Hugging Face's logo

BioMike
/
WikiLinking-bi-gliner-base-5000

PyTorch
Model card Files Community
BioMike commited on
Commit
f97bde0
•
1 Parent(s): e514049

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. added_tokens.json +3 -0
  2. gliner_config.json +216 -0
  3. optimizer.pt +3 -0
  4. pytorch_model.bin +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +15 -0
  8. spm.model +3 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +58 -0
  11. trainer_state.json +3533 -0
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[MASK]": 128000
3
+ }
gliner_config.json ADDED
@@ -0,0 +1,216 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "class_token_index": -1,
3
+ "dropout": 0.3,
4
+ "embed_ent_token": true,
5
+ "encoder_config": {
6
+ "_name_or_path": "microsoft/deberta-v3-base",
7
+ "add_cross_attention": false,
8
+ "architectures": null,
9
+ "attention_probs_dropout_prob": 0.1,
10
+ "bad_words_ids": null,
11
+ "begin_suppress_tokens": null,
12
+ "bos_token_id": null,
13
+ "chunk_size_feed_forward": 0,
14
+ "cross_attention_hidden_size": null,
15
+ "decoder_start_token_id": null,
16
+ "diversity_penalty": 0.0,
17
+ "do_sample": false,
18
+ "early_stopping": false,
19
+ "encoder_no_repeat_ngram_size": 0,
20
+ "eos_token_id": null,
21
+ "exponential_decay_length_penalty": null,
22
+ "finetuning_task": null,
23
+ "forced_bos_token_id": null,
24
+ "forced_eos_token_id": null,
25
+ "hidden_act": "gelu",
26
+ "hidden_dropout_prob": 0.1,
27
+ "hidden_size": 768,
28
+ "id2label": {
29
+ "0": "LABEL_0",
30
+ "1": "LABEL_1"
31
+ },
32
+ "initializer_range": 0.02,
33
+ "intermediate_size": 3072,
34
+ "is_decoder": false,
35
+ "is_encoder_decoder": false,
36
+ "label2id": {
37
+ "LABEL_0": 0,
38
+ "LABEL_1": 1
39
+ },
40
+ "layer_norm_eps": 1e-07,
41
+ "length_penalty": 1.0,
42
+ "max_length": 20,
43
+ "max_position_embeddings": 512,
44
+ "max_relative_positions": -1,
45
+ "min_length": 0,
46
+ "model_type": "deberta-v2",
47
+ "no_repeat_ngram_size": 0,
48
+ "norm_rel_ebd": "layer_norm",
49
+ "num_attention_heads": 12,
50
+ "num_beam_groups": 1,
51
+ "num_beams": 1,
52
+ "num_hidden_layers": 12,
53
+ "num_return_sequences": 1,
54
+ "output_attentions": false,
55
+ "output_hidden_states": false,
56
+ "output_scores": false,
57
+ "pad_token_id": 0,
58
+ "pooler_dropout": 0,
59
+ "pooler_hidden_act": "gelu",
60
+ "pooler_hidden_size": 768,
61
+ "pos_att_type": [
62
+ "p2c",
63
+ "c2p"
64
+ ],
65
+ "position_biased_input": false,
66
+ "position_buckets": 256,
67
+ "prefix": null,
68
+ "problem_type": null,
69
+ "pruned_heads": {},
70
+ "relative_attention": true,
71
+ "remove_invalid_values": false,
72
+ "repetition_penalty": 1.0,
73
+ "return_dict": true,
74
+ "return_dict_in_generate": false,
75
+ "sep_token_id": null,
76
+ "share_att_key": true,
77
+ "suppress_tokens": null,
78
+ "task_specific_params": null,
79
+ "temperature": 1.0,
80
+ "tf_legacy_loss": false,
81
+ "tie_encoder_decoder": false,
82
+ "tie_word_embeddings": true,
83
+ "tokenizer_class": null,
84
+ "top_k": 50,
85
+ "top_p": 1.0,
86
+ "torch_dtype": null,
87
+ "torchscript": false,
88
+ "type_vocab_size": 0,
89
+ "typical_p": 1.0,
90
+ "use_bfloat16": false,
91
+ "vocab_size": 128100
92
+ },
93
+ "ent_token": "<<ENT>>",
94
+ "eval_every": 1000,
95
+ "fine_tune": true,
96
+ "freeze_token_rep": false,
97
+ "fuse_layers": false,
98
+ "has_rnn": true,
99
+ "hidden_size": 768,
100
+ "label_smoothing": 0,
101
+ "labels_encoder": "BAAI/bge-small-en-v1.5",
102
+ "labels_encoder_config": {
103
+ "_name_or_path": "BAAI/bge-small-en-v1.5",
104
+ "add_cross_attention": false,
105
+ "architectures": [
106
+ "BertModel"
107
+ ],
108
+ "attention_probs_dropout_prob": 0.1,
109
+ "bad_words_ids": null,
110
+ "begin_suppress_tokens": null,
111
+ "bos_token_id": null,
112
+ "chunk_size_feed_forward": 0,
113
+ "classifier_dropout": null,
114
+ "cross_attention_hidden_size": null,
115
+ "decoder_start_token_id": null,
116
+ "diversity_penalty": 0.0,
117
+ "do_sample": false,
118
+ "early_stopping": false,
119
+ "encoder_no_repeat_ngram_size": 0,
120
+ "eos_token_id": null,
121
+ "exponential_decay_length_penalty": null,
122
+ "finetuning_task": null,
123
+ "forced_bos_token_id": null,
124
+ "forced_eos_token_id": null,
125
+ "hidden_act": "gelu",
126
+ "hidden_dropout_prob": 0.1,
127
+ "hidden_size": 384,
128
+ "id2label": {
129
+ "0": "LABEL_0"
130
+ },
131
+ "initializer_range": 0.02,
132
+ "intermediate_size": 1536,
133
+ "is_decoder": false,
134
+ "is_encoder_decoder": false,
135
+ "label2id": {
136
+ "LABEL_0": 0
137
+ },
138
+ "layer_norm_eps": 1e-12,
139
+ "length_penalty": 1.0,
140
+ "max_length": 20,
141
+ "max_position_embeddings": 512,
142
+ "min_length": 0,
143
+ "model_type": "bert",
144
+ "no_repeat_ngram_size": 0,
145
+ "num_attention_heads": 12,
146
+ "num_beam_groups": 1,
147
+ "num_beams": 1,
148
+ "num_hidden_layers": 12,
149
+ "num_return_sequences": 1,
150
+ "output_attentions": false,
151
+ "output_hidden_states": false,
152
+ "output_scores": false,
153
+ "pad_token_id": 0,
154
+ "position_embedding_type": "absolute",
155
+ "prefix": null,
156
+ "problem_type": null,
157
+ "pruned_heads": {},
158
+ "remove_invalid_values": false,
159
+ "repetition_penalty": 1.0,
160
+ "return_dict": true,
161
+ "return_dict_in_generate": false,
162
+ "sep_token_id": null,
163
+ "suppress_tokens": null,
164
+ "task_specific_params": null,
165
+ "temperature": 1.0,
166
+ "tf_legacy_loss": false,
167
+ "tie_encoder_decoder": false,
168
+ "tie_word_embeddings": true,
169
+ "tokenizer_class": null,
170
+ "top_k": 50,
171
+ "top_p": 1.0,
172
+ "torch_dtype": "float32",
173
+ "torchscript": false,
174
+ "type_vocab_size": 2,
175
+ "typical_p": 1.0,
176
+ "use_bfloat16": false,
177
+ "use_cache": true,
178
+ "vocab_size": 30522
179
+ },
180
+ "log_dir": "models-base_2/",
181
+ "loss_alpha": -1,
182
+ "loss_gamma": 0,
183
+ "loss_reduction": "sum",
184
+ "lr_encoder": "1e-5",
185
+ "lr_others": "5e-5",
186
+ "max_grad_norm": 10.0,
187
+ "max_len": 512,
188
+ "max_neg_type_ratio": 1,
189
+ "max_types": 128,
190
+ "max_width": 12,
191
+ "model_name": "microsoft/deberta-v3-base",
192
+ "model_type": "gliner",
193
+ "name": "span level gliner for entity linking",
194
+ "num_post_fusion_layers": 1,
195
+ "num_steps": 123750,
196
+ "post_fusion_schema": "",
197
+ "prev_path": null,
198
+ "random_drop": true,
199
+ "root_dir": "gliner_logs",
200
+ "save_total_limit": 3,
201
+ "scheduler_type": "cosine",
202
+ "sep_token": "<<SEP>>",
203
+ "shuffle_types": true,
204
+ "size_sup": -1,
205
+ "span_mode": "markerV0",
206
+ "subtoken_pooling": "first",
207
+ "train_batch_size": 8,
208
+ "train_data": "dataset_with_negatives.json",
209
+ "transformers_version": "4.45.2",
210
+ "val_data_dir": "none",
211
+ "vocab_size": -1,
212
+ "warmup_ratio": 0.1,
213
+ "weight_decay_encoder": 0.01,
214
+ "weight_decay_other": 0.01,
215
+ "words_splitter_type": "whitespace"
216
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce933941f779297392092b359c85f19951cecae1f3789bada7534e2f6918c25a
3
+ size 1909053417
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48aee6ce269baf4f97d647c30ff1a799acd33e83396e00e3e52d5b3b6378e24f
3
+ size 969281034
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:498578918d850667ef4f012c46594183cd770e1723d8a7cd23babf0675dca9ad
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d4f7fb3f84c9780fe38547b14bca86ededb8adbae4995dd223f8b1d131a68b6c
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "[CLS]",
3
+ "cls_token": "[CLS]",
4
+ "eos_token": "[SEP]",
5
+ "mask_token": "[MASK]",
6
+ "pad_token": "[PAD]",
7
+ "sep_token": "[SEP]",
8
+ "unk_token": {
9
+ "content": "[UNK]",
10
+ "lstrip": false,
11
+ "normalized": true,
12
+ "rstrip": false,
13
+ "single_word": false
14
+ }
15
+ }
spm.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
3
+ size 2464616
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "[PAD]",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "[CLS]",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "[SEP]",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "[UNK]",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "128000": {
36
+ "content": "[MASK]",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "bos_token": "[CLS]",
45
+ "clean_up_tokenization_spaces": false,
46
+ "cls_token": "[CLS]",
47
+ "do_lower_case": false,
48
+ "eos_token": "[SEP]",
49
+ "mask_token": "[MASK]",
50
+ "model_max_length": 1000000000000000019884624838656,
51
+ "pad_token": "[PAD]",
52
+ "sep_token": "[SEP]",
53
+ "sp_model_kwargs": {},
54
+ "split_by_punct": false,
55
+ "tokenizer_class": "DebertaV2Tokenizer",
56
+ "unk_token": "[UNK]",
57
+ "vocab_type": "spm"
58
+ }
trainer_state.json ADDED
@@ -0,0 +1,3533 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.040401102142066436,
5
+ "eval_steps": 500,
6
+ "global_step": 5000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 8.080220428413287e-05,
13
+ "grad_norm": 4661457.0,
14
+ "learning_rate": 4.040404040404041e-08,
15
+ "loss": 678012.1,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.00016160440856826573,
20
+ "grad_norm": 9865970.0,
21
+ "learning_rate": 8.080808080808082e-08,
22
+ "loss": 841980.1,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.00024240661285239863,
27
+ "grad_norm": 10027139.0,
28
+ "learning_rate": 1.2121212121212122e-07,
29
+ "loss": 816764.25,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.00032320881713653147,
34
+ "grad_norm": 16910428.0,
35
+ "learning_rate": 1.6161616161616163e-07,
36
+ "loss": 961656.7,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.00040401102142066436,
41
+ "grad_norm": 5068422.0,
42
+ "learning_rate": 2.0202020202020202e-07,
43
+ "loss": 679086.95,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.00048481322570479725,
48
+ "grad_norm": 2293057.5,
49
+ "learning_rate": 2.4242424242424244e-07,
50
+ "loss": 949613.2,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.0005656154299889301,
55
+ "grad_norm": 7849363.5,
56
+ "learning_rate": 2.8282828282828283e-07,
57
+ "loss": 859798.0,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0006464176342730629,
62
+ "grad_norm": 2669786.75,
63
+ "learning_rate": 3.2323232323232327e-07,
64
+ "loss": 863443.9,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.0007272198385571959,
69
+ "grad_norm": 6021406.5,
70
+ "learning_rate": 3.6363636363636366e-07,
71
+ "loss": 734401.3,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.0008080220428413287,
76
+ "grad_norm": 5000674.5,
77
+ "learning_rate": 4.0404040404040405e-07,
78
+ "loss": 526539.8,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.0008888242471254616,
83
+ "grad_norm": 4362635.5,
84
+ "learning_rate": 4.444444444444445e-07,
85
+ "loss": 422810.45,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0009696264514095945,
90
+ "grad_norm": 5343355.0,
91
+ "learning_rate": 4.848484848484849e-07,
92
+ "loss": 490438.55,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.0010504286556937273,
97
+ "grad_norm": 11366241.0,
98
+ "learning_rate": 5.252525252525253e-07,
99
+ "loss": 704809.3,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.0011312308599778602,
104
+ "grad_norm": 5520147.0,
105
+ "learning_rate": 5.656565656565657e-07,
106
+ "loss": 526687.5,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.001212033064261993,
111
+ "grad_norm": 3778741.0,
112
+ "learning_rate": 6.060606060606061e-07,
113
+ "loss": 322092.525,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.0012928352685461259,
118
+ "grad_norm": 3378473.0,
119
+ "learning_rate": 6.464646464646465e-07,
120
+ "loss": 516098.85,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.0013736374728302587,
125
+ "grad_norm": 4408531.0,
126
+ "learning_rate": 6.868686868686869e-07,
127
+ "loss": 410817.35,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.0014544396771143918,
132
+ "grad_norm": 766378.375,
133
+ "learning_rate": 7.272727272727273e-07,
134
+ "loss": 197623.1625,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.0015352418813985246,
139
+ "grad_norm": 5468394.5,
140
+ "learning_rate": 7.676767676767678e-07,
141
+ "loss": 238133.6,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.0016160440856826574,
146
+ "grad_norm": 7980194.5,
147
+ "learning_rate": 8.080808080808081e-07,
148
+ "loss": 182498.725,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.0016968462899667903,
153
+ "grad_norm": 737060.125,
154
+ "learning_rate": 8.484848484848486e-07,
155
+ "loss": 122880.075,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.0017776484942509231,
160
+ "grad_norm": 476725.8125,
161
+ "learning_rate": 8.88888888888889e-07,
162
+ "loss": 100649.7563,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.001858450698535056,
167
+ "grad_norm": 353791.875,
168
+ "learning_rate": 9.292929292929294e-07,
169
+ "loss": 58150.4563,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.001939252902819189,
174
+ "grad_norm": 684820.5625,
175
+ "learning_rate": 9.696969696969698e-07,
176
+ "loss": 44040.125,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.0020200551071033216,
181
+ "grad_norm": 228699.796875,
182
+ "learning_rate": 1.0101010101010103e-06,
183
+ "loss": 21505.6156,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.0021008573113874547,
188
+ "grad_norm": 91799.625,
189
+ "learning_rate": 1.0505050505050506e-06,
190
+ "loss": 11209.5648,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.0021816595156715873,
195
+ "grad_norm": 64077.56640625,
196
+ "learning_rate": 1.090909090909091e-06,
197
+ "loss": 5405.107,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.0022624617199557204,
202
+ "grad_norm": 68140.5078125,
203
+ "learning_rate": 1.1313131313131313e-06,
204
+ "loss": 2509.3971,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.0023432639242398534,
209
+ "grad_norm": 38560.15625,
210
+ "learning_rate": 1.1717171717171719e-06,
211
+ "loss": 1295.3724,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.002424066128523986,
216
+ "grad_norm": 2864.86474609375,
217
+ "learning_rate": 1.2121212121212122e-06,
218
+ "loss": 923.2405,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.002504868332808119,
223
+ "grad_norm": 2044.932861328125,
224
+ "learning_rate": 1.2525252525252527e-06,
225
+ "loss": 950.31,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.0025856705370922517,
230
+ "grad_norm": 2294.186279296875,
231
+ "learning_rate": 1.292929292929293e-06,
232
+ "loss": 830.1033,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.002666472741376385,
237
+ "grad_norm": 941.1571044921875,
238
+ "learning_rate": 1.3333333333333334e-06,
239
+ "loss": 711.3893,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.0027472749456605174,
244
+ "grad_norm": 6470.8056640625,
245
+ "learning_rate": 1.3737373737373738e-06,
246
+ "loss": 746.1014,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.0028280771499446505,
251
+ "grad_norm": 2412.6904296875,
252
+ "learning_rate": 1.4141414141414143e-06,
253
+ "loss": 774.8737,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.0029088793542287835,
258
+ "grad_norm": 2536.60791015625,
259
+ "learning_rate": 1.4545454545454546e-06,
260
+ "loss": 832.4265,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.002989681558512916,
265
+ "grad_norm": 5404.72265625,
266
+ "learning_rate": 1.4949494949494952e-06,
267
+ "loss": 676.5023,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.003070483762797049,
272
+ "grad_norm": 1271.8477783203125,
273
+ "learning_rate": 1.5353535353535355e-06,
274
+ "loss": 518.433,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.003151285967081182,
279
+ "grad_norm": 736.135009765625,
280
+ "learning_rate": 1.5757575757575759e-06,
281
+ "loss": 768.4572,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.003232088171365315,
286
+ "grad_norm": 1238.1878662109375,
287
+ "learning_rate": 1.6161616161616162e-06,
288
+ "loss": 696.9558,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.0033128903756494475,
293
+ "grad_norm": 1593.6436767578125,
294
+ "learning_rate": 1.6565656565656565e-06,
295
+ "loss": 670.7958,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.0033936925799335806,
300
+ "grad_norm": 819.6708374023438,
301
+ "learning_rate": 1.6969696969696973e-06,
302
+ "loss": 581.9092,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.0034744947842177136,
307
+ "grad_norm": 775.354248046875,
308
+ "learning_rate": 1.7373737373737376e-06,
309
+ "loss": 744.2987,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0035552969885018462,
314
+ "grad_norm": 744.1176147460938,
315
+ "learning_rate": 1.777777777777778e-06,
316
+ "loss": 665.0154,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.0036360991927859793,
321
+ "grad_norm": 1123.35986328125,
322
+ "learning_rate": 1.818181818181818e-06,
323
+ "loss": 681.7526,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.003716901397070112,
328
+ "grad_norm": 2373.274658203125,
329
+ "learning_rate": 1.8585858585858588e-06,
330
+ "loss": 654.0196,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.003797703601354245,
335
+ "grad_norm": 1098.4403076171875,
336
+ "learning_rate": 1.8989898989898992e-06,
337
+ "loss": 657.3389,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.003878505805638378,
342
+ "grad_norm": 1008.40966796875,
343
+ "learning_rate": 1.9393939393939395e-06,
344
+ "loss": 564.041,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.003959308009922511,
349
+ "grad_norm": 1227.1414794921875,
350
+ "learning_rate": 1.9797979797979796e-06,
351
+ "loss": 614.5699,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.004040110214206643,
356
+ "grad_norm": 57328.9609375,
357
+ "learning_rate": 2.0202020202020206e-06,
358
+ "loss": 596.2117,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.004120912418490777,
363
+ "grad_norm": 2744.44873046875,
364
+ "learning_rate": 2.0606060606060607e-06,
365
+ "loss": 676.9374,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.004201714622774909,
370
+ "grad_norm": 667.0958862304688,
371
+ "learning_rate": 2.1010101010101013e-06,
372
+ "loss": 812.1784,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.004282516827059042,
377
+ "grad_norm": 1447.203857421875,
378
+ "learning_rate": 2.1414141414141414e-06,
379
+ "loss": 640.1749,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.004363319031343175,
384
+ "grad_norm": 1186.56982421875,
385
+ "learning_rate": 2.181818181818182e-06,
386
+ "loss": 665.659,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.004444121235627308,
391
+ "grad_norm": 1263.9395751953125,
392
+ "learning_rate": 2.2222222222222225e-06,
393
+ "loss": 696.9437,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.004524923439911441,
398
+ "grad_norm": 1353.597412109375,
399
+ "learning_rate": 2.2626262626262626e-06,
400
+ "loss": 654.0818,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.004605725644195573,
405
+ "grad_norm": 876.721923828125,
406
+ "learning_rate": 2.303030303030303e-06,
407
+ "loss": 683.512,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.004686527848479707,
412
+ "grad_norm": 1866.14697265625,
413
+ "learning_rate": 2.3434343434343437e-06,
414
+ "loss": 727.845,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.0047673300527638395,
419
+ "grad_norm": 667.1629028320312,
420
+ "learning_rate": 2.383838383838384e-06,
421
+ "loss": 625.2918,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.004848132257047972,
426
+ "grad_norm": 614.8545532226562,
427
+ "learning_rate": 2.4242424242424244e-06,
428
+ "loss": 600.5201,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.004928934461332105,
433
+ "grad_norm": 879.8363647460938,
434
+ "learning_rate": 2.4646464646464645e-06,
435
+ "loss": 509.5772,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.005009736665616238,
440
+ "grad_norm": 1344.7303466796875,
441
+ "learning_rate": 2.5050505050505055e-06,
442
+ "loss": 639.9966,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.005090538869900371,
447
+ "grad_norm": 1223.520751953125,
448
+ "learning_rate": 2.5454545454545456e-06,
449
+ "loss": 706.7377,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.0051713410741845035,
454
+ "grad_norm": 1871.1324462890625,
455
+ "learning_rate": 2.585858585858586e-06,
456
+ "loss": 712.2218,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.005252143278468637,
461
+ "grad_norm": 804.082763671875,
462
+ "learning_rate": 2.6262626262626263e-06,
463
+ "loss": 666.623,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.00533294548275277,
468
+ "grad_norm": 10994.2314453125,
469
+ "learning_rate": 2.666666666666667e-06,
470
+ "loss": 726.5861,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.005413747687036902,
475
+ "grad_norm": 892.67919921875,
476
+ "learning_rate": 2.7070707070707074e-06,
477
+ "loss": 596.617,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.005494549891321035,
482
+ "grad_norm": 681.8887939453125,
483
+ "learning_rate": 2.7474747474747475e-06,
484
+ "loss": 496.8398,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.005575352095605168,
489
+ "grad_norm": 1190.68310546875,
490
+ "learning_rate": 2.787878787878788e-06,
491
+ "loss": 684.7522,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.005656154299889301,
496
+ "grad_norm": 1649.9376220703125,
497
+ "learning_rate": 2.8282828282828286e-06,
498
+ "loss": 717.8954,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.0057369565041734336,
503
+ "grad_norm": 2140.240234375,
504
+ "learning_rate": 2.8686868686868687e-06,
505
+ "loss": 633.3918,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.005817758708457567,
510
+ "grad_norm": 1765.3192138671875,
511
+ "learning_rate": 2.9090909090909093e-06,
512
+ "loss": 656.2145,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.0058985609127417,
517
+ "grad_norm": 2076.052001953125,
518
+ "learning_rate": 2.9494949494949494e-06,
519
+ "loss": 620.9487,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.005979363117025832,
524
+ "grad_norm": 817.6283569335938,
525
+ "learning_rate": 2.9898989898989904e-06,
526
+ "loss": 427.1178,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.006060165321309965,
531
+ "grad_norm": 1084.86767578125,
532
+ "learning_rate": 3.0303030303030305e-06,
533
+ "loss": 560.9343,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.006140967525594098,
538
+ "grad_norm": 786.8311767578125,
539
+ "learning_rate": 3.070707070707071e-06,
540
+ "loss": 520.7236,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.006221769729878231,
545
+ "grad_norm": 4558.533203125,
546
+ "learning_rate": 3.111111111111111e-06,
547
+ "loss": 555.48,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.006302571934162364,
552
+ "grad_norm": 785.0099487304688,
553
+ "learning_rate": 3.1515151515151517e-06,
554
+ "loss": 514.0267,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.006383374138446497,
559
+ "grad_norm": 1063.466796875,
560
+ "learning_rate": 3.191919191919192e-06,
561
+ "loss": 592.0211,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.00646417634273063,
566
+ "grad_norm": 2228.42041015625,
567
+ "learning_rate": 3.2323232323232324e-06,
568
+ "loss": 613.0331,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.006544978547014762,
573
+ "grad_norm": 1026.750732421875,
574
+ "learning_rate": 3.2727272727272733e-06,
575
+ "loss": 639.7108,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.006625780751298895,
580
+ "grad_norm": 2451.453369140625,
581
+ "learning_rate": 3.313131313131313e-06,
582
+ "loss": 659.9323,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.0067065829555830285,
587
+ "grad_norm": 1183.8045654296875,
588
+ "learning_rate": 3.3535353535353536e-06,
589
+ "loss": 601.1871,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.006787385159867161,
594
+ "grad_norm": 1861.18701171875,
595
+ "learning_rate": 3.3939393939393946e-06,
596
+ "loss": 589.3404,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.006868187364151294,
601
+ "grad_norm": 1340.5020751953125,
602
+ "learning_rate": 3.4343434343434343e-06,
603
+ "loss": 588.7653,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.006948989568435427,
608
+ "grad_norm": 8932.84375,
609
+ "learning_rate": 3.4747474747474752e-06,
610
+ "loss": 667.0873,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.00702979177271956,
615
+ "grad_norm": 1353.702392578125,
616
+ "learning_rate": 3.515151515151515e-06,
617
+ "loss": 574.4604,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.0071105939770036925,
622
+ "grad_norm": 1281.541748046875,
623
+ "learning_rate": 3.555555555555556e-06,
624
+ "loss": 683.6188,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.007191396181287825,
629
+ "grad_norm": 3347.4111328125,
630
+ "learning_rate": 3.5959595959595965e-06,
631
+ "loss": 591.1179,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.007272198385571959,
636
+ "grad_norm": 832.118896484375,
637
+ "learning_rate": 3.636363636363636e-06,
638
+ "loss": 509.1054,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.007353000589856091,
643
+ "grad_norm": 3215.1875,
644
+ "learning_rate": 3.676767676767677e-06,
645
+ "loss": 634.5958,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.007433802794140224,
650
+ "grad_norm": 1072.3865966796875,
651
+ "learning_rate": 3.7171717171717177e-06,
652
+ "loss": 525.3159,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.007514604998424357,
657
+ "grad_norm": 1002.2363891601562,
658
+ "learning_rate": 3.757575757575758e-06,
659
+ "loss": 659.7474,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.00759540720270849,
664
+ "grad_norm": 6355.693359375,
665
+ "learning_rate": 3.7979797979797984e-06,
666
+ "loss": 636.0396,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.007676209406992623,
671
+ "grad_norm": 1635.6080322265625,
672
+ "learning_rate": 3.8383838383838385e-06,
673
+ "loss": 624.7128,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.007757011611276756,
678
+ "grad_norm": 1284.3531494140625,
679
+ "learning_rate": 3.878787878787879e-06,
680
+ "loss": 637.0934,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.007837813815560889,
685
+ "grad_norm": 2921.176025390625,
686
+ "learning_rate": 3.9191919191919196e-06,
687
+ "loss": 648.8942,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.007918616019845021,
692
+ "grad_norm": 845.0542602539062,
693
+ "learning_rate": 3.959595959595959e-06,
694
+ "loss": 536.0546,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.007999418224129154,
699
+ "grad_norm": 986.8812866210938,
700
+ "learning_rate": 4.000000000000001e-06,
701
+ "loss": 643.056,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.008080220428413287,
706
+ "grad_norm": 748.7238159179688,
707
+ "learning_rate": 4.040404040404041e-06,
708
+ "loss": 587.7603,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.00816102263269742,
713
+ "grad_norm": 4340.18017578125,
714
+ "learning_rate": 4.080808080808081e-06,
715
+ "loss": 545.2501,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.008241824836981554,
720
+ "grad_norm": 1614.0966796875,
721
+ "learning_rate": 4.1212121212121215e-06,
722
+ "loss": 652.8924,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.008322627041265686,
727
+ "grad_norm": 10709.900390625,
728
+ "learning_rate": 4.161616161616161e-06,
729
+ "loss": 598.342,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.008403429245549819,
734
+ "grad_norm": 870.67578125,
735
+ "learning_rate": 4.2020202020202026e-06,
736
+ "loss": 612.1897,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.008484231449833951,
741
+ "grad_norm": 686.4441528320312,
742
+ "learning_rate": 4.242424242424243e-06,
743
+ "loss": 564.3604,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.008565033654118084,
748
+ "grad_norm": 800.3753051757812,
749
+ "learning_rate": 4.282828282828283e-06,
750
+ "loss": 515.465,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.008645835858402217,
755
+ "grad_norm": 1612.2467041015625,
756
+ "learning_rate": 4.323232323232323e-06,
757
+ "loss": 615.4043,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.00872663806268635,
762
+ "grad_norm": 1488.6268310546875,
763
+ "learning_rate": 4.363636363636364e-06,
764
+ "loss": 584.7924,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.008807440266970484,
769
+ "grad_norm": 1085.609619140625,
770
+ "learning_rate": 4.4040404040404044e-06,
771
+ "loss": 568.4938,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.008888242471254616,
776
+ "grad_norm": 1548.931884765625,
777
+ "learning_rate": 4.444444444444445e-06,
778
+ "loss": 515.5535,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.008969044675538749,
783
+ "grad_norm": 1010.5916137695312,
784
+ "learning_rate": 4.484848484848485e-06,
785
+ "loss": 496.3722,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.009049846879822881,
790
+ "grad_norm": 1959.3551025390625,
791
+ "learning_rate": 4.525252525252525e-06,
792
+ "loss": 697.1435,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.009130649084107014,
797
+ "grad_norm": 2370.032470703125,
798
+ "learning_rate": 4.565656565656566e-06,
799
+ "loss": 541.7345,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.009211451288391147,
804
+ "grad_norm": 1848.2596435546875,
805
+ "learning_rate": 4.606060606060606e-06,
806
+ "loss": 539.6271,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.00929225349267528,
811
+ "grad_norm": 2245.12548828125,
812
+ "learning_rate": 4.646464646464647e-06,
813
+ "loss": 485.2028,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.009373055696959414,
818
+ "grad_norm": 1303.244140625,
819
+ "learning_rate": 4.6868686868686874e-06,
820
+ "loss": 552.4869,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.009453857901243546,
825
+ "grad_norm": 813.8889770507812,
826
+ "learning_rate": 4.727272727272727e-06,
827
+ "loss": 566.426,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.009534660105527679,
832
+ "grad_norm": 2155.474609375,
833
+ "learning_rate": 4.767676767676768e-06,
834
+ "loss": 694.8829,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.009615462309811812,
839
+ "grad_norm": 3313.629150390625,
840
+ "learning_rate": 4.808080808080808e-06,
841
+ "loss": 541.8082,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.009696264514095944,
846
+ "grad_norm": 830.4761352539062,
847
+ "learning_rate": 4.848484848484849e-06,
848
+ "loss": 633.2839,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.009777066718380077,
853
+ "grad_norm": 2314.525634765625,
854
+ "learning_rate": 4.888888888888889e-06,
855
+ "loss": 465.8495,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.00985786892266421,
860
+ "grad_norm": 1821.2579345703125,
861
+ "learning_rate": 4.929292929292929e-06,
862
+ "loss": 543.8752,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.009938671126948344,
867
+ "grad_norm": 471.08380126953125,
868
+ "learning_rate": 4.96969696969697e-06,
869
+ "loss": 405.759,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.010019473331232476,
874
+ "grad_norm": 1476.8787841796875,
875
+ "learning_rate": 5.010101010101011e-06,
876
+ "loss": 617.1836,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.010100275535516609,
881
+ "grad_norm": 667.0234985351562,
882
+ "learning_rate": 5.050505050505051e-06,
883
+ "loss": 513.511,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.010181077739800742,
888
+ "grad_norm": 613.2576904296875,
889
+ "learning_rate": 5.090909090909091e-06,
890
+ "loss": 468.2697,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.010261879944084874,
895
+ "grad_norm": 1321.9991455078125,
896
+ "learning_rate": 5.131313131313131e-06,
897
+ "loss": 476.6512,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.010342682148369007,
902
+ "grad_norm": 840.0568237304688,
903
+ "learning_rate": 5.171717171717172e-06,
904
+ "loss": 607.1966,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.01042348435265314,
909
+ "grad_norm": 1686.292724609375,
910
+ "learning_rate": 5.212121212121213e-06,
911
+ "loss": 538.6932,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.010504286556937274,
916
+ "grad_norm": 2623.5302734375,
917
+ "learning_rate": 5.2525252525252526e-06,
918
+ "loss": 511.9114,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.010585088761221407,
923
+ "grad_norm": 1300.63671875,
924
+ "learning_rate": 5.292929292929293e-06,
925
+ "loss": 573.5399,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.01066589096550554,
930
+ "grad_norm": 828.8121948242188,
931
+ "learning_rate": 5.333333333333334e-06,
932
+ "loss": 617.9385,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.010746693169789672,
937
+ "grad_norm": 965.5888671875,
938
+ "learning_rate": 5.373737373737374e-06,
939
+ "loss": 428.5168,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.010827495374073804,
944
+ "grad_norm": 1123.3818359375,
945
+ "learning_rate": 5.414141414141415e-06,
946
+ "loss": 582.797,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.010908297578357937,
951
+ "grad_norm": 4098.775390625,
952
+ "learning_rate": 5.4545454545454545e-06,
953
+ "loss": 633.2868,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.01098909978264207,
958
+ "grad_norm": 792.64306640625,
959
+ "learning_rate": 5.494949494949495e-06,
960
+ "loss": 688.7412,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.011069901986926204,
965
+ "grad_norm": 1111.2113037109375,
966
+ "learning_rate": 5.5353535353535355e-06,
967
+ "loss": 473.5282,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.011150704191210337,
972
+ "grad_norm": 5420.9765625,
973
+ "learning_rate": 5.575757575757576e-06,
974
+ "loss": 516.3693,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.01123150639549447,
979
+ "grad_norm": 1263.883056640625,
980
+ "learning_rate": 5.616161616161617e-06,
981
+ "loss": 457.2747,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.011312308599778602,
986
+ "grad_norm": 944.7489624023438,
987
+ "learning_rate": 5.656565656565657e-06,
988
+ "loss": 539.8824,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.011393110804062734,
993
+ "grad_norm": 562.0636596679688,
994
+ "learning_rate": 5.696969696969697e-06,
995
+ "loss": 455.684,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.011473913008346867,
1000
+ "grad_norm": 1293.833984375,
1001
+ "learning_rate": 5.7373737373737374e-06,
1002
+ "loss": 527.1053,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.011554715212631,
1007
+ "grad_norm": 1857.3572998046875,
1008
+ "learning_rate": 5.777777777777778e-06,
1009
+ "loss": 519.1355,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.011635517416915134,
1014
+ "grad_norm": 1110.5677490234375,
1015
+ "learning_rate": 5.8181818181818185e-06,
1016
+ "loss": 470.4913,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.011716319621199267,
1021
+ "grad_norm": 1656.9898681640625,
1022
+ "learning_rate": 5.858585858585859e-06,
1023
+ "loss": 570.9559,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.0117971218254834,
1028
+ "grad_norm": 830.130126953125,
1029
+ "learning_rate": 5.898989898989899e-06,
1030
+ "loss": 551.6251,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.011877924029767532,
1035
+ "grad_norm": 718.1753540039062,
1036
+ "learning_rate": 5.93939393939394e-06,
1037
+ "loss": 495.0048,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.011958726234051665,
1042
+ "grad_norm": 2903.817138671875,
1043
+ "learning_rate": 5.979797979797981e-06,
1044
+ "loss": 742.8938,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.012039528438335797,
1049
+ "grad_norm": 893.6466674804688,
1050
+ "learning_rate": 6.0202020202020204e-06,
1051
+ "loss": 542.4396,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.01212033064261993,
1056
+ "grad_norm": 1576.75390625,
1057
+ "learning_rate": 6.060606060606061e-06,
1058
+ "loss": 513.3051,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.012201132846904064,
1063
+ "grad_norm": 1925.69140625,
1064
+ "learning_rate": 6.101010101010101e-06,
1065
+ "loss": 450.3941,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.012281935051188197,
1070
+ "grad_norm": 1998.0458984375,
1071
+ "learning_rate": 6.141414141414142e-06,
1072
+ "loss": 635.4773,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.01236273725547233,
1077
+ "grad_norm": 1258.9659423828125,
1078
+ "learning_rate": 6.181818181818183e-06,
1079
+ "loss": 595.3451,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.012443539459756462,
1084
+ "grad_norm": 1726.843017578125,
1085
+ "learning_rate": 6.222222222222222e-06,
1086
+ "loss": 556.3297,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.012524341664040595,
1091
+ "grad_norm": 2486.63134765625,
1092
+ "learning_rate": 6.262626262626263e-06,
1093
+ "loss": 647.7714,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.012605143868324727,
1098
+ "grad_norm": 1644.4110107421875,
1099
+ "learning_rate": 6.303030303030303e-06,
1100
+ "loss": 611.4104,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.01268594607260886,
1105
+ "grad_norm": 2169.43408203125,
1106
+ "learning_rate": 6.343434343434344e-06,
1107
+ "loss": 445.9718,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.012766748276892994,
1112
+ "grad_norm": 1554.9425048828125,
1113
+ "learning_rate": 6.383838383838384e-06,
1114
+ "loss": 561.5079,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.012847550481177127,
1119
+ "grad_norm": 891.0722045898438,
1120
+ "learning_rate": 6.424242424242424e-06,
1121
+ "loss": 510.2931,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.01292835268546126,
1126
+ "grad_norm": 22828.455078125,
1127
+ "learning_rate": 6.464646464646465e-06,
1128
+ "loss": 552.862,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.013009154889745392,
1133
+ "grad_norm": 722.2283325195312,
1134
+ "learning_rate": 6.505050505050505e-06,
1135
+ "loss": 512.3786,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.013089957094029525,
1140
+ "grad_norm": 1105.0716552734375,
1141
+ "learning_rate": 6.545454545454547e-06,
1142
+ "loss": 532.3522,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.013170759298313657,
1147
+ "grad_norm": 1146.0550537109375,
1148
+ "learning_rate": 6.5858585858585856e-06,
1149
+ "loss": 647.0639,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.01325156150259779,
1154
+ "grad_norm": 2632.366943359375,
1155
+ "learning_rate": 6.626262626262626e-06,
1156
+ "loss": 613.0408,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.013332363706881924,
1161
+ "grad_norm": 1236.950439453125,
1162
+ "learning_rate": 6.666666666666667e-06,
1163
+ "loss": 614.0946,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.013413165911166057,
1168
+ "grad_norm": 973.5294799804688,
1169
+ "learning_rate": 6.707070707070707e-06,
1170
+ "loss": 383.6397,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.01349396811545019,
1175
+ "grad_norm": 1101.2989501953125,
1176
+ "learning_rate": 6.747474747474749e-06,
1177
+ "loss": 556.5678,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.013574770319734322,
1182
+ "grad_norm": 694.1087646484375,
1183
+ "learning_rate": 6.787878787878789e-06,
1184
+ "loss": 415.0727,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.013655572524018455,
1189
+ "grad_norm": 1031.21240234375,
1190
+ "learning_rate": 6.828282828282828e-06,
1191
+ "loss": 508.7032,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.013736374728302587,
1196
+ "grad_norm": 734.2860717773438,
1197
+ "learning_rate": 6.8686868686868685e-06,
1198
+ "loss": 482.0532,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.01381717693258672,
1203
+ "grad_norm": 879.1790771484375,
1204
+ "learning_rate": 6.909090909090909e-06,
1205
+ "loss": 655.4877,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.013897979136870854,
1210
+ "grad_norm": 1699.49560546875,
1211
+ "learning_rate": 6.9494949494949505e-06,
1212
+ "loss": 400.7212,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.013978781341154987,
1217
+ "grad_norm": 1901.2413330078125,
1218
+ "learning_rate": 6.989898989898991e-06,
1219
+ "loss": 496.4101,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.01405958354543912,
1224
+ "grad_norm": 1368.836669921875,
1225
+ "learning_rate": 7.03030303030303e-06,
1226
+ "loss": 520.5295,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.014140385749723252,
1231
+ "grad_norm": 1587.8297119140625,
1232
+ "learning_rate": 7.0707070707070704e-06,
1233
+ "loss": 397.7875,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.014221187954007385,
1238
+ "grad_norm": 1054.773681640625,
1239
+ "learning_rate": 7.111111111111112e-06,
1240
+ "loss": 456.9818,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.014301990158291518,
1245
+ "grad_norm": 3368.4013671875,
1246
+ "learning_rate": 7.151515151515152e-06,
1247
+ "loss": 397.1987,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.01438279236257565,
1252
+ "grad_norm": 1371.4473876953125,
1253
+ "learning_rate": 7.191919191919193e-06,
1254
+ "loss": 433.1569,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.014463594566859785,
1259
+ "grad_norm": 1193.6456298828125,
1260
+ "learning_rate": 7.232323232323232e-06,
1261
+ "loss": 368.5404,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.014544396771143917,
1266
+ "grad_norm": 976.5448608398438,
1267
+ "learning_rate": 7.272727272727272e-06,
1268
+ "loss": 427.9085,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.01462519897542805,
1273
+ "grad_norm": 665.8095092773438,
1274
+ "learning_rate": 7.313131313131314e-06,
1275
+ "loss": 461.1592,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.014706001179712182,
1280
+ "grad_norm": 1352.773681640625,
1281
+ "learning_rate": 7.353535353535354e-06,
1282
+ "loss": 421.6953,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.014786803383996315,
1287
+ "grad_norm": 1284.1273193359375,
1288
+ "learning_rate": 7.393939393939395e-06,
1289
+ "loss": 486.5889,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.014867605588280448,
1294
+ "grad_norm": 3249.41650390625,
1295
+ "learning_rate": 7.434343434343435e-06,
1296
+ "loss": 534.9857,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.01494840779256458,
1301
+ "grad_norm": 1162.9168701171875,
1302
+ "learning_rate": 7.474747474747475e-06,
1303
+ "loss": 684.8136,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.015029209996848715,
1308
+ "grad_norm": 2700.0771484375,
1309
+ "learning_rate": 7.515151515151516e-06,
1310
+ "loss": 438.9466,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.015110012201132847,
1315
+ "grad_norm": 1472.51171875,
1316
+ "learning_rate": 7.555555555555556e-06,
1317
+ "loss": 541.1812,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.01519081440541698,
1322
+ "grad_norm": 763.5363159179688,
1323
+ "learning_rate": 7.595959595959597e-06,
1324
+ "loss": 574.8496,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.015271616609701113,
1329
+ "grad_norm": 1097.047607421875,
1330
+ "learning_rate": 7.636363636363638e-06,
1331
+ "loss": 495.2874,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.015352418813985245,
1336
+ "grad_norm": 1086.6981201171875,
1337
+ "learning_rate": 7.676767676767677e-06,
1338
+ "loss": 494.5146,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.015433221018269378,
1343
+ "grad_norm": 1003.9575805664062,
1344
+ "learning_rate": 7.717171717171717e-06,
1345
+ "loss": 493.7404,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.015514023222553512,
1350
+ "grad_norm": 1008.3009643554688,
1351
+ "learning_rate": 7.757575757575758e-06,
1352
+ "loss": 507.6875,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.015594825426837645,
1357
+ "grad_norm": 2552.92626953125,
1358
+ "learning_rate": 7.797979797979799e-06,
1359
+ "loss": 420.8823,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.015675627631121777,
1364
+ "grad_norm": 735.5524291992188,
1365
+ "learning_rate": 7.838383838383839e-06,
1366
+ "loss": 519.3292,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.01575642983540591,
1371
+ "grad_norm": 1559.79296875,
1372
+ "learning_rate": 7.878787878787878e-06,
1373
+ "loss": 468.1747,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.015837232039690043,
1378
+ "grad_norm": 1734.436767578125,
1379
+ "learning_rate": 7.919191919191919e-06,
1380
+ "loss": 417.3558,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.015918034243974175,
1385
+ "grad_norm": 2985.763671875,
1386
+ "learning_rate": 7.959595959595959e-06,
1387
+ "loss": 427.185,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.015998836448258308,
1392
+ "grad_norm": 1284.0130615234375,
1393
+ "learning_rate": 8.000000000000001e-06,
1394
+ "loss": 523.1269,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.01607963865254244,
1399
+ "grad_norm": 1079.0145263671875,
1400
+ "learning_rate": 8.040404040404042e-06,
1401
+ "loss": 497.3142,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.016160440856826573,
1406
+ "grad_norm": 1260.9786376953125,
1407
+ "learning_rate": 8.080808080808082e-06,
1408
+ "loss": 495.9584,
1409
+ "step": 2000
1410
+ },
1411
+ {
1412
+ "epoch": 0.016241243061110706,
1413
+ "grad_norm": 4998.38525390625,
1414
+ "learning_rate": 8.121212121212121e-06,
1415
+ "loss": 425.5371,
1416
+ "step": 2010
1417
+ },
1418
+ {
1419
+ "epoch": 0.01632204526539484,
1420
+ "grad_norm": 1627.860595703125,
1421
+ "learning_rate": 8.161616161616162e-06,
1422
+ "loss": 547.4274,
1423
+ "step": 2020
1424
+ },
1425
+ {
1426
+ "epoch": 0.016402847469678974,
1427
+ "grad_norm": 1790.0679931640625,
1428
+ "learning_rate": 8.202020202020202e-06,
1429
+ "loss": 519.9649,
1430
+ "step": 2030
1431
+ },
1432
+ {
1433
+ "epoch": 0.016483649673963107,
1434
+ "grad_norm": 1015.8907470703125,
1435
+ "learning_rate": 8.242424242424243e-06,
1436
+ "loss": 439.1214,
1437
+ "step": 2040
1438
+ },
1439
+ {
1440
+ "epoch": 0.01656445187824724,
1441
+ "grad_norm": 1249.7930908203125,
1442
+ "learning_rate": 8.282828282828283e-06,
1443
+ "loss": 493.0195,
1444
+ "step": 2050
1445
+ },
1446
+ {
1447
+ "epoch": 0.016645254082531372,
1448
+ "grad_norm": 7022.84423828125,
1449
+ "learning_rate": 8.323232323232322e-06,
1450
+ "loss": 497.0867,
1451
+ "step": 2060
1452
+ },
1453
+ {
1454
+ "epoch": 0.016726056286815505,
1455
+ "grad_norm": 985.5494384765625,
1456
+ "learning_rate": 8.363636363636365e-06,
1457
+ "loss": 464.8873,
1458
+ "step": 2070
1459
+ },
1460
+ {
1461
+ "epoch": 0.016806858491099638,
1462
+ "grad_norm": 977.4537353515625,
1463
+ "learning_rate": 8.404040404040405e-06,
1464
+ "loss": 548.6022,
1465
+ "step": 2080
1466
+ },
1467
+ {
1468
+ "epoch": 0.01688766069538377,
1469
+ "grad_norm": 1503.036376953125,
1470
+ "learning_rate": 8.444444444444446e-06,
1471
+ "loss": 535.7155,
1472
+ "step": 2090
1473
+ },
1474
+ {
1475
+ "epoch": 0.016968462899667903,
1476
+ "grad_norm": 876.1661987304688,
1477
+ "learning_rate": 8.484848484848486e-06,
1478
+ "loss": 439.9817,
1479
+ "step": 2100
1480
+ },
1481
+ {
1482
+ "epoch": 0.017049265103952035,
1483
+ "grad_norm": 631.0904541015625,
1484
+ "learning_rate": 8.525252525252525e-06,
1485
+ "loss": 432.0362,
1486
+ "step": 2110
1487
+ },
1488
+ {
1489
+ "epoch": 0.017130067308236168,
1490
+ "grad_norm": 1009.0225830078125,
1491
+ "learning_rate": 8.565656565656566e-06,
1492
+ "loss": 560.3087,
1493
+ "step": 2120
1494
+ },
1495
+ {
1496
+ "epoch": 0.0172108695125203,
1497
+ "grad_norm": 893.6896362304688,
1498
+ "learning_rate": 8.606060606060606e-06,
1499
+ "loss": 574.9824,
1500
+ "step": 2130
1501
+ },
1502
+ {
1503
+ "epoch": 0.017291671716804433,
1504
+ "grad_norm": 1586.4207763671875,
1505
+ "learning_rate": 8.646464646464647e-06,
1506
+ "loss": 541.0257,
1507
+ "step": 2140
1508
+ },
1509
+ {
1510
+ "epoch": 0.017372473921088566,
1511
+ "grad_norm": 1161.7142333984375,
1512
+ "learning_rate": 8.686868686868687e-06,
1513
+ "loss": 360.7528,
1514
+ "step": 2150
1515
+ },
1516
+ {
1517
+ "epoch": 0.0174532761253727,
1518
+ "grad_norm": 1364.39697265625,
1519
+ "learning_rate": 8.727272727272728e-06,
1520
+ "loss": 511.5766,
1521
+ "step": 2160
1522
+ },
1523
+ {
1524
+ "epoch": 0.017534078329656835,
1525
+ "grad_norm": 1430.6226806640625,
1526
+ "learning_rate": 8.767676767676768e-06,
1527
+ "loss": 374.0022,
1528
+ "step": 2170
1529
+ },
1530
+ {
1531
+ "epoch": 0.017614880533940967,
1532
+ "grad_norm": 944.2110595703125,
1533
+ "learning_rate": 8.808080808080809e-06,
1534
+ "loss": 385.8277,
1535
+ "step": 2180
1536
+ },
1537
+ {
1538
+ "epoch": 0.0176956827382251,
1539
+ "grad_norm": 1198.7069091796875,
1540
+ "learning_rate": 8.84848484848485e-06,
1541
+ "loss": 459.1448,
1542
+ "step": 2190
1543
+ },
1544
+ {
1545
+ "epoch": 0.017776484942509233,
1546
+ "grad_norm": 969.3467407226562,
1547
+ "learning_rate": 8.88888888888889e-06,
1548
+ "loss": 472.9384,
1549
+ "step": 2200
1550
+ },
1551
+ {
1552
+ "epoch": 0.017857287146793365,
1553
+ "grad_norm": 1518.4356689453125,
1554
+ "learning_rate": 8.92929292929293e-06,
1555
+ "loss": 389.709,
1556
+ "step": 2210
1557
+ },
1558
+ {
1559
+ "epoch": 0.017938089351077498,
1560
+ "grad_norm": 1016.8683471679688,
1561
+ "learning_rate": 8.96969696969697e-06,
1562
+ "loss": 416.4938,
1563
+ "step": 2220
1564
+ },
1565
+ {
1566
+ "epoch": 0.01801889155536163,
1567
+ "grad_norm": 1106.466796875,
1568
+ "learning_rate": 9.01010101010101e-06,
1569
+ "loss": 430.7052,
1570
+ "step": 2230
1571
+ },
1572
+ {
1573
+ "epoch": 0.018099693759645763,
1574
+ "grad_norm": 959.3176879882812,
1575
+ "learning_rate": 9.05050505050505e-06,
1576
+ "loss": 439.8072,
1577
+ "step": 2240
1578
+ },
1579
+ {
1580
+ "epoch": 0.018180495963929896,
1581
+ "grad_norm": 892.5261840820312,
1582
+ "learning_rate": 9.090909090909091e-06,
1583
+ "loss": 433.3271,
1584
+ "step": 2250
1585
+ },
1586
+ {
1587
+ "epoch": 0.018261298168214028,
1588
+ "grad_norm": 2410.990234375,
1589
+ "learning_rate": 9.131313131313132e-06,
1590
+ "loss": 393.4176,
1591
+ "step": 2260
1592
+ },
1593
+ {
1594
+ "epoch": 0.01834210037249816,
1595
+ "grad_norm": 1493.900390625,
1596
+ "learning_rate": 9.171717171717172e-06,
1597
+ "loss": 412.9764,
1598
+ "step": 2270
1599
+ },
1600
+ {
1601
+ "epoch": 0.018422902576782293,
1602
+ "grad_norm": 2805.1650390625,
1603
+ "learning_rate": 9.212121212121213e-06,
1604
+ "loss": 625.3552,
1605
+ "step": 2280
1606
+ },
1607
+ {
1608
+ "epoch": 0.018503704781066426,
1609
+ "grad_norm": 1327.8782958984375,
1610
+ "learning_rate": 9.252525252525253e-06,
1611
+ "loss": 473.8814,
1612
+ "step": 2290
1613
+ },
1614
+ {
1615
+ "epoch": 0.01858450698535056,
1616
+ "grad_norm": 1174.0333251953125,
1617
+ "learning_rate": 9.292929292929294e-06,
1618
+ "loss": 557.2924,
1619
+ "step": 2300
1620
+ },
1621
+ {
1622
+ "epoch": 0.018665309189634695,
1623
+ "grad_norm": 779.054931640625,
1624
+ "learning_rate": 9.333333333333334e-06,
1625
+ "loss": 340.7218,
1626
+ "step": 2310
1627
+ },
1628
+ {
1629
+ "epoch": 0.018746111393918827,
1630
+ "grad_norm": 1373.9344482421875,
1631
+ "learning_rate": 9.373737373737375e-06,
1632
+ "loss": 364.399,
1633
+ "step": 2320
1634
+ },
1635
+ {
1636
+ "epoch": 0.01882691359820296,
1637
+ "grad_norm": 1212.10595703125,
1638
+ "learning_rate": 9.414141414141414e-06,
1639
+ "loss": 532.5891,
1640
+ "step": 2330
1641
+ },
1642
+ {
1643
+ "epoch": 0.018907715802487093,
1644
+ "grad_norm": 1213.900146484375,
1645
+ "learning_rate": 9.454545454545454e-06,
1646
+ "loss": 452.1891,
1647
+ "step": 2340
1648
+ },
1649
+ {
1650
+ "epoch": 0.018988518006771225,
1651
+ "grad_norm": 5826.9482421875,
1652
+ "learning_rate": 9.494949494949495e-06,
1653
+ "loss": 416.3645,
1654
+ "step": 2350
1655
+ },
1656
+ {
1657
+ "epoch": 0.019069320211055358,
1658
+ "grad_norm": 6701.46630859375,
1659
+ "learning_rate": 9.535353535353535e-06,
1660
+ "loss": 441.6159,
1661
+ "step": 2360
1662
+ },
1663
+ {
1664
+ "epoch": 0.01915012241533949,
1665
+ "grad_norm": 2676.863525390625,
1666
+ "learning_rate": 9.575757575757578e-06,
1667
+ "loss": 418.8925,
1668
+ "step": 2370
1669
+ },
1670
+ {
1671
+ "epoch": 0.019230924619623623,
1672
+ "grad_norm": 1438.232177734375,
1673
+ "learning_rate": 9.616161616161616e-06,
1674
+ "loss": 484.5964,
1675
+ "step": 2380
1676
+ },
1677
+ {
1678
+ "epoch": 0.019311726823907756,
1679
+ "grad_norm": 2674.849853515625,
1680
+ "learning_rate": 9.656565656565657e-06,
1681
+ "loss": 448.1574,
1682
+ "step": 2390
1683
+ },
1684
+ {
1685
+ "epoch": 0.01939252902819189,
1686
+ "grad_norm": 1142.754638671875,
1687
+ "learning_rate": 9.696969696969698e-06,
1688
+ "loss": 368.5235,
1689
+ "step": 2400
1690
+ },
1691
+ {
1692
+ "epoch": 0.01947333123247602,
1693
+ "grad_norm": 1908.450927734375,
1694
+ "learning_rate": 9.737373737373738e-06,
1695
+ "loss": 448.1265,
1696
+ "step": 2410
1697
+ },
1698
+ {
1699
+ "epoch": 0.019554133436760154,
1700
+ "grad_norm": 930.4298706054688,
1701
+ "learning_rate": 9.777777777777779e-06,
1702
+ "loss": 412.3102,
1703
+ "step": 2420
1704
+ },
1705
+ {
1706
+ "epoch": 0.019634935641044286,
1707
+ "grad_norm": 885.7802124023438,
1708
+ "learning_rate": 9.818181818181818e-06,
1709
+ "loss": 425.3447,
1710
+ "step": 2430
1711
+ },
1712
+ {
1713
+ "epoch": 0.01971573784532842,
1714
+ "grad_norm": 1650.6605224609375,
1715
+ "learning_rate": 9.858585858585858e-06,
1716
+ "loss": 458.773,
1717
+ "step": 2440
1718
+ },
1719
+ {
1720
+ "epoch": 0.019796540049612555,
1721
+ "grad_norm": 1456.41943359375,
1722
+ "learning_rate": 9.898989898989899e-06,
1723
+ "loss": 453.6257,
1724
+ "step": 2450
1725
+ },
1726
+ {
1727
+ "epoch": 0.019877342253896688,
1728
+ "grad_norm": 1131.8419189453125,
1729
+ "learning_rate": 9.93939393939394e-06,
1730
+ "loss": 404.5182,
1731
+ "step": 2460
1732
+ },
1733
+ {
1734
+ "epoch": 0.01995814445818082,
1735
+ "grad_norm": 1003.8864135742188,
1736
+ "learning_rate": 9.979797979797981e-06,
1737
+ "loss": 494.8275,
1738
+ "step": 2470
1739
+ },
1740
+ {
1741
+ "epoch": 0.020038946662464953,
1742
+ "grad_norm": 3438.40673828125,
1743
+ "learning_rate": 1.0020202020202022e-05,
1744
+ "loss": 434.0848,
1745
+ "step": 2480
1746
+ },
1747
+ {
1748
+ "epoch": 0.020119748866749085,
1749
+ "grad_norm": 1796.56298828125,
1750
+ "learning_rate": 1.006060606060606e-05,
1751
+ "loss": 568.4981,
1752
+ "step": 2490
1753
+ },
1754
+ {
1755
+ "epoch": 0.020200551071033218,
1756
+ "grad_norm": 2394.705322265625,
1757
+ "learning_rate": 1.0101010101010101e-05,
1758
+ "loss": 468.4812,
1759
+ "step": 2500
1760
+ },
1761
+ {
1762
+ "epoch": 0.02028135327531735,
1763
+ "grad_norm": 1344.266845703125,
1764
+ "learning_rate": 1.0141414141414142e-05,
1765
+ "loss": 344.5415,
1766
+ "step": 2510
1767
+ },
1768
+ {
1769
+ "epoch": 0.020362155479601483,
1770
+ "grad_norm": 1845.0267333984375,
1771
+ "learning_rate": 1.0181818181818182e-05,
1772
+ "loss": 466.122,
1773
+ "step": 2520
1774
+ },
1775
+ {
1776
+ "epoch": 0.020442957683885616,
1777
+ "grad_norm": 2368.758544921875,
1778
+ "learning_rate": 1.0222222222222223e-05,
1779
+ "loss": 449.0622,
1780
+ "step": 2530
1781
+ },
1782
+ {
1783
+ "epoch": 0.02052375988816975,
1784
+ "grad_norm": 1838.78955078125,
1785
+ "learning_rate": 1.0262626262626262e-05,
1786
+ "loss": 408.0261,
1787
+ "step": 2540
1788
+ },
1789
+ {
1790
+ "epoch": 0.02060456209245388,
1791
+ "grad_norm": 1171.175537109375,
1792
+ "learning_rate": 1.0303030303030304e-05,
1793
+ "loss": 438.3216,
1794
+ "step": 2550
1795
+ },
1796
+ {
1797
+ "epoch": 0.020685364296738014,
1798
+ "grad_norm": 997.9658813476562,
1799
+ "learning_rate": 1.0343434343434345e-05,
1800
+ "loss": 396.0697,
1801
+ "step": 2560
1802
+ },
1803
+ {
1804
+ "epoch": 0.020766166501022146,
1805
+ "grad_norm": 955.8753662109375,
1806
+ "learning_rate": 1.0383838383838385e-05,
1807
+ "loss": 352.4154,
1808
+ "step": 2570
1809
+ },
1810
+ {
1811
+ "epoch": 0.02084696870530628,
1812
+ "grad_norm": 1111.3548583984375,
1813
+ "learning_rate": 1.0424242424242426e-05,
1814
+ "loss": 408.0275,
1815
+ "step": 2580
1816
+ },
1817
+ {
1818
+ "epoch": 0.020927770909590415,
1819
+ "grad_norm": 1489.833740234375,
1820
+ "learning_rate": 1.0464646464646465e-05,
1821
+ "loss": 442.6512,
1822
+ "step": 2590
1823
+ },
1824
+ {
1825
+ "epoch": 0.021008573113874548,
1826
+ "grad_norm": 1263.6846923828125,
1827
+ "learning_rate": 1.0505050505050505e-05,
1828
+ "loss": 451.8872,
1829
+ "step": 2600
1830
+ },
1831
+ {
1832
+ "epoch": 0.02108937531815868,
1833
+ "grad_norm": 1563.267333984375,
1834
+ "learning_rate": 1.0545454545454546e-05,
1835
+ "loss": 433.7399,
1836
+ "step": 2610
1837
+ },
1838
+ {
1839
+ "epoch": 0.021170177522442813,
1840
+ "grad_norm": 2102.853515625,
1841
+ "learning_rate": 1.0585858585858586e-05,
1842
+ "loss": 486.5411,
1843
+ "step": 2620
1844
+ },
1845
+ {
1846
+ "epoch": 0.021250979726726946,
1847
+ "grad_norm": 1215.5638427734375,
1848
+ "learning_rate": 1.0626262626262627e-05,
1849
+ "loss": 489.4264,
1850
+ "step": 2630
1851
+ },
1852
+ {
1853
+ "epoch": 0.02133178193101108,
1854
+ "grad_norm": 1669.6094970703125,
1855
+ "learning_rate": 1.0666666666666667e-05,
1856
+ "loss": 421.0313,
1857
+ "step": 2640
1858
+ },
1859
+ {
1860
+ "epoch": 0.02141258413529521,
1861
+ "grad_norm": 1581.879150390625,
1862
+ "learning_rate": 1.0707070707070708e-05,
1863
+ "loss": 446.2861,
1864
+ "step": 2650
1865
+ },
1866
+ {
1867
+ "epoch": 0.021493386339579344,
1868
+ "grad_norm": 3905.970703125,
1869
+ "learning_rate": 1.0747474747474748e-05,
1870
+ "loss": 455.1569,
1871
+ "step": 2660
1872
+ },
1873
+ {
1874
+ "epoch": 0.021574188543863476,
1875
+ "grad_norm": 3564.063232421875,
1876
+ "learning_rate": 1.0787878787878789e-05,
1877
+ "loss": 404.7379,
1878
+ "step": 2670
1879
+ },
1880
+ {
1881
+ "epoch": 0.02165499074814761,
1882
+ "grad_norm": 1134.174072265625,
1883
+ "learning_rate": 1.082828282828283e-05,
1884
+ "loss": 479.0893,
1885
+ "step": 2680
1886
+ },
1887
+ {
1888
+ "epoch": 0.02173579295243174,
1889
+ "grad_norm": 1493.994873046875,
1890
+ "learning_rate": 1.086868686868687e-05,
1891
+ "loss": 487.3219,
1892
+ "step": 2690
1893
+ },
1894
+ {
1895
+ "epoch": 0.021816595156715874,
1896
+ "grad_norm": 1302.6314697265625,
1897
+ "learning_rate": 1.0909090909090909e-05,
1898
+ "loss": 517.1011,
1899
+ "step": 2700
1900
+ },
1901
+ {
1902
+ "epoch": 0.021897397361000007,
1903
+ "grad_norm": 2952.864013671875,
1904
+ "learning_rate": 1.094949494949495e-05,
1905
+ "loss": 462.4752,
1906
+ "step": 2710
1907
+ },
1908
+ {
1909
+ "epoch": 0.02197819956528414,
1910
+ "grad_norm": 1392.331298828125,
1911
+ "learning_rate": 1.098989898989899e-05,
1912
+ "loss": 494.226,
1913
+ "step": 2720
1914
+ },
1915
+ {
1916
+ "epoch": 0.022059001769568275,
1917
+ "grad_norm": 2029.0225830078125,
1918
+ "learning_rate": 1.103030303030303e-05,
1919
+ "loss": 636.2813,
1920
+ "step": 2730
1921
+ },
1922
+ {
1923
+ "epoch": 0.022139803973852408,
1924
+ "grad_norm": 3201.144287109375,
1925
+ "learning_rate": 1.1070707070707071e-05,
1926
+ "loss": 470.682,
1927
+ "step": 2740
1928
+ },
1929
+ {
1930
+ "epoch": 0.02222060617813654,
1931
+ "grad_norm": 2171.435791015625,
1932
+ "learning_rate": 1.1111111111111112e-05,
1933
+ "loss": 514.9135,
1934
+ "step": 2750
1935
+ },
1936
+ {
1937
+ "epoch": 0.022301408382420673,
1938
+ "grad_norm": 760.407470703125,
1939
+ "learning_rate": 1.1151515151515152e-05,
1940
+ "loss": 396.2693,
1941
+ "step": 2760
1942
+ },
1943
+ {
1944
+ "epoch": 0.022382210586704806,
1945
+ "grad_norm": 5750.24951171875,
1946
+ "learning_rate": 1.1191919191919193e-05,
1947
+ "loss": 394.7851,
1948
+ "step": 2770
1949
+ },
1950
+ {
1951
+ "epoch": 0.02246301279098894,
1952
+ "grad_norm": 2005.41650390625,
1953
+ "learning_rate": 1.1232323232323233e-05,
1954
+ "loss": 399.2545,
1955
+ "step": 2780
1956
+ },
1957
+ {
1958
+ "epoch": 0.02254381499527307,
1959
+ "grad_norm": 2525.82080078125,
1960
+ "learning_rate": 1.1272727272727274e-05,
1961
+ "loss": 303.4952,
1962
+ "step": 2790
1963
+ },
1964
+ {
1965
+ "epoch": 0.022624617199557204,
1966
+ "grad_norm": 1201.04248046875,
1967
+ "learning_rate": 1.1313131313131314e-05,
1968
+ "loss": 421.5876,
1969
+ "step": 2800
1970
+ },
1971
+ {
1972
+ "epoch": 0.022705419403841336,
1973
+ "grad_norm": 1753.299560546875,
1974
+ "learning_rate": 1.1353535353535353e-05,
1975
+ "loss": 401.7698,
1976
+ "step": 2810
1977
+ },
1978
+ {
1979
+ "epoch": 0.02278622160812547,
1980
+ "grad_norm": 1007.8385009765625,
1981
+ "learning_rate": 1.1393939393939394e-05,
1982
+ "loss": 481.1457,
1983
+ "step": 2820
1984
+ },
1985
+ {
1986
+ "epoch": 0.0228670238124096,
1987
+ "grad_norm": 1377.8157958984375,
1988
+ "learning_rate": 1.1434343434343434e-05,
1989
+ "loss": 474.0957,
1990
+ "step": 2830
1991
+ },
1992
+ {
1993
+ "epoch": 0.022947826016693734,
1994
+ "grad_norm": 769.2996826171875,
1995
+ "learning_rate": 1.1474747474747475e-05,
1996
+ "loss": 358.7357,
1997
+ "step": 2840
1998
+ },
1999
+ {
2000
+ "epoch": 0.023028628220977867,
2001
+ "grad_norm": 1722.425048828125,
2002
+ "learning_rate": 1.1515151515151517e-05,
2003
+ "loss": 526.5532,
2004
+ "step": 2850
2005
+ },
2006
+ {
2007
+ "epoch": 0.023109430425262,
2008
+ "grad_norm": 890.7434692382812,
2009
+ "learning_rate": 1.1555555555555556e-05,
2010
+ "loss": 374.7328,
2011
+ "step": 2860
2012
+ },
2013
+ {
2014
+ "epoch": 0.023190232629546136,
2015
+ "grad_norm": 1437.0924072265625,
2016
+ "learning_rate": 1.1595959595959597e-05,
2017
+ "loss": 354.1128,
2018
+ "step": 2870
2019
+ },
2020
+ {
2021
+ "epoch": 0.023271034833830268,
2022
+ "grad_norm": 1823.4405517578125,
2023
+ "learning_rate": 1.1636363636363637e-05,
2024
+ "loss": 497.2729,
2025
+ "step": 2880
2026
+ },
2027
+ {
2028
+ "epoch": 0.0233518370381144,
2029
+ "grad_norm": 1081.6297607421875,
2030
+ "learning_rate": 1.1676767676767678e-05,
2031
+ "loss": 381.0187,
2032
+ "step": 2890
2033
+ },
2034
+ {
2035
+ "epoch": 0.023432639242398533,
2036
+ "grad_norm": 2291.82763671875,
2037
+ "learning_rate": 1.1717171717171718e-05,
2038
+ "loss": 439.2377,
2039
+ "step": 2900
2040
+ },
2041
+ {
2042
+ "epoch": 0.023513441446682666,
2043
+ "grad_norm": 1737.9361572265625,
2044
+ "learning_rate": 1.1757575757575757e-05,
2045
+ "loss": 363.3017,
2046
+ "step": 2910
2047
+ },
2048
+ {
2049
+ "epoch": 0.0235942436509668,
2050
+ "grad_norm": 1336.8502197265625,
2051
+ "learning_rate": 1.1797979797979798e-05,
2052
+ "loss": 420.6219,
2053
+ "step": 2920
2054
+ },
2055
+ {
2056
+ "epoch": 0.02367504585525093,
2057
+ "grad_norm": 1407.14208984375,
2058
+ "learning_rate": 1.1838383838383838e-05,
2059
+ "loss": 586.4723,
2060
+ "step": 2930
2061
+ },
2062
+ {
2063
+ "epoch": 0.023755848059535064,
2064
+ "grad_norm": 1774.9156494140625,
2065
+ "learning_rate": 1.187878787878788e-05,
2066
+ "loss": 407.5993,
2067
+ "step": 2940
2068
+ },
2069
+ {
2070
+ "epoch": 0.023836650263819197,
2071
+ "grad_norm": 3886.3212890625,
2072
+ "learning_rate": 1.1919191919191921e-05,
2073
+ "loss": 488.9637,
2074
+ "step": 2950
2075
+ },
2076
+ {
2077
+ "epoch": 0.02391745246810333,
2078
+ "grad_norm": 977.4032592773438,
2079
+ "learning_rate": 1.1959595959595961e-05,
2080
+ "loss": 349.8829,
2081
+ "step": 2960
2082
+ },
2083
+ {
2084
+ "epoch": 0.023998254672387462,
2085
+ "grad_norm": 1776.18994140625,
2086
+ "learning_rate": 1.2e-05,
2087
+ "loss": 378.9573,
2088
+ "step": 2970
2089
+ },
2090
+ {
2091
+ "epoch": 0.024079056876671594,
2092
+ "grad_norm": 982.6826171875,
2093
+ "learning_rate": 1.2040404040404041e-05,
2094
+ "loss": 385.4363,
2095
+ "step": 2980
2096
+ },
2097
+ {
2098
+ "epoch": 0.024159859080955727,
2099
+ "grad_norm": 1428.6072998046875,
2100
+ "learning_rate": 1.2080808080808081e-05,
2101
+ "loss": 411.3545,
2102
+ "step": 2990
2103
+ },
2104
+ {
2105
+ "epoch": 0.02424066128523986,
2106
+ "grad_norm": 2080.078125,
2107
+ "learning_rate": 1.2121212121212122e-05,
2108
+ "loss": 307.3638,
2109
+ "step": 3000
2110
+ },
2111
+ {
2112
+ "epoch": 0.024321463489523996,
2113
+ "grad_norm": 2622.278076171875,
2114
+ "learning_rate": 1.2161616161616162e-05,
2115
+ "loss": 336.3731,
2116
+ "step": 3010
2117
+ },
2118
+ {
2119
+ "epoch": 0.02440226569380813,
2120
+ "grad_norm": 1076.52587890625,
2121
+ "learning_rate": 1.2202020202020201e-05,
2122
+ "loss": 389.7547,
2123
+ "step": 3020
2124
+ },
2125
+ {
2126
+ "epoch": 0.02448306789809226,
2127
+ "grad_norm": 2400.458740234375,
2128
+ "learning_rate": 1.2242424242424242e-05,
2129
+ "loss": 401.2474,
2130
+ "step": 3030
2131
+ },
2132
+ {
2133
+ "epoch": 0.024563870102376394,
2134
+ "grad_norm": 2083.1240234375,
2135
+ "learning_rate": 1.2282828282828284e-05,
2136
+ "loss": 405.6909,
2137
+ "step": 3040
2138
+ },
2139
+ {
2140
+ "epoch": 0.024644672306660526,
2141
+ "grad_norm": 3678.456787109375,
2142
+ "learning_rate": 1.2323232323232325e-05,
2143
+ "loss": 425.2927,
2144
+ "step": 3050
2145
+ },
2146
+ {
2147
+ "epoch": 0.02472547451094466,
2148
+ "grad_norm": 2447.279296875,
2149
+ "learning_rate": 1.2363636363636365e-05,
2150
+ "loss": 536.6354,
2151
+ "step": 3060
2152
+ },
2153
+ {
2154
+ "epoch": 0.02480627671522879,
2155
+ "grad_norm": 1382.6463623046875,
2156
+ "learning_rate": 1.2404040404040404e-05,
2157
+ "loss": 417.6737,
2158
+ "step": 3070
2159
+ },
2160
+ {
2161
+ "epoch": 0.024887078919512924,
2162
+ "grad_norm": 878.1357421875,
2163
+ "learning_rate": 1.2444444444444445e-05,
2164
+ "loss": 403.2435,
2165
+ "step": 3080
2166
+ },
2167
+ {
2168
+ "epoch": 0.024967881123797057,
2169
+ "grad_norm": 1655.641845703125,
2170
+ "learning_rate": 1.2484848484848485e-05,
2171
+ "loss": 398.7163,
2172
+ "step": 3090
2173
+ },
2174
+ {
2175
+ "epoch": 0.02504868332808119,
2176
+ "grad_norm": 1916.8333740234375,
2177
+ "learning_rate": 1.2525252525252526e-05,
2178
+ "loss": 417.5105,
2179
+ "step": 3100
2180
+ },
2181
+ {
2182
+ "epoch": 0.025129485532365322,
2183
+ "grad_norm": 3886.349853515625,
2184
+ "learning_rate": 1.2565656565656566e-05,
2185
+ "loss": 481.2548,
2186
+ "step": 3110
2187
+ },
2188
+ {
2189
+ "epoch": 0.025210287736649455,
2190
+ "grad_norm": 1206.383056640625,
2191
+ "learning_rate": 1.2606060606060607e-05,
2192
+ "loss": 356.7375,
2193
+ "step": 3120
2194
+ },
2195
+ {
2196
+ "epoch": 0.025291089940933587,
2197
+ "grad_norm": 1333.0904541015625,
2198
+ "learning_rate": 1.2646464646464647e-05,
2199
+ "loss": 418.2427,
2200
+ "step": 3130
2201
+ },
2202
+ {
2203
+ "epoch": 0.02537189214521772,
2204
+ "grad_norm": 1487.70458984375,
2205
+ "learning_rate": 1.2686868686868688e-05,
2206
+ "loss": 496.2586,
2207
+ "step": 3140
2208
+ },
2209
+ {
2210
+ "epoch": 0.025452694349501856,
2211
+ "grad_norm": 1888.8553466796875,
2212
+ "learning_rate": 1.2727272727272727e-05,
2213
+ "loss": 463.9542,
2214
+ "step": 3150
2215
+ },
2216
+ {
2217
+ "epoch": 0.02553349655378599,
2218
+ "grad_norm": 1996.4786376953125,
2219
+ "learning_rate": 1.2767676767676767e-05,
2220
+ "loss": 355.7204,
2221
+ "step": 3160
2222
+ },
2223
+ {
2224
+ "epoch": 0.02561429875807012,
2225
+ "grad_norm": 3336.894775390625,
2226
+ "learning_rate": 1.2808080808080808e-05,
2227
+ "loss": 491.6457,
2228
+ "step": 3170
2229
+ },
2230
+ {
2231
+ "epoch": 0.025695100962354254,
2232
+ "grad_norm": 1442.059814453125,
2233
+ "learning_rate": 1.2848484848484848e-05,
2234
+ "loss": 440.1709,
2235
+ "step": 3180
2236
+ },
2237
+ {
2238
+ "epoch": 0.025775903166638386,
2239
+ "grad_norm": 1862.8482666015625,
2240
+ "learning_rate": 1.2888888888888889e-05,
2241
+ "loss": 432.363,
2242
+ "step": 3190
2243
+ },
2244
+ {
2245
+ "epoch": 0.02585670537092252,
2246
+ "grad_norm": 1185.944091796875,
2247
+ "learning_rate": 1.292929292929293e-05,
2248
+ "loss": 493.8533,
2249
+ "step": 3200
2250
+ },
2251
+ {
2252
+ "epoch": 0.02593750757520665,
2253
+ "grad_norm": 2231.38134765625,
2254
+ "learning_rate": 1.296969696969697e-05,
2255
+ "loss": 399.7351,
2256
+ "step": 3210
2257
+ },
2258
+ {
2259
+ "epoch": 0.026018309779490784,
2260
+ "grad_norm": 2386.878173828125,
2261
+ "learning_rate": 1.301010101010101e-05,
2262
+ "loss": 428.7118,
2263
+ "step": 3220
2264
+ },
2265
+ {
2266
+ "epoch": 0.026099111983774917,
2267
+ "grad_norm": 1025.3896484375,
2268
+ "learning_rate": 1.3050505050505051e-05,
2269
+ "loss": 400.1819,
2270
+ "step": 3230
2271
+ },
2272
+ {
2273
+ "epoch": 0.02617991418805905,
2274
+ "grad_norm": 1678.713134765625,
2275
+ "learning_rate": 1.3090909090909093e-05,
2276
+ "loss": 434.1818,
2277
+ "step": 3240
2278
+ },
2279
+ {
2280
+ "epoch": 0.026260716392343182,
2281
+ "grad_norm": 1201.3856201171875,
2282
+ "learning_rate": 1.3131313131313134e-05,
2283
+ "loss": 395.3099,
2284
+ "step": 3250
2285
+ },
2286
+ {
2287
+ "epoch": 0.026341518596627315,
2288
+ "grad_norm": 5197.4033203125,
2289
+ "learning_rate": 1.3171717171717171e-05,
2290
+ "loss": 396.583,
2291
+ "step": 3260
2292
+ },
2293
+ {
2294
+ "epoch": 0.026422320800911447,
2295
+ "grad_norm": 1856.5380859375,
2296
+ "learning_rate": 1.3212121212121212e-05,
2297
+ "loss": 440.6047,
2298
+ "step": 3270
2299
+ },
2300
+ {
2301
+ "epoch": 0.02650312300519558,
2302
+ "grad_norm": 1735.7449951171875,
2303
+ "learning_rate": 1.3252525252525252e-05,
2304
+ "loss": 355.502,
2305
+ "step": 3280
2306
+ },
2307
+ {
2308
+ "epoch": 0.026583925209479716,
2309
+ "grad_norm": 1101.4957275390625,
2310
+ "learning_rate": 1.3292929292929293e-05,
2311
+ "loss": 320.7389,
2312
+ "step": 3290
2313
+ },
2314
+ {
2315
+ "epoch": 0.02666472741376385,
2316
+ "grad_norm": 1054.079345703125,
2317
+ "learning_rate": 1.3333333333333333e-05,
2318
+ "loss": 356.3952,
2319
+ "step": 3300
2320
+ },
2321
+ {
2322
+ "epoch": 0.02674552961804798,
2323
+ "grad_norm": 2106.282958984375,
2324
+ "learning_rate": 1.3373737373737374e-05,
2325
+ "loss": 403.6069,
2326
+ "step": 3310
2327
+ },
2328
+ {
2329
+ "epoch": 0.026826331822332114,
2330
+ "grad_norm": 1672.3568115234375,
2331
+ "learning_rate": 1.3414141414141414e-05,
2332
+ "loss": 448.6192,
2333
+ "step": 3320
2334
+ },
2335
+ {
2336
+ "epoch": 0.026907134026616247,
2337
+ "grad_norm": 1342.961181640625,
2338
+ "learning_rate": 1.3454545454545457e-05,
2339
+ "loss": 332.1857,
2340
+ "step": 3330
2341
+ },
2342
+ {
2343
+ "epoch": 0.02698793623090038,
2344
+ "grad_norm": 991.6637573242188,
2345
+ "learning_rate": 1.3494949494949497e-05,
2346
+ "loss": 336.3228,
2347
+ "step": 3340
2348
+ },
2349
+ {
2350
+ "epoch": 0.027068738435184512,
2351
+ "grad_norm": 3500.915771484375,
2352
+ "learning_rate": 1.3535353535353538e-05,
2353
+ "loss": 469.8078,
2354
+ "step": 3350
2355
+ },
2356
+ {
2357
+ "epoch": 0.027149540639468644,
2358
+ "grad_norm": 1306.3658447265625,
2359
+ "learning_rate": 1.3575757575757578e-05,
2360
+ "loss": 412.32,
2361
+ "step": 3360
2362
+ },
2363
+ {
2364
+ "epoch": 0.027230342843752777,
2365
+ "grad_norm": 1543.1270751953125,
2366
+ "learning_rate": 1.3616161616161615e-05,
2367
+ "loss": 568.8118,
2368
+ "step": 3370
2369
+ },
2370
+ {
2371
+ "epoch": 0.02731114504803691,
2372
+ "grad_norm": 2236.9619140625,
2373
+ "learning_rate": 1.3656565656565656e-05,
2374
+ "loss": 499.9743,
2375
+ "step": 3380
2376
+ },
2377
+ {
2378
+ "epoch": 0.027391947252321042,
2379
+ "grad_norm": 2010.0272216796875,
2380
+ "learning_rate": 1.3696969696969697e-05,
2381
+ "loss": 390.202,
2382
+ "step": 3390
2383
+ },
2384
+ {
2385
+ "epoch": 0.027472749456605175,
2386
+ "grad_norm": 2621.369873046875,
2387
+ "learning_rate": 1.3737373737373737e-05,
2388
+ "loss": 447.2703,
2389
+ "step": 3400
2390
+ },
2391
+ {
2392
+ "epoch": 0.027553551660889308,
2393
+ "grad_norm": 1979.689697265625,
2394
+ "learning_rate": 1.3777777777777778e-05,
2395
+ "loss": 364.1719,
2396
+ "step": 3410
2397
+ },
2398
+ {
2399
+ "epoch": 0.02763435386517344,
2400
+ "grad_norm": 1638.1971435546875,
2401
+ "learning_rate": 1.3818181818181818e-05,
2402
+ "loss": 330.4022,
2403
+ "step": 3420
2404
+ },
2405
+ {
2406
+ "epoch": 0.027715156069457576,
2407
+ "grad_norm": 1491.321533203125,
2408
+ "learning_rate": 1.385858585858586e-05,
2409
+ "loss": 381.2672,
2410
+ "step": 3430
2411
+ },
2412
+ {
2413
+ "epoch": 0.02779595827374171,
2414
+ "grad_norm": 1468.6680908203125,
2415
+ "learning_rate": 1.3898989898989901e-05,
2416
+ "loss": 464.3973,
2417
+ "step": 3440
2418
+ },
2419
+ {
2420
+ "epoch": 0.02787676047802584,
2421
+ "grad_norm": 4992.0,
2422
+ "learning_rate": 1.3939393939393942e-05,
2423
+ "loss": 414.7291,
2424
+ "step": 3450
2425
+ },
2426
+ {
2427
+ "epoch": 0.027957562682309974,
2428
+ "grad_norm": 739.7650756835938,
2429
+ "learning_rate": 1.3979797979797982e-05,
2430
+ "loss": 369.1546,
2431
+ "step": 3460
2432
+ },
2433
+ {
2434
+ "epoch": 0.028038364886594107,
2435
+ "grad_norm": 1464.85009765625,
2436
+ "learning_rate": 1.402020202020202e-05,
2437
+ "loss": 346.7063,
2438
+ "step": 3470
2439
+ },
2440
+ {
2441
+ "epoch": 0.02811916709087824,
2442
+ "grad_norm": 1048.708984375,
2443
+ "learning_rate": 1.406060606060606e-05,
2444
+ "loss": 321.1962,
2445
+ "step": 3480
2446
+ },
2447
+ {
2448
+ "epoch": 0.028199969295162372,
2449
+ "grad_norm": 4313.70556640625,
2450
+ "learning_rate": 1.41010101010101e-05,
2451
+ "loss": 353.4636,
2452
+ "step": 3490
2453
+ },
2454
+ {
2455
+ "epoch": 0.028280771499446505,
2456
+ "grad_norm": 2037.3917236328125,
2457
+ "learning_rate": 1.4141414141414141e-05,
2458
+ "loss": 399.9984,
2459
+ "step": 3500
2460
+ },
2461
+ {
2462
+ "epoch": 0.028361573703730637,
2463
+ "grad_norm": 2155.431396484375,
2464
+ "learning_rate": 1.4181818181818181e-05,
2465
+ "loss": 370.1783,
2466
+ "step": 3510
2467
+ },
2468
+ {
2469
+ "epoch": 0.02844237590801477,
2470
+ "grad_norm": 4556.9873046875,
2471
+ "learning_rate": 1.4222222222222224e-05,
2472
+ "loss": 342.3656,
2473
+ "step": 3520
2474
+ },
2475
+ {
2476
+ "epoch": 0.028523178112298903,
2477
+ "grad_norm": 1619.715087890625,
2478
+ "learning_rate": 1.4262626262626264e-05,
2479
+ "loss": 399.8575,
2480
+ "step": 3530
2481
+ },
2482
+ {
2483
+ "epoch": 0.028603980316583035,
2484
+ "grad_norm": 1036.035400390625,
2485
+ "learning_rate": 1.4303030303030305e-05,
2486
+ "loss": 336.8583,
2487
+ "step": 3540
2488
+ },
2489
+ {
2490
+ "epoch": 0.028684782520867168,
2491
+ "grad_norm": 896.1290893554688,
2492
+ "learning_rate": 1.4343434343434345e-05,
2493
+ "loss": 397.4199,
2494
+ "step": 3550
2495
+ },
2496
+ {
2497
+ "epoch": 0.0287655847251513,
2498
+ "grad_norm": 871.9555053710938,
2499
+ "learning_rate": 1.4383838383838386e-05,
2500
+ "loss": 421.5666,
2501
+ "step": 3560
2502
+ },
2503
+ {
2504
+ "epoch": 0.028846386929435437,
2505
+ "grad_norm": 1713.01806640625,
2506
+ "learning_rate": 1.4424242424242426e-05,
2507
+ "loss": 385.2063,
2508
+ "step": 3570
2509
+ },
2510
+ {
2511
+ "epoch": 0.02892718913371957,
2512
+ "grad_norm": 6706.7744140625,
2513
+ "learning_rate": 1.4464646464646464e-05,
2514
+ "loss": 574.8361,
2515
+ "step": 3580
2516
+ },
2517
+ {
2518
+ "epoch": 0.029007991338003702,
2519
+ "grad_norm": 47397.72265625,
2520
+ "learning_rate": 1.4505050505050504e-05,
2521
+ "loss": 529.9805,
2522
+ "step": 3590
2523
+ },
2524
+ {
2525
+ "epoch": 0.029088793542287834,
2526
+ "grad_norm": 2835.7958984375,
2527
+ "learning_rate": 1.4545454545454545e-05,
2528
+ "loss": 392.688,
2529
+ "step": 3600
2530
+ },
2531
+ {
2532
+ "epoch": 0.029169595746571967,
2533
+ "grad_norm": 1097.8177490234375,
2534
+ "learning_rate": 1.4585858585858587e-05,
2535
+ "loss": 359.7729,
2536
+ "step": 3610
2537
+ },
2538
+ {
2539
+ "epoch": 0.0292503979508561,
2540
+ "grad_norm": 954.6292114257812,
2541
+ "learning_rate": 1.4626262626262627e-05,
2542
+ "loss": 392.3154,
2543
+ "step": 3620
2544
+ },
2545
+ {
2546
+ "epoch": 0.029331200155140232,
2547
+ "grad_norm": 3372.08251953125,
2548
+ "learning_rate": 1.4666666666666668e-05,
2549
+ "loss": 400.445,
2550
+ "step": 3630
2551
+ },
2552
+ {
2553
+ "epoch": 0.029412002359424365,
2554
+ "grad_norm": 3138.397705078125,
2555
+ "learning_rate": 1.4707070707070709e-05,
2556
+ "loss": 341.2362,
2557
+ "step": 3640
2558
+ },
2559
+ {
2560
+ "epoch": 0.029492804563708497,
2561
+ "grad_norm": 1593.2418212890625,
2562
+ "learning_rate": 1.4747474747474749e-05,
2563
+ "loss": 371.5777,
2564
+ "step": 3650
2565
+ },
2566
+ {
2567
+ "epoch": 0.02957360676799263,
2568
+ "grad_norm": 1874.13037109375,
2569
+ "learning_rate": 1.478787878787879e-05,
2570
+ "loss": 445.4564,
2571
+ "step": 3660
2572
+ },
2573
+ {
2574
+ "epoch": 0.029654408972276763,
2575
+ "grad_norm": 1240.8599853515625,
2576
+ "learning_rate": 1.482828282828283e-05,
2577
+ "loss": 383.5819,
2578
+ "step": 3670
2579
+ },
2580
+ {
2581
+ "epoch": 0.029735211176560895,
2582
+ "grad_norm": 6037.068359375,
2583
+ "learning_rate": 1.486868686868687e-05,
2584
+ "loss": 345.8714,
2585
+ "step": 3680
2586
+ },
2587
+ {
2588
+ "epoch": 0.029816013380845028,
2589
+ "grad_norm": 6898.64990234375,
2590
+ "learning_rate": 1.4909090909090908e-05,
2591
+ "loss": 316.7964,
2592
+ "step": 3690
2593
+ },
2594
+ {
2595
+ "epoch": 0.02989681558512916,
2596
+ "grad_norm": 1123.4033203125,
2597
+ "learning_rate": 1.494949494949495e-05,
2598
+ "loss": 469.9616,
2599
+ "step": 3700
2600
+ },
2601
+ {
2602
+ "epoch": 0.029977617789413297,
2603
+ "grad_norm": 1321.8525390625,
2604
+ "learning_rate": 1.498989898989899e-05,
2605
+ "loss": 391.1738,
2606
+ "step": 3710
2607
+ },
2608
+ {
2609
+ "epoch": 0.03005841999369743,
2610
+ "grad_norm": 2863.68212890625,
2611
+ "learning_rate": 1.5030303030303031e-05,
2612
+ "loss": 295.6113,
2613
+ "step": 3720
2614
+ },
2615
+ {
2616
+ "epoch": 0.030139222197981562,
2617
+ "grad_norm": 1591.6097412109375,
2618
+ "learning_rate": 1.5070707070707072e-05,
2619
+ "loss": 446.7923,
2620
+ "step": 3730
2621
+ },
2622
+ {
2623
+ "epoch": 0.030220024402265695,
2624
+ "grad_norm": 2339.865234375,
2625
+ "learning_rate": 1.5111111111111112e-05,
2626
+ "loss": 401.7253,
2627
+ "step": 3740
2628
+ },
2629
+ {
2630
+ "epoch": 0.030300826606549827,
2631
+ "grad_norm": 1764.2125244140625,
2632
+ "learning_rate": 1.5151515151515153e-05,
2633
+ "loss": 439.8055,
2634
+ "step": 3750
2635
+ },
2636
+ {
2637
+ "epoch": 0.03038162881083396,
2638
+ "grad_norm": 1147.5810546875,
2639
+ "learning_rate": 1.5191919191919193e-05,
2640
+ "loss": 404.343,
2641
+ "step": 3760
2642
+ },
2643
+ {
2644
+ "epoch": 0.030462431015118092,
2645
+ "grad_norm": 1645.48095703125,
2646
+ "learning_rate": 1.5232323232323234e-05,
2647
+ "loss": 393.5273,
2648
+ "step": 3770
2649
+ },
2650
+ {
2651
+ "epoch": 0.030543233219402225,
2652
+ "grad_norm": 1111.396484375,
2653
+ "learning_rate": 1.5272727272727276e-05,
2654
+ "loss": 339.2941,
2655
+ "step": 3780
2656
+ },
2657
+ {
2658
+ "epoch": 0.030624035423686358,
2659
+ "grad_norm": 1468.22216796875,
2660
+ "learning_rate": 1.531313131313131e-05,
2661
+ "loss": 428.423,
2662
+ "step": 3790
2663
+ },
2664
+ {
2665
+ "epoch": 0.03070483762797049,
2666
+ "grad_norm": 6860.38134765625,
2667
+ "learning_rate": 1.5353535353535354e-05,
2668
+ "loss": 480.3668,
2669
+ "step": 3800
2670
+ },
2671
+ {
2672
+ "epoch": 0.030785639832254623,
2673
+ "grad_norm": 1157.9761962890625,
2674
+ "learning_rate": 1.5393939393939393e-05,
2675
+ "loss": 397.3216,
2676
+ "step": 3810
2677
+ },
2678
+ {
2679
+ "epoch": 0.030866442036538756,
2680
+ "grad_norm": 751.1998291015625,
2681
+ "learning_rate": 1.5434343434343435e-05,
2682
+ "loss": 314.7172,
2683
+ "step": 3820
2684
+ },
2685
+ {
2686
+ "epoch": 0.030947244240822888,
2687
+ "grad_norm": 862.4395141601562,
2688
+ "learning_rate": 1.5474747474747474e-05,
2689
+ "loss": 372.4114,
2690
+ "step": 3830
2691
+ },
2692
+ {
2693
+ "epoch": 0.031028046445107024,
2694
+ "grad_norm": 1985.647216796875,
2695
+ "learning_rate": 1.5515151515151516e-05,
2696
+ "loss": 410.9164,
2697
+ "step": 3840
2698
+ },
2699
+ {
2700
+ "epoch": 0.031108848649391157,
2701
+ "grad_norm": 1223.5931396484375,
2702
+ "learning_rate": 1.5555555555555555e-05,
2703
+ "loss": 409.5396,
2704
+ "step": 3850
2705
+ },
2706
+ {
2707
+ "epoch": 0.03118965085367529,
2708
+ "grad_norm": 1147.3355712890625,
2709
+ "learning_rate": 1.5595959595959597e-05,
2710
+ "loss": 343.1769,
2711
+ "step": 3860
2712
+ },
2713
+ {
2714
+ "epoch": 0.03127045305795942,
2715
+ "grad_norm": 931.1632080078125,
2716
+ "learning_rate": 1.563636363636364e-05,
2717
+ "loss": 287.52,
2718
+ "step": 3870
2719
+ },
2720
+ {
2721
+ "epoch": 0.031351255262243555,
2722
+ "grad_norm": 2921.90478515625,
2723
+ "learning_rate": 1.5676767676767678e-05,
2724
+ "loss": 466.8278,
2725
+ "step": 3880
2726
+ },
2727
+ {
2728
+ "epoch": 0.031432057466527684,
2729
+ "grad_norm": 1351.41064453125,
2730
+ "learning_rate": 1.571717171717172e-05,
2731
+ "loss": 367.0911,
2732
+ "step": 3890
2733
+ },
2734
+ {
2735
+ "epoch": 0.03151285967081182,
2736
+ "grad_norm": 2405.606689453125,
2737
+ "learning_rate": 1.5757575757575756e-05,
2738
+ "loss": 368.1811,
2739
+ "step": 3900
2740
+ },
2741
+ {
2742
+ "epoch": 0.031593661875095956,
2743
+ "grad_norm": 4332.23095703125,
2744
+ "learning_rate": 1.5797979797979798e-05,
2745
+ "loss": 346.5809,
2746
+ "step": 3910
2747
+ },
2748
+ {
2749
+ "epoch": 0.031674464079380085,
2750
+ "grad_norm": 2268.54736328125,
2751
+ "learning_rate": 1.5838383838383837e-05,
2752
+ "loss": 332.8164,
2753
+ "step": 3920
2754
+ },
2755
+ {
2756
+ "epoch": 0.03175526628366422,
2757
+ "grad_norm": 1450.4573974609375,
2758
+ "learning_rate": 1.587878787878788e-05,
2759
+ "loss": 360.3381,
2760
+ "step": 3930
2761
+ },
2762
+ {
2763
+ "epoch": 0.03183606848794835,
2764
+ "grad_norm": 7813.84912109375,
2765
+ "learning_rate": 1.5919191919191918e-05,
2766
+ "loss": 463.5723,
2767
+ "step": 3940
2768
+ },
2769
+ {
2770
+ "epoch": 0.03191687069223249,
2771
+ "grad_norm": 1389.827880859375,
2772
+ "learning_rate": 1.595959595959596e-05,
2773
+ "loss": 272.0756,
2774
+ "step": 3950
2775
+ },
2776
+ {
2777
+ "epoch": 0.031997672896516616,
2778
+ "grad_norm": 1054.016357421875,
2779
+ "learning_rate": 1.6000000000000003e-05,
2780
+ "loss": 410.4074,
2781
+ "step": 3960
2782
+ },
2783
+ {
2784
+ "epoch": 0.03207847510080075,
2785
+ "grad_norm": 2336.52294921875,
2786
+ "learning_rate": 1.604040404040404e-05,
2787
+ "loss": 374.9348,
2788
+ "step": 3970
2789
+ },
2790
+ {
2791
+ "epoch": 0.03215927730508488,
2792
+ "grad_norm": 1440.7698974609375,
2793
+ "learning_rate": 1.6080808080808084e-05,
2794
+ "loss": 433.5792,
2795
+ "step": 3980
2796
+ },
2797
+ {
2798
+ "epoch": 0.03224007950936902,
2799
+ "grad_norm": 1275.298095703125,
2800
+ "learning_rate": 1.6121212121212123e-05,
2801
+ "loss": 407.9875,
2802
+ "step": 3990
2803
+ },
2804
+ {
2805
+ "epoch": 0.032320881713653146,
2806
+ "grad_norm": 2616.390869140625,
2807
+ "learning_rate": 1.6161616161616165e-05,
2808
+ "loss": 345.8577,
2809
+ "step": 4000
2810
+ },
2811
+ {
2812
+ "epoch": 0.03240168391793728,
2813
+ "grad_norm": 4405.9677734375,
2814
+ "learning_rate": 1.62020202020202e-05,
2815
+ "loss": 563.3746,
2816
+ "step": 4010
2817
+ },
2818
+ {
2819
+ "epoch": 0.03248248612222141,
2820
+ "grad_norm": 1293.5716552734375,
2821
+ "learning_rate": 1.6242424242424243e-05,
2822
+ "loss": 438.8694,
2823
+ "step": 4020
2824
+ },
2825
+ {
2826
+ "epoch": 0.03256328832650555,
2827
+ "grad_norm": 1728.333251953125,
2828
+ "learning_rate": 1.628282828282828e-05,
2829
+ "loss": 436.4476,
2830
+ "step": 4030
2831
+ },
2832
+ {
2833
+ "epoch": 0.03264409053078968,
2834
+ "grad_norm": 1484.926025390625,
2835
+ "learning_rate": 1.6323232323232324e-05,
2836
+ "loss": 338.0915,
2837
+ "step": 4040
2838
+ },
2839
+ {
2840
+ "epoch": 0.03272489273507381,
2841
+ "grad_norm": 1212.051025390625,
2842
+ "learning_rate": 1.6363636363636366e-05,
2843
+ "loss": 345.7399,
2844
+ "step": 4050
2845
+ },
2846
+ {
2847
+ "epoch": 0.03280569493935795,
2848
+ "grad_norm": 2545.336669921875,
2849
+ "learning_rate": 1.6404040404040405e-05,
2850
+ "loss": 383.8622,
2851
+ "step": 4060
2852
+ },
2853
+ {
2854
+ "epoch": 0.03288649714364208,
2855
+ "grad_norm": 1114.4097900390625,
2856
+ "learning_rate": 1.6444444444444447e-05,
2857
+ "loss": 329.3188,
2858
+ "step": 4070
2859
+ },
2860
+ {
2861
+ "epoch": 0.032967299347926214,
2862
+ "grad_norm": 1056.169677734375,
2863
+ "learning_rate": 1.6484848484848486e-05,
2864
+ "loss": 311.8675,
2865
+ "step": 4080
2866
+ },
2867
+ {
2868
+ "epoch": 0.03304810155221034,
2869
+ "grad_norm": 1482.0853271484375,
2870
+ "learning_rate": 1.6525252525252528e-05,
2871
+ "loss": 299.9676,
2872
+ "step": 4090
2873
+ },
2874
+ {
2875
+ "epoch": 0.03312890375649448,
2876
+ "grad_norm": 7475.61279296875,
2877
+ "learning_rate": 1.6565656565656567e-05,
2878
+ "loss": 404.3645,
2879
+ "step": 4100
2880
+ },
2881
+ {
2882
+ "epoch": 0.03320970596077861,
2883
+ "grad_norm": 1250.048583984375,
2884
+ "learning_rate": 1.6606060606060606e-05,
2885
+ "loss": 353.2959,
2886
+ "step": 4110
2887
+ },
2888
+ {
2889
+ "epoch": 0.033290508165062745,
2890
+ "grad_norm": 6574.8798828125,
2891
+ "learning_rate": 1.6646464646464645e-05,
2892
+ "loss": 332.5623,
2893
+ "step": 4120
2894
+ },
2895
+ {
2896
+ "epoch": 0.033371310369346874,
2897
+ "grad_norm": 884.6202392578125,
2898
+ "learning_rate": 1.6686868686868687e-05,
2899
+ "loss": 393.6061,
2900
+ "step": 4130
2901
+ },
2902
+ {
2903
+ "epoch": 0.03345211257363101,
2904
+ "grad_norm": 1503.19384765625,
2905
+ "learning_rate": 1.672727272727273e-05,
2906
+ "loss": 471.0581,
2907
+ "step": 4140
2908
+ },
2909
+ {
2910
+ "epoch": 0.03353291477791514,
2911
+ "grad_norm": 2390.65380859375,
2912
+ "learning_rate": 1.6767676767676768e-05,
2913
+ "loss": 379.6377,
2914
+ "step": 4150
2915
+ },
2916
+ {
2917
+ "epoch": 0.033613716982199275,
2918
+ "grad_norm": 2654.139892578125,
2919
+ "learning_rate": 1.680808080808081e-05,
2920
+ "loss": 361.57,
2921
+ "step": 4160
2922
+ },
2923
+ {
2924
+ "epoch": 0.033694519186483404,
2925
+ "grad_norm": 1114.0318603515625,
2926
+ "learning_rate": 1.684848484848485e-05,
2927
+ "loss": 315.6078,
2928
+ "step": 4170
2929
+ },
2930
+ {
2931
+ "epoch": 0.03377532139076754,
2932
+ "grad_norm": 2269.401611328125,
2933
+ "learning_rate": 1.688888888888889e-05,
2934
+ "loss": 387.0298,
2935
+ "step": 4180
2936
+ },
2937
+ {
2938
+ "epoch": 0.033856123595051676,
2939
+ "grad_norm": 2904.967529296875,
2940
+ "learning_rate": 1.692929292929293e-05,
2941
+ "loss": 286.9893,
2942
+ "step": 4190
2943
+ },
2944
+ {
2945
+ "epoch": 0.033936925799335806,
2946
+ "grad_norm": 1229.243408203125,
2947
+ "learning_rate": 1.6969696969696972e-05,
2948
+ "loss": 371.6422,
2949
+ "step": 4200
2950
+ },
2951
+ {
2952
+ "epoch": 0.03401772800361994,
2953
+ "grad_norm": 1940.672607421875,
2954
+ "learning_rate": 1.701010101010101e-05,
2955
+ "loss": 358.8707,
2956
+ "step": 4210
2957
+ },
2958
+ {
2959
+ "epoch": 0.03409853020790407,
2960
+ "grad_norm": 2709.007568359375,
2961
+ "learning_rate": 1.705050505050505e-05,
2962
+ "loss": 333.9397,
2963
+ "step": 4220
2964
+ },
2965
+ {
2966
+ "epoch": 0.03417933241218821,
2967
+ "grad_norm": 1473.49609375,
2968
+ "learning_rate": 1.7090909090909092e-05,
2969
+ "loss": 276.9469,
2970
+ "step": 4230
2971
+ },
2972
+ {
2973
+ "epoch": 0.034260134616472336,
2974
+ "grad_norm": 2211.987060546875,
2975
+ "learning_rate": 1.713131313131313e-05,
2976
+ "loss": 343.1185,
2977
+ "step": 4240
2978
+ },
2979
+ {
2980
+ "epoch": 0.03434093682075647,
2981
+ "grad_norm": 918.4265747070312,
2982
+ "learning_rate": 1.7171717171717173e-05,
2983
+ "loss": 441.9232,
2984
+ "step": 4250
2985
+ },
2986
+ {
2987
+ "epoch": 0.0344217390250406,
2988
+ "grad_norm": 1585.6217041015625,
2989
+ "learning_rate": 1.7212121212121212e-05,
2990
+ "loss": 401.2611,
2991
+ "step": 4260
2992
+ },
2993
+ {
2994
+ "epoch": 0.03450254122932474,
2995
+ "grad_norm": 2644.159912109375,
2996
+ "learning_rate": 1.7252525252525255e-05,
2997
+ "loss": 319.812,
2998
+ "step": 4270
2999
+ },
3000
+ {
3001
+ "epoch": 0.03458334343360887,
3002
+ "grad_norm": 1674.13623046875,
3003
+ "learning_rate": 1.7292929292929293e-05,
3004
+ "loss": 410.6561,
3005
+ "step": 4280
3006
+ },
3007
+ {
3008
+ "epoch": 0.034664145637893,
3009
+ "grad_norm": 7976.99609375,
3010
+ "learning_rate": 1.7333333333333336e-05,
3011
+ "loss": 384.7048,
3012
+ "step": 4290
3013
+ },
3014
+ {
3015
+ "epoch": 0.03474494784217713,
3016
+ "grad_norm": 1133.471435546875,
3017
+ "learning_rate": 1.7373737373737375e-05,
3018
+ "loss": 518.7885,
3019
+ "step": 4300
3020
+ },
3021
+ {
3022
+ "epoch": 0.03482575004646127,
3023
+ "grad_norm": 1117.8463134765625,
3024
+ "learning_rate": 1.7414141414141417e-05,
3025
+ "loss": 373.2396,
3026
+ "step": 4310
3027
+ },
3028
+ {
3029
+ "epoch": 0.0349065522507454,
3030
+ "grad_norm": 2111.07373046875,
3031
+ "learning_rate": 1.7454545454545456e-05,
3032
+ "loss": 380.1865,
3033
+ "step": 4320
3034
+ },
3035
+ {
3036
+ "epoch": 0.03498735445502953,
3037
+ "grad_norm": 3885.1787109375,
3038
+ "learning_rate": 1.7494949494949494e-05,
3039
+ "loss": 334.7283,
3040
+ "step": 4330
3041
+ },
3042
+ {
3043
+ "epoch": 0.03506815665931367,
3044
+ "grad_norm": 28903.310546875,
3045
+ "learning_rate": 1.7535353535353537e-05,
3046
+ "loss": 333.2706,
3047
+ "step": 4340
3048
+ },
3049
+ {
3050
+ "epoch": 0.0351489588635978,
3051
+ "grad_norm": 2059.6552734375,
3052
+ "learning_rate": 1.7575757575757576e-05,
3053
+ "loss": 358.8692,
3054
+ "step": 4350
3055
+ },
3056
+ {
3057
+ "epoch": 0.035229761067881935,
3058
+ "grad_norm": 1722.4095458984375,
3059
+ "learning_rate": 1.7616161616161618e-05,
3060
+ "loss": 488.4147,
3061
+ "step": 4360
3062
+ },
3063
+ {
3064
+ "epoch": 0.035310563272166064,
3065
+ "grad_norm": 1890.9517822265625,
3066
+ "learning_rate": 1.7656565656565657e-05,
3067
+ "loss": 432.1608,
3068
+ "step": 4370
3069
+ },
3070
+ {
3071
+ "epoch": 0.0353913654764502,
3072
+ "grad_norm": 1202.87646484375,
3073
+ "learning_rate": 1.76969696969697e-05,
3074
+ "loss": 358.1542,
3075
+ "step": 4380
3076
+ },
3077
+ {
3078
+ "epoch": 0.03547216768073433,
3079
+ "grad_norm": 2918.1181640625,
3080
+ "learning_rate": 1.7737373737373738e-05,
3081
+ "loss": 326.9036,
3082
+ "step": 4390
3083
+ },
3084
+ {
3085
+ "epoch": 0.035552969885018465,
3086
+ "grad_norm": 901.55419921875,
3087
+ "learning_rate": 1.777777777777778e-05,
3088
+ "loss": 337.8409,
3089
+ "step": 4400
3090
+ },
3091
+ {
3092
+ "epoch": 0.035633772089302594,
3093
+ "grad_norm": 1185.158447265625,
3094
+ "learning_rate": 1.781818181818182e-05,
3095
+ "loss": 359.4807,
3096
+ "step": 4410
3097
+ },
3098
+ {
3099
+ "epoch": 0.03571457429358673,
3100
+ "grad_norm": 1455.7763671875,
3101
+ "learning_rate": 1.785858585858586e-05,
3102
+ "loss": 352.2927,
3103
+ "step": 4420
3104
+ },
3105
+ {
3106
+ "epoch": 0.03579537649787086,
3107
+ "grad_norm": 2693.363037109375,
3108
+ "learning_rate": 1.78989898989899e-05,
3109
+ "loss": 370.5259,
3110
+ "step": 4430
3111
+ },
3112
+ {
3113
+ "epoch": 0.035876178702154995,
3114
+ "grad_norm": 2017.5230712890625,
3115
+ "learning_rate": 1.793939393939394e-05,
3116
+ "loss": 323.6158,
3117
+ "step": 4440
3118
+ },
3119
+ {
3120
+ "epoch": 0.035956980906439125,
3121
+ "grad_norm": 2855.155029296875,
3122
+ "learning_rate": 1.797979797979798e-05,
3123
+ "loss": 356.3111,
3124
+ "step": 4450
3125
+ },
3126
+ {
3127
+ "epoch": 0.03603778311072326,
3128
+ "grad_norm": 2315.664306640625,
3129
+ "learning_rate": 1.802020202020202e-05,
3130
+ "loss": 319.9107,
3131
+ "step": 4460
3132
+ },
3133
+ {
3134
+ "epoch": 0.0361185853150074,
3135
+ "grad_norm": 1550.8470458984375,
3136
+ "learning_rate": 1.8060606060606062e-05,
3137
+ "loss": 314.4013,
3138
+ "step": 4470
3139
+ },
3140
+ {
3141
+ "epoch": 0.036199387519291526,
3142
+ "grad_norm": 2441.33984375,
3143
+ "learning_rate": 1.81010101010101e-05,
3144
+ "loss": 392.892,
3145
+ "step": 4480
3146
+ },
3147
+ {
3148
+ "epoch": 0.03628018972357566,
3149
+ "grad_norm": 4503.03662109375,
3150
+ "learning_rate": 1.8141414141414143e-05,
3151
+ "loss": 314.955,
3152
+ "step": 4490
3153
+ },
3154
+ {
3155
+ "epoch": 0.03636099192785979,
3156
+ "grad_norm": 1784.7449951171875,
3157
+ "learning_rate": 1.8181818181818182e-05,
3158
+ "loss": 350.5214,
3159
+ "step": 4500
3160
+ },
3161
+ {
3162
+ "epoch": 0.03644179413214393,
3163
+ "grad_norm": 2955.252685546875,
3164
+ "learning_rate": 1.8222222222222224e-05,
3165
+ "loss": 340.7913,
3166
+ "step": 4510
3167
+ },
3168
+ {
3169
+ "epoch": 0.036522596336428056,
3170
+ "grad_norm": 2789.01513671875,
3171
+ "learning_rate": 1.8262626262626263e-05,
3172
+ "loss": 317.8641,
3173
+ "step": 4520
3174
+ },
3175
+ {
3176
+ "epoch": 0.03660339854071219,
3177
+ "grad_norm": 2411.71728515625,
3178
+ "learning_rate": 1.8303030303030305e-05,
3179
+ "loss": 407.7839,
3180
+ "step": 4530
3181
+ },
3182
+ {
3183
+ "epoch": 0.03668420074499632,
3184
+ "grad_norm": 2479.978759765625,
3185
+ "learning_rate": 1.8343434343434344e-05,
3186
+ "loss": 373.7947,
3187
+ "step": 4540
3188
+ },
3189
+ {
3190
+ "epoch": 0.03676500294928046,
3191
+ "grad_norm": 4516.0986328125,
3192
+ "learning_rate": 1.8383838383838383e-05,
3193
+ "loss": 353.7791,
3194
+ "step": 4550
3195
+ },
3196
+ {
3197
+ "epoch": 0.03684580515356459,
3198
+ "grad_norm": 2898.03955078125,
3199
+ "learning_rate": 1.8424242424242425e-05,
3200
+ "loss": 272.4948,
3201
+ "step": 4560
3202
+ },
3203
+ {
3204
+ "epoch": 0.03692660735784872,
3205
+ "grad_norm": 1040.892822265625,
3206
+ "learning_rate": 1.8464646464646464e-05,
3207
+ "loss": 271.458,
3208
+ "step": 4570
3209
+ },
3210
+ {
3211
+ "epoch": 0.03700740956213285,
3212
+ "grad_norm": 4451.0947265625,
3213
+ "learning_rate": 1.8505050505050506e-05,
3214
+ "loss": 402.6701,
3215
+ "step": 4580
3216
+ },
3217
+ {
3218
+ "epoch": 0.03708821176641699,
3219
+ "grad_norm": 3623.4384765625,
3220
+ "learning_rate": 1.8545454545454545e-05,
3221
+ "loss": 311.2422,
3222
+ "step": 4590
3223
+ },
3224
+ {
3225
+ "epoch": 0.03716901397070112,
3226
+ "grad_norm": 2795.01806640625,
3227
+ "learning_rate": 1.8585858585858588e-05,
3228
+ "loss": 337.2056,
3229
+ "step": 4600
3230
+ },
3231
+ {
3232
+ "epoch": 0.037249816174985254,
3233
+ "grad_norm": 1317.880859375,
3234
+ "learning_rate": 1.8626262626262626e-05,
3235
+ "loss": 367.2568,
3236
+ "step": 4610
3237
+ },
3238
+ {
3239
+ "epoch": 0.03733061837926939,
3240
+ "grad_norm": 3477.01953125,
3241
+ "learning_rate": 1.866666666666667e-05,
3242
+ "loss": 364.4255,
3243
+ "step": 4620
3244
+ },
3245
+ {
3246
+ "epoch": 0.03741142058355352,
3247
+ "grad_norm": 5694.70068359375,
3248
+ "learning_rate": 1.8707070707070707e-05,
3249
+ "loss": 356.1188,
3250
+ "step": 4630
3251
+ },
3252
+ {
3253
+ "epoch": 0.037492222787837655,
3254
+ "grad_norm": 1367.3173828125,
3255
+ "learning_rate": 1.874747474747475e-05,
3256
+ "loss": 370.9894,
3257
+ "step": 4640
3258
+ },
3259
+ {
3260
+ "epoch": 0.037573024992121784,
3261
+ "grad_norm": 2475.6630859375,
3262
+ "learning_rate": 1.878787878787879e-05,
3263
+ "loss": 362.2273,
3264
+ "step": 4650
3265
+ },
3266
+ {
3267
+ "epoch": 0.03765382719640592,
3268
+ "grad_norm": 2650.109619140625,
3269
+ "learning_rate": 1.8828282828282827e-05,
3270
+ "loss": 374.3637,
3271
+ "step": 4660
3272
+ },
3273
+ {
3274
+ "epoch": 0.03773462940069005,
3275
+ "grad_norm": 1204.2740478515625,
3276
+ "learning_rate": 1.886868686868687e-05,
3277
+ "loss": 392.7774,
3278
+ "step": 4670
3279
+ },
3280
+ {
3281
+ "epoch": 0.037815431604974185,
3282
+ "grad_norm": 1503.53076171875,
3283
+ "learning_rate": 1.890909090909091e-05,
3284
+ "loss": 324.107,
3285
+ "step": 4680
3286
+ },
3287
+ {
3288
+ "epoch": 0.037896233809258315,
3289
+ "grad_norm": 1724.01953125,
3290
+ "learning_rate": 1.894949494949495e-05,
3291
+ "loss": 327.8625,
3292
+ "step": 4690
3293
+ },
3294
+ {
3295
+ "epoch": 0.03797703601354245,
3296
+ "grad_norm": 879.3681640625,
3297
+ "learning_rate": 1.898989898989899e-05,
3298
+ "loss": 294.3035,
3299
+ "step": 4700
3300
+ },
3301
+ {
3302
+ "epoch": 0.03805783821782658,
3303
+ "grad_norm": 1925.4000244140625,
3304
+ "learning_rate": 1.9030303030303032e-05,
3305
+ "loss": 322.031,
3306
+ "step": 4710
3307
+ },
3308
+ {
3309
+ "epoch": 0.038138640422110716,
3310
+ "grad_norm": 1489.126708984375,
3311
+ "learning_rate": 1.907070707070707e-05,
3312
+ "loss": 304.9248,
3313
+ "step": 4720
3314
+ },
3315
+ {
3316
+ "epoch": 0.038219442626394845,
3317
+ "grad_norm": 1604.568603515625,
3318
+ "learning_rate": 1.9111111111111113e-05,
3319
+ "loss": 373.274,
3320
+ "step": 4730
3321
+ },
3322
+ {
3323
+ "epoch": 0.03830024483067898,
3324
+ "grad_norm": 4657.24755859375,
3325
+ "learning_rate": 1.9151515151515155e-05,
3326
+ "loss": 390.6698,
3327
+ "step": 4740
3328
+ },
3329
+ {
3330
+ "epoch": 0.03838104703496312,
3331
+ "grad_norm": 2019.6849365234375,
3332
+ "learning_rate": 1.919191919191919e-05,
3333
+ "loss": 361.781,
3334
+ "step": 4750
3335
+ },
3336
+ {
3337
+ "epoch": 0.038461849239247246,
3338
+ "grad_norm": 3976.860107421875,
3339
+ "learning_rate": 1.9232323232323233e-05,
3340
+ "loss": 309.9517,
3341
+ "step": 4760
3342
+ },
3343
+ {
3344
+ "epoch": 0.03854265144353138,
3345
+ "grad_norm": 996.7420043945312,
3346
+ "learning_rate": 1.9272727272727272e-05,
3347
+ "loss": 273.8845,
3348
+ "step": 4770
3349
+ },
3350
+ {
3351
+ "epoch": 0.03862345364781551,
3352
+ "grad_norm": 5474.6142578125,
3353
+ "learning_rate": 1.9313131313131314e-05,
3354
+ "loss": 345.1392,
3355
+ "step": 4780
3356
+ },
3357
+ {
3358
+ "epoch": 0.03870425585209965,
3359
+ "grad_norm": 1052.18115234375,
3360
+ "learning_rate": 1.9353535353535353e-05,
3361
+ "loss": 308.6369,
3362
+ "step": 4790
3363
+ },
3364
+ {
3365
+ "epoch": 0.03878505805638378,
3366
+ "grad_norm": 2130.4462890625,
3367
+ "learning_rate": 1.9393939393939395e-05,
3368
+ "loss": 298.5647,
3369
+ "step": 4800
3370
+ },
3371
+ {
3372
+ "epoch": 0.03886586026066791,
3373
+ "grad_norm": 4150.1962890625,
3374
+ "learning_rate": 1.9434343434343434e-05,
3375
+ "loss": 285.5174,
3376
+ "step": 4810
3377
+ },
3378
+ {
3379
+ "epoch": 0.03894666246495204,
3380
+ "grad_norm": 1532.9091796875,
3381
+ "learning_rate": 1.9474747474747476e-05,
3382
+ "loss": 382.989,
3383
+ "step": 4820
3384
+ },
3385
+ {
3386
+ "epoch": 0.03902746466923618,
3387
+ "grad_norm": 2456.46875,
3388
+ "learning_rate": 1.951515151515152e-05,
3389
+ "loss": 373.614,
3390
+ "step": 4830
3391
+ },
3392
+ {
3393
+ "epoch": 0.03910826687352031,
3394
+ "grad_norm": 16273.7587890625,
3395
+ "learning_rate": 1.9555555555555557e-05,
3396
+ "loss": 271.4283,
3397
+ "step": 4840
3398
+ },
3399
+ {
3400
+ "epoch": 0.03918906907780444,
3401
+ "grad_norm": 1657.758544921875,
3402
+ "learning_rate": 1.95959595959596e-05,
3403
+ "loss": 341.233,
3404
+ "step": 4850
3405
+ },
3406
+ {
3407
+ "epoch": 0.03926987128208857,
3408
+ "grad_norm": 1495.0029296875,
3409
+ "learning_rate": 1.9636363636363635e-05,
3410
+ "loss": 396.9126,
3411
+ "step": 4860
3412
+ },
3413
+ {
3414
+ "epoch": 0.03935067348637271,
3415
+ "grad_norm": 911.6842651367188,
3416
+ "learning_rate": 1.9676767676767677e-05,
3417
+ "loss": 259.0603,
3418
+ "step": 4870
3419
+ },
3420
+ {
3421
+ "epoch": 0.03943147569065684,
3422
+ "grad_norm": 3205.702392578125,
3423
+ "learning_rate": 1.9717171717171716e-05,
3424
+ "loss": 345.96,
3425
+ "step": 4880
3426
+ },
3427
+ {
3428
+ "epoch": 0.039512277894940974,
3429
+ "grad_norm": 1126.9405517578125,
3430
+ "learning_rate": 1.975757575757576e-05,
3431
+ "loss": 277.1656,
3432
+ "step": 4890
3433
+ },
3434
+ {
3435
+ "epoch": 0.03959308009922511,
3436
+ "grad_norm": 1073.0098876953125,
3437
+ "learning_rate": 1.9797979797979797e-05,
3438
+ "loss": 314.1923,
3439
+ "step": 4900
3440
+ },
3441
+ {
3442
+ "epoch": 0.03967388230350924,
3443
+ "grad_norm": 2321.18408203125,
3444
+ "learning_rate": 1.983838383838384e-05,
3445
+ "loss": 352.1233,
3446
+ "step": 4910
3447
+ },
3448
+ {
3449
+ "epoch": 0.039754684507793375,
3450
+ "grad_norm": 3440.341064453125,
3451
+ "learning_rate": 1.987878787878788e-05,
3452
+ "loss": 387.7612,
3453
+ "step": 4920
3454
+ },
3455
+ {
3456
+ "epoch": 0.039835486712077504,
3457
+ "grad_norm": 2081.38134765625,
3458
+ "learning_rate": 1.991919191919192e-05,
3459
+ "loss": 331.0657,
3460
+ "step": 4930
3461
+ },
3462
+ {
3463
+ "epoch": 0.03991628891636164,
3464
+ "grad_norm": 1204.45703125,
3465
+ "learning_rate": 1.9959595959595963e-05,
3466
+ "loss": 310.2021,
3467
+ "step": 4940
3468
+ },
3469
+ {
3470
+ "epoch": 0.03999709112064577,
3471
+ "grad_norm": 1972.9552001953125,
3472
+ "learning_rate": 2e-05,
3473
+ "loss": 284.6502,
3474
+ "step": 4950
3475
+ },
3476
+ {
3477
+ "epoch": 0.040077893324929906,
3478
+ "grad_norm": 2157.6552734375,
3479
+ "learning_rate": 2.0040404040404044e-05,
3480
+ "loss": 374.1642,
3481
+ "step": 4960
3482
+ },
3483
+ {
3484
+ "epoch": 0.040158695529214035,
3485
+ "grad_norm": 1071.7281494140625,
3486
+ "learning_rate": 2.008080808080808e-05,
3487
+ "loss": 315.1676,
3488
+ "step": 4970
3489
+ },
3490
+ {
3491
+ "epoch": 0.04023949773349817,
3492
+ "grad_norm": 2088.307861328125,
3493
+ "learning_rate": 2.012121212121212e-05,
3494
+ "loss": 298.2418,
3495
+ "step": 4980
3496
+ },
3497
+ {
3498
+ "epoch": 0.0403202999377823,
3499
+ "grad_norm": 1532.578857421875,
3500
+ "learning_rate": 2.016161616161616e-05,
3501
+ "loss": 343.915,
3502
+ "step": 4990
3503
+ },
3504
+ {
3505
+ "epoch": 0.040401102142066436,
3506
+ "grad_norm": 1117.131591796875,
3507
+ "learning_rate": 2.0202020202020203e-05,
3508
+ "loss": 266.2881,
3509
+ "step": 5000
3510
+ }
3511
+ ],
3512
+ "logging_steps": 10,
3513
+ "max_steps": 123750,
3514
+ "num_input_tokens_seen": 0,
3515
+ "num_train_epochs": 1,
3516
+ "save_steps": 1000,
3517
+ "stateful_callbacks": {
3518
+ "TrainerControl": {
3519
+ "args": {
3520
+ "should_epoch_stop": false,
3521
+ "should_evaluate": false,
3522
+ "should_log": false,
3523
+ "should_save": true,
3524
+ "should_training_stop": false
3525
+ },
3526
+ "attributes": {}
3527
+ }
3528
+ },
3529
+ "total_flos": 0.0,
3530
+ "train_batch_size": 8,
3531
+ "trial_name": null,
3532
+ "trial_params": null
3533
+ }