
2201
+ "epoch": 3.5,
2202
+ "learning_rate": 6.563069850100448e-06,
2203
+ "loss": 1.6187,
2204
+ "step": 183000
2205
+ },
2206
+ {
2207
+ "epoch": 3.51,
2208
+ "learning_rate": 6.553411373821666e-06,
2209
+ "loss": 1.6152,
2210
+ "step": 183500
2211
+ },
2212
+ {
2213
+ "epoch": 3.52,
2214
+ "learning_rate": 6.543752897542884e-06,
2215
+ "loss": 1.6162,
2216
+ "step": 184000
2217
+ },
2218
+ {
2219
+ "epoch": 3.53,
2220
+ "learning_rate": 6.5341137382166595e-06,
2221
+ "loss": 1.6145,
2222
+ "step": 184500
2223
+ },
2224
+ {
2225
+ "epoch": 3.54,
2226
+ "learning_rate": 6.524455261937877e-06,
2227
+ "loss": 1.6189,
2228
+ "step": 185000
2229
+ },
2230
+ {
2231
+ "epoch": 3.55,
2232
+ "learning_rate": 6.514796785659095e-06,
2233
+ "loss": 1.6163,
2234
+ "step": 185500
2235
+ },
2236
+ {
2237
+ "epoch": 3.56,
2238
+ "learning_rate": 6.5051383093803124e-06,
2239
+ "loss": 1.6172,
2240
+ "step": 186000
2241
+ },
2242
+ {
2243
+ "epoch": 3.57,
2244
+ "learning_rate": 6.49547983310153e-06,
2245
+ "loss": 1.617,
2246
+ "step": 186500
2247
+ },
2248
+ {
2249
+ "epoch": 3.58,
2250
+ "learning_rate": 6.485840673775306e-06,
2251
+ "loss": 1.6148,
2252
+ "step": 187000
2253
+ },
2254
+ {
2255
+ "epoch": 3.59,
2256
+ "learning_rate": 6.4761821974965236e-06,
2257
+ "loss": 1.617,
2258
+ "step": 187500
2259
+ },
2260
+ {
2261
+ "epoch": 3.6,
2262
+ "learning_rate": 6.4665237212177415e-06,
2263
+ "loss": 1.6119,
2264
+ "step": 188000
2265
+ },
2266
+ {
2267
+ "epoch": 3.61,
2268
+ "learning_rate": 6.456865244938959e-06,
2269
+ "loss": 1.6143,
2270
+ "step": 188500
2271
+ },
2272
+ {
2273
+ "epoch": 3.62,
2274
+ "learning_rate": 6.447226085612735e-06,
2275
+ "loss": 1.6165,
2276
+ "step": 189000
2277
+ },
2278
+ {
2279
+ "epoch": 3.63,
2280
+ "learning_rate": 6.43758692628651e-06,
2281
+ "loss": 1.6142,
2282
+ "step": 189500
2283
+ },
2284
+ {
2285
+ "epoch": 3.64,
2286
+ "learning_rate": 6.427928450007728e-06,
2287
+ "loss": 1.6163,
2288
+ "step": 190000
2289
+ },
2290
+ {
2291
+ "epoch": 3.64,
2292
+ "learning_rate": 6.418269973728946e-06,
2293
+ "loss": 1.6151,
2294
+ "step": 190500
2295
+ },
2296
+ {
2297
+ "epoch": 3.65,
2298
+ "learning_rate": 6.408611497450162e-06,
2299
+ "loss": 1.6152,
2300
+ "step": 191000
2301
+ },
2302
+ {
2303
+ "epoch": 3.66,
2304
+ "learning_rate": 6.39895302117138e-06,
2305
+ "loss": 1.6158,
2306
+ "step": 191500
2307
+ },
2308
+ {
2309
+ "epoch": 3.67,
2310
+ "learning_rate": 6.389294544892598e-06,
2311
+ "loss": 1.6178,
2312
+ "step": 192000
2313
+ },
2314
+ {
2315
+ "epoch": 3.68,
2316
+ "learning_rate": 6.379636068613816e-06,
2317
+ "loss": 1.6151,
2318
+ "step": 192500
2319
+ },
2320
+ {
2321
+ "epoch": 3.69,
2322
+ "learning_rate": 6.369977592335034e-06,
2323
+ "loss": 1.6157,
2324
+ "step": 193000
2325
+ },
2326
+ {
2327
+ "epoch": 3.7,
2328
+ "learning_rate": 6.360338433008809e-06,
2329
+ "loss": 1.6147,
2330
+ "step": 193500
2331
+ },
2332
+ {
2333
+ "epoch": 3.71,
2334
+ "learning_rate": 6.350679956730027e-06,
2335
+ "loss": 1.62,
2336
+ "step": 194000
2337
+ },
2338
+ {
2339
+ "epoch": 3.72,
2340
+ "learning_rate": 6.341021480451245e-06,
2341
+ "loss": 1.6171,
2342
+ "step": 194500
2343
+ },
2344
+ {
2345
+ "epoch": 3.73,
2346
+ "learning_rate": 6.331363004172463e-06,
2347
+ "loss": 1.613,
2348
+ "step": 195000
2349
+ },
2350
+ {
2351
+ "epoch": 3.74,
2352
+ "learning_rate": 6.321723844846237e-06,
2353
+ "loss": 1.6155,
2354
+ "step": 195500
2355
+ },
2356
+ {
2357
+ "epoch": 3.75,
2358
+ "learning_rate": 6.312065368567455e-06,
2359
+ "loss": 1.6122,
2360
+ "step": 196000
2361
+ },
2362
+ {
2363
+ "epoch": 3.76,
2364
+ "learning_rate": 6.302406892288673e-06,
2365
+ "loss": 1.6157,
2366
+ "step": 196500
2367
+ },
2368
+ {
2369
+ "epoch": 3.77,
2370
+ "learning_rate": 6.292748416009891e-06,
2371
+ "loss": 1.6173,
2372
+ "step": 197000
2373
+ },
2374
+ {
2375
+ "epoch": 3.78,
2376
+ "learning_rate": 6.283089939731108e-06,
2377
+ "loss": 1.6127,
2378
+ "step": 197500
2379
+ },
2380
+ {
2381
+ "epoch": 3.79,
2382
+ "learning_rate": 6.273450780404884e-06,
2383
+ "loss": 1.6149,
2384
+ "step": 198000
2385
+ },
2386
+ {
2387
+ "epoch": 3.8,
2388
+ "learning_rate": 6.263792304126102e-06,
2389
+ "loss": 1.6183,
2390
+ "step": 198500
2391
+ },
2392
+ {
2393
+ "epoch": 3.81,
2394
+ "learning_rate": 6.254133827847319e-06,
2395
+ "loss": 1.6152,
2396
+ "step": 199000
2397
+ },
2398
+ {
2399
+ "epoch": 3.82,
2400
+ "learning_rate": 6.244475351568537e-06,
2401
+ "loss": 1.6135,
2402
+ "step": 199500
2403
+ },
2404
+ {
2405
+ "epoch": 3.83,
2406
+ "learning_rate": 6.234816875289755e-06,
2407
+ "loss": 1.6153,
2408
+ "step": 200000
2409
+ },
2410
+ {
2411
+ "epoch": 3.84,
2412
+ "learning_rate": 6.225158399010972e-06,
2413
+ "loss": 1.6148,
2414
+ "step": 200500
2415
+ },
2416
+ {
2417
+ "epoch": 3.85,
2418
+ "learning_rate": 6.215519239684747e-06,
2419
+ "loss": 1.6136,
2420
+ "step": 201000
2421
+ },
2422
+ {
2423
+ "epoch": 3.86,
2424
+ "learning_rate": 6.205860763405965e-06,
2425
+ "loss": 1.616,
2426
+ "step": 201500
2427
+ },
2428
+ {
2429
+ "epoch": 3.86,
2430
+ "learning_rate": 6.196202287127183e-06,
2431
+ "loss": 1.6146,
2432
+ "step": 202000
2433
+ },
2434
+ {
2435
+ "epoch": 3.87,
2436
+ "learning_rate": 6.186543810848401e-06,
2437
+ "loss": 1.6133,
2438
+ "step": 202500
2439
+ },
2440
+ {
2441
+ "epoch": 3.88,
2442
+ "learning_rate": 6.176885334569619e-06,
2443
+ "loss": 1.6134,
2444
+ "step": 203000
2445
+ },
2446
+ {
2447
+ "epoch": 3.89,
2448
+ "learning_rate": 6.167246175243394e-06,
2449
+ "loss": 1.6138,
2450
+ "step": 203500
2451
+ },
2452
+ {
2453
+ "epoch": 3.9,
2454
+ "learning_rate": 6.157587698964612e-06,
2455
+ "loss": 1.6158,
2456
+ "step": 204000
2457
+ },
2458
+ {
2459
+ "epoch": 3.91,
2460
+ "learning_rate": 6.14792922268583e-06,
2461
+ "loss": 1.6148,
2462
+ "step": 204500
2463
+ },
2464
+ {
2465
+ "epoch": 3.92,
2466
+ "learning_rate": 6.1382707464070465e-06,
2467
+ "loss": 1.6155,
2468
+ "step": 205000
2469
+ },
2470
+ {
2471
+ "epoch": 3.93,
2472
+ "learning_rate": 6.128631587080822e-06,
2473
+ "loss": 1.6139,
2474
+ "step": 205500
2475
+ },
2476
+ {
2477
+ "epoch": 3.94,
2478
+ "learning_rate": 6.11897311080204e-06,
2479
+ "loss": 1.6175,
2480
+ "step": 206000
2481
+ },
2482
+ {
2483
+ "epoch": 3.95,
2484
+ "learning_rate": 6.109314634523258e-06,
2485
+ "loss": 1.6148,
2486
+ "step": 206500
2487
+ },
2488
+ {
2489
+ "epoch": 3.96,
2490
+ "learning_rate": 6.0996561582444755e-06,
2491
+ "loss": 1.6117,
2492
+ "step": 207000
2493
+ },
2494
+ {
2495
+ "epoch": 3.97,
2496
+ "learning_rate": 6.090016998918251e-06,
2497
+ "loss": 1.617,
2498
+ "step": 207500
2499
+ },
2500
+ {
2501
+ "epoch": 3.98,
2502
+ "learning_rate": 6.080358522639469e-06,
2503
+ "loss": 1.615,
2504
+ "step": 208000
2505
+ },
2506
+ {
2507
+ "epoch": 3.99,
2508
+ "learning_rate": 6.070700046360687e-06,
2509
+ "loss": 1.615,
2510
+ "step": 208500
2511
+ },
2512
+ {
2513
+ "epoch": 4.0,
2514
+ "learning_rate": 6.061041570081905e-06,
2515
+ "loss": 1.6126,
2516
+ "step": 209000
2517
+ },
2518
+ {
2519
+ "epoch": 4.01,
2520
+ "learning_rate": 6.0513830938031225e-06,
2521
+ "loss": 1.6145,
2522
+ "step": 209500
2523
+ },
2524
+ {
2525
+ "epoch": 4.02,
2526
+ "learning_rate": 6.04172461752434e-06,
2527
+ "loss": 1.6145,
2528
+ "step": 210000
2529
+ },
2530
+ {
2531
+ "epoch": 4.03,
2532
+ "learning_rate": 6.032085458198115e-06,
2533
+ "loss": 1.6169,
2534
+ "step": 210500
2535
+ },
2536
+ {
2537
+ "epoch": 4.04,
2538
+ "learning_rate": 6.022426981919333e-06,
2539
+ "loss": 1.6144,
2540
+ "step": 211000
2541
+ },
2542
+ {
2543
+ "epoch": 4.05,
2544
+ "learning_rate": 6.012768505640551e-06,
2545
+ "loss": 1.6122,
2546
+ "step": 211500
2547
+ },
2548
+ {
2549
+ "epoch": 4.06,
2550
+ "learning_rate": 6.003110029361769e-06,
2551
+ "loss": 1.6162,
2552
+ "step": 212000
2553
+ },
2554
+ {
2555
+ "epoch": 4.07,
2556
+ "learning_rate": 5.9934515530829866e-06,
2557
+ "loss": 1.6116,
2558
+ "step": 212500
2559
+ },
2560
+ {
2561
+ "epoch": 4.08,
2562
+ "learning_rate": 5.983812393756762e-06,
2563
+ "loss": 1.6115,
2564
+ "step": 213000
2565
+ },
2566
+ {
2567
+ "epoch": 4.08,
2568
+ "learning_rate": 5.97415391747798e-06,
2569
+ "loss": 1.6114,
2570
+ "step": 213500
2571
+ },
2572
+ {
2573
+ "epoch": 4.09,
2574
+ "learning_rate": 5.964495441199198e-06,
2575
+ "loss": 1.6111,
2576
+ "step": 214000
2577
+ },
2578
+ {
2579
+ "epoch": 4.1,
2580
+ "learning_rate": 5.954836964920415e-06,
2581
+ "loss": 1.6119,
2582
+ "step": 214500
2583
+ },
2584
+ {
2585
+ "epoch": 4.11,
2586
+ "learning_rate": 5.945197805594189e-06,
2587
+ "loss": 1.6192,
2588
+ "step": 215000
2589
+ },
2590
+ {
2591
+ "epoch": 4.12,
2592
+ "learning_rate": 5.935539329315407e-06,
2593
+ "loss": 1.6117,
2594
+ "step": 215500
2595
+ },
2596
+ {
2597
+ "epoch": 4.13,
2598
+ "learning_rate": 5.925880853036625e-06,
2599
+ "loss": 1.6133,
2600
+ "step": 216000
2601
+ },
2602
+ {
2603
+ "epoch": 4.14,
2604
+ "learning_rate": 5.916222376757843e-06,
2605
+ "loss": 1.6125,
2606
+ "step": 216500
2607
+ },
2608
+ {
2609
+ "epoch": 4.15,
2610
+ "learning_rate": 5.906583217431618e-06,
2611
+ "loss": 1.6106,
2612
+ "step": 217000
2613
+ },
2614
+ {
2615
+ "epoch": 4.16,
2616
+ "learning_rate": 5.896924741152836e-06,
2617
+ "loss": 1.6106,
2618
+ "step": 217500
2619
+ },
2620
+ {
2621
+ "epoch": 4.17,
2622
+ "learning_rate": 5.887266264874054e-06,
2623
+ "loss": 1.6127,
2624
+ "step": 218000
2625
+ },
2626
+ {
2627
+ "epoch": 4.18,
2628
+ "learning_rate": 5.877607788595272e-06,
2629
+ "loss": 1.6112,
2630
+ "step": 218500
2631
+ },
2632
+ {
2633
+ "epoch": 4.19,
2634
+ "learning_rate": 5.86794931231649e-06,
2635
+ "loss": 1.6127,
2636
+ "step": 219000
2637
+ },
2638
+ {
2639
+ "epoch": 4.2,
2640
+ "learning_rate": 5.858310152990265e-06,
2641
+ "loss": 1.6168,
2642
+ "step": 219500
2643
+ },
2644
+ {
2645
+ "epoch": 4.21,
2646
+ "learning_rate": 5.848651676711482e-06,
2647
+ "loss": 1.6077,
2648
+ "step": 220000
2649
+ },
2650
+ {
2651
+ "epoch": 4.22,
2652
+ "learning_rate": 5.8389932004327e-06,
2653
+ "loss": 1.6129,
2654
+ "step": 220500
2655
+ },
2656
+ {
2657
+ "epoch": 4.23,
2658
+ "learning_rate": 5.829334724153918e-06,
2659
+ "loss": 1.6152,
2660
+ "step": 221000
2661
+ },
2662
+ {
2663
+ "epoch": 4.24,
2664
+ "learning_rate": 5.8196955648276934e-06,
2665
+ "loss": 1.6168,
2666
+ "step": 221500
2667
+ },
2668
+ {
2669
+ "epoch": 4.25,
2670
+ "learning_rate": 5.810037088548911e-06,
2671
+ "loss": 1.6144,
2672
+ "step": 222000
2673
+ },
2674
+ {
2675
+ "epoch": 4.26,
2676
+ "learning_rate": 5.800378612270129e-06,
2677
+ "loss": 1.613,
2678
+ "step": 222500
2679
+ },
2680
+ {
2681
+ "epoch": 4.27,
2682
+ "learning_rate": 5.790720135991346e-06,
2683
+ "loss": 1.6119,
2684
+ "step": 223000
2685
+ },
2686
+ {
2687
+ "epoch": 4.28,
2688
+ "learning_rate": 5.781061659712564e-06,
2689
+ "loss": 1.614,
2690
+ "step": 223500
2691
+ },
2692
+ {
2693
+ "epoch": 4.29,
2694
+ "learning_rate": 5.77142250038634e-06,
2695
+ "loss": 1.6163,
2696
+ "step": 224000
2697
+ },
2698
+ {
2699
+ "epoch": 4.3,
2700
+ "learning_rate": 5.761764024107557e-06,
2701
+ "loss": 1.6141,
2702
+ "step": 224500
2703
+ },
2704
+ {
2705
+ "epoch": 4.3,
2706
+ "learning_rate": 5.752105547828775e-06,
2707
+ "loss": 1.6147,
2708
+ "step": 225000
2709
+ },
2710
+ {
2711
+ "epoch": 4.31,
2712
+ "learning_rate": 5.7424470715499925e-06,
2713
+ "loss": 1.6119,
2714
+ "step": 225500
2715
+ },
2716
+ {
2717
+ "epoch": 4.32,
2718
+ "learning_rate": 5.732807912223768e-06,
2719
+ "loss": 1.6122,
2720
+ "step": 226000
2721
+ },
2722
+ {
2723
+ "epoch": 4.33,
2724
+ "learning_rate": 5.723149435944986e-06,
2725
+ "loss": 1.6125,
2726
+ "step": 226500
2727
+ },
2728
+ {
2729
+ "epoch": 4.34,
2730
+ "learning_rate": 5.713490959666204e-06,
2731
+ "loss": 1.6177,
2732
+ "step": 227000
2733
+ },
2734
+ {
2735
+ "epoch": 4.35,
2736
+ "learning_rate": 5.7038324833874216e-06,
2737
+ "loss": 1.6126,
2738
+ "step": 227500
2739
+ },
2740
+ {
2741
+ "epoch": 4.36,
2742
+ "learning_rate": 5.6941740071086395e-06,
2743
+ "loss": 1.6151,
2744
+ "step": 228000
2745
+ },
2746
+ {
2747
+ "epoch": 4.37,
2748
+ "learning_rate": 5.684534847782415e-06,
2749
+ "loss": 1.613,
2750
+ "step": 228500
2751
+ },
2752
+ {
2753
+ "epoch": 4.38,
2754
+ "learning_rate": 5.674876371503633e-06,
2755
+ "loss": 1.6142,
2756
+ "step": 229000
2757
+ },
2758
+ {
2759
+ "epoch": 4.39,
2760
+ "learning_rate": 5.665217895224849e-06,
2761
+ "loss": 1.6158,
2762
+ "step": 229500
2763
+ },
2764
+ {
2765
+ "epoch": 4.4,
2766
+ "learning_rate": 5.655559418946067e-06,
2767
+ "loss": 1.6134,
2768
+ "step": 230000
2769
+ },
2770
+ {
2771
+ "epoch": 4.41,
2772
+ "learning_rate": 5.645900942667285e-06,
2773
+ "loss": 1.6123,
2774
+ "step": 230500
2775
+ },
2776
+ {
2777
+ "epoch": 4.42,
2778
+ "learning_rate": 5.63626178334106e-06,
2779
+ "loss": 1.6152,
2780
+ "step": 231000
2781
+ },
2782
+ {
2783
+ "epoch": 4.43,
2784
+ "learning_rate": 5.626603307062278e-06,
2785
+ "loss": 1.6142,
2786
+ "step": 231500
2787
+ },
2788
+ {
2789
+ "epoch": 4.44,
2790
+ "learning_rate": 5.616944830783496e-06,
2791
+ "loss": 1.6137,
2792
+ "step": 232000
2793
+ },
2794
+ {
2795
+ "epoch": 4.45,
2796
+ "learning_rate": 5.607286354504714e-06,
2797
+ "loss": 1.6111,
2798
+ "step": 232500
2799
+ },
2800
+ {
2801
+ "epoch": 4.46,
2802
+ "learning_rate": 5.597647195178489e-06,
2803
+ "loss": 1.6121,
2804
+ "step": 233000
2805
+ },
2806
+ {
2807
+ "epoch": 4.47,
2808
+ "learning_rate": 5.587988718899707e-06,
2809
+ "loss": 1.6171,
2810
+ "step": 233500
2811
+ },
2812
+ {
2813
+ "epoch": 4.48,
2814
+ "learning_rate": 5.578330242620924e-06,
2815
+ "loss": 1.6144,
2816
+ "step": 234000
2817
+ },
2818
+ {
2819
+ "epoch": 4.49,
2820
+ "learning_rate": 5.568671766342142e-06,
2821
+ "loss": 1.6173,
2822
+ "step": 234500
2823
+ },
2824
+ {
2825
+ "epoch": 4.5,
2826
+ "learning_rate": 5.55901329006336e-06,
2827
+ "loss": 1.6124,
2828
+ "step": 235000
2829
+ },
2830
+ {
2831
+ "epoch": 4.51,
2832
+ "learning_rate": 5.549374130737135e-06,
2833
+ "loss": 1.6104,
2834
+ "step": 235500
2835
+ },
2836
+ {
2837
+ "epoch": 4.52,
2838
+ "learning_rate": 5.539715654458353e-06,
2839
+ "loss": 1.6095,
2840
+ "step": 236000
2841
+ },
2842
+ {
2843
+ "epoch": 4.52,
2844
+ "learning_rate": 5.530057178179571e-06,
2845
+ "loss": 1.608,
2846
+ "step": 236500
2847
+ },
2848
+ {
2849
+ "epoch": 4.53,
2850
+ "learning_rate": 5.520398701900789e-06,
2851
+ "loss": 1.6132,
2852
+ "step": 237000
2853
+ },
2854
+ {
2855
+ "epoch": 4.54,
2856
+ "learning_rate": 5.510759542574564e-06,
2857
+ "loss": 1.6132,
2858
+ "step": 237500
2859
+ },
2860
+ {
2861
+ "epoch": 4.55,
2862
+ "learning_rate": 5.501101066295782e-06,
2863
+ "loss": 1.6147,
2864
+ "step": 238000
2865
+ },
2866
+ {
2867
+ "epoch": 4.56,
2868
+ "learning_rate": 5.491442590017e-06,
2869
+ "loss": 1.6135,
2870
+ "step": 238500
2871
+ },
2872
+ {
2873
+ "epoch": 4.57,
2874
+ "learning_rate": 5.481784113738216e-06,
2875
+ "loss": 1.6129,
2876
+ "step": 239000
2877
+ },
2878
+ {
2879
+ "epoch": 4.58,
2880
+ "learning_rate": 5.472125637459434e-06,
2881
+ "loss": 1.6141,
2882
+ "step": 239500
2883
+ },
2884
+ {
2885
+ "epoch": 4.59,
2886
+ "learning_rate": 5.4624864781332096e-06,
2887
+ "loss": 1.6139,
2888
+ "step": 240000
2889
+ },
2890
+ {
2891
+ "epoch": 4.6,
2892
+ "learning_rate": 5.4528280018544275e-06,
2893
+ "loss": 1.6125,
2894
+ "step": 240500
2895
+ },
2896
+ {
2897
+ "epoch": 4.61,
2898
+ "learning_rate": 5.443169525575645e-06,
2899
+ "loss": 1.613,
2900
+ "step": 241000
2901
+ },
2902
+ {
2903
+ "epoch": 4.62,
2904
+ "learning_rate": 5.433511049296863e-06,
2905
+ "loss": 1.6123,
2906
+ "step": 241500
2907
+ },
2908
+ {
2909
+ "epoch": 4.63,
2910
+ "learning_rate": 5.423871889970639e-06,
2911
+ "loss": 1.6149,
2912
+ "step": 242000
2913
+ },
2914
+ {
2915
+ "epoch": 4.64,
2916
+ "learning_rate": 5.4142134136918565e-06,
2917
+ "loss": 1.6131,
2918
+ "step": 242500
2919
+ },
2920
+ {
2921
+ "epoch": 4.65,
2922
+ "learning_rate": 5.4045549374130745e-06,
2923
+ "loss": 1.6118,
2924
+ "step": 243000
2925
+ },
2926
+ {
2927
+ "epoch": 4.66,
2928
+ "learning_rate": 5.394896461134292e-06,
2929
+ "loss": 1.6102,
2930
+ "step": 243500
2931
+ },
2932
+ {
2933
+ "epoch": 4.67,
2934
+ "learning_rate": 5.3852379848555095e-06,
2935
+ "loss": 1.6146,
2936
+ "step": 244000
2937
+ },
2938
+ {
2939
+ "epoch": 4.68,
2940
+ "learning_rate": 5.375579508576727e-06,
2941
+ "loss": 1.6123,
2942
+ "step": 244500
2943
+ },
2944
+ {
2945
+ "epoch": 4.69,
2946
+ "learning_rate": 5.365921032297945e-06,
2947
+ "loss": 1.6144,
2948
+ "step": 245000
2949
+ },
2950
+ {
2951
+ "epoch": 4.7,
2952
+ "learning_rate": 5.356281872971721e-06,
2953
+ "loss": 1.6161,
2954
+ "step": 245500
2955
+ },
2956
+ {
2957
+ "epoch": 4.71,
2958
+ "learning_rate": 5.3466233966929385e-06,
2959
+ "loss": 1.6171,
2960
+ "step": 246000
2961
+ },
2962
+ {
2963
+ "epoch": 4.72,
2964
+ "learning_rate": 5.3369649204141565e-06,
2965
+ "loss": 1.611,
2966
+ "step": 246500
2967
+ },
2968
+ {
2969
+ "epoch": 4.73,
2970
+ "learning_rate": 5.3273064441353735e-06,
2971
+ "loss": 1.6102,
2972
+ "step": 247000
2973
+ },
2974
+ {
2975
+ "epoch": 4.74,
2976
+ "learning_rate": 5.3176479678565915e-06,
2977
+ "loss": 1.6107,
2978
+ "step": 247500
2979
+ },
2980
+ {
2981
+ "epoch": 4.74,
2982
+ "learning_rate": 5.308008808530367e-06,
2983
+ "loss": 1.6101,
2984
+ "step": 248000
2985
+ },
2986
+ {
2987
+ "epoch": 4.75,
2988
+ "learning_rate": 5.298350332251584e-06,
2989
+ "loss": 1.6115,
2990
+ "step": 248500
2991
+ },
2992
+ {
2993
+ "epoch": 4.76,
2994
+ "learning_rate": 5.288691855972802e-06,
2995
+ "loss": 1.6109,
2996
+ "step": 249000
2997
+ },
2998
+ {
2999
+ "epoch": 4.77,
3000
+ "learning_rate": 5.27903337969402e-06,
3001
+ "loss": 1.6103,
3002
+ "step": 249500
3003
+ },
3004
+ {
3005
+ "epoch": 4.78,
3006
+ "learning_rate": 5.269374903415238e-06,
3007
+ "loss": 1.6111,
3008
+ "step": 250000
3009
+ },
3010
+ {
3011
+ "epoch": 4.79,
3012
+ "learning_rate": 5.259735744089013e-06,
3013
+ "loss": 1.6134,
3014
+ "step": 250500
3015
+ },
3016
+ {
3017
+ "epoch": 4.8,
3018
+ "learning_rate": 5.250077267810231e-06,
3019
+ "loss": 1.6124,
3020
+ "step": 251000
3021
+ },
3022
+ {
3023
+ "epoch": 4.81,
3024
+ "learning_rate": 5.240418791531449e-06,
3025
+ "loss": 1.6117,
3026
+ "step": 251500
3027
+ },
3028
+ {
3029
+ "epoch": 4.82,
3030
+ "learning_rate": 5.230760315252667e-06,
3031
+ "loss": 1.6151,
3032
+ "step": 252000
3033
+ },
3034
+ {
3035
+ "epoch": 4.83,
3036
+ "learning_rate": 5.2211018389738846e-06,
3037
+ "loss": 1.612,
3038
+ "step": 252500
3039
+ },
3040
+ {
3041
+ "epoch": 4.84,
3042
+ "learning_rate": 5.211443362695101e-06,
3043
+ "loss": 1.6137,
3044
+ "step": 253000
3045
+ },
3046
+ {
3047
+ "epoch": 4.85,
3048
+ "learning_rate": 5.201804203368876e-06,
3049
+ "loss": 1.6147,
3050
+ "step": 253500
3051
+ },
3052
+ {
3053
+ "epoch": 4.86,
3054
+ "learning_rate": 5.192145727090094e-06,
3055
+ "loss": 1.6138,
3056
+ "step": 254000
3057
+ },
3058
+ {
3059
+ "epoch": 4.87,
3060
+ "learning_rate": 5.182487250811312e-06,
3061
+ "loss": 1.6151,
3062
+ "step": 254500
3063
+ },
3064
+ {
3065
+ "epoch": 4.88,
3066
+ "learning_rate": 5.17282877453253e-06,
3067
+ "loss": 1.6146,
3068
+ "step": 255000
3069
+ },
3070
+ {
3071
+ "epoch": 4.89,
3072
+ "learning_rate": 5.163170298253748e-06,
3073
+ "loss": 1.6127,
3074
+ "step": 255500
3075
+ },
3076
+ {
3077
+ "epoch": 4.9,
3078
+ "learning_rate": 5.153531138927523e-06,
3079
+ "loss": 1.612,
3080
+ "step": 256000
3081
+ },
3082
+ {
3083
+ "epoch": 4.91,
3084
+ "learning_rate": 5.143872662648741e-06,
3085
+ "loss": 1.6127,
3086
+ "step": 256500
3087
+ },
3088
+ {
3089
+ "epoch": 4.92,
3090
+ "learning_rate": 5.134214186369959e-06,
3091
+ "loss": 1.6124,
3092
+ "step": 257000
3093
+ },
3094
+ {
3095
+ "epoch": 4.93,
3096
+ "learning_rate": 5.124555710091177e-06,
3097
+ "loss": 1.6141,
3098
+ "step": 257500
3099
+ },
3100
+ {
3101
+ "epoch": 4.94,
3102
+ "learning_rate": 5.114897233812394e-06,
3103
+ "loss": 1.6145,
3104
+ "step": 258000
3105
+ },
3106
+ {
3107
+ "epoch": 4.95,
3108
+ "learning_rate": 5.105258074486169e-06,
3109
+ "loss": 1.6109,
3110
+ "step": 258500
3111
+ },
3112
+ {
3113
+ "epoch": 4.96,
3114
+ "learning_rate": 5.095599598207387e-06,
3115
+ "loss": 1.6079,
3116
+ "step": 259000
3117
+ },
3118
+ {
3119
+ "epoch": 4.96,
3120
+ "learning_rate": 5.085941121928605e-06,
3121
+ "loss": 1.6126,
3122
+ "step": 259500
3123
+ },
3124
+ {
3125
+ "epoch": 4.97,
3126
+ "learning_rate": 5.076282645649823e-06,
3127
+ "loss": 1.6127,
3128
+ "step": 260000
3129
+ },
3130
+ {
3131
+ "epoch": 4.98,
3132
+ "learning_rate": 5.066624169371041e-06,
3133
+ "loss": 1.6095,
3134
+ "step": 260500
3135
+ },
3136
+ {
3137
+ "epoch": 4.99,
3138
+ "learning_rate": 5.056985010044816e-06,
3139
+ "loss": 1.613,
3140
+ "step": 261000
3141
+ },
3142
+ {
3143
+ "epoch": 5.0,
3144
+ "learning_rate": 5.047326533766034e-06,
3145
+ "loss": 1.6118,
3146
+ "step": 261500
3147
+ },
3148
+ {
3149
+ "epoch": 5.01,
3150
+ "learning_rate": 5.037668057487252e-06,
3151
+ "loss": 1.6124,
3152
+ "step": 262000
3153
+ },
3154
+ {
3155
+ "epoch": 5.02,
3156
+ "learning_rate": 5.028009581208469e-06,
3157
+ "loss": 1.6107,
3158
+ "step": 262500
3159
+ },
3160
+ {
3161
+ "epoch": 5.03,
3162
+ "learning_rate": 5.0183704218822435e-06,
3163
+ "loss": 1.6127,
3164
+ "step": 263000
3165
+ },
3166
+ {
3167
+ "epoch": 5.04,
3168
+ "learning_rate": 5.0087119456034615e-06,
3169
+ "loss": 1.6145,
3170
+ "step": 263500
3171
+ },
3172
+ {
3173
+ "epoch": 5.05,
3174
+ "learning_rate": 4.999053469324679e-06,
3175
+ "loss": 1.6085,
3176
+ "step": 264000
3177
+ },
3178
+ {
3179
+ "epoch": 5.06,
3180
+ "learning_rate": 4.989394993045897e-06,
3181
+ "loss": 1.6128,
3182
+ "step": 264500
3183
+ },
3184
+ {
3185
+ "epoch": 5.07,
3186
+ "learning_rate": 4.979755833719673e-06,
3187
+ "loss": 1.6102,
3188
+ "step": 265000
3189
+ },
3190
+ {
3191
+ "epoch": 5.08,
3192
+ "learning_rate": 4.9700973574408905e-06,
3193
+ "loss": 1.6097,
3194
+ "step": 265500
3195
+ },
3196
+ {
3197
+ "epoch": 5.09,
3198
+ "learning_rate": 4.9604388811621084e-06,
3199
+ "loss": 1.6089,
3200
+ "step": 266000
3201
+ },
3202
+ {
3203
+ "epoch": 5.1,
3204
+ "learning_rate": 4.9507804048833255e-06,
3205
+ "loss": 1.6106,
3206
+ "step": 266500
3207
+ },
3208
+ {
3209
+ "epoch": 5.11,
3210
+ "learning_rate": 4.9411219286045434e-06,
3211
+ "loss": 1.6085,
3212
+ "step": 267000
3213
+ },
3214
+ {
3215
+ "epoch": 5.12,
3216
+ "learning_rate": 4.931482769278319e-06,
3217
+ "loss": 1.6115,
3218
+ "step": 267500
3219
+ },
3220
+ {
3221
+ "epoch": 5.13,
3222
+ "learning_rate": 4.921824292999537e-06,
3223
+ "loss": 1.6096,
3224
+ "step": 268000
3225
+ },
3226
+ {
3227
+ "epoch": 5.14,
3228
+ "learning_rate": 4.9121658167207546e-06,
3229
+ "loss": 1.61,
3230
+ "step": 268500
3231
+ },
3232
+ {
3233
+ "epoch": 5.15,
3234
+ "learning_rate": 4.9025073404419725e-06,
3235
+ "loss": 1.6103,
3236
+ "step": 269000
3237
+ },
3238
+ {
3239
+ "epoch": 5.16,
3240
+ "learning_rate": 4.8928488641631896e-06,
3241
+ "loss": 1.6106,
3242
+ "step": 269500
3243
+ },
3244
+ {
3245
+ "epoch": 5.17,
3246
+ "learning_rate": 4.883209704836966e-06,
3247
+ "loss": 1.6139,
3248
+ "step": 270000
3249
+ },
3250
+ {
3251
+ "epoch": 5.18,
3252
+ "learning_rate": 4.873551228558183e-06,
3253
+ "loss": 1.6095,
3254
+ "step": 270500
3255
+ },
3256
+ {
3257
+ "epoch": 5.18,
3258
+ "learning_rate": 4.863892752279401e-06,
3259
+ "loss": 1.6117,
3260
+ "step": 271000
3261
+ },
3262
+ {
3263
+ "epoch": 5.19,
3264
+ "learning_rate": 4.854234276000619e-06,
3265
+ "loss": 1.6126,
3266
+ "step": 271500
3267
+ },
3268
+ {
3269
+ "epoch": 5.2,
3270
+ "learning_rate": 4.844595116674394e-06,
3271
+ "loss": 1.6104,
3272
+ "step": 272000
3273
+ },
3274
+ {
3275
+ "epoch": 5.21,
3276
+ "learning_rate": 4.834936640395612e-06,
3277
+ "loss": 1.6073,
3278
+ "step": 272500
3279
+ },
3280
+ {
3281
+ "epoch": 5.22,
3282
+ "learning_rate": 4.82527816411683e-06,
3283
+ "loss": 1.6107,
3284
+ "step": 273000
3285
+ },
3286
+ {
3287
+ "epoch": 5.23,
3288
+ "learning_rate": 4.815619687838048e-06,
3289
+ "loss": 1.6116,
3290
+ "step": 273500
3291
+ },
3292
+ {
3293
+ "epoch": 5.24,
3294
+ "learning_rate": 4.805961211559265e-06,
3295
+ "loss": 1.6106,
3296
+ "step": 274000
3297
+ },
3298
+ {
3299
+ "epoch": 5.25,
3300
+ "learning_rate": 4.79632205223304e-06,
3301
+ "loss": 1.61,
3302
+ "step": 274500
3303
+ },
3304
+ {
3305
+ "epoch": 5.26,
3306
+ "learning_rate": 4.786663575954258e-06,
3307
+ "loss": 1.608,
3308
+ "step": 275000
3309
+ },
3310
+ {
3311
+ "epoch": 5.27,
3312
+ "learning_rate": 4.777005099675476e-06,
3313
+ "loss": 1.6133,
3314
+ "step": 275500
3315
+ },
3316
+ {
3317
+ "epoch": 5.28,
3318
+ "learning_rate": 4.767346623396693e-06,
3319
+ "loss": 1.6104,
3320
+ "step": 276000
3321
+ },
3322
+ {
3323
+ "epoch": 5.29,
3324
+ "learning_rate": 4.757707464070468e-06,
3325
+ "loss": 1.6099,
3326
+ "step": 276500
3327
+ },
3328
+ {
3329
+ "epoch": 5.3,
3330
+ "learning_rate": 4.748048987791686e-06,
3331
+ "loss": 1.6101,
3332
+ "step": 277000
3333
+ },
3334
+ {
3335
+ "epoch": 5.31,
3336
+ "learning_rate": 4.738390511512904e-06,
3337
+ "loss": 1.6115,
3338
+ "step": 277500
3339
+ },
3340
+ {
3341
+ "epoch": 5.32,
3342
+ "learning_rate": 4.728732035234122e-06,
3343
+ "loss": 1.6121,
3344
+ "step": 278000
3345
+ },
3346
+ {
3347
+ "epoch": 5.33,
3348
+ "learning_rate": 4.719073558955339e-06,
3349
+ "loss": 1.6103,
3350
+ "step": 278500
3351
+ },
3352
+ {
3353
+ "epoch": 5.34,
3354
+ "learning_rate": 4.709434399629114e-06,
3355
+ "loss": 1.6097,
3356
+ "step": 279000
3357
+ },
3358
+ {
3359
+ "epoch": 5.35,
3360
+ "learning_rate": 4.699775923350332e-06,
3361
+ "loss": 1.6123,
3362
+ "step": 279500
3363
+ },
3364
+ {
3365
+ "epoch": 5.36,
3366
+ "learning_rate": 4.69011744707155e-06,
3367
+ "loss": 1.6097,
3368
+ "step": 280000
3369
+ },
3370
+ {
3371
+ "epoch": 5.37,
3372
+ "learning_rate": 4.680458970792768e-06,
3373
+ "loss": 1.6104,
3374
+ "step": 280500
3375
+ },
3376
+ {
3377
+ "epoch": 5.38,
3378
+ "learning_rate": 4.670800494513986e-06,
3379
+ "loss": 1.6107,
3380
+ "step": 281000
3381
+ },
3382
+ {
3383
+ "epoch": 5.39,
3384
+ "learning_rate": 4.661161335187761e-06,
3385
+ "loss": 1.6106,
3386
+ "step": 281500
3387
+ },
3388
+ {
3389
+ "epoch": 5.4,
3390
+ "learning_rate": 4.651502858908979e-06,
3391
+ "loss": 1.6077,
3392
+ "step": 282000
3393
+ },
3394
+ {
3395
+ "epoch": 5.4,
3396
+ "learning_rate": 4.641844382630196e-06,
3397
+ "loss": 1.6108,
3398
+ "step": 282500
3399
+ },
3400
+ {
3401
+ "epoch": 5.41,
3402
+ "learning_rate": 4.632185906351414e-06,
3403
+ "loss": 1.6107,
3404
+ "step": 283000
3405
+ },
3406
+ {
3407
+ "epoch": 5.42,
3408
+ "learning_rate": 4.6225467470251896e-06,
3409
+ "loss": 1.6144,
3410
+ "step": 283500
3411
+ },
3412
+ {
3413
+ "epoch": 5.43,
3414
+ "learning_rate": 4.6128882707464075e-06,
3415
+ "loss": 1.6144,
3416
+ "step": 284000
3417
+ },
3418
+ {
3419
+ "epoch": 5.44,
3420
+ "learning_rate": 4.603229794467625e-06,
3421
+ "loss": 1.6138,
3422
+ "step": 284500
3423
+ },
3424
+ {
3425
+ "epoch": 5.45,
3426
+ "learning_rate": 4.593571318188843e-06,
3427
+ "loss": 1.6136,
3428
+ "step": 285000
3429
+ },
3430
+ {
3431
+ "epoch": 5.46,
3432
+ "learning_rate": 4.583932158862619e-06,
3433
+ "loss": 1.6099,
3434
+ "step": 285500
3435
+ },
3436
+ {
3437
+ "epoch": 5.47,
3438
+ "learning_rate": 4.574273682583836e-06,
3439
+ "loss": 1.6091,
3440
+ "step": 286000
3441
+ },
3442
+ {
3443
+ "epoch": 5.48,
3444
+ "learning_rate": 4.564615206305054e-06,
3445
+ "loss": 1.6072,
3446
+ "step": 286500
3447
+ },
3448
+ {
3449
+ "epoch": 5.49,
3450
+ "learning_rate": 4.5549567300262715e-06,
3451
+ "loss": 1.6103,
3452
+ "step": 287000
3453
+ },
3454
+ {
3455
+ "epoch": 5.5,
3456
+ "learning_rate": 4.5452982537474895e-06,
3457
+ "loss": 1.613,
3458
+ "step": 287500
3459
+ },
3460
+ {
3461
+ "epoch": 5.51,
3462
+ "learning_rate": 4.535659094421265e-06,
3463
+ "loss": 1.6097,
3464
+ "step": 288000
3465
+ },
3466
+ {
3467
+ "epoch": 5.52,
3468
+ "learning_rate": 4.526000618142482e-06,
3469
+ "loss": 1.6116,
3470
+ "step": 288500
3471
+ },
3472
+ {
3473
+ "epoch": 5.53,
3474
+ "learning_rate": 4.5163421418637e-06,
3475
+ "loss": 1.6107,
3476
+ "step": 289000
3477
+ },
3478
+ {
3479
+ "epoch": 5.54,
3480
+ "learning_rate": 4.506683665584918e-06,
3481
+ "loss": 1.6067,
3482
+ "step": 289500
3483
+ },
3484
+ {
3485
+ "epoch": 5.55,
3486
+ "learning_rate": 4.497025189306136e-06,
3487
+ "loss": 1.6122,
3488
+ "step": 290000
3489
+ },
3490
+ {
3491
+ "epoch": 5.56,
3492
+ "learning_rate": 4.487386029979911e-06,
3493
+ "loss": 1.6097,
3494
+ "step": 290500
3495
+ },
3496
+ {
3497
+ "epoch": 5.57,
3498
+ "learning_rate": 4.477727553701128e-06,
3499
+ "loss": 1.6078,
3500
+ "step": 291000
3501
+ },
3502
+ {
3503
+ "epoch": 5.58,
3504
+ "learning_rate": 4.468069077422346e-06,
3505
+ "loss": 1.6107,
3506
+ "step": 291500
3507
+ },
3508
+ {
3509
+ "epoch": 5.59,
3510
+ "learning_rate": 4.458410601143564e-06,
3511
+ "loss": 1.6091,
3512
+ "step": 292000
3513
+ },
3514
+ {
3515
+ "epoch": 5.6,
3516
+ "learning_rate": 4.448752124864782e-06,
3517
+ "loss": 1.6121,
3518
+ "step": 292500
3519
+ },
3520
+ {
3521
+ "epoch": 5.61,
3522
+ "learning_rate": 4.439112965538557e-06,
3523
+ "loss": 1.6121,
3524
+ "step": 293000
3525
+ },
3526
+ {
3527
+ "epoch": 5.62,
3528
+ "learning_rate": 4.429454489259775e-06,
3529
+ "loss": 1.6126,
3530
+ "step": 293500
3531
+ },
3532
+ {
3533
+ "epoch": 5.62,
3534
+ "learning_rate": 4.419796012980993e-06,
3535
+ "loss": 1.6132,
3536
+ "step": 294000
3537
+ },
3538
+ {
3539
+ "epoch": 5.63,
3540
+ "learning_rate": 4.41013753670221e-06,
3541
+ "loss": 1.6111,
3542
+ "step": 294500
3543
+ },
3544
+ {
3545
+ "epoch": 5.64,
3546
+ "learning_rate": 4.400498377375986e-06,
3547
+ "loss": 1.6112,
3548
+ "step": 295000
3549
+ },
3550
+ {
3551
+ "epoch": 5.65,
3552
+ "learning_rate": 4.390839901097203e-06,
3553
+ "loss": 1.6102,
3554
+ "step": 295500
3555
+ },
3556
+ {
3557
+ "epoch": 5.66,
3558
+ "learning_rate": 4.381181424818421e-06,
3559
+ "loss": 1.6114,
3560
+ "step": 296000
3561
+ },
3562
+ {
3563
+ "epoch": 5.67,
3564
+ "learning_rate": 4.371522948539639e-06,
3565
+ "loss": 1.6093,
3566
+ "step": 296500
3567
+ },
3568
+ {
3569
+ "epoch": 5.68,
3570
+ "learning_rate": 4.361864472260857e-06,
3571
+ "loss": 1.6125,
3572
+ "step": 297000
3573
+ },
3574
+ {
3575
+ "epoch": 5.69,
3576
+ "learning_rate": 4.352225312934632e-06,
3577
+ "loss": 1.6101,
3578
+ "step": 297500
3579
+ },
3580
+ {
3581
+ "epoch": 5.7,
3582
+ "learning_rate": 4.342566836655849e-06,
3583
+ "loss": 1.611,
3584
+ "step": 298000
3585
+ },
3586
+ {
3587
+ "epoch": 5.71,
3588
+ "learning_rate": 4.332908360377067e-06,
3589
+ "loss": 1.6111,
3590
+ "step": 298500
3591
+ },
3592
+ {
3593
+ "epoch": 5.72,
3594
+ "learning_rate": 4.323249884098285e-06,
3595
+ "loss": 1.6117,
3596
+ "step": 299000
3597
+ },
3598
+ {
3599
+ "epoch": 5.73,
3600
+ "learning_rate": 4.31361072477206e-06,
3601
+ "loss": 1.6118,
3602
+ "step": 299500
3603
+ },
3604
+ {
3605
+ "epoch": 5.74,
3606
+ "learning_rate": 4.303952248493278e-06,
3607
+ "loss": 1.611,
3608
+ "step": 300000
3609
+ },
3610
+ {
3611
+ "epoch": 5.75,
3612
+ "learning_rate": 4.294293772214495e-06,
3613
+ "loss": 1.6147,
3614
+ "step": 300500
3615
+ },
3616
+ {
3617
+ "epoch": 5.76,
3618
+ "learning_rate": 4.284635295935713e-06,
3619
+ "loss": 1.6097,
3620
+ "step": 301000
3621
+ },
3622
+ {
3623
+ "epoch": 5.77,
3624
+ "learning_rate": 4.274976819656931e-06,
3625
+ "loss": 1.6082,
3626
+ "step": 301500
3627
+ },
3628
+ {
3629
+ "epoch": 5.78,
3630
+ "learning_rate": 4.265318343378149e-06,
3631
+ "loss": 1.6096,
3632
+ "step": 302000
3633
+ },
3634
+ {
3635
+ "epoch": 5.79,
3636
+ "learning_rate": 4.2556791840519244e-06,
3637
+ "loss": 1.6054,
3638
+ "step": 302500
3639
+ },
3640
+ {
3641
+ "epoch": 5.8,
3642
+ "learning_rate": 4.2460207077731415e-06,
3643
+ "loss": 1.6098,
3644
+ "step": 303000
3645
+ },
3646
+ {
3647
+ "epoch": 5.81,
3648
+ "learning_rate": 4.2363622314943595e-06,
3649
+ "loss": 1.6085,
3650
+ "step": 303500
3651
+ },
3652
+ {
3653
+ "epoch": 5.82,
3654
+ "learning_rate": 4.226703755215577e-06,
3655
+ "loss": 1.615,
3656
+ "step": 304000
3657
+ },
3658
+ {
3659
+ "epoch": 5.83,
3660
+ "learning_rate": 4.217064595889353e-06,
3661
+ "loss": 1.61,
3662
+ "step": 304500
3663
+ },
3664
+ {
3665
+ "epoch": 5.84,
3666
+ "learning_rate": 4.207406119610571e-06,
3667
+ "loss": 1.6108,
3668
+ "step": 305000
3669
+ },
3670
+ {
3671
+ "epoch": 5.84,
3672
+ "learning_rate": 4.1977476433317885e-06,
3673
+ "loss": 1.6102,
3674
+ "step": 305500
3675
+ },
3676
+ {
3677
+ "epoch": 5.85,
3678
+ "learning_rate": 4.1880891670530064e-06,
3679
+ "loss": 1.6127,
3680
+ "step": 306000
3681
+ },
3682
+ {
3683
+ "epoch": 5.86,
3684
+ "learning_rate": 4.1784306907742235e-06,
3685
+ "loss": 1.609,
3686
+ "step": 306500
3687
+ },
3688
+ {
3689
+ "epoch": 5.87,
3690
+ "learning_rate": 4.168791531448e-06,
3691
+ "loss": 1.6085,
3692
+ "step": 307000
3693
+ },
3694
+ {
3695
+ "epoch": 5.88,
3696
+ "learning_rate": 4.159133055169217e-06,
3697
+ "loss": 1.607,
3698
+ "step": 307500
3699
+ },
3700
+ {
3701
+ "epoch": 5.89,
3702
+ "learning_rate": 4.149474578890435e-06,
3703
+ "loss": 1.6097,
3704
+ "step": 308000
3705
+ },
3706
+ {
3707
+ "epoch": 5.9,
3708
+ "learning_rate": 4.1398161026116526e-06,
3709
+ "loss": 1.6093,
3710
+ "step": 308500
3711
+ },
3712
+ {
3713
+ "epoch": 5.91,
3714
+ "learning_rate": 4.130176943285428e-06,
3715
+ "loss": 1.6127,
3716
+ "step": 309000
3717
+ },
3718
+ {
3719
+ "epoch": 5.92,
3720
+ "learning_rate": 4.120518467006646e-06,
3721
+ "loss": 1.611,
3722
+ "step": 309500
3723
+ },
3724
+ {
3725
+ "epoch": 5.93,
3726
+ "learning_rate": 4.110859990727863e-06,
3727
+ "loss": 1.6101,
3728
+ "step": 310000
3729
+ }
3730
+ ],
3731
+ "max_steps": 522680,
3732
+ "num_train_epochs": 10,
3733
+ "total_flos": 6.527459946085057e+18,
3734
+ "trial_name": null,
3735
+ "trial_params": null
3736
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b0dce33054797c89a32155c9042339f77f179913ccd439e9bf3478e5b3b68a3
3
+ size 3503
vocab.txt ADDED
The diff for this file is too large to render. See raw diff