Hugging Face's logo

336labs
/
VisionEncoder-to-MLLM-ModelZoo

Safetensors
vision-encoder-evaluation
multimodal
visual-tokenizer-benchmark
checkpoints
Model card Files
xet
 Community
lyunee commited on
Commit
e8e6470
·
verified ·
1 Parent(s): 7a70fea

Upload continuous evaluation checkpoints, batch 44

Browse files
Files changed (42) hide show
  1. continuous/smollm2/siglip2_sm14_224/finetune/model.safetensors +3 -0
  2. continuous/smollm2/siglip2_sm14_224/finetune/special_tokens_map.json +28 -0
  3. continuous/smollm2/siglip2_sm14_224/finetune/tokenizer.json +0 -0
  4. continuous/smollm2/siglip2_sm14_224/finetune/tokenizer_config.json +155 -0
  5. continuous/smollm2/siglip2_sm14_224/finetune/trainer_state.json +1856 -0
  6. continuous/smollm2/siglip2_sm14_224/finetune/training_args.bin +3 -0
  7. continuous/smollm2/siglip2_sm14_224/finetune/vocab.json +0 -0
  8. continuous/smollm2/siglip2_sm14_224/pretrain/config.json +70 -0
  9. continuous/smollm2/siglip2_sm14_224/pretrain/mm_projector.bin +3 -0
  10. continuous/smollm2/siglip2_sm14_224/pretrain/trainer_state.json +421 -0
  11. continuous/smollm2/siglip2_sm14_384/finetune/chat_template.jinja +6 -0
  12. continuous/smollm2/siglip2_sm14_384/finetune/config.json +70 -0
  13. continuous/smollm2/siglip2_sm14_384/finetune/generation_config.json +9 -0
  14. continuous/smollm2/siglip2_sm14_384/finetune/merges.txt +0 -0
  15. continuous/smollm2/siglip2_sm14_384/finetune/model.safetensors +3 -0
  16. continuous/smollm2/siglip2_sm14_384/finetune/special_tokens_map.json +28 -0
  17. continuous/smollm2/siglip2_sm14_384/finetune/tokenizer.json +0 -0
  18. continuous/smollm2/siglip2_sm14_384/finetune/tokenizer_config.json +155 -0
  19. continuous/smollm2/siglip2_sm14_384/finetune/trainer_state.json +1856 -0
  20. continuous/smollm2/siglip2_sm14_384/finetune/training_args.bin +3 -0
  21. continuous/smollm2/siglip2_sm14_384/finetune/vocab.json +0 -0
  22. continuous/smollm2/siglip2_sm14_384/pretrain/config.json +70 -0
  23. continuous/smollm2/siglip2_sm14_384/pretrain/mm_projector.bin +3 -0
  24. continuous/smollm2/siglip2_sm14_384/pretrain/trainer_state.json +421 -0
  25. continuous/smollm2/siglip2_sm16_256/finetune/chat_template.jinja +6 -0
  26. continuous/smollm2/siglip2_sm16_256/finetune/config.json +70 -0
  27. continuous/smollm2/siglip2_sm16_256/finetune/generation_config.json +9 -0
  28. continuous/smollm2/siglip2_sm16_256/finetune/merges.txt +0 -0
  29. continuous/smollm2/siglip2_sm16_256/finetune/model.safetensors +3 -0
  30. continuous/smollm2/siglip2_sm16_256/finetune/special_tokens_map.json +28 -0
  31. continuous/smollm2/siglip2_sm16_256/finetune/tokenizer.json +0 -0
  32. continuous/smollm2/siglip2_sm16_256/finetune/tokenizer_config.json +155 -0
  33. continuous/smollm2/siglip2_sm16_256/finetune/trainer_state.json +1856 -0
  34. continuous/smollm2/siglip2_sm16_256/finetune/training_args.bin +3 -0
  35. continuous/smollm2/siglip2_sm16_256/finetune/vocab.json +0 -0
  36. continuous/smollm2/siglip2_sm16_256/pretrain/config.json +70 -0
  37. continuous/smollm2/siglip2_sm16_256/pretrain/mm_projector.bin +3 -0
  38. continuous/smollm2/siglip2_sm16_256/pretrain/trainer_state.json +421 -0
  39. continuous/smollm2/siglip2_sm16_384/finetune/chat_template.jinja +6 -0
  40. continuous/smollm2/siglip2_sm16_384/finetune/config.json +70 -0
  41. continuous/smollm2/siglip2_sm16_384/finetune/generation_config.json +9 -0
  42. continuous/smollm2/siglip2_sm16_384/finetune/merges.txt +0 -0
continuous/smollm2/siglip2_sm14_224/finetune/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5fda18a871758977fac95ea9d1fbc2f10f4fd893ee2935d58ef96f987b7f1ee
3
+ size 4505743736
continuous/smollm2/siglip2_sm14_224/finetune/special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|im_start|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|im_end|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|endoftext|>",
21
+ "unk_token": {
22
+ "content": "<|endoftext|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
continuous/smollm2/siglip2_sm14_224/finetune/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm14_224/finetune/tokenizer_config.json ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<repo_name>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "4": {
37
+ "content": "<reponame>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "5": {
45
+ "content": "<file_sep>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "6": {
53
+ "content": "<filename>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "7": {
61
+ "content": "<gh_stars>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "8": {
69
+ "content": "<issue_start>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "9": {
77
+ "content": "<issue_comment>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "10": {
85
+ "content": "<issue_closed>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "11": {
93
+ "content": "<jupyter_start>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "12": {
101
+ "content": "<jupyter_text>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "13": {
109
+ "content": "<jupyter_code>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "14": {
117
+ "content": "<jupyter_output>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": true
123
+ },
124
+ "15": {
125
+ "content": "<jupyter_script>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": true
131
+ },
132
+ "16": {
133
+ "content": "<empty_output>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": true
139
+ }
140
+ },
141
+ "additional_special_tokens": [
142
+ "<|im_start|>",
143
+ "<|im_end|>"
144
+ ],
145
+ "bos_token": "<|im_start|>",
146
+ "clean_up_tokenization_spaces": false,
147
+ "eos_token": "<|im_end|>",
148
+ "extra_special_tokens": {},
149
+ "model_max_length": 8192,
150
+ "pad_token": "<|endoftext|>",
151
+ "padding_side": "right",
152
+ "tokenizer_class": "GPT2Tokenizer",
153
+ "unk_token": "<|endoftext|>",
154
+ "vocab_size": 49152
155
+ }
continuous/smollm2/siglip2_sm14_224/finetune/trainer_state.json ADDED
@@ -0,0 +1,1856 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2597,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.003850596842510589,
14
+ "grad_norm": 3.410252809524536,
15
+ "learning_rate": 1.153846153846154e-06,
16
+ "loss": 1.3183,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.007701193685021178,
21
+ "grad_norm": 2.246640205383301,
22
+ "learning_rate": 2.435897435897436e-06,
23
+ "loss": 1.1553,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.011551790527531768,
28
+ "grad_norm": 1.860365390777588,
29
+ "learning_rate": 3.7179487179487184e-06,
30
+ "loss": 0.9841,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.015402387370042356,
35
+ "grad_norm": 1.3328458070755005,
36
+ "learning_rate": 5e-06,
37
+ "loss": 0.9245,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.019252984212552945,
42
+ "grad_norm": 1.2704187631607056,
43
+ "learning_rate": 6.282051282051282e-06,
44
+ "loss": 0.9118,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.023103581055063535,
49
+ "grad_norm": 0.260476678609848,
50
+ "learning_rate": 7.564102564102564e-06,
51
+ "loss": 0.8412,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.026954177897574125,
56
+ "grad_norm": 0.2262180596590042,
57
+ "learning_rate": 8.846153846153847e-06,
58
+ "loss": 0.7975,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.030804774740084712,
63
+ "grad_norm": 0.21982882916927338,
64
+ "learning_rate": 9.999996111488723e-06,
65
+ "loss": 0.8115,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.0346553715825953,
70
+ "grad_norm": 0.2095281332731247,
71
+ "learning_rate": 9.999529497453782e-06,
72
+ "loss": 0.7972,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.03850596842510589,
77
+ "grad_norm": 0.21203096210956573,
78
+ "learning_rate": 9.998285264324193e-06,
79
+ "loss": 0.7619,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.04235656526761648,
84
+ "grad_norm": 0.22813068330287933,
85
+ "learning_rate": 9.996263605626052e-06,
86
+ "loss": 0.7576,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.04620716211012707,
91
+ "grad_norm": 0.20328658819198608,
92
+ "learning_rate": 9.99346483580503e-06,
93
+ "loss": 0.7761,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.05005775895263766,
98
+ "grad_norm": 0.20365795493125916,
99
+ "learning_rate": 9.989889390177467e-06,
100
+ "loss": 0.7501,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.05390835579514825,
105
+ "grad_norm": 0.21199211478233337,
106
+ "learning_rate": 9.985537824862647e-06,
107
+ "loss": 0.7657,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.05775895263765884,
112
+ "grad_norm": 0.212788388133049,
113
+ "learning_rate": 9.980410816696317e-06,
114
+ "loss": 0.7423,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.061609549480169425,
119
+ "grad_norm": 0.19180867075920105,
120
+ "learning_rate": 9.97450916312541e-06,
121
+ "loss": 0.7497,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.06546014632268002,
126
+ "grad_norm": 0.16682106256484985,
127
+ "learning_rate": 9.967833782084004e-06,
128
+ "loss": 0.751,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.0693107431651906,
133
+ "grad_norm": 0.18626518547534943,
134
+ "learning_rate": 9.960385711850554e-06,
135
+ "loss": 0.7806,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.07316134000770119,
140
+ "grad_norm": 0.20952846109867096,
141
+ "learning_rate": 9.952166110886394e-06,
142
+ "loss": 0.7669,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.07701193685021178,
147
+ "grad_norm": 0.1880127489566803,
148
+ "learning_rate": 9.943176257655567e-06,
149
+ "loss": 0.7653,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.08086253369272237,
154
+ "grad_norm": 0.21062295138835907,
155
+ "learning_rate": 9.933417550425952e-06,
156
+ "loss": 0.7564,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.08471313053523297,
161
+ "grad_norm": 0.17835523188114166,
162
+ "learning_rate": 9.9228915070518e-06,
163
+ "loss": 0.7415,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.08856372737774355,
168
+ "grad_norm": 0.183059960603714,
169
+ "learning_rate": 9.911599764737632e-06,
170
+ "loss": 0.7384,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.09241432422025414,
175
+ "grad_norm": 0.17969390749931335,
176
+ "learning_rate": 9.899544079783606e-06,
177
+ "loss": 0.7341,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.09626492106276473,
182
+ "grad_norm": 0.1874191015958786,
183
+ "learning_rate": 9.886726327312338e-06,
184
+ "loss": 0.7514,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.10011551790527531,
189
+ "grad_norm": 0.1626584529876709,
190
+ "learning_rate": 9.87314850097724e-06,
191
+ "loss": 0.7491,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.1039661147477859,
196
+ "grad_norm": 0.1819298416376114,
197
+ "learning_rate": 9.858812712652452e-06,
198
+ "loss": 0.7317,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.1078167115902965,
203
+ "grad_norm": 0.1637846827507019,
204
+ "learning_rate": 9.843721192104341e-06,
205
+ "loss": 0.7292,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.11166730843280709,
210
+ "grad_norm": 0.17745709419250488,
211
+ "learning_rate": 9.827876286644698e-06,
212
+ "loss": 0.7606,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.11551790527531768,
217
+ "grad_norm": 0.16406483948230743,
218
+ "learning_rate": 9.811280460765643e-06,
219
+ "loss": 0.7196,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.11936850211782826,
224
+ "grad_norm": 0.16925553977489471,
225
+ "learning_rate": 9.793936295756292e-06,
226
+ "loss": 0.7239,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.12321909896033885,
231
+ "grad_norm": 0.17721088230609894,
232
+ "learning_rate": 9.775846489301279e-06,
233
+ "loss": 0.7364,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.12706969580284944,
238
+ "grad_norm": 0.1944226622581482,
239
+ "learning_rate": 9.757013855061157e-06,
240
+ "loss": 0.7346,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.13092029264536004,
245
+ "grad_norm": 0.17596177756786346,
246
+ "learning_rate": 9.737441322234758e-06,
247
+ "loss": 0.7382,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.1347708894878706,
252
+ "grad_norm": 0.18825531005859375,
253
+ "learning_rate": 9.717131935103605e-06,
254
+ "loss": 0.745,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.1386214863303812,
259
+ "grad_norm": 0.17958585917949677,
260
+ "learning_rate": 9.696088852558398e-06,
261
+ "loss": 0.7269,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.1424720831728918,
266
+ "grad_norm": 0.2009432464838028,
267
+ "learning_rate": 9.674315347607686e-06,
268
+ "loss": 0.7554,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.14632268001540238,
273
+ "grad_norm": 0.16520164906978607,
274
+ "learning_rate": 9.651814806868796e-06,
275
+ "loss": 0.7209,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.15017327685791299,
280
+ "grad_norm": 0.17699997127056122,
281
+ "learning_rate": 9.62859073004107e-06,
282
+ "loss": 0.7203,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.15402387370042356,
287
+ "grad_norm": 0.16259215772151947,
288
+ "learning_rate": 9.604646729361547e-06,
289
+ "loss": 0.7277,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.15787447054293416,
294
+ "grad_norm": 0.18949541449546814,
295
+ "learning_rate": 9.579986529043099e-06,
296
+ "loss": 0.7231,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.16172506738544473,
301
+ "grad_norm": 0.1809382140636444,
302
+ "learning_rate": 9.554613964695189e-06,
303
+ "loss": 0.7417,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.16557566422795533,
308
+ "grad_norm": 0.18302281200885773,
309
+ "learning_rate": 9.528532982727281e-06,
310
+ "loss": 0.7247,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.16942626107046593,
315
+ "grad_norm": 0.1747412532567978,
316
+ "learning_rate": 9.501747639735018e-06,
317
+ "loss": 0.7275,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.1732768579129765,
322
+ "grad_norm": 0.1819590926170349,
323
+ "learning_rate": 9.474262101869272e-06,
324
+ "loss": 0.7143,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.1771274547554871,
329
+ "grad_norm": 0.18618161976337433,
330
+ "learning_rate": 9.446080644188141e-06,
331
+ "loss": 0.7242,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.18097805159799768,
336
+ "grad_norm": 0.1761256754398346,
337
+ "learning_rate": 9.417207649992017e-06,
338
+ "loss": 0.7265,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.18482864844050828,
343
+ "grad_norm": 0.18174435198307037,
344
+ "learning_rate": 9.387647610141816e-06,
345
+ "loss": 0.7225,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.18867924528301888,
350
+ "grad_norm": 0.1641910970211029,
351
+ "learning_rate": 9.35740512236047e-06,
352
+ "loss": 0.7196,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.19252984212552945,
357
+ "grad_norm": 0.1642703413963318,
358
+ "learning_rate": 9.326484890517809e-06,
359
+ "loss": 0.7246,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.19638043896804006,
364
+ "grad_norm": 0.1761842519044876,
365
+ "learning_rate": 9.294891723898926e-06,
366
+ "loss": 0.7129,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.20023103581055063,
371
+ "grad_norm": 0.1762363761663437,
372
+ "learning_rate": 9.262630536456153e-06,
373
+ "loss": 0.7354,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.20408163265306123,
378
+ "grad_norm": 0.17389236390590668,
379
+ "learning_rate": 9.229706346044749e-06,
380
+ "loss": 0.7207,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.2079322294955718,
385
+ "grad_norm": 0.17771634459495544,
386
+ "learning_rate": 9.196124273642422e-06,
387
+ "loss": 0.7294,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 0.2117828263380824,
392
+ "grad_norm": 0.17216582596302032,
393
+ "learning_rate": 9.161889542552842e-06,
394
+ "loss": 0.7372,
395
+ "step": 550
396
+ },
397
+ {
398
+ "epoch": 0.215633423180593,
399
+ "grad_norm": 0.1693568378686905,
400
+ "learning_rate": 9.127007477593193e-06,
401
+ "loss": 0.7374,
402
+ "step": 560
403
+ },
404
+ {
405
+ "epoch": 0.21948402002310358,
406
+ "grad_norm": 0.1783280074596405,
407
+ "learning_rate": 9.091483504265974e-06,
408
+ "loss": 0.7474,
409
+ "step": 570
410
+ },
411
+ {
412
+ "epoch": 0.22333461686561418,
413
+ "grad_norm": 0.18218481540679932,
414
+ "learning_rate": 9.055323147915116e-06,
415
+ "loss": 0.7184,
416
+ "step": 580
417
+ },
418
+ {
419
+ "epoch": 0.22718521370812475,
420
+ "grad_norm": 0.17550291121006012,
421
+ "learning_rate": 9.018532032866585e-06,
422
+ "loss": 0.7314,
423
+ "step": 590
424
+ },
425
+ {
426
+ "epoch": 0.23103581055063535,
427
+ "grad_norm": 0.16836154460906982,
428
+ "learning_rate": 8.981115881553579e-06,
429
+ "loss": 0.7295,
430
+ "step": 600
431
+ },
432
+ {
433
+ "epoch": 0.23488640739314592,
434
+ "grad_norm": 0.1895703822374344,
435
+ "learning_rate": 8.943080513626476e-06,
436
+ "loss": 0.7489,
437
+ "step": 610
438
+ },
439
+ {
440
+ "epoch": 0.23873700423565652,
441
+ "grad_norm": 0.16716930270195007,
442
+ "learning_rate": 8.904431845047641e-06,
443
+ "loss": 0.7278,
444
+ "step": 620
445
+ },
446
+ {
447
+ "epoch": 0.24258760107816713,
448
+ "grad_norm": 0.17479395866394043,
449
+ "learning_rate": 8.86517588717128e-06,
450
+ "loss": 0.7395,
451
+ "step": 630
452
+ },
453
+ {
454
+ "epoch": 0.2464381979206777,
455
+ "grad_norm": 0.1768491119146347,
456
+ "learning_rate": 8.82531874580844e-06,
457
+ "loss": 0.7415,
458
+ "step": 640
459
+ },
460
+ {
461
+ "epoch": 0.25028879476318827,
462
+ "grad_norm": 0.16241413354873657,
463
+ "learning_rate": 8.784866620277314e-06,
464
+ "loss": 0.7001,
465
+ "step": 650
466
+ },
467
+ {
468
+ "epoch": 0.25413939160569887,
469
+ "grad_norm": 0.18252688646316528,
470
+ "learning_rate": 8.743825802439018e-06,
471
+ "loss": 0.7218,
472
+ "step": 660
473
+ },
474
+ {
475
+ "epoch": 0.2579899884482095,
476
+ "grad_norm": 0.1709984540939331,
477
+ "learning_rate": 8.702202675718956e-06,
478
+ "loss": 0.7355,
479
+ "step": 670
480
+ },
481
+ {
482
+ "epoch": 0.2618405852907201,
483
+ "grad_norm": 0.1637190580368042,
484
+ "learning_rate": 8.660003714113959e-06,
485
+ "loss": 0.7138,
486
+ "step": 680
487
+ },
488
+ {
489
+ "epoch": 0.2656911821332307,
490
+ "grad_norm": 0.17103135585784912,
491
+ "learning_rate": 8.617235481185323e-06,
492
+ "loss": 0.7237,
493
+ "step": 690
494
+ },
495
+ {
496
+ "epoch": 0.2695417789757412,
497
+ "grad_norm": 0.17852146923542023,
498
+ "learning_rate": 8.573904629037931e-06,
499
+ "loss": 0.7373,
500
+ "step": 700
501
+ },
502
+ {
503
+ "epoch": 0.2733923758182518,
504
+ "grad_norm": 0.16110314428806305,
505
+ "learning_rate": 8.530017897285585e-06,
506
+ "loss": 0.7201,
507
+ "step": 710
508
+ },
509
+ {
510
+ "epoch": 0.2772429726607624,
511
+ "grad_norm": 0.1659325212240219,
512
+ "learning_rate": 8.485582112002746e-06,
513
+ "loss": 0.7214,
514
+ "step": 720
515
+ },
516
+ {
517
+ "epoch": 0.281093569503273,
518
+ "grad_norm": 0.169901505112648,
519
+ "learning_rate": 8.440604184662811e-06,
520
+ "loss": 0.7279,
521
+ "step": 730
522
+ },
523
+ {
524
+ "epoch": 0.2849441663457836,
525
+ "grad_norm": 0.17225155234336853,
526
+ "learning_rate": 8.395091111063118e-06,
527
+ "loss": 0.7146,
528
+ "step": 740
529
+ },
530
+ {
531
+ "epoch": 0.28879476318829417,
532
+ "grad_norm": 0.15956944227218628,
533
+ "learning_rate": 8.349049970236822e-06,
534
+ "loss": 0.7132,
535
+ "step": 750
536
+ },
537
+ {
538
+ "epoch": 0.29264536003080477,
539
+ "grad_norm": 0.1736423522233963,
540
+ "learning_rate": 8.302487923351842e-06,
541
+ "loss": 0.7088,
542
+ "step": 760
543
+ },
544
+ {
545
+ "epoch": 0.29649595687331537,
546
+ "grad_norm": 0.15401533246040344,
547
+ "learning_rate": 8.255412212597021e-06,
548
+ "loss": 0.7275,
549
+ "step": 770
550
+ },
551
+ {
552
+ "epoch": 0.30034655371582597,
553
+ "grad_norm": 0.17427247762680054,
554
+ "learning_rate": 8.207830160055688e-06,
555
+ "loss": 0.724,
556
+ "step": 780
557
+ },
558
+ {
559
+ "epoch": 0.3041971505583365,
560
+ "grad_norm": 0.16461236774921417,
561
+ "learning_rate": 8.159749166566786e-06,
562
+ "loss": 0.7054,
563
+ "step": 790
564
+ },
565
+ {
566
+ "epoch": 0.3080477474008471,
567
+ "grad_norm": 0.174094557762146,
568
+ "learning_rate": 8.11117671057377e-06,
569
+ "loss": 0.7271,
570
+ "step": 800
571
+ },
572
+ {
573
+ "epoch": 0.3118983442433577,
574
+ "grad_norm": 0.1690794825553894,
575
+ "learning_rate": 8.062120346961404e-06,
576
+ "loss": 0.7247,
577
+ "step": 810
578
+ },
579
+ {
580
+ "epoch": 0.3157489410858683,
581
+ "grad_norm": 0.1610432118177414,
582
+ "learning_rate": 8.012587705880705e-06,
583
+ "loss": 0.7211,
584
+ "step": 820
585
+ },
586
+ {
587
+ "epoch": 0.3195995379283789,
588
+ "grad_norm": 0.17133329808712006,
589
+ "learning_rate": 7.962586491562146e-06,
590
+ "loss": 0.7089,
591
+ "step": 830
592
+ },
593
+ {
594
+ "epoch": 0.32345013477088946,
595
+ "grad_norm": 0.16634228825569153,
596
+ "learning_rate": 7.912124481117357e-06,
597
+ "loss": 0.7088,
598
+ "step": 840
599
+ },
600
+ {
601
+ "epoch": 0.32730073161340006,
602
+ "grad_norm": 0.16256579756736755,
603
+ "learning_rate": 7.861209523329494e-06,
604
+ "loss": 0.7216,
605
+ "step": 850
606
+ },
607
+ {
608
+ "epoch": 0.33115132845591067,
609
+ "grad_norm": 0.16937249898910522,
610
+ "learning_rate": 7.809849537432432e-06,
611
+ "loss": 0.7103,
612
+ "step": 860
613
+ },
614
+ {
615
+ "epoch": 0.33500192529842127,
616
+ "grad_norm": 0.16090841591358185,
617
+ "learning_rate": 7.758052511879035e-06,
618
+ "loss": 0.7135,
619
+ "step": 870
620
+ },
621
+ {
622
+ "epoch": 0.33885252214093187,
623
+ "grad_norm": 0.16991138458251953,
624
+ "learning_rate": 7.705826503098642e-06,
625
+ "loss": 0.7172,
626
+ "step": 880
627
+ },
628
+ {
629
+ "epoch": 0.3427031189834424,
630
+ "grad_norm": 0.1767956018447876,
631
+ "learning_rate": 7.653179634243979e-06,
632
+ "loss": 0.7118,
633
+ "step": 890
634
+ },
635
+ {
636
+ "epoch": 0.346553715825953,
637
+ "grad_norm": 0.159182608127594,
638
+ "learning_rate": 7.600120093927696e-06,
639
+ "loss": 0.7099,
640
+ "step": 900
641
+ },
642
+ {
643
+ "epoch": 0.3504043126684636,
644
+ "grad_norm": 0.1927134245634079,
645
+ "learning_rate": 7.546656134948729e-06,
646
+ "loss": 0.7021,
647
+ "step": 910
648
+ },
649
+ {
650
+ "epoch": 0.3542549095109742,
651
+ "grad_norm": 0.16242843866348267,
652
+ "learning_rate": 7.492796073008661e-06,
653
+ "loss": 0.7128,
654
+ "step": 920
655
+ },
656
+ {
657
+ "epoch": 0.3581055063534848,
658
+ "grad_norm": 0.18781474232673645,
659
+ "learning_rate": 7.438548285418326e-06,
660
+ "loss": 0.6934,
661
+ "step": 930
662
+ },
663
+ {
664
+ "epoch": 0.36195610319599536,
665
+ "grad_norm": 0.16656655073165894,
666
+ "learning_rate": 7.3839212097948e-06,
667
+ "loss": 0.7241,
668
+ "step": 940
669
+ },
670
+ {
671
+ "epoch": 0.36580670003850596,
672
+ "grad_norm": 0.17641980946063995,
673
+ "learning_rate": 7.328923342749039e-06,
674
+ "loss": 0.7415,
675
+ "step": 950
676
+ },
677
+ {
678
+ "epoch": 0.36965729688101656,
679
+ "grad_norm": 0.1705055683851242,
680
+ "learning_rate": 7.273563238564319e-06,
681
+ "loss": 0.736,
682
+ "step": 960
683
+ },
684
+ {
685
+ "epoch": 0.37350789372352716,
686
+ "grad_norm": 0.16295592486858368,
687
+ "learning_rate": 7.217849507865724e-06,
688
+ "loss": 0.7023,
689
+ "step": 970
690
+ },
691
+ {
692
+ "epoch": 0.37735849056603776,
693
+ "grad_norm": 0.16591008007526398,
694
+ "learning_rate": 7.161790816280858e-06,
695
+ "loss": 0.715,
696
+ "step": 980
697
+ },
698
+ {
699
+ "epoch": 0.3812090874085483,
700
+ "grad_norm": 0.1771365851163864,
701
+ "learning_rate": 7.1053958830920125e-06,
702
+ "loss": 0.6977,
703
+ "step": 990
704
+ },
705
+ {
706
+ "epoch": 0.3850596842510589,
707
+ "grad_norm": 0.17414148151874542,
708
+ "learning_rate": 7.0486734798799715e-06,
709
+ "loss": 0.7132,
710
+ "step": 1000
711
+ },
712
+ {
713
+ "epoch": 0.3889102810935695,
714
+ "grad_norm": 0.16304875910282135,
715
+ "learning_rate": 6.991632429159708e-06,
716
+ "loss": 0.7326,
717
+ "step": 1010
718
+ },
719
+ {
720
+ "epoch": 0.3927608779360801,
721
+ "grad_norm": 0.17977264523506165,
722
+ "learning_rate": 6.934281603008132e-06,
723
+ "loss": 0.7287,
724
+ "step": 1020
725
+ },
726
+ {
727
+ "epoch": 0.39661147477859066,
728
+ "grad_norm": 0.15143145620822906,
729
+ "learning_rate": 6.876629921684143e-06,
730
+ "loss": 0.7251,
731
+ "step": 1030
732
+ },
733
+ {
734
+ "epoch": 0.40046207162110126,
735
+ "grad_norm": 0.1835201531648636,
736
+ "learning_rate": 6.818686352241199e-06,
737
+ "loss": 0.7124,
738
+ "step": 1040
739
+ },
740
+ {
741
+ "epoch": 0.40431266846361186,
742
+ "grad_norm": 0.15366476774215698,
743
+ "learning_rate": 6.760459907132573e-06,
744
+ "loss": 0.7069,
745
+ "step": 1050
746
+ },
747
+ {
748
+ "epoch": 0.40816326530612246,
749
+ "grad_norm": 0.15979094803333282,
750
+ "learning_rate": 6.701959642809589e-06,
751
+ "loss": 0.7133,
752
+ "step": 1060
753
+ },
754
+ {
755
+ "epoch": 0.41201386214863306,
756
+ "grad_norm": 0.16519398987293243,
757
+ "learning_rate": 6.643194658312987e-06,
758
+ "loss": 0.7031,
759
+ "step": 1070
760
+ },
761
+ {
762
+ "epoch": 0.4158644589911436,
763
+ "grad_norm": 0.15494659543037415,
764
+ "learning_rate": 6.584174093857676e-06,
765
+ "loss": 0.732,
766
+ "step": 1080
767
+ },
768
+ {
769
+ "epoch": 0.4197150558336542,
770
+ "grad_norm": 0.1748020499944687,
771
+ "learning_rate": 6.5249071294110735e-06,
772
+ "loss": 0.7147,
773
+ "step": 1090
774
+ },
775
+ {
776
+ "epoch": 0.4235656526761648,
777
+ "grad_norm": 0.164225772023201,
778
+ "learning_rate": 6.465402983265274e-06,
779
+ "loss": 0.726,
780
+ "step": 1100
781
+ },
782
+ {
783
+ "epoch": 0.4274162495186754,
784
+ "grad_norm": 0.1706341654062271,
785
+ "learning_rate": 6.405670910603243e-06,
786
+ "loss": 0.7177,
787
+ "step": 1110
788
+ },
789
+ {
790
+ "epoch": 0.431266846361186,
791
+ "grad_norm": 0.1794327050447464,
792
+ "learning_rate": 6.345720202059291e-06,
793
+ "loss": 0.715,
794
+ "step": 1120
795
+ },
796
+ {
797
+ "epoch": 0.43511744320369655,
798
+ "grad_norm": 0.1683606058359146,
799
+ "learning_rate": 6.285560182274008e-06,
800
+ "loss": 0.6956,
801
+ "step": 1130
802
+ },
803
+ {
804
+ "epoch": 0.43896804004620715,
805
+ "grad_norm": 0.18095041811466217,
806
+ "learning_rate": 6.225200208443936e-06,
807
+ "loss": 0.7212,
808
+ "step": 1140
809
+ },
810
+ {
811
+ "epoch": 0.44281863688871775,
812
+ "grad_norm": 0.17024852335453033,
813
+ "learning_rate": 6.16464966886616e-06,
814
+ "loss": 0.7043,
815
+ "step": 1150
816
+ },
817
+ {
818
+ "epoch": 0.44666923373122835,
819
+ "grad_norm": 0.20930354297161102,
820
+ "learning_rate": 6.103917981478065e-06,
821
+ "loss": 0.7357,
822
+ "step": 1160
823
+ },
824
+ {
825
+ "epoch": 0.45051983057373896,
826
+ "grad_norm": 0.1631799191236496,
827
+ "learning_rate": 6.043014592392489e-06,
828
+ "loss": 0.7412,
829
+ "step": 1170
830
+ },
831
+ {
832
+ "epoch": 0.4543704274162495,
833
+ "grad_norm": 0.16626891493797302,
834
+ "learning_rate": 5.981948974428485e-06,
835
+ "loss": 0.7229,
836
+ "step": 1180
837
+ },
838
+ {
839
+ "epoch": 0.4582210242587601,
840
+ "grad_norm": 0.15446509420871735,
841
+ "learning_rate": 5.920730625637934e-06,
842
+ "loss": 0.696,
843
+ "step": 1190
844
+ },
845
+ {
846
+ "epoch": 0.4620716211012707,
847
+ "grad_norm": 0.1683821678161621,
848
+ "learning_rate": 5.8593690678282355e-06,
849
+ "loss": 0.708,
850
+ "step": 1200
851
+ },
852
+ {
853
+ "epoch": 0.4659222179437813,
854
+ "grad_norm": 0.16211453080177307,
855
+ "learning_rate": 5.797873845081301e-06,
856
+ "loss": 0.7064,
857
+ "step": 1210
858
+ },
859
+ {
860
+ "epoch": 0.46977281478629185,
861
+ "grad_norm": 0.16853362321853638,
862
+ "learning_rate": 5.736254522269075e-06,
863
+ "loss": 0.7295,
864
+ "step": 1220
865
+ },
866
+ {
867
+ "epoch": 0.47362341162880245,
868
+ "grad_norm": 0.16036058962345123,
869
+ "learning_rate": 5.6745206835658386e-06,
870
+ "loss": 0.7274,
871
+ "step": 1230
872
+ },
873
+ {
874
+ "epoch": 0.47747400847131305,
875
+ "grad_norm": 0.15929721295833588,
876
+ "learning_rate": 5.612681930957495e-06,
877
+ "loss": 0.7065,
878
+ "step": 1240
879
+ },
880
+ {
881
+ "epoch": 0.48132460531382365,
882
+ "grad_norm": 0.1578306406736374,
883
+ "learning_rate": 5.550747882748095e-06,
884
+ "loss": 0.6962,
885
+ "step": 1250
886
+ },
887
+ {
888
+ "epoch": 0.48517520215633425,
889
+ "grad_norm": 0.18160882592201233,
890
+ "learning_rate": 5.4887281720638195e-06,
891
+ "loss": 0.7131,
892
+ "step": 1260
893
+ },
894
+ {
895
+ "epoch": 0.4890257989988448,
896
+ "grad_norm": 0.18188680708408356,
897
+ "learning_rate": 5.42663244535466e-06,
898
+ "loss": 0.6872,
899
+ "step": 1270
900
+ },
901
+ {
902
+ "epoch": 0.4928763958413554,
903
+ "grad_norm": 0.17423401772975922,
904
+ "learning_rate": 5.3644703608940205e-06,
905
+ "loss": 0.7063,
906
+ "step": 1280
907
+ },
908
+ {
909
+ "epoch": 0.496726992683866,
910
+ "grad_norm": 0.15617011487483978,
911
+ "learning_rate": 5.302251587276491e-06,
912
+ "loss": 0.717,
913
+ "step": 1290
914
+ },
915
+ {
916
+ "epoch": 0.5005775895263765,
917
+ "grad_norm": 0.15636976063251495,
918
+ "learning_rate": 5.2399858019140005e-06,
919
+ "loss": 0.7078,
920
+ "step": 1300
921
+ },
922
+ {
923
+ "epoch": 0.5044281863688872,
924
+ "grad_norm": 0.16704416275024414,
925
+ "learning_rate": 5.177682689530615e-06,
926
+ "loss": 0.7297,
927
+ "step": 1310
928
+ },
929
+ {
930
+ "epoch": 0.5082787832113977,
931
+ "grad_norm": 0.17475445568561554,
932
+ "learning_rate": 5.11535194065619e-06,
933
+ "loss": 0.6958,
934
+ "step": 1320
935
+ },
936
+ {
937
+ "epoch": 0.5121293800539084,
938
+ "grad_norm": 0.17944228649139404,
939
+ "learning_rate": 5.053003250119113e-06,
940
+ "loss": 0.7038,
941
+ "step": 1330
942
+ },
943
+ {
944
+ "epoch": 0.515979976896419,
945
+ "grad_norm": 0.17061328887939453,
946
+ "learning_rate": 4.9906463155383925e-06,
947
+ "loss": 0.711,
948
+ "step": 1340
949
+ },
950
+ {
951
+ "epoch": 0.5198305737389295,
952
+ "grad_norm": 0.15828372538089752,
953
+ "learning_rate": 4.9282908358153046e-06,
954
+ "loss": 0.7108,
955
+ "step": 1350
956
+ },
957
+ {
958
+ "epoch": 0.5236811705814401,
959
+ "grad_norm": 0.18121634423732758,
960
+ "learning_rate": 4.8659465096248345e-06,
961
+ "loss": 0.7233,
962
+ "step": 1360
963
+ },
964
+ {
965
+ "epoch": 0.5275317674239507,
966
+ "grad_norm": 0.17322197556495667,
967
+ "learning_rate": 4.803623033907169e-06,
968
+ "loss": 0.7141,
969
+ "step": 1370
970
+ },
971
+ {
972
+ "epoch": 0.5313823642664613,
973
+ "grad_norm": 0.16196571290493011,
974
+ "learning_rate": 4.741330102359441e-06,
975
+ "loss": 0.7295,
976
+ "step": 1380
977
+ },
978
+ {
979
+ "epoch": 0.5352329611089719,
980
+ "grad_norm": 0.17091745138168335,
981
+ "learning_rate": 4.679077403927996e-06,
982
+ "loss": 0.7046,
983
+ "step": 1390
984
+ },
985
+ {
986
+ "epoch": 0.5390835579514824,
987
+ "grad_norm": 0.16337157785892487,
988
+ "learning_rate": 4.616874621301373e-06,
989
+ "loss": 0.7074,
990
+ "step": 1400
991
+ },
992
+ {
993
+ "epoch": 0.5429341547939931,
994
+ "grad_norm": 0.17702145874500275,
995
+ "learning_rate": 4.554731429404293e-06,
996
+ "loss": 0.7274,
997
+ "step": 1410
998
+ },
999
+ {
1000
+ "epoch": 0.5467847516365036,
1001
+ "grad_norm": 0.17445914447307587,
1002
+ "learning_rate": 4.492657493892822e-06,
1003
+ "loss": 0.706,
1004
+ "step": 1420
1005
+ },
1006
+ {
1007
+ "epoch": 0.5506353484790143,
1008
+ "grad_norm": 0.1756899505853653,
1009
+ "learning_rate": 4.430662469650999e-06,
1010
+ "loss": 0.7202,
1011
+ "step": 1430
1012
+ },
1013
+ {
1014
+ "epoch": 0.5544859453215248,
1015
+ "grad_norm": 0.17556408047676086,
1016
+ "learning_rate": 4.368755999289118e-06,
1017
+ "loss": 0.7234,
1018
+ "step": 1440
1019
+ },
1020
+ {
1021
+ "epoch": 0.5583365421640354,
1022
+ "grad_norm": 0.17832975089550018,
1023
+ "learning_rate": 4.306947711643949e-06,
1024
+ "loss": 0.7061,
1025
+ "step": 1450
1026
+ },
1027
+ {
1028
+ "epoch": 0.562187139006546,
1029
+ "grad_norm": 0.16593042016029358,
1030
+ "learning_rate": 4.245247220281068e-06,
1031
+ "loss": 0.7111,
1032
+ "step": 1460
1033
+ },
1034
+ {
1035
+ "epoch": 0.5660377358490566,
1036
+ "grad_norm": 0.16543225944042206,
1037
+ "learning_rate": 4.183664121999587e-06,
1038
+ "loss": 0.7049,
1039
+ "step": 1470
1040
+ },
1041
+ {
1042
+ "epoch": 0.5698883326915672,
1043
+ "grad_norm": 0.1694854199886322,
1044
+ "learning_rate": 4.122207995339474e-06,
1045
+ "loss": 0.7003,
1046
+ "step": 1480
1047
+ },
1048
+ {
1049
+ "epoch": 0.5737389295340778,
1050
+ "grad_norm": 0.16000093519687653,
1051
+ "learning_rate": 4.060888399091734e-06,
1052
+ "loss": 0.7078,
1053
+ "step": 1490
1054
+ },
1055
+ {
1056
+ "epoch": 0.5775895263765883,
1057
+ "grad_norm": 0.16897840797901154,
1058
+ "learning_rate": 3.999714870811639e-06,
1059
+ "loss": 0.7214,
1060
+ "step": 1500
1061
+ },
1062
+ {
1063
+ "epoch": 0.581440123219099,
1064
+ "grad_norm": 0.16999736428260803,
1065
+ "learning_rate": 3.938696925335275e-06,
1066
+ "loss": 0.714,
1067
+ "step": 1510
1068
+ },
1069
+ {
1070
+ "epoch": 0.5852907200616095,
1071
+ "grad_norm": 0.1655653566122055,
1072
+ "learning_rate": 3.87784405329962e-06,
1073
+ "loss": 0.7221,
1074
+ "step": 1520
1075
+ },
1076
+ {
1077
+ "epoch": 0.5891413169041202,
1078
+ "grad_norm": 0.15528897941112518,
1079
+ "learning_rate": 3.817165719666382e-06,
1080
+ "loss": 0.7203,
1081
+ "step": 1530
1082
+ },
1083
+ {
1084
+ "epoch": 0.5929919137466307,
1085
+ "grad_norm": 0.16718459129333496,
1086
+ "learning_rate": 3.756671362249841e-06,
1087
+ "loss": 0.726,
1088
+ "step": 1540
1089
+ },
1090
+ {
1091
+ "epoch": 0.5968425105891413,
1092
+ "grad_norm": 0.1574758142232895,
1093
+ "learning_rate": 3.6963703902488907e-06,
1094
+ "loss": 0.7116,
1095
+ "step": 1550
1096
+ },
1097
+ {
1098
+ "epoch": 0.6006931074316519,
1099
+ "grad_norm": 0.16119393706321716,
1100
+ "learning_rate": 3.6362721827835655e-06,
1101
+ "loss": 0.72,
1102
+ "step": 1560
1103
+ },
1104
+ {
1105
+ "epoch": 0.6045437042741625,
1106
+ "grad_norm": 0.17839756608009338,
1107
+ "learning_rate": 3.576386087436212e-06,
1108
+ "loss": 0.7284,
1109
+ "step": 1570
1110
+ },
1111
+ {
1112
+ "epoch": 0.608394301116673,
1113
+ "grad_norm": 0.17771956324577332,
1114
+ "learning_rate": 3.5167214187975913e-06,
1115
+ "loss": 0.7196,
1116
+ "step": 1580
1117
+ },
1118
+ {
1119
+ "epoch": 0.6122448979591837,
1120
+ "grad_norm": 0.15544916689395905,
1121
+ "learning_rate": 3.457287457018086e-06,
1122
+ "loss": 0.7159,
1123
+ "step": 1590
1124
+ },
1125
+ {
1126
+ "epoch": 0.6160954948016942,
1127
+ "grad_norm": 0.16711726784706116,
1128
+ "learning_rate": 3.3980934463643003e-06,
1129
+ "loss": 0.7127,
1130
+ "step": 1600
1131
+ },
1132
+ {
1133
+ "epoch": 0.6199460916442049,
1134
+ "grad_norm": 0.1567486971616745,
1135
+ "learning_rate": 3.3391485937812053e-06,
1136
+ "loss": 0.6893,
1137
+ "step": 1610
1138
+ },
1139
+ {
1140
+ "epoch": 0.6237966884867154,
1141
+ "grad_norm": 0.17628176510334015,
1142
+ "learning_rate": 3.280462067460115e-06,
1143
+ "loss": 0.6906,
1144
+ "step": 1620
1145
+ },
1146
+ {
1147
+ "epoch": 0.627647285329226,
1148
+ "grad_norm": 0.17002440989017487,
1149
+ "learning_rate": 3.222042995412669e-06,
1150
+ "loss": 0.7158,
1151
+ "step": 1630
1152
+ },
1153
+ {
1154
+ "epoch": 0.6314978821717366,
1155
+ "grad_norm": 0.1608469933271408,
1156
+ "learning_rate": 3.163900464051089e-06,
1157
+ "loss": 0.7065,
1158
+ "step": 1640
1159
+ },
1160
+ {
1161
+ "epoch": 0.6353484790142472,
1162
+ "grad_norm": 0.17013967037200928,
1163
+ "learning_rate": 3.1060435167748826e-06,
1164
+ "loss": 0.7218,
1165
+ "step": 1650
1166
+ },
1167
+ {
1168
+ "epoch": 0.6391990758567578,
1169
+ "grad_norm": 0.158955380320549,
1170
+ "learning_rate": 3.048481152564253e-06,
1171
+ "loss": 0.6997,
1172
+ "step": 1660
1173
+ },
1174
+ {
1175
+ "epoch": 0.6430496726992684,
1176
+ "grad_norm": 0.1767643541097641,
1177
+ "learning_rate": 2.9912223245804017e-06,
1178
+ "loss": 0.7174,
1179
+ "step": 1670
1180
+ },
1181
+ {
1182
+ "epoch": 0.6469002695417789,
1183
+ "grad_norm": 0.17139631509780884,
1184
+ "learning_rate": 2.934275938772981e-06,
1185
+ "loss": 0.7072,
1186
+ "step": 1680
1187
+ },
1188
+ {
1189
+ "epoch": 0.6507508663842896,
1190
+ "grad_norm": 0.17709286510944366,
1191
+ "learning_rate": 2.8776508524948676e-06,
1192
+ "loss": 0.7031,
1193
+ "step": 1690
1194
+ },
1195
+ {
1196
+ "epoch": 0.6546014632268001,
1197
+ "grad_norm": 0.1649220585823059,
1198
+ "learning_rate": 2.821355873124503e-06,
1199
+ "loss": 0.7044,
1200
+ "step": 1700
1201
+ },
1202
+ {
1203
+ "epoch": 0.6584520600693108,
1204
+ "grad_norm": 0.1720547080039978,
1205
+ "learning_rate": 2.7653997566960144e-06,
1206
+ "loss": 0.7132,
1207
+ "step": 1710
1208
+ },
1209
+ {
1210
+ "epoch": 0.6623026569118213,
1211
+ "grad_norm": 0.18356113135814667,
1212
+ "learning_rate": 2.709791206537309e-06,
1213
+ "loss": 0.7059,
1214
+ "step": 1720
1215
+ },
1216
+ {
1217
+ "epoch": 0.6661532537543319,
1218
+ "grad_norm": 0.1893559992313385,
1219
+ "learning_rate": 2.6545388719163747e-06,
1220
+ "loss": 0.7076,
1221
+ "step": 1730
1222
+ },
1223
+ {
1224
+ "epoch": 0.6700038505968425,
1225
+ "grad_norm": 0.16616623103618622,
1226
+ "learning_rate": 2.599651346695979e-06,
1227
+ "loss": 0.7174,
1228
+ "step": 1740
1229
+ },
1230
+ {
1231
+ "epoch": 0.6738544474393531,
1232
+ "grad_norm": 0.1615504026412964,
1233
+ "learning_rate": 2.545137167997007e-06,
1234
+ "loss": 0.7146,
1235
+ "step": 1750
1236
+ },
1237
+ {
1238
+ "epoch": 0.6777050442818637,
1239
+ "grad_norm": 0.16182774305343628,
1240
+ "learning_rate": 2.4910048148705972e-06,
1241
+ "loss": 0.704,
1242
+ "step": 1760
1243
+ },
1244
+ {
1245
+ "epoch": 0.6815556411243743,
1246
+ "grad_norm": 0.17074298858642578,
1247
+ "learning_rate": 2.4372627069793327e-06,
1248
+ "loss": 0.7409,
1249
+ "step": 1770
1250
+ },
1251
+ {
1252
+ "epoch": 0.6854062379668848,
1253
+ "grad_norm": 0.163547083735466,
1254
+ "learning_rate": 2.3839192032876486e-06,
1255
+ "loss": 0.7165,
1256
+ "step": 1780
1257
+ },
1258
+ {
1259
+ "epoch": 0.6892568348093955,
1260
+ "grad_norm": 0.171365886926651,
1261
+ "learning_rate": 2.3309826007617047e-06,
1262
+ "loss": 0.7325,
1263
+ "step": 1790
1264
+ },
1265
+ {
1266
+ "epoch": 0.693107431651906,
1267
+ "grad_norm": 0.16179928183555603,
1268
+ "learning_rate": 2.278461133078881e-06,
1269
+ "loss": 0.7007,
1270
+ "step": 1800
1271
+ },
1272
+ {
1273
+ "epoch": 0.6969580284944167,
1274
+ "grad_norm": 0.16905726492404938,
1275
+ "learning_rate": 2.2263629693471258e-06,
1276
+ "loss": 0.6943,
1277
+ "step": 1810
1278
+ },
1279
+ {
1280
+ "epoch": 0.7008086253369272,
1281
+ "grad_norm": 0.16048197448253632,
1282
+ "learning_rate": 2.1746962128343414e-06,
1283
+ "loss": 0.7162,
1284
+ "step": 1820
1285
+ },
1286
+ {
1287
+ "epoch": 0.7046592221794378,
1288
+ "grad_norm": 0.17191454768180847,
1289
+ "learning_rate": 2.1234688997080183e-06,
1290
+ "loss": 0.7011,
1291
+ "step": 1830
1292
+ },
1293
+ {
1294
+ "epoch": 0.7085098190219484,
1295
+ "grad_norm": 0.16227976977825165,
1296
+ "learning_rate": 2.0726889977853053e-06,
1297
+ "loss": 0.7068,
1298
+ "step": 1840
1299
+ },
1300
+ {
1301
+ "epoch": 0.712360415864459,
1302
+ "grad_norm": 0.16832232475280762,
1303
+ "learning_rate": 2.022364405293703e-06,
1304
+ "loss": 0.7213,
1305
+ "step": 1850
1306
+ },
1307
+ {
1308
+ "epoch": 0.7162110127069696,
1309
+ "grad_norm": 0.16463162004947662,
1310
+ "learning_rate": 1.9725029496425814e-06,
1311
+ "loss": 0.7191,
1312
+ "step": 1860
1313
+ },
1314
+ {
1315
+ "epoch": 0.7200616095494802,
1316
+ "grad_norm": 0.16694504022598267,
1317
+ "learning_rate": 1.923112386205727e-06,
1318
+ "loss": 0.71,
1319
+ "step": 1870
1320
+ },
1321
+ {
1322
+ "epoch": 0.7239122063919907,
1323
+ "grad_norm": 0.16297578811645508,
1324
+ "learning_rate": 1.874200397115079e-06,
1325
+ "loss": 0.7206,
1326
+ "step": 1880
1327
+ },
1328
+ {
1329
+ "epoch": 0.7277628032345014,
1330
+ "grad_norm": 0.1687304675579071,
1331
+ "learning_rate": 1.825774590065857e-06,
1332
+ "loss": 0.7053,
1333
+ "step": 1890
1334
+ },
1335
+ {
1336
+ "epoch": 0.7316134000770119,
1337
+ "grad_norm": 0.1779334396123886,
1338
+ "learning_rate": 1.7778424971332813e-06,
1339
+ "loss": 0.7164,
1340
+ "step": 1900
1341
+ },
1342
+ {
1343
+ "epoch": 0.7354639969195226,
1344
+ "grad_norm": 0.1609511375427246,
1345
+ "learning_rate": 1.7304115736010407e-06,
1346
+ "loss": 0.7117,
1347
+ "step": 1910
1348
+ },
1349
+ {
1350
+ "epoch": 0.7393145937620331,
1351
+ "grad_norm": 0.16286328434944153,
1352
+ "learning_rate": 1.6834891968017108e-06,
1353
+ "loss": 0.7039,
1354
+ "step": 1920
1355
+ },
1356
+ {
1357
+ "epoch": 0.7431651906045437,
1358
+ "grad_norm": 0.1737114042043686,
1359
+ "learning_rate": 1.6370826649692884e-06,
1360
+ "loss": 0.7189,
1361
+ "step": 1930
1362
+ },
1363
+ {
1364
+ "epoch": 0.7470157874470543,
1365
+ "grad_norm": 0.16243602335453033,
1366
+ "learning_rate": 1.5911991961040463e-06,
1367
+ "loss": 0.7222,
1368
+ "step": 1940
1369
+ },
1370
+ {
1371
+ "epoch": 0.7508663842895649,
1372
+ "grad_norm": 0.17490696907043457,
1373
+ "learning_rate": 1.545845926849851e-06,
1374
+ "loss": 0.7054,
1375
+ "step": 1950
1376
+ },
1377
+ {
1378
+ "epoch": 0.7547169811320755,
1379
+ "grad_norm": 0.178118035197258,
1380
+ "learning_rate": 1.5010299113841397e-06,
1381
+ "loss": 0.7293,
1382
+ "step": 1960
1383
+ },
1384
+ {
1385
+ "epoch": 0.7585675779745861,
1386
+ "grad_norm": 0.16985875368118286,
1387
+ "learning_rate": 1.4567581203207238e-06,
1388
+ "loss": 0.7189,
1389
+ "step": 1970
1390
+ },
1391
+ {
1392
+ "epoch": 0.7624181748170966,
1393
+ "grad_norm": 0.18562936782836914,
1394
+ "learning_rate": 1.413037439625593e-06,
1395
+ "loss": 0.7191,
1396
+ "step": 1980
1397
+ },
1398
+ {
1399
+ "epoch": 0.7662687716596073,
1400
+ "grad_norm": 0.18047331273555756,
1401
+ "learning_rate": 1.3698746695458847e-06,
1402
+ "loss": 0.708,
1403
+ "step": 1990
1404
+ },
1405
+ {
1406
+ "epoch": 0.7701193685021178,
1407
+ "grad_norm": 0.16733287274837494,
1408
+ "learning_rate": 1.3272765235521745e-06,
1409
+ "loss": 0.7088,
1410
+ "step": 2000
1411
+ },
1412
+ {
1413
+ "epoch": 0.7739699653446284,
1414
+ "grad_norm": 0.16080398857593536,
1415
+ "learning_rate": 1.2852496272942877e-06,
1416
+ "loss": 0.7164,
1417
+ "step": 2010
1418
+ },
1419
+ {
1420
+ "epoch": 0.777820562187139,
1421
+ "grad_norm": 0.15744544565677643,
1422
+ "learning_rate": 1.2438005175707396e-06,
1423
+ "loss": 0.6955,
1424
+ "step": 2020
1425
+ },
1426
+ {
1427
+ "epoch": 0.7816711590296496,
1428
+ "grad_norm": 0.16594071686267853,
1429
+ "learning_rate": 1.2029356413120297e-06,
1430
+ "loss": 0.69,
1431
+ "step": 2030
1432
+ },
1433
+ {
1434
+ "epoch": 0.7855217558721602,
1435
+ "grad_norm": 0.16892139613628387,
1436
+ "learning_rate": 1.1626613545778786e-06,
1437
+ "loss": 0.706,
1438
+ "step": 2040
1439
+ },
1440
+ {
1441
+ "epoch": 0.7893723527146708,
1442
+ "grad_norm": 0.1616152673959732,
1443
+ "learning_rate": 1.1229839215686294e-06,
1444
+ "loss": 0.7088,
1445
+ "step": 2050
1446
+ },
1447
+ {
1448
+ "epoch": 0.7932229495571813,
1449
+ "grad_norm": 0.15014874935150146,
1450
+ "learning_rate": 1.0839095136509188e-06,
1451
+ "loss": 0.6931,
1452
+ "step": 2060
1453
+ },
1454
+ {
1455
+ "epoch": 0.797073546399692,
1456
+ "grad_norm": 0.1673061102628708,
1457
+ "learning_rate": 1.045444208397791e-06,
1458
+ "loss": 0.7085,
1459
+ "step": 2070
1460
+ },
1461
+ {
1462
+ "epoch": 0.8009241432422025,
1463
+ "grad_norm": 0.15214619040489197,
1464
+ "learning_rate": 1.007593988643402e-06,
1465
+ "loss": 0.6864,
1466
+ "step": 2080
1467
+ },
1468
+ {
1469
+ "epoch": 0.8047747400847132,
1470
+ "grad_norm": 0.15270189940929413,
1471
+ "learning_rate": 9.703647415524613e-07,
1472
+ "loss": 0.7142,
1473
+ "step": 2090
1474
+ },
1475
+ {
1476
+ "epoch": 0.8086253369272237,
1477
+ "grad_norm": 0.15630601346492767,
1478
+ "learning_rate": 9.337622577045485e-07,
1479
+ "loss": 0.7033,
1480
+ "step": 2100
1481
+ },
1482
+ {
1483
+ "epoch": 0.8124759337697343,
1484
+ "grad_norm": 0.16377653181552887,
1485
+ "learning_rate": 8.977922301934588e-07,
1486
+ "loss": 0.7112,
1487
+ "step": 2110
1488
+ },
1489
+ {
1490
+ "epoch": 0.8163265306122449,
1491
+ "grad_norm": 0.1806977540254593,
1492
+ "learning_rate": 8.624602537417009e-07,
1493
+ "loss": 0.7259,
1494
+ "step": 2120
1495
+ },
1496
+ {
1497
+ "epoch": 0.8201771274547555,
1498
+ "grad_norm": 0.165560781955719,
1499
+ "learning_rate": 8.277718238303123e-07,
1500
+ "loss": 0.7061,
1501
+ "step": 2130
1502
+ },
1503
+ {
1504
+ "epoch": 0.8240277242972661,
1505
+ "grad_norm": 0.17147821187973022,
1506
+ "learning_rate": 7.937323358440935e-07,
1507
+ "loss": 0.7091,
1508
+ "step": 2140
1509
+ },
1510
+ {
1511
+ "epoch": 0.8278783211397767,
1512
+ "grad_norm": 0.16392134130001068,
1513
+ "learning_rate": 7.603470842324223e-07,
1514
+ "loss": 0.6954,
1515
+ "step": 2150
1516
+ },
1517
+ {
1518
+ "epoch": 0.8317289179822872,
1519
+ "grad_norm": 0.1647264063358307,
1520
+ "learning_rate": 7.276212616857575e-07,
1521
+ "loss": 0.7184,
1522
+ "step": 2160
1523
+ },
1524
+ {
1525
+ "epoch": 0.8355795148247979,
1526
+ "grad_norm": 0.1683371663093567,
1527
+ "learning_rate": 6.955599583279837e-07,
1528
+ "loss": 0.711,
1529
+ "step": 2170
1530
+ },
1531
+ {
1532
+ "epoch": 0.8394301116673084,
1533
+ "grad_norm": 0.1631588488817215,
1534
+ "learning_rate": 6.641681609246981e-07,
1535
+ "loss": 0.7081,
1536
+ "step": 2180
1537
+ },
1538
+ {
1539
+ "epoch": 0.8432807085098191,
1540
+ "grad_norm": 0.17289875447750092,
1541
+ "learning_rate": 6.334507521075711e-07,
1542
+ "loss": 0.7061,
1543
+ "step": 2190
1544
+ },
1545
+ {
1546
+ "epoch": 0.8471313053523296,
1547
+ "grad_norm": 0.1706165373325348,
1548
+ "learning_rate": 6.034125096149179e-07,
1549
+ "loss": 0.7101,
1550
+ "step": 2200
1551
+ },
1552
+ {
1553
+ "epoch": 0.8509819021948402,
1554
+ "grad_norm": 0.1599239557981491,
1555
+ "learning_rate": 5.740581055485722e-07,
1556
+ "loss": 0.6809,
1557
+ "step": 2210
1558
+ },
1559
+ {
1560
+ "epoch": 0.8548324990373508,
1561
+ "grad_norm": 0.15503591299057007,
1562
+ "learning_rate": 5.453921056471961e-07,
1563
+ "loss": 0.7136,
1564
+ "step": 2220
1565
+ },
1566
+ {
1567
+ "epoch": 0.8586830958798614,
1568
+ "grad_norm": 0.17559321224689484,
1569
+ "learning_rate": 5.174189685761288e-07,
1570
+ "loss": 0.7071,
1571
+ "step": 2230
1572
+ },
1573
+ {
1574
+ "epoch": 0.862533692722372,
1575
+ "grad_norm": 0.16481715440750122,
1576
+ "learning_rate": 4.901430452338979e-07,
1577
+ "loss": 0.7202,
1578
+ "step": 2240
1579
+ },
1580
+ {
1581
+ "epoch": 0.8663842895648826,
1582
+ "grad_norm": 0.16638413071632385,
1583
+ "learning_rate": 4.6356857807548286e-07,
1584
+ "loss": 0.7063,
1585
+ "step": 2250
1586
+ },
1587
+ {
1588
+ "epoch": 0.8702348864073931,
1589
+ "grad_norm": 0.19296357035636902,
1590
+ "learning_rate": 4.376997004524541e-07,
1591
+ "loss": 0.6898,
1592
+ "step": 2260
1593
+ },
1594
+ {
1595
+ "epoch": 0.8740854832499038,
1596
+ "grad_norm": 0.16456134617328644,
1597
+ "learning_rate": 4.12540435970073e-07,
1598
+ "loss": 0.7062,
1599
+ "step": 2270
1600
+ },
1601
+ {
1602
+ "epoch": 0.8779360800924143,
1603
+ "grad_norm": 0.15853694081306458,
1604
+ "learning_rate": 3.880946978614703e-07,
1605
+ "loss": 0.7162,
1606
+ "step": 2280
1607
+ },
1608
+ {
1609
+ "epoch": 0.881786676934925,
1610
+ "grad_norm": 0.16481667757034302,
1611
+ "learning_rate": 3.643662883789878e-07,
1612
+ "loss": 0.7122,
1613
+ "step": 2290
1614
+ },
1615
+ {
1616
+ "epoch": 0.8856372737774355,
1617
+ "grad_norm": 0.16149882972240448,
1618
+ "learning_rate": 3.41358898202781e-07,
1619
+ "loss": 0.6968,
1620
+ "step": 2300
1621
+ },
1622
+ {
1623
+ "epoch": 0.889487870619946,
1624
+ "grad_norm": 0.15941961109638214,
1625
+ "learning_rate": 3.190761058667741e-07,
1626
+ "loss": 0.7084,
1627
+ "step": 2310
1628
+ },
1629
+ {
1630
+ "epoch": 0.8933384674624567,
1631
+ "grad_norm": 0.15761925280094147,
1632
+ "learning_rate": 2.9752137720206463e-07,
1633
+ "loss": 0.7093,
1634
+ "step": 2320
1635
+ },
1636
+ {
1637
+ "epoch": 0.8971890643049673,
1638
+ "grad_norm": 0.1690497100353241,
1639
+ "learning_rate": 2.7669806479785067e-07,
1640
+ "loss": 0.7206,
1641
+ "step": 2330
1642
+ },
1643
+ {
1644
+ "epoch": 0.9010396611474779,
1645
+ "grad_norm": 0.1605324149131775,
1646
+ "learning_rate": 2.566094074799763e-07,
1647
+ "loss": 0.7103,
1648
+ "step": 2340
1649
+ },
1650
+ {
1651
+ "epoch": 0.9048902579899885,
1652
+ "grad_norm": 0.1666429340839386,
1653
+ "learning_rate": 2.3725852980716557e-07,
1654
+ "loss": 0.6949,
1655
+ "step": 2350
1656
+ },
1657
+ {
1658
+ "epoch": 0.908740854832499,
1659
+ "grad_norm": 0.16167500615119934,
1660
+ "learning_rate": 2.1864844158503983e-07,
1661
+ "loss": 0.7076,
1662
+ "step": 2360
1663
+ },
1664
+ {
1665
+ "epoch": 0.9125914516750097,
1666
+ "grad_norm": 0.18248993158340454,
1667
+ "learning_rate": 2.007820373979702e-07,
1668
+ "loss": 0.7,
1669
+ "step": 2370
1670
+ },
1671
+ {
1672
+ "epoch": 0.9164420485175202,
1673
+ "grad_norm": 0.1618565022945404,
1674
+ "learning_rate": 1.8366209615886532e-07,
1675
+ "loss": 0.7132,
1676
+ "step": 2380
1677
+ },
1678
+ {
1679
+ "epoch": 0.9202926453600309,
1680
+ "grad_norm": 0.16353757679462433,
1681
+ "learning_rate": 1.6729128067693613e-07,
1682
+ "loss": 0.7259,
1683
+ "step": 2390
1684
+ },
1685
+ {
1686
+ "epoch": 0.9241432422025414,
1687
+ "grad_norm": 0.18218517303466797,
1688
+ "learning_rate": 1.5167213724353426e-07,
1689
+ "loss": 0.7192,
1690
+ "step": 2400
1691
+ },
1692
+ {
1693
+ "epoch": 0.927993839045052,
1694
+ "grad_norm": 0.159896582365036,
1695
+ "learning_rate": 1.3680709523610058e-07,
1696
+ "loss": 0.6955,
1697
+ "step": 2410
1698
+ },
1699
+ {
1700
+ "epoch": 0.9318444358875626,
1701
+ "grad_norm": 0.15993618965148926,
1702
+ "learning_rate": 1.2269846674030418e-07,
1703
+ "loss": 0.7247,
1704
+ "step": 2420
1705
+ },
1706
+ {
1707
+ "epoch": 0.9356950327300732,
1708
+ "grad_norm": 0.16899316012859344,
1709
+ "learning_rate": 1.0934844619042595e-07,
1710
+ "loss": 0.7196,
1711
+ "step": 2430
1712
+ },
1713
+ {
1714
+ "epoch": 0.9395456295725837,
1715
+ "grad_norm": 0.16290944814682007,
1716
+ "learning_rate": 9.675911002803961e-08,
1717
+ "loss": 0.7064,
1718
+ "step": 2440
1719
+ },
1720
+ {
1721
+ "epoch": 0.9433962264150944,
1722
+ "grad_norm": 0.16544076800346375,
1723
+ "learning_rate": 8.493241637904381e-08,
1724
+ "loss": 0.6956,
1725
+ "step": 2450
1726
+ },
1727
+ {
1728
+ "epoch": 0.9472468232576049,
1729
+ "grad_norm": 0.1915256679058075,
1730
+ "learning_rate": 7.387020474909689e-08,
1731
+ "loss": 0.7233,
1732
+ "step": 2460
1733
+ },
1734
+ {
1735
+ "epoch": 0.9510974201001156,
1736
+ "grad_norm": 0.17133891582489014,
1737
+ "learning_rate": 6.357419573750856e-08,
1738
+ "loss": 0.7187,
1739
+ "step": 2470
1740
+ },
1741
+ {
1742
+ "epoch": 0.9549480169426261,
1743
+ "grad_norm": 0.16470745205879211,
1744
+ "learning_rate": 5.4045990769613365e-08,
1745
+ "loss": 0.7047,
1746
+ "step": 2480
1747
+ },
1748
+ {
1749
+ "epoch": 0.9587986137851366,
1750
+ "grad_norm": 0.16229769587516785,
1751
+ "learning_rate": 4.5287071847694434e-08,
1752
+ "loss": 0.7054,
1753
+ "step": 2490
1754
+ },
1755
+ {
1756
+ "epoch": 0.9626492106276473,
1757
+ "grad_norm": 0.16431164741516113,
1758
+ "learning_rate": 3.7298801320467817e-08,
1759
+ "loss": 0.7097,
1760
+ "step": 2500
1761
+ },
1762
+ {
1763
+ "epoch": 0.9664998074701578,
1764
+ "grad_norm": 0.1572009027004242,
1765
+ "learning_rate": 3.0082421671192576e-08,
1766
+ "loss": 0.7071,
1767
+ "step": 2510
1768
+ },
1769
+ {
1770
+ "epoch": 0.9703504043126685,
1771
+ "grad_norm": 0.1646891087293625,
1772
+ "learning_rate": 2.3639055324410355e-08,
1773
+ "loss": 0.7234,
1774
+ "step": 2520
1775
+ },
1776
+ {
1777
+ "epoch": 0.974201001155179,
1778
+ "grad_norm": 0.1656169891357422,
1779
+ "learning_rate": 1.7969704471369477e-08,
1780
+ "loss": 0.7019,
1781
+ "step": 2530
1782
+ },
1783
+ {
1784
+ "epoch": 0.9780515979976896,
1785
+ "grad_norm": 0.15622319281101227,
1786
+ "learning_rate": 1.3075250914142413e-08,
1787
+ "loss": 0.7045,
1788
+ "step": 2540
1789
+ },
1790
+ {
1791
+ "epoch": 0.9819021948402002,
1792
+ "grad_norm": 0.1527712643146515,
1793
+ "learning_rate": 8.956455928476049e-09,
1794
+ "loss": 0.7181,
1795
+ "step": 2550
1796
+ },
1797
+ {
1798
+ "epoch": 0.9857527916827108,
1799
+ "grad_norm": 0.16184458136558533,
1800
+ "learning_rate": 5.6139601453786275e-09,
1801
+ "loss": 0.7018,
1802
+ "step": 2560
1803
+ },
1804
+ {
1805
+ "epoch": 0.9896033885252215,
1806
+ "grad_norm": 0.1573844850063324,
1807
+ "learning_rate": 3.0482834514805602e-09,
1808
+ "loss": 0.7,
1809
+ "step": 2570
1810
+ },
1811
+ {
1812
+ "epoch": 0.993453985367732,
1813
+ "grad_norm": 0.1611800342798233,
1814
+ "learning_rate": 1.2598249081713366e-09,
1815
+ "loss": 0.6882,
1816
+ "step": 2580
1817
+ },
1818
+ {
1819
+ "epoch": 0.9973045822102425,
1820
+ "grad_norm": 0.1730285882949829,
1821
+ "learning_rate": 2.488626895302826e-10,
1822
+ "loss": 0.7238,
1823
+ "step": 2590
1824
+ },
1825
+ {
1826
+ "epoch": 1.0,
1827
+ "step": 2597,
1828
+ "total_flos": 7.169401182782226e+18,
1829
+ "train_loss": 0.7259787457421691,
1830
+ "train_runtime": 14923.7613,
1831
+ "train_samples_per_second": 44.553,
1832
+ "train_steps_per_second": 0.174
1833
+ }
1834
+ ],
1835
+ "logging_steps": 10,
1836
+ "max_steps": 2597,
1837
+ "num_input_tokens_seen": 584317136,
1838
+ "num_train_epochs": 1,
1839
+ "save_steps": 500,
1840
+ "stateful_callbacks": {
1841
+ "TrainerControl": {
1842
+ "args": {
1843
+ "should_epoch_stop": false,
1844
+ "should_evaluate": false,
1845
+ "should_log": false,
1846
+ "should_save": true,
1847
+ "should_training_stop": true
1848
+ },
1849
+ "attributes": {}
1850
+ }
1851
+ },
1852
+ "total_flos": 7.169401182782226e+18,
1853
+ "train_batch_size": 16,
1854
+ "trial_name": null,
1855
+ "trial_params": null
1856
+ }
continuous/smollm2/siglip2_sm14_224/finetune/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52c5996de4ba26c371cb7fee49c8a60c38e22145501a4c6c01e977285499729e
3
+ size 6264
continuous/smollm2/siglip2_sm14_224/finetune/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm14_224/pretrain/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": null,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-14-SigLIP2",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm14_224.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm14_224/pretrain/mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:076f5c3b23682ee9e86ddf8c68c785978852dee91ba78440818468952f17077a
3
+ size 26232888
continuous/smollm2/siglip2_sm14_224/pretrain/trainer_state.json ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 545,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.01834862385321101,
14
+ "grad_norm": 1.4414466619491577,
15
+ "learning_rate": 0.00032142857142857147,
16
+ "loss": 3.6188,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.03669724770642202,
21
+ "grad_norm": 0.11616519838571548,
22
+ "learning_rate": 0.0006785714285714287,
23
+ "loss": 2.8714,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.05504587155963303,
28
+ "grad_norm": 0.08332502096891403,
29
+ "learning_rate": 0.0009999907688176173,
30
+ "loss": 2.7751,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.07339449541284404,
35
+ "grad_norm": 0.10312360525131226,
36
+ "learning_rate": 0.0009988834393115766,
37
+ "loss": 2.7254,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.09174311926605505,
42
+ "grad_norm": 0.10000826418399811,
43
+ "learning_rate": 0.0009959345572940772,
44
+ "loss": 2.6729,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.11009174311926606,
49
+ "grad_norm": 0.09005548059940338,
50
+ "learning_rate": 0.0009911550081155983,
51
+ "loss": 2.5669,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.12844036697247707,
56
+ "grad_norm": 0.11410067230463028,
57
+ "learning_rate": 0.0009845624347566433,
58
+ "loss": 2.4833,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.14678899082568808,
63
+ "grad_norm": 0.12443874776363373,
64
+ "learning_rate": 0.0009761811727013547,
65
+ "loss": 2.4152,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.1651376146788991,
70
+ "grad_norm": 0.09679734706878662,
71
+ "learning_rate": 0.0009660421601067666,
72
+ "loss": 2.3646,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.1834862385321101,
77
+ "grad_norm": 0.1259429156780243,
78
+ "learning_rate": 0.0009541828235992926,
79
+ "loss": 2.3321,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.2018348623853211,
84
+ "grad_norm": 0.15712870657444,
85
+ "learning_rate": 0.0009406469401200151,
86
+ "loss": 2.2898,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.22018348623853212,
91
+ "grad_norm": 0.1036045029759407,
92
+ "learning_rate": 0.0009254844753287492,
93
+ "loss": 2.2657,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.23853211009174313,
98
+ "grad_norm": 0.16349664330482483,
99
+ "learning_rate": 0.0009087513991633924,
100
+ "loss": 2.2289,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.25688073394495414,
105
+ "grad_norm": 0.12344957143068314,
106
+ "learning_rate": 0.0008905094792353917,
107
+ "loss": 2.1984,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.27522935779816515,
112
+ "grad_norm": 0.12201276421546936,
113
+ "learning_rate": 0.0008708260528239789,
114
+ "loss": 2.1734,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.29357798165137616,
119
+ "grad_norm": 0.1025775671005249,
120
+ "learning_rate": 0.0008497737783108237,
121
+ "loss": 2.1343,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.3119266055045872,
126
+ "grad_norm": 0.0995042473077774,
127
+ "learning_rate": 0.0008274303669726426,
128
+ "loss": 2.1164,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.3302752293577982,
133
+ "grad_norm": 0.10612590610980988,
134
+ "learning_rate": 0.0008038782961218137,
135
+ "loss": 2.1028,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.3486238532110092,
140
+ "grad_norm": 0.10388248413801193,
141
+ "learning_rate": 0.000779204504653894,
142
+ "loss": 2.0915,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.3669724770642202,
147
+ "grad_norm": 0.10103309899568558,
148
+ "learning_rate": 0.0007535000721258791,
149
+ "loss": 2.0659,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.3853211009174312,
154
+ "grad_norm": 0.10418111085891724,
155
+ "learning_rate": 0.00072685988254984,
156
+ "loss": 2.0588,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.4036697247706422,
161
+ "grad_norm": 0.10390561074018478,
162
+ "learning_rate": 0.0006993822741429906,
163
+ "loss": 2.0665,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.42201834862385323,
168
+ "grad_norm": 0.09681087732315063,
169
+ "learning_rate": 0.0006711686763270818,
170
+ "loss": 2.0525,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.44036697247706424,
175
+ "grad_norm": 0.09364120662212372,
176
+ "learning_rate": 0.0006423232353170831,
177
+ "loss": 2.0462,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.45871559633027525,
182
+ "grad_norm": 0.11022620648145676,
183
+ "learning_rate": 0.0006129524296812335,
184
+ "loss": 2.0483,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.47706422018348627,
189
+ "grad_norm": 0.09428097307682037,
190
+ "learning_rate": 0.0005831646772915651,
191
+ "loss": 2.0509,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.4954128440366973,
196
+ "grad_norm": 0.09154687076807022,
197
+ "learning_rate": 0.0005530699351157809,
198
+ "loss": 2.028,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.5137614678899083,
203
+ "grad_norm": 0.09811349213123322,
204
+ "learning_rate": 0.0005227792933277943,
205
+ "loss": 2.0317,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.5321100917431193,
210
+ "grad_norm": 0.09124897420406342,
211
+ "learning_rate": 0.0004924045652352092,
212
+ "loss": 2.022,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.5504587155963303,
217
+ "grad_norm": 0.09375732392072678,
218
+ "learning_rate": 0.0004620578745374633,
219
+ "loss": 2.0211,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.5688073394495413,
224
+ "grad_norm": 0.12656466662883759,
225
+ "learning_rate": 0.0004318512414382059,
226
+ "loss": 2.0259,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.5871559633027523,
231
+ "grad_norm": 0.08702733367681503,
232
+ "learning_rate": 0.0004018961691397148,
233
+ "loss": 2.0262,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.6055045871559633,
238
+ "grad_norm": 0.11043024808168411,
239
+ "learning_rate": 0.0003723032322457458,
240
+ "loss": 2.0133,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.6238532110091743,
245
+ "grad_norm": 0.1079900860786438,
246
+ "learning_rate": 0.0003431816685921625,
247
+ "loss": 2.0144,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.6422018348623854,
252
+ "grad_norm": 0.08872897177934647,
253
+ "learning_rate": 0.0003146389760120416,
254
+ "loss": 2.0108,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.6605504587155964,
259
+ "grad_norm": 0.09089556336402893,
260
+ "learning_rate": 0.00028678051552373485,
261
+ "loss": 2.0117,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.6788990825688074,
266
+ "grad_norm": 0.08325676620006561,
267
+ "learning_rate": 0.0002597091224066581,
268
+ "loss": 2.0047,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.6972477064220184,
273
+ "grad_norm": 0.08584984391927719,
274
+ "learning_rate": 0.0002335247266004629,
275
+ "loss": 2.0124,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.7155963302752294,
280
+ "grad_norm": 0.08932308107614517,
281
+ "learning_rate": 0.000208323983828828,
282
+ "loss": 2.0018,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.7339449541284404,
287
+ "grad_norm": 0.08564908057451248,
288
+ "learning_rate": 0.0001841999188095224,
289
+ "loss": 1.9957,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.7522935779816514,
294
+ "grad_norm": 0.09345333278179169,
295
+ "learning_rate": 0.00016124158186777676,
296
+ "loss": 2.0044,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.7706422018348624,
301
+ "grad_norm": 0.09944067150354385,
302
+ "learning_rate": 0.0001395337202205211,
303
+ "loss": 1.9975,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.7889908256880734,
308
+ "grad_norm": 0.08176930993795395,
309
+ "learning_rate": 0.00011915646514489292,
310
+ "loss": 1.9957,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.8073394495412844,
315
+ "grad_norm": 0.09043989330530167,
316
+ "learning_rate": 0.00010018503618578818,
317
+ "loss": 2.0013,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.8256880733944955,
322
+ "grad_norm": 0.08676031231880188,
323
+ "learning_rate": 8.268946349432583e-05,
324
+ "loss": 1.9949,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.8440366972477065,
329
+ "grad_norm": 0.08266797661781311,
330
+ "learning_rate": 6.673432932217117e-05,
331
+ "loss": 1.9898,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.8623853211009175,
336
+ "grad_norm": 0.08629272133111954,
337
+ "learning_rate": 5.2378529625954684e-05,
338
+ "loss": 1.9836,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.8807339449541285,
343
+ "grad_norm": 0.08169011026620865,
344
+ "learning_rate": 3.9675056661785556e-05,
345
+ "loss": 1.9861,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.8990825688073395,
350
+ "grad_norm": 0.08260252326726913,
351
+ "learning_rate": 2.8670803372379538e-05,
352
+ "loss": 1.9918,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.9174311926605505,
357
+ "grad_norm": 0.08261581510305405,
358
+ "learning_rate": 1.9406390288876586e-05,
359
+ "loss": 1.9963,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.9357798165137615,
364
+ "grad_norm": 0.07998602837324142,
365
+ "learning_rate": 1.191601558631461e-05,
366
+ "loss": 1.9881,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.9541284403669725,
371
+ "grad_norm": 0.0814804881811142,
372
+ "learning_rate": 6.2273288462559305e-06,
373
+ "loss": 1.9836,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.9724770642201835,
378
+ "grad_norm": 0.08208692073822021,
379
+ "learning_rate": 2.361328992552314e-06,
380
+ "loss": 1.9875,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.9908256880733946,
385
+ "grad_norm": 0.08227529376745224,
386
+ "learning_rate": 3.322867770044202e-07,
387
+ "loss": 1.9867,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 1.0,
392
+ "step": 545,
393
+ "total_flos": 4.252457314948219e+17,
394
+ "train_loss": 2.163891466823193,
395
+ "train_runtime": 1688.2311,
396
+ "train_samples_per_second": 330.599,
397
+ "train_steps_per_second": 0.323
398
+ }
399
+ ],
400
+ "logging_steps": 10,
401
+ "max_steps": 545,
402
+ "num_input_tokens_seen": 34658176,
403
+ "num_train_epochs": 1,
404
+ "save_steps": 500,
405
+ "stateful_callbacks": {
406
+ "TrainerControl": {
407
+ "args": {
408
+ "should_epoch_stop": false,
409
+ "should_evaluate": false,
410
+ "should_log": false,
411
+ "should_save": false,
412
+ "should_training_stop": false
413
+ },
414
+ "attributes": {}
415
+ }
416
+ },
417
+ "total_flos": 4.252457314948219e+17,
418
+ "train_batch_size": 64,
419
+ "trial_name": null,
420
+ "trial_params": null
421
+ }
continuous/smollm2/siglip2_sm14_384/finetune/chat_template.jinja ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
2
+ You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
3
+ ' }}{% endif %}{{'<|im_start|>' + message['role'] + '
4
+ ' + message['content'] + '<|im_end|>' + '
5
+ '}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
6
+ ' }}{% endif %}
continuous/smollm2/siglip2_sm14_384/finetune/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlavaLlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": 0.0001,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-14-SigLIP2-378",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm14_384.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm14_384/finetune/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": [
5
+ 2
6
+ ],
7
+ "pad_token_id": 0,
8
+ "transformers_version": "4.57.6"
9
+ }
continuous/smollm2/siglip2_sm14_384/finetune/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm14_384/finetune/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec7e80c6c669ca7b92665167d35b13d10b6fe482dea77570794ef30f9507de96
3
+ size 4506833528
continuous/smollm2/siglip2_sm14_384/finetune/special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|im_start|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|im_end|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|endoftext|>",
21
+ "unk_token": {
22
+ "content": "<|endoftext|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
continuous/smollm2/siglip2_sm14_384/finetune/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm14_384/finetune/tokenizer_config.json ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<repo_name>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "4": {
37
+ "content": "<reponame>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "5": {
45
+ "content": "<file_sep>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "6": {
53
+ "content": "<filename>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "7": {
61
+ "content": "<gh_stars>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "8": {
69
+ "content": "<issue_start>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "9": {
77
+ "content": "<issue_comment>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "10": {
85
+ "content": "<issue_closed>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "11": {
93
+ "content": "<jupyter_start>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "12": {
101
+ "content": "<jupyter_text>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "13": {
109
+ "content": "<jupyter_code>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "14": {
117
+ "content": "<jupyter_output>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": true
123
+ },
124
+ "15": {
125
+ "content": "<jupyter_script>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": true
131
+ },
132
+ "16": {
133
+ "content": "<empty_output>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": true
139
+ }
140
+ },
141
+ "additional_special_tokens": [
142
+ "<|im_start|>",
143
+ "<|im_end|>"
144
+ ],
145
+ "bos_token": "<|im_start|>",
146
+ "clean_up_tokenization_spaces": false,
147
+ "eos_token": "<|im_end|>",
148
+ "extra_special_tokens": {},
149
+ "model_max_length": 8192,
150
+ "pad_token": "<|endoftext|>",
151
+ "padding_side": "right",
152
+ "tokenizer_class": "GPT2Tokenizer",
153
+ "unk_token": "<|endoftext|>",
154
+ "vocab_size": 49152
155
+ }
continuous/smollm2/siglip2_sm14_384/finetune/trainer_state.json ADDED
@@ -0,0 +1,1856 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2597,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.003850596842510589,
14
+ "grad_norm": 3.0950262546539307,
15
+ "learning_rate": 1.153846153846154e-06,
16
+ "loss": 1.2447,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.007701193685021178,
21
+ "grad_norm": 1.9394880533218384,
22
+ "learning_rate": 2.435897435897436e-06,
23
+ "loss": 1.0874,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.011551790527531768,
28
+ "grad_norm": 1.873196005821228,
29
+ "learning_rate": 3.7179487179487184e-06,
30
+ "loss": 0.9658,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.015402387370042356,
35
+ "grad_norm": 1.3697388172149658,
36
+ "learning_rate": 5e-06,
37
+ "loss": 0.9209,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.019252984212552945,
42
+ "grad_norm": 1.3363012075424194,
43
+ "learning_rate": 6.282051282051282e-06,
44
+ "loss": 0.9094,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.023103581055063535,
49
+ "grad_norm": 0.287982702255249,
50
+ "learning_rate": 7.564102564102564e-06,
51
+ "loss": 0.8394,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.026954177897574125,
56
+ "grad_norm": 0.2364620715379715,
57
+ "learning_rate": 8.846153846153847e-06,
58
+ "loss": 0.7926,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.030804774740084712,
63
+ "grad_norm": 0.2519861161708832,
64
+ "learning_rate": 9.999996111488723e-06,
65
+ "loss": 0.8063,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.0346553715825953,
70
+ "grad_norm": 0.21470877528190613,
71
+ "learning_rate": 9.999529497453782e-06,
72
+ "loss": 0.7912,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.03850596842510589,
77
+ "grad_norm": 0.2332296222448349,
78
+ "learning_rate": 9.998285264324193e-06,
79
+ "loss": 0.7575,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.04235656526761648,
84
+ "grad_norm": 0.22654838860034943,
85
+ "learning_rate": 9.996263605626052e-06,
86
+ "loss": 0.7527,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.04620716211012707,
91
+ "grad_norm": 0.2080432027578354,
92
+ "learning_rate": 9.99346483580503e-06,
93
+ "loss": 0.7709,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.05005775895263766,
98
+ "grad_norm": 0.21305350959300995,
99
+ "learning_rate": 9.989889390177467e-06,
100
+ "loss": 0.7444,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.05390835579514825,
105
+ "grad_norm": 0.23992770910263062,
106
+ "learning_rate": 9.985537824862647e-06,
107
+ "loss": 0.7618,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.05775895263765884,
112
+ "grad_norm": 0.2190324068069458,
113
+ "learning_rate": 9.980410816696317e-06,
114
+ "loss": 0.7369,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.061609549480169425,
119
+ "grad_norm": 0.20278184115886688,
120
+ "learning_rate": 9.97450916312541e-06,
121
+ "loss": 0.7451,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.06546014632268002,
126
+ "grad_norm": 0.17835144698619843,
127
+ "learning_rate": 9.967833782084004e-06,
128
+ "loss": 0.7471,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.0693107431651906,
133
+ "grad_norm": 0.21779058873653412,
134
+ "learning_rate": 9.960385711850554e-06,
135
+ "loss": 0.7759,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.07316134000770119,
140
+ "grad_norm": 0.2302885502576828,
141
+ "learning_rate": 9.952166110886394e-06,
142
+ "loss": 0.762,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.07701193685021178,
147
+ "grad_norm": 0.19389261305332184,
148
+ "learning_rate": 9.943176257655567e-06,
149
+ "loss": 0.7603,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.08086253369272237,
154
+ "grad_norm": 0.20007243752479553,
155
+ "learning_rate": 9.933417550425952e-06,
156
+ "loss": 0.7516,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.08471313053523297,
161
+ "grad_norm": 0.19343504309654236,
162
+ "learning_rate": 9.9228915070518e-06,
163
+ "loss": 0.7364,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.08856372737774355,
168
+ "grad_norm": 0.18626917898654938,
169
+ "learning_rate": 9.911599764737632e-06,
170
+ "loss": 0.7339,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.09241432422025414,
175
+ "grad_norm": 0.18712322413921356,
176
+ "learning_rate": 9.899544079783606e-06,
177
+ "loss": 0.7303,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.09626492106276473,
182
+ "grad_norm": 0.20052842795848846,
183
+ "learning_rate": 9.886726327312338e-06,
184
+ "loss": 0.747,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.10011551790527531,
189
+ "grad_norm": 0.18433290719985962,
190
+ "learning_rate": 9.87314850097724e-06,
191
+ "loss": 0.7444,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.1039661147477859,
196
+ "grad_norm": 0.1874879151582718,
197
+ "learning_rate": 9.858812712652452e-06,
198
+ "loss": 0.7276,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.1078167115902965,
203
+ "grad_norm": 0.1703278124332428,
204
+ "learning_rate": 9.843721192104341e-06,
205
+ "loss": 0.7243,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.11166730843280709,
210
+ "grad_norm": 0.18889667093753815,
211
+ "learning_rate": 9.827876286644698e-06,
212
+ "loss": 0.7556,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.11551790527531768,
217
+ "grad_norm": 0.18423715233802795,
218
+ "learning_rate": 9.811280460765643e-06,
219
+ "loss": 0.7148,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.11936850211782826,
224
+ "grad_norm": 0.19556981325149536,
225
+ "learning_rate": 9.793936295756292e-06,
226
+ "loss": 0.7199,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.12321909896033885,
231
+ "grad_norm": 0.18121841549873352,
232
+ "learning_rate": 9.775846489301279e-06,
233
+ "loss": 0.7318,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.12706969580284944,
238
+ "grad_norm": 0.18317541480064392,
239
+ "learning_rate": 9.757013855061157e-06,
240
+ "loss": 0.7299,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.13092029264536004,
245
+ "grad_norm": 0.17842233180999756,
246
+ "learning_rate": 9.737441322234758e-06,
247
+ "loss": 0.7336,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.1347708894878706,
252
+ "grad_norm": 0.2015552520751953,
253
+ "learning_rate": 9.717131935103605e-06,
254
+ "loss": 0.7403,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.1386214863303812,
259
+ "grad_norm": 0.18337400257587433,
260
+ "learning_rate": 9.696088852558398e-06,
261
+ "loss": 0.7226,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.1424720831728918,
266
+ "grad_norm": 0.21495500206947327,
267
+ "learning_rate": 9.674315347607686e-06,
268
+ "loss": 0.7519,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.14632268001540238,
273
+ "grad_norm": 0.19869428873062134,
274
+ "learning_rate": 9.651814806868796e-06,
275
+ "loss": 0.7159,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.15017327685791299,
280
+ "grad_norm": 0.1836620718240738,
281
+ "learning_rate": 9.62859073004107e-06,
282
+ "loss": 0.7164,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.15402387370042356,
287
+ "grad_norm": 0.17674340307712555,
288
+ "learning_rate": 9.604646729361547e-06,
289
+ "loss": 0.7233,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.15787447054293416,
294
+ "grad_norm": 0.19943110644817352,
295
+ "learning_rate": 9.579986529043099e-06,
296
+ "loss": 0.719,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.16172506738544473,
301
+ "grad_norm": 0.19508565962314606,
302
+ "learning_rate": 9.554613964695189e-06,
303
+ "loss": 0.7368,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.16557566422795533,
308
+ "grad_norm": 0.19070926308631897,
309
+ "learning_rate": 9.528532982727281e-06,
310
+ "loss": 0.7205,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.16942626107046593,
315
+ "grad_norm": 0.18043412268161774,
316
+ "learning_rate": 9.501747639735018e-06,
317
+ "loss": 0.723,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.1732768579129765,
322
+ "grad_norm": 0.19123156368732452,
323
+ "learning_rate": 9.474262101869272e-06,
324
+ "loss": 0.7109,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.1771274547554871,
329
+ "grad_norm": 0.2005036473274231,
330
+ "learning_rate": 9.446080644188141e-06,
331
+ "loss": 0.7207,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.18097805159799768,
336
+ "grad_norm": 0.18703444302082062,
337
+ "learning_rate": 9.417207649992017e-06,
338
+ "loss": 0.722,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.18482864844050828,
343
+ "grad_norm": 0.1858997642993927,
344
+ "learning_rate": 9.387647610141816e-06,
345
+ "loss": 0.7175,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.18867924528301888,
350
+ "grad_norm": 0.17180319130420685,
351
+ "learning_rate": 9.35740512236047e-06,
352
+ "loss": 0.7148,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.19252984212552945,
357
+ "grad_norm": 0.17571838200092316,
358
+ "learning_rate": 9.326484890517809e-06,
359
+ "loss": 0.7209,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.19638043896804006,
364
+ "grad_norm": 0.18318626284599304,
365
+ "learning_rate": 9.294891723898926e-06,
366
+ "loss": 0.7085,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.20023103581055063,
371
+ "grad_norm": 0.17966823279857635,
372
+ "learning_rate": 9.262630536456153e-06,
373
+ "loss": 0.7303,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.20408163265306123,
378
+ "grad_norm": 0.1843014657497406,
379
+ "learning_rate": 9.229706346044749e-06,
380
+ "loss": 0.7162,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.2079322294955718,
385
+ "grad_norm": 0.18524780869483948,
386
+ "learning_rate": 9.196124273642422e-06,
387
+ "loss": 0.7253,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 0.2117828263380824,
392
+ "grad_norm": 0.17711922526359558,
393
+ "learning_rate": 9.161889542552842e-06,
394
+ "loss": 0.7336,
395
+ "step": 550
396
+ },
397
+ {
398
+ "epoch": 0.215633423180593,
399
+ "grad_norm": 0.1875663697719574,
400
+ "learning_rate": 9.127007477593193e-06,
401
+ "loss": 0.733,
402
+ "step": 560
403
+ },
404
+ {
405
+ "epoch": 0.21948402002310358,
406
+ "grad_norm": 0.18302255868911743,
407
+ "learning_rate": 9.091483504265974e-06,
408
+ "loss": 0.7426,
409
+ "step": 570
410
+ },
411
+ {
412
+ "epoch": 0.22333461686561418,
413
+ "grad_norm": 0.19478768110275269,
414
+ "learning_rate": 9.055323147915116e-06,
415
+ "loss": 0.7147,
416
+ "step": 580
417
+ },
418
+ {
419
+ "epoch": 0.22718521370812475,
420
+ "grad_norm": 0.1949205994606018,
421
+ "learning_rate": 9.018532032866585e-06,
422
+ "loss": 0.7269,
423
+ "step": 590
424
+ },
425
+ {
426
+ "epoch": 0.23103581055063535,
427
+ "grad_norm": 0.17177525162696838,
428
+ "learning_rate": 8.981115881553579e-06,
429
+ "loss": 0.7242,
430
+ "step": 600
431
+ },
432
+ {
433
+ "epoch": 0.23488640739314592,
434
+ "grad_norm": 0.19594444334506989,
435
+ "learning_rate": 8.943080513626476e-06,
436
+ "loss": 0.7443,
437
+ "step": 610
438
+ },
439
+ {
440
+ "epoch": 0.23873700423565652,
441
+ "grad_norm": 0.17567043006420135,
442
+ "learning_rate": 8.904431845047641e-06,
443
+ "loss": 0.7239,
444
+ "step": 620
445
+ },
446
+ {
447
+ "epoch": 0.24258760107816713,
448
+ "grad_norm": 0.1756083220243454,
449
+ "learning_rate": 8.86517588717128e-06,
450
+ "loss": 0.7352,
451
+ "step": 630
452
+ },
453
+ {
454
+ "epoch": 0.2464381979206777,
455
+ "grad_norm": 0.1943541318178177,
456
+ "learning_rate": 8.82531874580844e-06,
457
+ "loss": 0.7368,
458
+ "step": 640
459
+ },
460
+ {
461
+ "epoch": 0.25028879476318827,
462
+ "grad_norm": 0.17658770084381104,
463
+ "learning_rate": 8.784866620277314e-06,
464
+ "loss": 0.6954,
465
+ "step": 650
466
+ },
467
+ {
468
+ "epoch": 0.25413939160569887,
469
+ "grad_norm": 0.18532630801200867,
470
+ "learning_rate": 8.743825802439018e-06,
471
+ "loss": 0.7176,
472
+ "step": 660
473
+ },
474
+ {
475
+ "epoch": 0.2579899884482095,
476
+ "grad_norm": 0.17297881841659546,
477
+ "learning_rate": 8.702202675718956e-06,
478
+ "loss": 0.7323,
479
+ "step": 670
480
+ },
481
+ {
482
+ "epoch": 0.2618405852907201,
483
+ "grad_norm": 0.17080391943454742,
484
+ "learning_rate": 8.660003714113959e-06,
485
+ "loss": 0.7096,
486
+ "step": 680
487
+ },
488
+ {
489
+ "epoch": 0.2656911821332307,
490
+ "grad_norm": 0.18693123757839203,
491
+ "learning_rate": 8.617235481185323e-06,
492
+ "loss": 0.7196,
493
+ "step": 690
494
+ },
495
+ {
496
+ "epoch": 0.2695417789757412,
497
+ "grad_norm": 0.20461197197437286,
498
+ "learning_rate": 8.573904629037931e-06,
499
+ "loss": 0.7336,
500
+ "step": 700
501
+ },
502
+ {
503
+ "epoch": 0.2733923758182518,
504
+ "grad_norm": 0.16472092270851135,
505
+ "learning_rate": 8.530017897285585e-06,
506
+ "loss": 0.7158,
507
+ "step": 710
508
+ },
509
+ {
510
+ "epoch": 0.2772429726607624,
511
+ "grad_norm": 0.17247265577316284,
512
+ "learning_rate": 8.485582112002746e-06,
513
+ "loss": 0.7181,
514
+ "step": 720
515
+ },
516
+ {
517
+ "epoch": 0.281093569503273,
518
+ "grad_norm": 0.17503874003887177,
519
+ "learning_rate": 8.440604184662811e-06,
520
+ "loss": 0.724,
521
+ "step": 730
522
+ },
523
+ {
524
+ "epoch": 0.2849441663457836,
525
+ "grad_norm": 0.18170703947544098,
526
+ "learning_rate": 8.395091111063118e-06,
527
+ "loss": 0.7105,
528
+ "step": 740
529
+ },
530
+ {
531
+ "epoch": 0.28879476318829417,
532
+ "grad_norm": 0.15658243000507355,
533
+ "learning_rate": 8.349049970236822e-06,
534
+ "loss": 0.7102,
535
+ "step": 750
536
+ },
537
+ {
538
+ "epoch": 0.29264536003080477,
539
+ "grad_norm": 0.16886980831623077,
540
+ "learning_rate": 8.302487923351842e-06,
541
+ "loss": 0.703,
542
+ "step": 760
543
+ },
544
+ {
545
+ "epoch": 0.29649595687331537,
546
+ "grad_norm": 0.16262605786323547,
547
+ "learning_rate": 8.255412212597021e-06,
548
+ "loss": 0.7237,
549
+ "step": 770
550
+ },
551
+ {
552
+ "epoch": 0.30034655371582597,
553
+ "grad_norm": 0.18170657753944397,
554
+ "learning_rate": 8.207830160055688e-06,
555
+ "loss": 0.7203,
556
+ "step": 780
557
+ },
558
+ {
559
+ "epoch": 0.3041971505583365,
560
+ "grad_norm": 0.1729075312614441,
561
+ "learning_rate": 8.159749166566786e-06,
562
+ "loss": 0.7015,
563
+ "step": 790
564
+ },
565
+ {
566
+ "epoch": 0.3080477474008471,
567
+ "grad_norm": 0.17293652892112732,
568
+ "learning_rate": 8.11117671057377e-06,
569
+ "loss": 0.722,
570
+ "step": 800
571
+ },
572
+ {
573
+ "epoch": 0.3118983442433577,
574
+ "grad_norm": 0.1841427981853485,
575
+ "learning_rate": 8.062120346961404e-06,
576
+ "loss": 0.7201,
577
+ "step": 810
578
+ },
579
+ {
580
+ "epoch": 0.3157489410858683,
581
+ "grad_norm": 0.1667763739824295,
582
+ "learning_rate": 8.012587705880705e-06,
583
+ "loss": 0.7168,
584
+ "step": 820
585
+ },
586
+ {
587
+ "epoch": 0.3195995379283789,
588
+ "grad_norm": 0.18048426508903503,
589
+ "learning_rate": 7.962586491562146e-06,
590
+ "loss": 0.7044,
591
+ "step": 830
592
+ },
593
+ {
594
+ "epoch": 0.32345013477088946,
595
+ "grad_norm": 0.17600588500499725,
596
+ "learning_rate": 7.912124481117357e-06,
597
+ "loss": 0.7043,
598
+ "step": 840
599
+ },
600
+ {
601
+ "epoch": 0.32730073161340006,
602
+ "grad_norm": 0.17007315158843994,
603
+ "learning_rate": 7.861209523329494e-06,
604
+ "loss": 0.7173,
605
+ "step": 850
606
+ },
607
+ {
608
+ "epoch": 0.33115132845591067,
609
+ "grad_norm": 0.17885972559452057,
610
+ "learning_rate": 7.809849537432432e-06,
611
+ "loss": 0.7066,
612
+ "step": 860
613
+ },
614
+ {
615
+ "epoch": 0.33500192529842127,
616
+ "grad_norm": 0.1664038449525833,
617
+ "learning_rate": 7.758052511879035e-06,
618
+ "loss": 0.7092,
619
+ "step": 870
620
+ },
621
+ {
622
+ "epoch": 0.33885252214093187,
623
+ "grad_norm": 0.1865394115447998,
624
+ "learning_rate": 7.705826503098642e-06,
625
+ "loss": 0.7127,
626
+ "step": 880
627
+ },
628
+ {
629
+ "epoch": 0.3427031189834424,
630
+ "grad_norm": 0.18686555325984955,
631
+ "learning_rate": 7.653179634243979e-06,
632
+ "loss": 0.7077,
633
+ "step": 890
634
+ },
635
+ {
636
+ "epoch": 0.346553715825953,
637
+ "grad_norm": 0.173833966255188,
638
+ "learning_rate": 7.600120093927696e-06,
639
+ "loss": 0.7052,
640
+ "step": 900
641
+ },
642
+ {
643
+ "epoch": 0.3504043126684636,
644
+ "grad_norm": 0.20005890727043152,
645
+ "learning_rate": 7.546656134948729e-06,
646
+ "loss": 0.6975,
647
+ "step": 910
648
+ },
649
+ {
650
+ "epoch": 0.3542549095109742,
651
+ "grad_norm": 0.16714885830879211,
652
+ "learning_rate": 7.492796073008661e-06,
653
+ "loss": 0.7088,
654
+ "step": 920
655
+ },
656
+ {
657
+ "epoch": 0.3581055063534848,
658
+ "grad_norm": 0.19144658744335175,
659
+ "learning_rate": 7.438548285418326e-06,
660
+ "loss": 0.6885,
661
+ "step": 930
662
+ },
663
+ {
664
+ "epoch": 0.36195610319599536,
665
+ "grad_norm": 0.17240391671657562,
666
+ "learning_rate": 7.3839212097948e-06,
667
+ "loss": 0.7197,
668
+ "step": 940
669
+ },
670
+ {
671
+ "epoch": 0.36580670003850596,
672
+ "grad_norm": 0.18145813047885895,
673
+ "learning_rate": 7.328923342749039e-06,
674
+ "loss": 0.7374,
675
+ "step": 950
676
+ },
677
+ {
678
+ "epoch": 0.36965729688101656,
679
+ "grad_norm": 0.17329685389995575,
680
+ "learning_rate": 7.273563238564319e-06,
681
+ "loss": 0.7309,
682
+ "step": 960
683
+ },
684
+ {
685
+ "epoch": 0.37350789372352716,
686
+ "grad_norm": 0.1664896160364151,
687
+ "learning_rate": 7.217849507865724e-06,
688
+ "loss": 0.6978,
689
+ "step": 970
690
+ },
691
+ {
692
+ "epoch": 0.37735849056603776,
693
+ "grad_norm": 0.1705247312784195,
694
+ "learning_rate": 7.161790816280858e-06,
695
+ "loss": 0.7118,
696
+ "step": 980
697
+ },
698
+ {
699
+ "epoch": 0.3812090874085483,
700
+ "grad_norm": 0.17965954542160034,
701
+ "learning_rate": 7.1053958830920125e-06,
702
+ "loss": 0.6927,
703
+ "step": 990
704
+ },
705
+ {
706
+ "epoch": 0.3850596842510589,
707
+ "grad_norm": 0.1812833696603775,
708
+ "learning_rate": 7.0486734798799715e-06,
709
+ "loss": 0.7086,
710
+ "step": 1000
711
+ },
712
+ {
713
+ "epoch": 0.3889102810935695,
714
+ "grad_norm": 0.1721523553133011,
715
+ "learning_rate": 6.991632429159708e-06,
716
+ "loss": 0.7288,
717
+ "step": 1010
718
+ },
719
+ {
720
+ "epoch": 0.3927608779360801,
721
+ "grad_norm": 0.18749043345451355,
722
+ "learning_rate": 6.934281603008132e-06,
723
+ "loss": 0.7252,
724
+ "step": 1020
725
+ },
726
+ {
727
+ "epoch": 0.39661147477859066,
728
+ "grad_norm": 0.15639224648475647,
729
+ "learning_rate": 6.876629921684143e-06,
730
+ "loss": 0.7209,
731
+ "step": 1030
732
+ },
733
+ {
734
+ "epoch": 0.40046207162110126,
735
+ "grad_norm": 0.17905481159687042,
736
+ "learning_rate": 6.818686352241199e-06,
737
+ "loss": 0.7085,
738
+ "step": 1040
739
+ },
740
+ {
741
+ "epoch": 0.40431266846361186,
742
+ "grad_norm": 0.1619415581226349,
743
+ "learning_rate": 6.760459907132573e-06,
744
+ "loss": 0.7029,
745
+ "step": 1050
746
+ },
747
+ {
748
+ "epoch": 0.40816326530612246,
749
+ "grad_norm": 0.1631655991077423,
750
+ "learning_rate": 6.701959642809589e-06,
751
+ "loss": 0.7085,
752
+ "step": 1060
753
+ },
754
+ {
755
+ "epoch": 0.41201386214863306,
756
+ "grad_norm": 0.17090685665607452,
757
+ "learning_rate": 6.643194658312987e-06,
758
+ "loss": 0.6992,
759
+ "step": 1070
760
+ },
761
+ {
762
+ "epoch": 0.4158644589911436,
763
+ "grad_norm": 0.1643698811531067,
764
+ "learning_rate": 6.584174093857676e-06,
765
+ "loss": 0.7278,
766
+ "step": 1080
767
+ },
768
+ {
769
+ "epoch": 0.4197150558336542,
770
+ "grad_norm": 0.18299128115177155,
771
+ "learning_rate": 6.5249071294110735e-06,
772
+ "loss": 0.7104,
773
+ "step": 1090
774
+ },
775
+ {
776
+ "epoch": 0.4235656526761648,
777
+ "grad_norm": 0.1796012818813324,
778
+ "learning_rate": 6.465402983265274e-06,
779
+ "loss": 0.7222,
780
+ "step": 1100
781
+ },
782
+ {
783
+ "epoch": 0.4274162495186754,
784
+ "grad_norm": 0.18283215165138245,
785
+ "learning_rate": 6.405670910603243e-06,
786
+ "loss": 0.7124,
787
+ "step": 1110
788
+ },
789
+ {
790
+ "epoch": 0.431266846361186,
791
+ "grad_norm": 0.18329812586307526,
792
+ "learning_rate": 6.345720202059291e-06,
793
+ "loss": 0.7106,
794
+ "step": 1120
795
+ },
796
+ {
797
+ "epoch": 0.43511744320369655,
798
+ "grad_norm": 0.1835220605134964,
799
+ "learning_rate": 6.285560182274008e-06,
800
+ "loss": 0.6908,
801
+ "step": 1130
802
+ },
803
+ {
804
+ "epoch": 0.43896804004620715,
805
+ "grad_norm": 0.1829475462436676,
806
+ "learning_rate": 6.225200208443936e-06,
807
+ "loss": 0.7165,
808
+ "step": 1140
809
+ },
810
+ {
811
+ "epoch": 0.44281863688871775,
812
+ "grad_norm": 0.1748993843793869,
813
+ "learning_rate": 6.16464966886616e-06,
814
+ "loss": 0.6999,
815
+ "step": 1150
816
+ },
817
+ {
818
+ "epoch": 0.44666923373122835,
819
+ "grad_norm": 0.19866006076335907,
820
+ "learning_rate": 6.103917981478065e-06,
821
+ "loss": 0.7306,
822
+ "step": 1160
823
+ },
824
+ {
825
+ "epoch": 0.45051983057373896,
826
+ "grad_norm": 0.16905611753463745,
827
+ "learning_rate": 6.043014592392489e-06,
828
+ "loss": 0.7372,
829
+ "step": 1170
830
+ },
831
+ {
832
+ "epoch": 0.4543704274162495,
833
+ "grad_norm": 0.17134757339954376,
834
+ "learning_rate": 5.981948974428485e-06,
835
+ "loss": 0.7183,
836
+ "step": 1180
837
+ },
838
+ {
839
+ "epoch": 0.4582210242587601,
840
+ "grad_norm": 0.16037195920944214,
841
+ "learning_rate": 5.920730625637934e-06,
842
+ "loss": 0.6913,
843
+ "step": 1190
844
+ },
845
+ {
846
+ "epoch": 0.4620716211012707,
847
+ "grad_norm": 0.17638260126113892,
848
+ "learning_rate": 5.8593690678282355e-06,
849
+ "loss": 0.7034,
850
+ "step": 1200
851
+ },
852
+ {
853
+ "epoch": 0.4659222179437813,
854
+ "grad_norm": 0.17149432003498077,
855
+ "learning_rate": 5.797873845081301e-06,
856
+ "loss": 0.7015,
857
+ "step": 1210
858
+ },
859
+ {
860
+ "epoch": 0.46977281478629185,
861
+ "grad_norm": 0.18346697092056274,
862
+ "learning_rate": 5.736254522269075e-06,
863
+ "loss": 0.7249,
864
+ "step": 1220
865
+ },
866
+ {
867
+ "epoch": 0.47362341162880245,
868
+ "grad_norm": 0.16656143963336945,
869
+ "learning_rate": 5.6745206835658386e-06,
870
+ "loss": 0.723,
871
+ "step": 1230
872
+ },
873
+ {
874
+ "epoch": 0.47747400847131305,
875
+ "grad_norm": 0.17568273842334747,
876
+ "learning_rate": 5.612681930957495e-06,
877
+ "loss": 0.702,
878
+ "step": 1240
879
+ },
880
+ {
881
+ "epoch": 0.48132460531382365,
882
+ "grad_norm": 0.1690274029970169,
883
+ "learning_rate": 5.550747882748095e-06,
884
+ "loss": 0.6915,
885
+ "step": 1250
886
+ },
887
+ {
888
+ "epoch": 0.48517520215633425,
889
+ "grad_norm": 0.196041539311409,
890
+ "learning_rate": 5.4887281720638195e-06,
891
+ "loss": 0.7075,
892
+ "step": 1260
893
+ },
894
+ {
895
+ "epoch": 0.4890257989988448,
896
+ "grad_norm": 0.19659490883350372,
897
+ "learning_rate": 5.42663244535466e-06,
898
+ "loss": 0.6827,
899
+ "step": 1270
900
+ },
901
+ {
902
+ "epoch": 0.4928763958413554,
903
+ "grad_norm": 0.1741725355386734,
904
+ "learning_rate": 5.3644703608940205e-06,
905
+ "loss": 0.7013,
906
+ "step": 1280
907
+ },
908
+ {
909
+ "epoch": 0.496726992683866,
910
+ "grad_norm": 0.1608942300081253,
911
+ "learning_rate": 5.302251587276491e-06,
912
+ "loss": 0.7123,
913
+ "step": 1290
914
+ },
915
+ {
916
+ "epoch": 0.5005775895263765,
917
+ "grad_norm": 0.1742510199546814,
918
+ "learning_rate": 5.2399858019140005e-06,
919
+ "loss": 0.7031,
920
+ "step": 1300
921
+ },
922
+ {
923
+ "epoch": 0.5044281863688872,
924
+ "grad_norm": 0.1791425496339798,
925
+ "learning_rate": 5.177682689530615e-06,
926
+ "loss": 0.726,
927
+ "step": 1310
928
+ },
929
+ {
930
+ "epoch": 0.5082787832113977,
931
+ "grad_norm": 0.18308281898498535,
932
+ "learning_rate": 5.11535194065619e-06,
933
+ "loss": 0.6917,
934
+ "step": 1320
935
+ },
936
+ {
937
+ "epoch": 0.5121293800539084,
938
+ "grad_norm": 0.1858581304550171,
939
+ "learning_rate": 5.053003250119113e-06,
940
+ "loss": 0.6989,
941
+ "step": 1330
942
+ },
943
+ {
944
+ "epoch": 0.515979976896419,
945
+ "grad_norm": 0.17110225558280945,
946
+ "learning_rate": 4.9906463155383925e-06,
947
+ "loss": 0.706,
948
+ "step": 1340
949
+ },
950
+ {
951
+ "epoch": 0.5198305737389295,
952
+ "grad_norm": 0.17445145547389984,
953
+ "learning_rate": 4.9282908358153046e-06,
954
+ "loss": 0.7071,
955
+ "step": 1350
956
+ },
957
+ {
958
+ "epoch": 0.5236811705814401,
959
+ "grad_norm": 0.183705672621727,
960
+ "learning_rate": 4.8659465096248345e-06,
961
+ "loss": 0.7189,
962
+ "step": 1360
963
+ },
964
+ {
965
+ "epoch": 0.5275317674239507,
966
+ "grad_norm": 0.17757634818553925,
967
+ "learning_rate": 4.803623033907169e-06,
968
+ "loss": 0.7091,
969
+ "step": 1370
970
+ },
971
+ {
972
+ "epoch": 0.5313823642664613,
973
+ "grad_norm": 0.1673879772424698,
974
+ "learning_rate": 4.741330102359441e-06,
975
+ "loss": 0.7252,
976
+ "step": 1380
977
+ },
978
+ {
979
+ "epoch": 0.5352329611089719,
980
+ "grad_norm": 0.1792771816253662,
981
+ "learning_rate": 4.679077403927996e-06,
982
+ "loss": 0.6988,
983
+ "step": 1390
984
+ },
985
+ {
986
+ "epoch": 0.5390835579514824,
987
+ "grad_norm": 0.16861207783222198,
988
+ "learning_rate": 4.616874621301373e-06,
989
+ "loss": 0.7024,
990
+ "step": 1400
991
+ },
992
+ {
993
+ "epoch": 0.5429341547939931,
994
+ "grad_norm": 0.1909550428390503,
995
+ "learning_rate": 4.554731429404293e-06,
996
+ "loss": 0.7222,
997
+ "step": 1410
998
+ },
999
+ {
1000
+ "epoch": 0.5467847516365036,
1001
+ "grad_norm": 0.1797027587890625,
1002
+ "learning_rate": 4.492657493892822e-06,
1003
+ "loss": 0.701,
1004
+ "step": 1420
1005
+ },
1006
+ {
1007
+ "epoch": 0.5506353484790143,
1008
+ "grad_norm": 0.18553535640239716,
1009
+ "learning_rate": 4.430662469650999e-06,
1010
+ "loss": 0.7145,
1011
+ "step": 1430
1012
+ },
1013
+ {
1014
+ "epoch": 0.5544859453215248,
1015
+ "grad_norm": 0.18707172572612762,
1016
+ "learning_rate": 4.368755999289118e-06,
1017
+ "loss": 0.7187,
1018
+ "step": 1440
1019
+ },
1020
+ {
1021
+ "epoch": 0.5583365421640354,
1022
+ "grad_norm": 0.1872454136610031,
1023
+ "learning_rate": 4.306947711643949e-06,
1024
+ "loss": 0.7013,
1025
+ "step": 1450
1026
+ },
1027
+ {
1028
+ "epoch": 0.562187139006546,
1029
+ "grad_norm": 0.1724030077457428,
1030
+ "learning_rate": 4.245247220281068e-06,
1031
+ "loss": 0.706,
1032
+ "step": 1460
1033
+ },
1034
+ {
1035
+ "epoch": 0.5660377358490566,
1036
+ "grad_norm": 0.1706014722585678,
1037
+ "learning_rate": 4.183664121999587e-06,
1038
+ "loss": 0.6998,
1039
+ "step": 1470
1040
+ },
1041
+ {
1042
+ "epoch": 0.5698883326915672,
1043
+ "grad_norm": 0.180300772190094,
1044
+ "learning_rate": 4.122207995339474e-06,
1045
+ "loss": 0.6957,
1046
+ "step": 1480
1047
+ },
1048
+ {
1049
+ "epoch": 0.5737389295340778,
1050
+ "grad_norm": 0.1741262823343277,
1051
+ "learning_rate": 4.060888399091734e-06,
1052
+ "loss": 0.7031,
1053
+ "step": 1490
1054
+ },
1055
+ {
1056
+ "epoch": 0.5775895263765883,
1057
+ "grad_norm": 0.18754544854164124,
1058
+ "learning_rate": 3.999714870811639e-06,
1059
+ "loss": 0.7167,
1060
+ "step": 1500
1061
+ },
1062
+ {
1063
+ "epoch": 0.581440123219099,
1064
+ "grad_norm": 0.17907388508319855,
1065
+ "learning_rate": 3.938696925335275e-06,
1066
+ "loss": 0.7095,
1067
+ "step": 1510
1068
+ },
1069
+ {
1070
+ "epoch": 0.5852907200616095,
1071
+ "grad_norm": 0.17697542905807495,
1072
+ "learning_rate": 3.87784405329962e-06,
1073
+ "loss": 0.7164,
1074
+ "step": 1520
1075
+ },
1076
+ {
1077
+ "epoch": 0.5891413169041202,
1078
+ "grad_norm": 0.16688750684261322,
1079
+ "learning_rate": 3.817165719666382e-06,
1080
+ "loss": 0.7156,
1081
+ "step": 1530
1082
+ },
1083
+ {
1084
+ "epoch": 0.5929919137466307,
1085
+ "grad_norm": 0.17989963293075562,
1086
+ "learning_rate": 3.756671362249841e-06,
1087
+ "loss": 0.721,
1088
+ "step": 1540
1089
+ },
1090
+ {
1091
+ "epoch": 0.5968425105891413,
1092
+ "grad_norm": 0.16907528042793274,
1093
+ "learning_rate": 3.6963703902488907e-06,
1094
+ "loss": 0.7071,
1095
+ "step": 1550
1096
+ },
1097
+ {
1098
+ "epoch": 0.6006931074316519,
1099
+ "grad_norm": 0.17311136424541473,
1100
+ "learning_rate": 3.6362721827835655e-06,
1101
+ "loss": 0.7145,
1102
+ "step": 1560
1103
+ },
1104
+ {
1105
+ "epoch": 0.6045437042741625,
1106
+ "grad_norm": 0.18077132105827332,
1107
+ "learning_rate": 3.576386087436212e-06,
1108
+ "loss": 0.7239,
1109
+ "step": 1570
1110
+ },
1111
+ {
1112
+ "epoch": 0.608394301116673,
1113
+ "grad_norm": 0.1893598586320877,
1114
+ "learning_rate": 3.5167214187975913e-06,
1115
+ "loss": 0.7147,
1116
+ "step": 1580
1117
+ },
1118
+ {
1119
+ "epoch": 0.6122448979591837,
1120
+ "grad_norm": 0.16440844535827637,
1121
+ "learning_rate": 3.457287457018086e-06,
1122
+ "loss": 0.7111,
1123
+ "step": 1590
1124
+ },
1125
+ {
1126
+ "epoch": 0.6160954948016942,
1127
+ "grad_norm": 0.17374873161315918,
1128
+ "learning_rate": 3.3980934463643003e-06,
1129
+ "loss": 0.7071,
1130
+ "step": 1600
1131
+ },
1132
+ {
1133
+ "epoch": 0.6199460916442049,
1134
+ "grad_norm": 0.16687820851802826,
1135
+ "learning_rate": 3.3391485937812053e-06,
1136
+ "loss": 0.684,
1137
+ "step": 1610
1138
+ },
1139
+ {
1140
+ "epoch": 0.6237966884867154,
1141
+ "grad_norm": 0.17627935111522675,
1142
+ "learning_rate": 3.280462067460115e-06,
1143
+ "loss": 0.686,
1144
+ "step": 1620
1145
+ },
1146
+ {
1147
+ "epoch": 0.627647285329226,
1148
+ "grad_norm": 0.17719842493534088,
1149
+ "learning_rate": 3.222042995412669e-06,
1150
+ "loss": 0.7106,
1151
+ "step": 1630
1152
+ },
1153
+ {
1154
+ "epoch": 0.6314978821717366,
1155
+ "grad_norm": 0.16091850399971008,
1156
+ "learning_rate": 3.163900464051089e-06,
1157
+ "loss": 0.7013,
1158
+ "step": 1640
1159
+ },
1160
+ {
1161
+ "epoch": 0.6353484790142472,
1162
+ "grad_norm": 0.18081437051296234,
1163
+ "learning_rate": 3.1060435167748826e-06,
1164
+ "loss": 0.7177,
1165
+ "step": 1650
1166
+ },
1167
+ {
1168
+ "epoch": 0.6391990758567578,
1169
+ "grad_norm": 0.1700412780046463,
1170
+ "learning_rate": 3.048481152564253e-06,
1171
+ "loss": 0.6955,
1172
+ "step": 1660
1173
+ },
1174
+ {
1175
+ "epoch": 0.6430496726992684,
1176
+ "grad_norm": 0.19202637672424316,
1177
+ "learning_rate": 2.9912223245804017e-06,
1178
+ "loss": 0.7124,
1179
+ "step": 1670
1180
+ },
1181
+ {
1182
+ "epoch": 0.6469002695417789,
1183
+ "grad_norm": 0.1716102957725525,
1184
+ "learning_rate": 2.934275938772981e-06,
1185
+ "loss": 0.7021,
1186
+ "step": 1680
1187
+ },
1188
+ {
1189
+ "epoch": 0.6507508663842896,
1190
+ "grad_norm": 0.17713803052902222,
1191
+ "learning_rate": 2.8776508524948676e-06,
1192
+ "loss": 0.6976,
1193
+ "step": 1690
1194
+ },
1195
+ {
1196
+ "epoch": 0.6546014632268001,
1197
+ "grad_norm": 0.17363609373569489,
1198
+ "learning_rate": 2.821355873124503e-06,
1199
+ "loss": 0.6992,
1200
+ "step": 1700
1201
+ },
1202
+ {
1203
+ "epoch": 0.6584520600693108,
1204
+ "grad_norm": 0.17823483049869537,
1205
+ "learning_rate": 2.7653997566960144e-06,
1206
+ "loss": 0.7083,
1207
+ "step": 1710
1208
+ },
1209
+ {
1210
+ "epoch": 0.6623026569118213,
1211
+ "grad_norm": 0.18773998320102692,
1212
+ "learning_rate": 2.709791206537309e-06,
1213
+ "loss": 0.7012,
1214
+ "step": 1720
1215
+ },
1216
+ {
1217
+ "epoch": 0.6661532537543319,
1218
+ "grad_norm": 0.19326363503932953,
1219
+ "learning_rate": 2.6545388719163747e-06,
1220
+ "loss": 0.7019,
1221
+ "step": 1730
1222
+ },
1223
+ {
1224
+ "epoch": 0.6700038505968425,
1225
+ "grad_norm": 0.17272448539733887,
1226
+ "learning_rate": 2.599651346695979e-06,
1227
+ "loss": 0.7122,
1228
+ "step": 1740
1229
+ },
1230
+ {
1231
+ "epoch": 0.6738544474393531,
1232
+ "grad_norm": 0.17249658703804016,
1233
+ "learning_rate": 2.545137167997007e-06,
1234
+ "loss": 0.7096,
1235
+ "step": 1750
1236
+ },
1237
+ {
1238
+ "epoch": 0.6777050442818637,
1239
+ "grad_norm": 0.1703721582889557,
1240
+ "learning_rate": 2.4910048148705972e-06,
1241
+ "loss": 0.698,
1242
+ "step": 1760
1243
+ },
1244
+ {
1245
+ "epoch": 0.6815556411243743,
1246
+ "grad_norm": 0.1806936264038086,
1247
+ "learning_rate": 2.4372627069793327e-06,
1248
+ "loss": 0.7358,
1249
+ "step": 1770
1250
+ },
1251
+ {
1252
+ "epoch": 0.6854062379668848,
1253
+ "grad_norm": 0.17266897857189178,
1254
+ "learning_rate": 2.3839192032876486e-06,
1255
+ "loss": 0.7117,
1256
+ "step": 1780
1257
+ },
1258
+ {
1259
+ "epoch": 0.6892568348093955,
1260
+ "grad_norm": 0.18332628905773163,
1261
+ "learning_rate": 2.3309826007617047e-06,
1262
+ "loss": 0.7278,
1263
+ "step": 1790
1264
+ },
1265
+ {
1266
+ "epoch": 0.693107431651906,
1267
+ "grad_norm": 0.1662522554397583,
1268
+ "learning_rate": 2.278461133078881e-06,
1269
+ "loss": 0.6959,
1270
+ "step": 1800
1271
+ },
1272
+ {
1273
+ "epoch": 0.6969580284944167,
1274
+ "grad_norm": 0.18044349551200867,
1275
+ "learning_rate": 2.2263629693471258e-06,
1276
+ "loss": 0.6896,
1277
+ "step": 1810
1278
+ },
1279
+ {
1280
+ "epoch": 0.7008086253369272,
1281
+ "grad_norm": 0.17569559812545776,
1282
+ "learning_rate": 2.1746962128343414e-06,
1283
+ "loss": 0.71,
1284
+ "step": 1820
1285
+ },
1286
+ {
1287
+ "epoch": 0.7046592221794378,
1288
+ "grad_norm": 0.1831965446472168,
1289
+ "learning_rate": 2.1234688997080183e-06,
1290
+ "loss": 0.696,
1291
+ "step": 1830
1292
+ },
1293
+ {
1294
+ "epoch": 0.7085098190219484,
1295
+ "grad_norm": 0.17632266879081726,
1296
+ "learning_rate": 2.0726889977853053e-06,
1297
+ "loss": 0.7014,
1298
+ "step": 1840
1299
+ },
1300
+ {
1301
+ "epoch": 0.712360415864459,
1302
+ "grad_norm": 0.1805836409330368,
1303
+ "learning_rate": 2.022364405293703e-06,
1304
+ "loss": 0.7173,
1305
+ "step": 1850
1306
+ },
1307
+ {
1308
+ "epoch": 0.7162110127069696,
1309
+ "grad_norm": 0.16981081664562225,
1310
+ "learning_rate": 1.9725029496425814e-06,
1311
+ "loss": 0.7136,
1312
+ "step": 1860
1313
+ },
1314
+ {
1315
+ "epoch": 0.7200616095494802,
1316
+ "grad_norm": 0.17112630605697632,
1317
+ "learning_rate": 1.923112386205727e-06,
1318
+ "loss": 0.7052,
1319
+ "step": 1870
1320
+ },
1321
+ {
1322
+ "epoch": 0.7239122063919907,
1323
+ "grad_norm": 0.17008112370967865,
1324
+ "learning_rate": 1.874200397115079e-06,
1325
+ "loss": 0.7151,
1326
+ "step": 1880
1327
+ },
1328
+ {
1329
+ "epoch": 0.7277628032345014,
1330
+ "grad_norm": 0.17560604214668274,
1331
+ "learning_rate": 1.825774590065857e-06,
1332
+ "loss": 0.6998,
1333
+ "step": 1890
1334
+ },
1335
+ {
1336
+ "epoch": 0.7316134000770119,
1337
+ "grad_norm": 0.18375921249389648,
1338
+ "learning_rate": 1.7778424971332813e-06,
1339
+ "loss": 0.7112,
1340
+ "step": 1900
1341
+ },
1342
+ {
1343
+ "epoch": 0.7354639969195226,
1344
+ "grad_norm": 0.16855861246585846,
1345
+ "learning_rate": 1.7304115736010407e-06,
1346
+ "loss": 0.7066,
1347
+ "step": 1910
1348
+ },
1349
+ {
1350
+ "epoch": 0.7393145937620331,
1351
+ "grad_norm": 0.17428883910179138,
1352
+ "learning_rate": 1.6834891968017108e-06,
1353
+ "loss": 0.6989,
1354
+ "step": 1920
1355
+ },
1356
+ {
1357
+ "epoch": 0.7431651906045437,
1358
+ "grad_norm": 0.182221919298172,
1359
+ "learning_rate": 1.6370826649692884e-06,
1360
+ "loss": 0.7127,
1361
+ "step": 1930
1362
+ },
1363
+ {
1364
+ "epoch": 0.7470157874470543,
1365
+ "grad_norm": 0.17560556530952454,
1366
+ "learning_rate": 1.5911991961040463e-06,
1367
+ "loss": 0.7173,
1368
+ "step": 1940
1369
+ },
1370
+ {
1371
+ "epoch": 0.7508663842895649,
1372
+ "grad_norm": 0.17544133961200714,
1373
+ "learning_rate": 1.545845926849851e-06,
1374
+ "loss": 0.7012,
1375
+ "step": 1950
1376
+ },
1377
+ {
1378
+ "epoch": 0.7547169811320755,
1379
+ "grad_norm": 0.18214553594589233,
1380
+ "learning_rate": 1.5010299113841397e-06,
1381
+ "loss": 0.7241,
1382
+ "step": 1960
1383
+ },
1384
+ {
1385
+ "epoch": 0.7585675779745861,
1386
+ "grad_norm": 0.17534007132053375,
1387
+ "learning_rate": 1.4567581203207238e-06,
1388
+ "loss": 0.7143,
1389
+ "step": 1970
1390
+ },
1391
+ {
1392
+ "epoch": 0.7624181748170966,
1393
+ "grad_norm": 0.1895742565393448,
1394
+ "learning_rate": 1.413037439625593e-06,
1395
+ "loss": 0.7139,
1396
+ "step": 1980
1397
+ },
1398
+ {
1399
+ "epoch": 0.7662687716596073,
1400
+ "grad_norm": 0.18715301156044006,
1401
+ "learning_rate": 1.3698746695458847e-06,
1402
+ "loss": 0.7024,
1403
+ "step": 1990
1404
+ },
1405
+ {
1406
+ "epoch": 0.7701193685021178,
1407
+ "grad_norm": 0.1722307801246643,
1408
+ "learning_rate": 1.3272765235521745e-06,
1409
+ "loss": 0.703,
1410
+ "step": 2000
1411
+ },
1412
+ {
1413
+ "epoch": 0.7739699653446284,
1414
+ "grad_norm": 0.16585290431976318,
1415
+ "learning_rate": 1.2852496272942877e-06,
1416
+ "loss": 0.7126,
1417
+ "step": 2010
1418
+ },
1419
+ {
1420
+ "epoch": 0.777820562187139,
1421
+ "grad_norm": 0.16949434578418732,
1422
+ "learning_rate": 1.2438005175707396e-06,
1423
+ "loss": 0.6905,
1424
+ "step": 2020
1425
+ },
1426
+ {
1427
+ "epoch": 0.7816711590296496,
1428
+ "grad_norm": 0.17125190794467926,
1429
+ "learning_rate": 1.2029356413120297e-06,
1430
+ "loss": 0.6846,
1431
+ "step": 2030
1432
+ },
1433
+ {
1434
+ "epoch": 0.7855217558721602,
1435
+ "grad_norm": 0.17474661767482758,
1436
+ "learning_rate": 1.1626613545778786e-06,
1437
+ "loss": 0.7,
1438
+ "step": 2040
1439
+ },
1440
+ {
1441
+ "epoch": 0.7893723527146708,
1442
+ "grad_norm": 0.17034736275672913,
1443
+ "learning_rate": 1.1229839215686294e-06,
1444
+ "loss": 0.7026,
1445
+ "step": 2050
1446
+ },
1447
+ {
1448
+ "epoch": 0.7932229495571813,
1449
+ "grad_norm": 0.15978913009166718,
1450
+ "learning_rate": 1.0839095136509188e-06,
1451
+ "loss": 0.6877,
1452
+ "step": 2060
1453
+ },
1454
+ {
1455
+ "epoch": 0.797073546399692,
1456
+ "grad_norm": 0.18265946209430695,
1457
+ "learning_rate": 1.045444208397791e-06,
1458
+ "loss": 0.7028,
1459
+ "step": 2070
1460
+ },
1461
+ {
1462
+ "epoch": 0.8009241432422025,
1463
+ "grad_norm": 0.164162278175354,
1464
+ "learning_rate": 1.007593988643402e-06,
1465
+ "loss": 0.6816,
1466
+ "step": 2080
1467
+ },
1468
+ {
1469
+ "epoch": 0.8047747400847132,
1470
+ "grad_norm": 0.16214220225811005,
1471
+ "learning_rate": 9.703647415524613e-07,
1472
+ "loss": 0.7089,
1473
+ "step": 2090
1474
+ },
1475
+ {
1476
+ "epoch": 0.8086253369272237,
1477
+ "grad_norm": 0.1653766930103302,
1478
+ "learning_rate": 9.337622577045485e-07,
1479
+ "loss": 0.6983,
1480
+ "step": 2100
1481
+ },
1482
+ {
1483
+ "epoch": 0.8124759337697343,
1484
+ "grad_norm": 0.1727655529975891,
1485
+ "learning_rate": 8.977922301934588e-07,
1486
+ "loss": 0.7063,
1487
+ "step": 2110
1488
+ },
1489
+ {
1490
+ "epoch": 0.8163265306122449,
1491
+ "grad_norm": 0.1893244981765747,
1492
+ "learning_rate": 8.624602537417009e-07,
1493
+ "loss": 0.7208,
1494
+ "step": 2120
1495
+ },
1496
+ {
1497
+ "epoch": 0.8201771274547555,
1498
+ "grad_norm": 0.17647957801818848,
1499
+ "learning_rate": 8.277718238303123e-07,
1500
+ "loss": 0.7,
1501
+ "step": 2130
1502
+ },
1503
+ {
1504
+ "epoch": 0.8240277242972661,
1505
+ "grad_norm": 0.1758110076189041,
1506
+ "learning_rate": 7.937323358440935e-07,
1507
+ "loss": 0.7037,
1508
+ "step": 2140
1509
+ },
1510
+ {
1511
+ "epoch": 0.8278783211397767,
1512
+ "grad_norm": 0.16953715682029724,
1513
+ "learning_rate": 7.603470842324223e-07,
1514
+ "loss": 0.6894,
1515
+ "step": 2150
1516
+ },
1517
+ {
1518
+ "epoch": 0.8317289179822872,
1519
+ "grad_norm": 0.1704743206501007,
1520
+ "learning_rate": 7.276212616857575e-07,
1521
+ "loss": 0.7127,
1522
+ "step": 2160
1523
+ },
1524
+ {
1525
+ "epoch": 0.8355795148247979,
1526
+ "grad_norm": 0.17713814973831177,
1527
+ "learning_rate": 6.955599583279837e-07,
1528
+ "loss": 0.705,
1529
+ "step": 2170
1530
+ },
1531
+ {
1532
+ "epoch": 0.8394301116673084,
1533
+ "grad_norm": 0.1707342565059662,
1534
+ "learning_rate": 6.641681609246981e-07,
1535
+ "loss": 0.703,
1536
+ "step": 2180
1537
+ },
1538
+ {
1539
+ "epoch": 0.8432807085098191,
1540
+ "grad_norm": 0.18225648999214172,
1541
+ "learning_rate": 6.334507521075711e-07,
1542
+ "loss": 0.7002,
1543
+ "step": 2190
1544
+ },
1545
+ {
1546
+ "epoch": 0.8471313053523296,
1547
+ "grad_norm": 0.18060645461082458,
1548
+ "learning_rate": 6.034125096149179e-07,
1549
+ "loss": 0.7052,
1550
+ "step": 2200
1551
+ },
1552
+ {
1553
+ "epoch": 0.8509819021948402,
1554
+ "grad_norm": 0.1652495115995407,
1555
+ "learning_rate": 5.740581055485722e-07,
1556
+ "loss": 0.6762,
1557
+ "step": 2210
1558
+ },
1559
+ {
1560
+ "epoch": 0.8548324990373508,
1561
+ "grad_norm": 0.16007845103740692,
1562
+ "learning_rate": 5.453921056471961e-07,
1563
+ "loss": 0.7084,
1564
+ "step": 2220
1565
+ },
1566
+ {
1567
+ "epoch": 0.8586830958798614,
1568
+ "grad_norm": 0.18703590333461761,
1569
+ "learning_rate": 5.174189685761288e-07,
1570
+ "loss": 0.7024,
1571
+ "step": 2230
1572
+ },
1573
+ {
1574
+ "epoch": 0.862533692722372,
1575
+ "grad_norm": 0.1679428666830063,
1576
+ "learning_rate": 4.901430452338979e-07,
1577
+ "loss": 0.7146,
1578
+ "step": 2240
1579
+ },
1580
+ {
1581
+ "epoch": 0.8663842895648826,
1582
+ "grad_norm": 0.1708674430847168,
1583
+ "learning_rate": 4.6356857807548286e-07,
1584
+ "loss": 0.7014,
1585
+ "step": 2250
1586
+ },
1587
+ {
1588
+ "epoch": 0.8702348864073931,
1589
+ "grad_norm": 0.2018357664346695,
1590
+ "learning_rate": 4.376997004524541e-07,
1591
+ "loss": 0.6839,
1592
+ "step": 2260
1593
+ },
1594
+ {
1595
+ "epoch": 0.8740854832499038,
1596
+ "grad_norm": 0.16331219673156738,
1597
+ "learning_rate": 4.12540435970073e-07,
1598
+ "loss": 0.6999,
1599
+ "step": 2270
1600
+ },
1601
+ {
1602
+ "epoch": 0.8779360800924143,
1603
+ "grad_norm": 0.16613022983074188,
1604
+ "learning_rate": 3.880946978614703e-07,
1605
+ "loss": 0.7117,
1606
+ "step": 2280
1607
+ },
1608
+ {
1609
+ "epoch": 0.881786676934925,
1610
+ "grad_norm": 0.1681394726037979,
1611
+ "learning_rate": 3.643662883789878e-07,
1612
+ "loss": 0.7059,
1613
+ "step": 2290
1614
+ },
1615
+ {
1616
+ "epoch": 0.8856372737774355,
1617
+ "grad_norm": 0.1642981469631195,
1618
+ "learning_rate": 3.41358898202781e-07,
1619
+ "loss": 0.6918,
1620
+ "step": 2300
1621
+ },
1622
+ {
1623
+ "epoch": 0.889487870619946,
1624
+ "grad_norm": 0.167175754904747,
1625
+ "learning_rate": 3.190761058667741e-07,
1626
+ "loss": 0.7037,
1627
+ "step": 2310
1628
+ },
1629
+ {
1630
+ "epoch": 0.8933384674624567,
1631
+ "grad_norm": 0.1593972146511078,
1632
+ "learning_rate": 2.9752137720206463e-07,
1633
+ "loss": 0.7041,
1634
+ "step": 2320
1635
+ },
1636
+ {
1637
+ "epoch": 0.8971890643049673,
1638
+ "grad_norm": 0.17676116526126862,
1639
+ "learning_rate": 2.7669806479785067e-07,
1640
+ "loss": 0.7152,
1641
+ "step": 2330
1642
+ },
1643
+ {
1644
+ "epoch": 0.9010396611474779,
1645
+ "grad_norm": 0.16829048097133636,
1646
+ "learning_rate": 2.566094074799763e-07,
1647
+ "loss": 0.7057,
1648
+ "step": 2340
1649
+ },
1650
+ {
1651
+ "epoch": 0.9048902579899885,
1652
+ "grad_norm": 0.17307069897651672,
1653
+ "learning_rate": 2.3725852980716557e-07,
1654
+ "loss": 0.6895,
1655
+ "step": 2350
1656
+ },
1657
+ {
1658
+ "epoch": 0.908740854832499,
1659
+ "grad_norm": 0.16956764459609985,
1660
+ "learning_rate": 2.1864844158503983e-07,
1661
+ "loss": 0.7022,
1662
+ "step": 2360
1663
+ },
1664
+ {
1665
+ "epoch": 0.9125914516750097,
1666
+ "grad_norm": 0.18180802464485168,
1667
+ "learning_rate": 2.007820373979702e-07,
1668
+ "loss": 0.6949,
1669
+ "step": 2370
1670
+ },
1671
+ {
1672
+ "epoch": 0.9164420485175202,
1673
+ "grad_norm": 0.1688234806060791,
1674
+ "learning_rate": 1.8366209615886532e-07,
1675
+ "loss": 0.7078,
1676
+ "step": 2380
1677
+ },
1678
+ {
1679
+ "epoch": 0.9202926453600309,
1680
+ "grad_norm": 0.17595353722572327,
1681
+ "learning_rate": 1.6729128067693613e-07,
1682
+ "loss": 0.7204,
1683
+ "step": 2390
1684
+ },
1685
+ {
1686
+ "epoch": 0.9241432422025414,
1687
+ "grad_norm": 0.18783344328403473,
1688
+ "learning_rate": 1.5167213724353426e-07,
1689
+ "loss": 0.7135,
1690
+ "step": 2400
1691
+ },
1692
+ {
1693
+ "epoch": 0.927993839045052,
1694
+ "grad_norm": 0.16535428166389465,
1695
+ "learning_rate": 1.3680709523610058e-07,
1696
+ "loss": 0.6907,
1697
+ "step": 2410
1698
+ },
1699
+ {
1700
+ "epoch": 0.9318444358875626,
1701
+ "grad_norm": 0.16660408675670624,
1702
+ "learning_rate": 1.2269846674030418e-07,
1703
+ "loss": 0.7187,
1704
+ "step": 2420
1705
+ },
1706
+ {
1707
+ "epoch": 0.9356950327300732,
1708
+ "grad_norm": 0.17548833787441254,
1709
+ "learning_rate": 1.0934844619042595e-07,
1710
+ "loss": 0.7142,
1711
+ "step": 2430
1712
+ },
1713
+ {
1714
+ "epoch": 0.9395456295725837,
1715
+ "grad_norm": 0.17332340776920319,
1716
+ "learning_rate": 9.675911002803961e-08,
1717
+ "loss": 0.7007,
1718
+ "step": 2440
1719
+ },
1720
+ {
1721
+ "epoch": 0.9433962264150944,
1722
+ "grad_norm": 0.17920458316802979,
1723
+ "learning_rate": 8.493241637904381e-08,
1724
+ "loss": 0.6898,
1725
+ "step": 2450
1726
+ },
1727
+ {
1728
+ "epoch": 0.9472468232576049,
1729
+ "grad_norm": 0.18417583405971527,
1730
+ "learning_rate": 7.387020474909689e-08,
1731
+ "loss": 0.7175,
1732
+ "step": 2460
1733
+ },
1734
+ {
1735
+ "epoch": 0.9510974201001156,
1736
+ "grad_norm": 0.1807834804058075,
1737
+ "learning_rate": 6.357419573750856e-08,
1738
+ "loss": 0.7129,
1739
+ "step": 2470
1740
+ },
1741
+ {
1742
+ "epoch": 0.9549480169426261,
1743
+ "grad_norm": 0.17384229600429535,
1744
+ "learning_rate": 5.4045990769613365e-08,
1745
+ "loss": 0.6983,
1746
+ "step": 2480
1747
+ },
1748
+ {
1749
+ "epoch": 0.9587986137851366,
1750
+ "grad_norm": 0.16703028976917267,
1751
+ "learning_rate": 4.5287071847694434e-08,
1752
+ "loss": 0.6996,
1753
+ "step": 2490
1754
+ },
1755
+ {
1756
+ "epoch": 0.9626492106276473,
1757
+ "grad_norm": 0.17462129890918732,
1758
+ "learning_rate": 3.7298801320467817e-08,
1759
+ "loss": 0.7051,
1760
+ "step": 2500
1761
+ },
1762
+ {
1763
+ "epoch": 0.9664998074701578,
1764
+ "grad_norm": 0.1667340099811554,
1765
+ "learning_rate": 3.0082421671192576e-08,
1766
+ "loss": 0.7025,
1767
+ "step": 2510
1768
+ },
1769
+ {
1770
+ "epoch": 0.9703504043126685,
1771
+ "grad_norm": 0.17370162904262543,
1772
+ "learning_rate": 2.3639055324410355e-08,
1773
+ "loss": 0.7175,
1774
+ "step": 2520
1775
+ },
1776
+ {
1777
+ "epoch": 0.974201001155179,
1778
+ "grad_norm": 0.1803671419620514,
1779
+ "learning_rate": 1.7969704471369477e-08,
1780
+ "loss": 0.6964,
1781
+ "step": 2530
1782
+ },
1783
+ {
1784
+ "epoch": 0.9780515979976896,
1785
+ "grad_norm": 0.1619933694601059,
1786
+ "learning_rate": 1.3075250914142413e-08,
1787
+ "loss": 0.6991,
1788
+ "step": 2540
1789
+ },
1790
+ {
1791
+ "epoch": 0.9819021948402002,
1792
+ "grad_norm": 0.16015507280826569,
1793
+ "learning_rate": 8.956455928476049e-09,
1794
+ "loss": 0.7122,
1795
+ "step": 2550
1796
+ },
1797
+ {
1798
+ "epoch": 0.9857527916827108,
1799
+ "grad_norm": 0.1679883748292923,
1800
+ "learning_rate": 5.6139601453786275e-09,
1801
+ "loss": 0.6957,
1802
+ "step": 2560
1803
+ },
1804
+ {
1805
+ "epoch": 0.9896033885252215,
1806
+ "grad_norm": 0.17142683267593384,
1807
+ "learning_rate": 3.0482834514805602e-09,
1808
+ "loss": 0.6947,
1809
+ "step": 2570
1810
+ },
1811
+ {
1812
+ "epoch": 0.993453985367732,
1813
+ "grad_norm": 0.1635570079088211,
1814
+ "learning_rate": 1.2598249081713366e-09,
1815
+ "loss": 0.6822,
1816
+ "step": 2580
1817
+ },
1818
+ {
1819
+ "epoch": 0.9973045822102425,
1820
+ "grad_norm": 0.17791008949279785,
1821
+ "learning_rate": 2.488626895302826e-10,
1822
+ "loss": 0.7197,
1823
+ "step": 2590
1824
+ },
1825
+ {
1826
+ "epoch": 1.0,
1827
+ "step": 2597,
1828
+ "total_flos": 7.171311541285814e+18,
1829
+ "train_loss": 0.7206280151605147,
1830
+ "train_runtime": 16955.1801,
1831
+ "train_samples_per_second": 39.215,
1832
+ "train_steps_per_second": 0.153
1833
+ }
1834
+ ],
1835
+ "logging_steps": 10,
1836
+ "max_steps": 2597,
1837
+ "num_input_tokens_seen": 584317136,
1838
+ "num_train_epochs": 1,
1839
+ "save_steps": 500,
1840
+ "stateful_callbacks": {
1841
+ "TrainerControl": {
1842
+ "args": {
1843
+ "should_epoch_stop": false,
1844
+ "should_evaluate": false,
1845
+ "should_log": false,
1846
+ "should_save": true,
1847
+ "should_training_stop": true
1848
+ },
1849
+ "attributes": {}
1850
+ }
1851
+ },
1852
+ "total_flos": 7.171311541285814e+18,
1853
+ "train_batch_size": 16,
1854
+ "trial_name": null,
1855
+ "trial_params": null
1856
+ }
continuous/smollm2/siglip2_sm14_384/finetune/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a993897a3e496b968e67caaaf1623dd3749de4e9791ab4fdad9e310fa3ba47e8
3
+ size 6264
continuous/smollm2/siglip2_sm14_384/finetune/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm14_384/pretrain/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": null,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-14-SigLIP2-378",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm14_384.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm14_384/pretrain/mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c671e1c2bb74cf1421362b5cedb80d35692fc56667f86119e5adbf78b29d6a94
3
+ size 26232888
continuous/smollm2/siglip2_sm14_384/pretrain/trainer_state.json ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 545,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.01834862385321101,
14
+ "grad_norm": 1.0944809913635254,
15
+ "learning_rate": 0.00032142857142857147,
16
+ "loss": 3.7357,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.03669724770642202,
21
+ "grad_norm": 0.15402820706367493,
22
+ "learning_rate": 0.0006785714285714287,
23
+ "loss": 3.0123,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.05504587155963303,
28
+ "grad_norm": 0.06196267902851105,
29
+ "learning_rate": 0.0009999907688176173,
30
+ "loss": 2.8529,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.07339449541284404,
35
+ "grad_norm": 0.04546934738755226,
36
+ "learning_rate": 0.0009988834393115766,
37
+ "loss": 2.7937,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.09174311926605505,
42
+ "grad_norm": 0.04514617100358009,
43
+ "learning_rate": 0.0009959345572940772,
44
+ "loss": 2.7737,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.11009174311926606,
49
+ "grad_norm": 0.11118846386671066,
50
+ "learning_rate": 0.0009911550081155983,
51
+ "loss": 2.7434,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.12844036697247707,
56
+ "grad_norm": 0.15055929124355316,
57
+ "learning_rate": 0.0009845624347566433,
58
+ "loss": 2.7078,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.14678899082568808,
63
+ "grad_norm": 0.06847454607486725,
64
+ "learning_rate": 0.0009761811727013547,
65
+ "loss": 2.6661,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.1651376146788991,
70
+ "grad_norm": 0.07623093575239182,
71
+ "learning_rate": 0.0009660421601067666,
72
+ "loss": 2.6156,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.1834862385321101,
77
+ "grad_norm": 0.07163584232330322,
78
+ "learning_rate": 0.0009541828235992926,
79
+ "loss": 2.5684,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.2018348623853211,
84
+ "grad_norm": 0.0763249322772026,
85
+ "learning_rate": 0.0009406469401200151,
86
+ "loss": 2.5116,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.22018348623853212,
91
+ "grad_norm": 0.10594922304153442,
92
+ "learning_rate": 0.0009254844753287492,
93
+ "loss": 2.4725,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.23853211009174313,
98
+ "grad_norm": 0.07520003616809845,
99
+ "learning_rate": 0.0009087513991633924,
100
+ "loss": 2.4258,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.25688073394495414,
105
+ "grad_norm": 0.06923802196979523,
106
+ "learning_rate": 0.0008905094792353917,
107
+ "loss": 2.3916,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.27522935779816515,
112
+ "grad_norm": 0.08765542507171631,
113
+ "learning_rate": 0.0008708260528239789,
114
+ "loss": 2.3764,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.29357798165137616,
119
+ "grad_norm": 0.07028616964817047,
120
+ "learning_rate": 0.0008497737783108237,
121
+ "loss": 2.3436,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.3119266055045872,
126
+ "grad_norm": 0.08140672743320465,
127
+ "learning_rate": 0.0008274303669726426,
128
+ "loss": 2.326,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.3302752293577982,
133
+ "grad_norm": 0.07616031169891357,
134
+ "learning_rate": 0.0008038782961218137,
135
+ "loss": 2.2943,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.3486238532110092,
140
+ "grad_norm": 0.08688798546791077,
141
+ "learning_rate": 0.000779204504653894,
142
+ "loss": 2.2724,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.3669724770642202,
147
+ "grad_norm": 0.08126090466976166,
148
+ "learning_rate": 0.0007535000721258791,
149
+ "loss": 2.2337,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.3853211009174312,
154
+ "grad_norm": 0.07281481474637985,
155
+ "learning_rate": 0.00072685988254984,
156
+ "loss": 2.2097,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.4036697247706422,
161
+ "grad_norm": 0.07721654325723648,
162
+ "learning_rate": 0.0006993822741429906,
163
+ "loss": 2.1964,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.42201834862385323,
168
+ "grad_norm": 0.07494688034057617,
169
+ "learning_rate": 0.0006711686763270818,
170
+ "loss": 2.1567,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.44036697247706424,
175
+ "grad_norm": 0.0644783228635788,
176
+ "learning_rate": 0.0006423232353170831,
177
+ "loss": 2.1307,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.45871559633027525,
182
+ "grad_norm": 0.06788823753595352,
183
+ "learning_rate": 0.0006129524296812335,
184
+ "loss": 2.1209,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.47706422018348627,
189
+ "grad_norm": 0.06672422587871552,
190
+ "learning_rate": 0.0005831646772915651,
191
+ "loss": 2.1123,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.4954128440366973,
196
+ "grad_norm": 0.06705792248249054,
197
+ "learning_rate": 0.0005530699351157809,
198
+ "loss": 2.0884,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.5137614678899083,
203
+ "grad_norm": 0.05863376334309578,
204
+ "learning_rate": 0.0005227792933277943,
205
+ "loss": 2.086,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.5321100917431193,
210
+ "grad_norm": 0.06323550641536713,
211
+ "learning_rate": 0.0004924045652352092,
212
+ "loss": 2.0702,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.5504587155963303,
217
+ "grad_norm": 0.06973816454410553,
218
+ "learning_rate": 0.0004620578745374633,
219
+ "loss": 2.0701,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.5688073394495413,
224
+ "grad_norm": 0.07226919382810593,
225
+ "learning_rate": 0.0004318512414382059,
226
+ "loss": 2.0726,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.5871559633027523,
231
+ "grad_norm": 0.05667873099446297,
232
+ "learning_rate": 0.0004018961691397148,
233
+ "loss": 2.0699,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.6055045871559633,
238
+ "grad_norm": 0.06372054666280746,
239
+ "learning_rate": 0.0003723032322457458,
240
+ "loss": 2.0534,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.6238532110091743,
245
+ "grad_norm": 0.07519938051700592,
246
+ "learning_rate": 0.0003431816685921625,
247
+ "loss": 2.0566,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.6422018348623854,
252
+ "grad_norm": 0.07054085284471512,
253
+ "learning_rate": 0.0003146389760120416,
254
+ "loss": 2.0514,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.6605504587155964,
259
+ "grad_norm": 0.07887300848960876,
260
+ "learning_rate": 0.00028678051552373485,
261
+ "loss": 2.0512,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.6788990825688074,
266
+ "grad_norm": 0.05539758875966072,
267
+ "learning_rate": 0.0002597091224066581,
268
+ "loss": 2.0451,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.6972477064220184,
273
+ "grad_norm": 0.05682758241891861,
274
+ "learning_rate": 0.0002335247266004629,
275
+ "loss": 2.0493,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.7155963302752294,
280
+ "grad_norm": 0.06325507909059525,
281
+ "learning_rate": 0.000208323983828828,
282
+ "loss": 2.04,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.7339449541284404,
287
+ "grad_norm": 0.057230085134506226,
288
+ "learning_rate": 0.0001841999188095224,
289
+ "loss": 2.0331,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.7522935779816514,
294
+ "grad_norm": 0.05684232339262962,
295
+ "learning_rate": 0.00016124158186777676,
296
+ "loss": 2.0383,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.7706422018348624,
301
+ "grad_norm": 0.05775289982557297,
302
+ "learning_rate": 0.0001395337202205211,
303
+ "loss": 2.0319,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.7889908256880734,
308
+ "grad_norm": 0.05256415158510208,
309
+ "learning_rate": 0.00011915646514489292,
310
+ "loss": 2.0309,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.8073394495412844,
315
+ "grad_norm": 0.054987646639347076,
316
+ "learning_rate": 0.00010018503618578818,
317
+ "loss": 2.0358,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.8256880733944955,
322
+ "grad_norm": 0.054050445556640625,
323
+ "learning_rate": 8.268946349432583e-05,
324
+ "loss": 2.0284,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.8440366972477065,
329
+ "grad_norm": 0.05575854703783989,
330
+ "learning_rate": 6.673432932217117e-05,
331
+ "loss": 2.0229,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.8623853211009175,
336
+ "grad_norm": 0.054133038967847824,
337
+ "learning_rate": 5.2378529625954684e-05,
338
+ "loss": 2.0161,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.8807339449541285,
343
+ "grad_norm": 0.052059952169656754,
344
+ "learning_rate": 3.9675056661785556e-05,
345
+ "loss": 2.0208,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.8990825688073395,
350
+ "grad_norm": 0.05479845777153969,
351
+ "learning_rate": 2.8670803372379538e-05,
352
+ "loss": 2.0247,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.9174311926605505,
357
+ "grad_norm": 0.0511295422911644,
358
+ "learning_rate": 1.9406390288876586e-05,
359
+ "loss": 2.029,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.9357798165137615,
364
+ "grad_norm": 0.05231447145342827,
365
+ "learning_rate": 1.191601558631461e-05,
366
+ "loss": 2.0215,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.9541284403669725,
371
+ "grad_norm": 0.0515405498445034,
372
+ "learning_rate": 6.2273288462559305e-06,
373
+ "loss": 2.0172,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.9724770642201835,
378
+ "grad_norm": 0.053000256419181824,
379
+ "learning_rate": 2.361328992552314e-06,
380
+ "loss": 2.019,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.9908256880733946,
385
+ "grad_norm": 0.052711524069309235,
386
+ "learning_rate": 3.322867770044202e-07,
387
+ "loss": 2.0178,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 1.0,
392
+ "step": 545,
393
+ "total_flos": 4.0818551392567296e+17,
394
+ "train_loss": 2.2614786585536573,
395
+ "train_runtime": 3771.4877,
396
+ "train_samples_per_second": 147.986,
397
+ "train_steps_per_second": 0.145
398
+ }
399
+ ],
400
+ "logging_steps": 10,
401
+ "max_steps": 545,
402
+ "num_input_tokens_seen": 33258880,
403
+ "num_train_epochs": 1,
404
+ "save_steps": 500,
405
+ "stateful_callbacks": {
406
+ "TrainerControl": {
407
+ "args": {
408
+ "should_epoch_stop": false,
409
+ "should_evaluate": false,
410
+ "should_log": false,
411
+ "should_save": false,
412
+ "should_training_stop": false
413
+ },
414
+ "attributes": {}
415
+ }
416
+ },
417
+ "total_flos": 4.0818551392567296e+17,
418
+ "train_batch_size": 32,
419
+ "trial_name": null,
420
+ "trial_params": null
421
+ }
continuous/smollm2/siglip2_sm16_256/finetune/chat_template.jinja ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
2
+ You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
3
+ ' }}{% endif %}{{'<|im_start|>' + message['role'] + '
4
+ ' + message['content'] + '<|im_end|>' + '
5
+ '}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
6
+ ' }}{% endif %}
continuous/smollm2/siglip2_sm16_256/finetune/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlavaLlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": 0.0001,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-16-SigLIP2-256",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm16_256.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm16_256/finetune/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": [
5
+ 2
6
+ ],
7
+ "pad_token_id": 0,
8
+ "transformers_version": "4.57.6"
9
+ }
continuous/smollm2/siglip2_sm16_256/finetune/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm16_256/finetune/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0c7c3b627f30cb22c8323f44fac714ece6dc86b1d0783ebafbe664ff25a6e33
3
+ size 4506158456
continuous/smollm2/siglip2_sm16_256/finetune/special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>"
5
+ ],
6
+ "bos_token": {
7
+ "content": "<|im_start|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "eos_token": {
14
+ "content": "<|im_end|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ },
20
+ "pad_token": "<|endoftext|>",
21
+ "unk_token": {
22
+ "content": "<|endoftext|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
continuous/smollm2/siglip2_sm16_256/finetune/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm16_256/finetune/tokenizer_config.json ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<repo_name>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "4": {
37
+ "content": "<reponame>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "5": {
45
+ "content": "<file_sep>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "6": {
53
+ "content": "<filename>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "7": {
61
+ "content": "<gh_stars>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "8": {
69
+ "content": "<issue_start>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "9": {
77
+ "content": "<issue_comment>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "10": {
85
+ "content": "<issue_closed>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "11": {
93
+ "content": "<jupyter_start>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "12": {
101
+ "content": "<jupyter_text>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "13": {
109
+ "content": "<jupyter_code>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "14": {
117
+ "content": "<jupyter_output>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": true
123
+ },
124
+ "15": {
125
+ "content": "<jupyter_script>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": true
131
+ },
132
+ "16": {
133
+ "content": "<empty_output>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": true
139
+ }
140
+ },
141
+ "additional_special_tokens": [
142
+ "<|im_start|>",
143
+ "<|im_end|>"
144
+ ],
145
+ "bos_token": "<|im_start|>",
146
+ "clean_up_tokenization_spaces": false,
147
+ "eos_token": "<|im_end|>",
148
+ "extra_special_tokens": {},
149
+ "model_max_length": 8192,
150
+ "pad_token": "<|endoftext|>",
151
+ "padding_side": "right",
152
+ "tokenizer_class": "GPT2Tokenizer",
153
+ "unk_token": "<|endoftext|>",
154
+ "vocab_size": 49152
155
+ }
continuous/smollm2/siglip2_sm16_256/finetune/trainer_state.json ADDED
@@ -0,0 +1,1856 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2597,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.003850596842510589,
14
+ "grad_norm": 3.2362518310546875,
15
+ "learning_rate": 1.153846153846154e-06,
16
+ "loss": 1.3038,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.007701193685021178,
21
+ "grad_norm": 2.0755655765533447,
22
+ "learning_rate": 2.435897435897436e-06,
23
+ "loss": 1.1384,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.011551790527531768,
28
+ "grad_norm": 1.8305031061172485,
29
+ "learning_rate": 3.7179487179487184e-06,
30
+ "loss": 0.9746,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.015402387370042356,
35
+ "grad_norm": 1.3183321952819824,
36
+ "learning_rate": 5e-06,
37
+ "loss": 0.9202,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.019252984212552945,
42
+ "grad_norm": 1.285186529159546,
43
+ "learning_rate": 6.282051282051282e-06,
44
+ "loss": 0.9053,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.023103581055063535,
49
+ "grad_norm": 0.25575950741767883,
50
+ "learning_rate": 7.564102564102564e-06,
51
+ "loss": 0.8337,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.026954177897574125,
56
+ "grad_norm": 0.21645641326904297,
57
+ "learning_rate": 8.846153846153847e-06,
58
+ "loss": 0.7894,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.030804774740084712,
63
+ "grad_norm": 0.22981122136116028,
64
+ "learning_rate": 9.999996111488723e-06,
65
+ "loss": 0.8034,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.0346553715825953,
70
+ "grad_norm": 0.20269900560379028,
71
+ "learning_rate": 9.999529497453782e-06,
72
+ "loss": 0.789,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.03850596842510589,
77
+ "grad_norm": 0.1891729235649109,
78
+ "learning_rate": 9.998285264324193e-06,
79
+ "loss": 0.7555,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.04235656526761648,
84
+ "grad_norm": 0.21256673336029053,
85
+ "learning_rate": 9.996263605626052e-06,
86
+ "loss": 0.7519,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.04620716211012707,
91
+ "grad_norm": 0.20167961716651917,
92
+ "learning_rate": 9.99346483580503e-06,
93
+ "loss": 0.7714,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.05005775895263766,
98
+ "grad_norm": 0.20235253870487213,
99
+ "learning_rate": 9.989889390177467e-06,
100
+ "loss": 0.7458,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.05390835579514825,
105
+ "grad_norm": 0.21410733461380005,
106
+ "learning_rate": 9.985537824862647e-06,
107
+ "loss": 0.7631,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.05775895263765884,
112
+ "grad_norm": 0.21909059584140778,
113
+ "learning_rate": 9.980410816696317e-06,
114
+ "loss": 0.7392,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.061609549480169425,
119
+ "grad_norm": 0.19526247680187225,
120
+ "learning_rate": 9.97450916312541e-06,
121
+ "loss": 0.7472,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.06546014632268002,
126
+ "grad_norm": 0.18816564977169037,
127
+ "learning_rate": 9.967833782084004e-06,
128
+ "loss": 0.7484,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.0693107431651906,
133
+ "grad_norm": 0.1872934252023697,
134
+ "learning_rate": 9.960385711850554e-06,
135
+ "loss": 0.7782,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.07316134000770119,
140
+ "grad_norm": 0.20371013879776,
141
+ "learning_rate": 9.952166110886394e-06,
142
+ "loss": 0.7648,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.07701193685021178,
147
+ "grad_norm": 0.19243837893009186,
148
+ "learning_rate": 9.943176257655567e-06,
149
+ "loss": 0.7632,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.08086253369272237,
154
+ "grad_norm": 0.19267037510871887,
155
+ "learning_rate": 9.933417550425952e-06,
156
+ "loss": 0.7551,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.08471313053523297,
161
+ "grad_norm": 0.1776629239320755,
162
+ "learning_rate": 9.9228915070518e-06,
163
+ "loss": 0.7393,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.08856372737774355,
168
+ "grad_norm": 0.18421541154384613,
169
+ "learning_rate": 9.911599764737632e-06,
170
+ "loss": 0.737,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.09241432422025414,
175
+ "grad_norm": 0.17718125879764557,
176
+ "learning_rate": 9.899544079783606e-06,
177
+ "loss": 0.7333,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.09626492106276473,
182
+ "grad_norm": 0.18649478256702423,
183
+ "learning_rate": 9.886726327312338e-06,
184
+ "loss": 0.7495,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.10011551790527531,
189
+ "grad_norm": 0.17066852748394012,
190
+ "learning_rate": 9.87314850097724e-06,
191
+ "loss": 0.7472,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.1039661147477859,
196
+ "grad_norm": 0.18825526535511017,
197
+ "learning_rate": 9.858812712652452e-06,
198
+ "loss": 0.7303,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.1078167115902965,
203
+ "grad_norm": 0.16291743516921997,
204
+ "learning_rate": 9.843721192104341e-06,
205
+ "loss": 0.7273,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.11166730843280709,
210
+ "grad_norm": 0.18610134720802307,
211
+ "learning_rate": 9.827876286644698e-06,
212
+ "loss": 0.7591,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.11551790527531768,
217
+ "grad_norm": 0.1641893982887268,
218
+ "learning_rate": 9.811280460765643e-06,
219
+ "loss": 0.7182,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.11936850211782826,
224
+ "grad_norm": 0.17247921228408813,
225
+ "learning_rate": 9.793936295756292e-06,
226
+ "loss": 0.7229,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.12321909896033885,
231
+ "grad_norm": 0.18496102094650269,
232
+ "learning_rate": 9.775846489301279e-06,
233
+ "loss": 0.7347,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.12706969580284944,
238
+ "grad_norm": 0.17405153810977936,
239
+ "learning_rate": 9.757013855061157e-06,
240
+ "loss": 0.7341,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.13092029264536004,
245
+ "grad_norm": 0.17510853707790375,
246
+ "learning_rate": 9.737441322234758e-06,
247
+ "loss": 0.737,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.1347708894878706,
252
+ "grad_norm": 0.19826702773571014,
253
+ "learning_rate": 9.717131935103605e-06,
254
+ "loss": 0.7436,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.1386214863303812,
259
+ "grad_norm": 0.18945351243019104,
260
+ "learning_rate": 9.696088852558398e-06,
261
+ "loss": 0.726,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.1424720831728918,
266
+ "grad_norm": 0.22103139758110046,
267
+ "learning_rate": 9.674315347607686e-06,
268
+ "loss": 0.7554,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.14632268001540238,
273
+ "grad_norm": 0.17334991693496704,
274
+ "learning_rate": 9.651814806868796e-06,
275
+ "loss": 0.7194,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.15017327685791299,
280
+ "grad_norm": 0.17932377755641937,
281
+ "learning_rate": 9.62859073004107e-06,
282
+ "loss": 0.7193,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.15402387370042356,
287
+ "grad_norm": 0.163352370262146,
288
+ "learning_rate": 9.604646729361547e-06,
289
+ "loss": 0.7267,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.15787447054293416,
294
+ "grad_norm": 0.2012857347726822,
295
+ "learning_rate": 9.579986529043099e-06,
296
+ "loss": 0.7226,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.16172506738544473,
301
+ "grad_norm": 0.18819113075733185,
302
+ "learning_rate": 9.554613964695189e-06,
303
+ "loss": 0.7408,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.16557566422795533,
308
+ "grad_norm": 0.18501833081245422,
309
+ "learning_rate": 9.528532982727281e-06,
310
+ "loss": 0.724,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.16942626107046593,
315
+ "grad_norm": 0.1765623241662979,
316
+ "learning_rate": 9.501747639735018e-06,
317
+ "loss": 0.7263,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.1732768579129765,
322
+ "grad_norm": 0.17408870160579681,
323
+ "learning_rate": 9.474262101869272e-06,
324
+ "loss": 0.714,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.1771274547554871,
329
+ "grad_norm": 0.1869042068719864,
330
+ "learning_rate": 9.446080644188141e-06,
331
+ "loss": 0.7231,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.18097805159799768,
336
+ "grad_norm": 0.1705714613199234,
337
+ "learning_rate": 9.417207649992017e-06,
338
+ "loss": 0.725,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.18482864844050828,
343
+ "grad_norm": 0.17837807536125183,
344
+ "learning_rate": 9.387647610141816e-06,
345
+ "loss": 0.7213,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.18867924528301888,
350
+ "grad_norm": 0.1750699132680893,
351
+ "learning_rate": 9.35740512236047e-06,
352
+ "loss": 0.7188,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.19252984212552945,
357
+ "grad_norm": 0.16972890496253967,
358
+ "learning_rate": 9.326484890517809e-06,
359
+ "loss": 0.7235,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.19638043896804006,
364
+ "grad_norm": 0.18386512994766235,
365
+ "learning_rate": 9.294891723898926e-06,
366
+ "loss": 0.7123,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.20023103581055063,
371
+ "grad_norm": 0.18339453637599945,
372
+ "learning_rate": 9.262630536456153e-06,
373
+ "loss": 0.7345,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.20408163265306123,
378
+ "grad_norm": 0.18077656626701355,
379
+ "learning_rate": 9.229706346044749e-06,
380
+ "loss": 0.7199,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.2079322294955718,
385
+ "grad_norm": 0.18190598487854004,
386
+ "learning_rate": 9.196124273642422e-06,
387
+ "loss": 0.728,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 0.2117828263380824,
392
+ "grad_norm": 0.18333284556865692,
393
+ "learning_rate": 9.161889542552842e-06,
394
+ "loss": 0.7368,
395
+ "step": 550
396
+ },
397
+ {
398
+ "epoch": 0.215633423180593,
399
+ "grad_norm": 0.1707497090101242,
400
+ "learning_rate": 9.127007477593193e-06,
401
+ "loss": 0.7368,
402
+ "step": 560
403
+ },
404
+ {
405
+ "epoch": 0.21948402002310358,
406
+ "grad_norm": 0.18913555145263672,
407
+ "learning_rate": 9.091483504265974e-06,
408
+ "loss": 0.747,
409
+ "step": 570
410
+ },
411
+ {
412
+ "epoch": 0.22333461686561418,
413
+ "grad_norm": 0.18646511435508728,
414
+ "learning_rate": 9.055323147915116e-06,
415
+ "loss": 0.7183,
416
+ "step": 580
417
+ },
418
+ {
419
+ "epoch": 0.22718521370812475,
420
+ "grad_norm": 0.17434298992156982,
421
+ "learning_rate": 9.018532032866585e-06,
422
+ "loss": 0.7306,
423
+ "step": 590
424
+ },
425
+ {
426
+ "epoch": 0.23103581055063535,
427
+ "grad_norm": 0.16352099180221558,
428
+ "learning_rate": 8.981115881553579e-06,
429
+ "loss": 0.7282,
430
+ "step": 600
431
+ },
432
+ {
433
+ "epoch": 0.23488640739314592,
434
+ "grad_norm": 0.1897197961807251,
435
+ "learning_rate": 8.943080513626476e-06,
436
+ "loss": 0.7476,
437
+ "step": 610
438
+ },
439
+ {
440
+ "epoch": 0.23873700423565652,
441
+ "grad_norm": 0.16767485439777374,
442
+ "learning_rate": 8.904431845047641e-06,
443
+ "loss": 0.7277,
444
+ "step": 620
445
+ },
446
+ {
447
+ "epoch": 0.24258760107816713,
448
+ "grad_norm": 0.17626842856407166,
449
+ "learning_rate": 8.86517588717128e-06,
450
+ "loss": 0.7383,
451
+ "step": 630
452
+ },
453
+ {
454
+ "epoch": 0.2464381979206777,
455
+ "grad_norm": 0.18332673609256744,
456
+ "learning_rate": 8.82531874580844e-06,
457
+ "loss": 0.7405,
458
+ "step": 640
459
+ },
460
+ {
461
+ "epoch": 0.25028879476318827,
462
+ "grad_norm": 0.16625626385211945,
463
+ "learning_rate": 8.784866620277314e-06,
464
+ "loss": 0.6992,
465
+ "step": 650
466
+ },
467
+ {
468
+ "epoch": 0.25413939160569887,
469
+ "grad_norm": 0.17100585997104645,
470
+ "learning_rate": 8.743825802439018e-06,
471
+ "loss": 0.7217,
472
+ "step": 660
473
+ },
474
+ {
475
+ "epoch": 0.2579899884482095,
476
+ "grad_norm": 0.1694749891757965,
477
+ "learning_rate": 8.702202675718956e-06,
478
+ "loss": 0.735,
479
+ "step": 670
480
+ },
481
+ {
482
+ "epoch": 0.2618405852907201,
483
+ "grad_norm": 0.16253964602947235,
484
+ "learning_rate": 8.660003714113959e-06,
485
+ "loss": 0.7135,
486
+ "step": 680
487
+ },
488
+ {
489
+ "epoch": 0.2656911821332307,
490
+ "grad_norm": 0.17892444133758545,
491
+ "learning_rate": 8.617235481185323e-06,
492
+ "loss": 0.7232,
493
+ "step": 690
494
+ },
495
+ {
496
+ "epoch": 0.2695417789757412,
497
+ "grad_norm": 0.18108415603637695,
498
+ "learning_rate": 8.573904629037931e-06,
499
+ "loss": 0.7367,
500
+ "step": 700
501
+ },
502
+ {
503
+ "epoch": 0.2733923758182518,
504
+ "grad_norm": 0.16438351571559906,
505
+ "learning_rate": 8.530017897285585e-06,
506
+ "loss": 0.7196,
507
+ "step": 710
508
+ },
509
+ {
510
+ "epoch": 0.2772429726607624,
511
+ "grad_norm": 0.17046120762825012,
512
+ "learning_rate": 8.485582112002746e-06,
513
+ "loss": 0.7204,
514
+ "step": 720
515
+ },
516
+ {
517
+ "epoch": 0.281093569503273,
518
+ "grad_norm": 0.17339079082012177,
519
+ "learning_rate": 8.440604184662811e-06,
520
+ "loss": 0.7274,
521
+ "step": 730
522
+ },
523
+ {
524
+ "epoch": 0.2849441663457836,
525
+ "grad_norm": 0.1738845705986023,
526
+ "learning_rate": 8.395091111063118e-06,
527
+ "loss": 0.7141,
528
+ "step": 740
529
+ },
530
+ {
531
+ "epoch": 0.28879476318829417,
532
+ "grad_norm": 0.16067427396774292,
533
+ "learning_rate": 8.349049970236822e-06,
534
+ "loss": 0.7126,
535
+ "step": 750
536
+ },
537
+ {
538
+ "epoch": 0.29264536003080477,
539
+ "grad_norm": 0.1670185923576355,
540
+ "learning_rate": 8.302487923351842e-06,
541
+ "loss": 0.7077,
542
+ "step": 760
543
+ },
544
+ {
545
+ "epoch": 0.29649595687331537,
546
+ "grad_norm": 0.15916778147220612,
547
+ "learning_rate": 8.255412212597021e-06,
548
+ "loss": 0.727,
549
+ "step": 770
550
+ },
551
+ {
552
+ "epoch": 0.30034655371582597,
553
+ "grad_norm": 0.17583651840686798,
554
+ "learning_rate": 8.207830160055688e-06,
555
+ "loss": 0.7236,
556
+ "step": 780
557
+ },
558
+ {
559
+ "epoch": 0.3041971505583365,
560
+ "grad_norm": 0.1704917699098587,
561
+ "learning_rate": 8.159749166566786e-06,
562
+ "loss": 0.7049,
563
+ "step": 790
564
+ },
565
+ {
566
+ "epoch": 0.3080477474008471,
567
+ "grad_norm": 0.1804777830839157,
568
+ "learning_rate": 8.11117671057377e-06,
569
+ "loss": 0.7264,
570
+ "step": 800
571
+ },
572
+ {
573
+ "epoch": 0.3118983442433577,
574
+ "grad_norm": 0.16584521532058716,
575
+ "learning_rate": 8.062120346961404e-06,
576
+ "loss": 0.724,
577
+ "step": 810
578
+ },
579
+ {
580
+ "epoch": 0.3157489410858683,
581
+ "grad_norm": 0.1655588299036026,
582
+ "learning_rate": 8.012587705880705e-06,
583
+ "loss": 0.7204,
584
+ "step": 820
585
+ },
586
+ {
587
+ "epoch": 0.3195995379283789,
588
+ "grad_norm": 0.18444493412971497,
589
+ "learning_rate": 7.962586491562146e-06,
590
+ "loss": 0.7084,
591
+ "step": 830
592
+ },
593
+ {
594
+ "epoch": 0.32345013477088946,
595
+ "grad_norm": 0.16677528619766235,
596
+ "learning_rate": 7.912124481117357e-06,
597
+ "loss": 0.7079,
598
+ "step": 840
599
+ },
600
+ {
601
+ "epoch": 0.32730073161340006,
602
+ "grad_norm": 0.16463375091552734,
603
+ "learning_rate": 7.861209523329494e-06,
604
+ "loss": 0.7209,
605
+ "step": 850
606
+ },
607
+ {
608
+ "epoch": 0.33115132845591067,
609
+ "grad_norm": 0.17583778500556946,
610
+ "learning_rate": 7.809849537432432e-06,
611
+ "loss": 0.7099,
612
+ "step": 860
613
+ },
614
+ {
615
+ "epoch": 0.33500192529842127,
616
+ "grad_norm": 0.16385096311569214,
617
+ "learning_rate": 7.758052511879035e-06,
618
+ "loss": 0.7124,
619
+ "step": 870
620
+ },
621
+ {
622
+ "epoch": 0.33885252214093187,
623
+ "grad_norm": 0.17329329252243042,
624
+ "learning_rate": 7.705826503098642e-06,
625
+ "loss": 0.7167,
626
+ "step": 880
627
+ },
628
+ {
629
+ "epoch": 0.3427031189834424,
630
+ "grad_norm": 0.17180101573467255,
631
+ "learning_rate": 7.653179634243979e-06,
632
+ "loss": 0.7113,
633
+ "step": 890
634
+ },
635
+ {
636
+ "epoch": 0.346553715825953,
637
+ "grad_norm": 0.16084113717079163,
638
+ "learning_rate": 7.600120093927696e-06,
639
+ "loss": 0.7097,
640
+ "step": 900
641
+ },
642
+ {
643
+ "epoch": 0.3504043126684636,
644
+ "grad_norm": 0.195755273103714,
645
+ "learning_rate": 7.546656134948729e-06,
646
+ "loss": 0.7012,
647
+ "step": 910
648
+ },
649
+ {
650
+ "epoch": 0.3542549095109742,
651
+ "grad_norm": 0.165066659450531,
652
+ "learning_rate": 7.492796073008661e-06,
653
+ "loss": 0.7126,
654
+ "step": 920
655
+ },
656
+ {
657
+ "epoch": 0.3581055063534848,
658
+ "grad_norm": 0.18719570338726044,
659
+ "learning_rate": 7.438548285418326e-06,
660
+ "loss": 0.6923,
661
+ "step": 930
662
+ },
663
+ {
664
+ "epoch": 0.36195610319599536,
665
+ "grad_norm": 0.17265130579471588,
666
+ "learning_rate": 7.3839212097948e-06,
667
+ "loss": 0.7235,
668
+ "step": 940
669
+ },
670
+ {
671
+ "epoch": 0.36580670003850596,
672
+ "grad_norm": 0.17333748936653137,
673
+ "learning_rate": 7.328923342749039e-06,
674
+ "loss": 0.7411,
675
+ "step": 950
676
+ },
677
+ {
678
+ "epoch": 0.36965729688101656,
679
+ "grad_norm": 0.16900794208049774,
680
+ "learning_rate": 7.273563238564319e-06,
681
+ "loss": 0.7355,
682
+ "step": 960
683
+ },
684
+ {
685
+ "epoch": 0.37350789372352716,
686
+ "grad_norm": 0.1611638069152832,
687
+ "learning_rate": 7.217849507865724e-06,
688
+ "loss": 0.7014,
689
+ "step": 970
690
+ },
691
+ {
692
+ "epoch": 0.37735849056603776,
693
+ "grad_norm": 0.16548675298690796,
694
+ "learning_rate": 7.161790816280858e-06,
695
+ "loss": 0.7146,
696
+ "step": 980
697
+ },
698
+ {
699
+ "epoch": 0.3812090874085483,
700
+ "grad_norm": 0.17532268166542053,
701
+ "learning_rate": 7.1053958830920125e-06,
702
+ "loss": 0.6968,
703
+ "step": 990
704
+ },
705
+ {
706
+ "epoch": 0.3850596842510589,
707
+ "grad_norm": 0.17464636266231537,
708
+ "learning_rate": 7.0486734798799715e-06,
709
+ "loss": 0.7123,
710
+ "step": 1000
711
+ },
712
+ {
713
+ "epoch": 0.3889102810935695,
714
+ "grad_norm": 0.16563694179058075,
715
+ "learning_rate": 6.991632429159708e-06,
716
+ "loss": 0.7321,
717
+ "step": 1010
718
+ },
719
+ {
720
+ "epoch": 0.3927608779360801,
721
+ "grad_norm": 0.1795864999294281,
722
+ "learning_rate": 6.934281603008132e-06,
723
+ "loss": 0.7283,
724
+ "step": 1020
725
+ },
726
+ {
727
+ "epoch": 0.39661147477859066,
728
+ "grad_norm": 0.1535368263721466,
729
+ "learning_rate": 6.876629921684143e-06,
730
+ "loss": 0.7246,
731
+ "step": 1030
732
+ },
733
+ {
734
+ "epoch": 0.40046207162110126,
735
+ "grad_norm": 0.1832147240638733,
736
+ "learning_rate": 6.818686352241199e-06,
737
+ "loss": 0.712,
738
+ "step": 1040
739
+ },
740
+ {
741
+ "epoch": 0.40431266846361186,
742
+ "grad_norm": 0.157639741897583,
743
+ "learning_rate": 6.760459907132573e-06,
744
+ "loss": 0.706,
745
+ "step": 1050
746
+ },
747
+ {
748
+ "epoch": 0.40816326530612246,
749
+ "grad_norm": 0.16029301285743713,
750
+ "learning_rate": 6.701959642809589e-06,
751
+ "loss": 0.7123,
752
+ "step": 1060
753
+ },
754
+ {
755
+ "epoch": 0.41201386214863306,
756
+ "grad_norm": 0.17438718676567078,
757
+ "learning_rate": 6.643194658312987e-06,
758
+ "loss": 0.7026,
759
+ "step": 1070
760
+ },
761
+ {
762
+ "epoch": 0.4158644589911436,
763
+ "grad_norm": 0.1521425098180771,
764
+ "learning_rate": 6.584174093857676e-06,
765
+ "loss": 0.7317,
766
+ "step": 1080
767
+ },
768
+ {
769
+ "epoch": 0.4197150558336542,
770
+ "grad_norm": 0.1870277374982834,
771
+ "learning_rate": 6.5249071294110735e-06,
772
+ "loss": 0.7145,
773
+ "step": 1090
774
+ },
775
+ {
776
+ "epoch": 0.4235656526761648,
777
+ "grad_norm": 0.17341867089271545,
778
+ "learning_rate": 6.465402983265274e-06,
779
+ "loss": 0.7253,
780
+ "step": 1100
781
+ },
782
+ {
783
+ "epoch": 0.4274162495186754,
784
+ "grad_norm": 0.17737151682376862,
785
+ "learning_rate": 6.405670910603243e-06,
786
+ "loss": 0.7169,
787
+ "step": 1110
788
+ },
789
+ {
790
+ "epoch": 0.431266846361186,
791
+ "grad_norm": 0.17929796874523163,
792
+ "learning_rate": 6.345720202059291e-06,
793
+ "loss": 0.7146,
794
+ "step": 1120
795
+ },
796
+ {
797
+ "epoch": 0.43511744320369655,
798
+ "grad_norm": 0.17237421870231628,
799
+ "learning_rate": 6.285560182274008e-06,
800
+ "loss": 0.6952,
801
+ "step": 1130
802
+ },
803
+ {
804
+ "epoch": 0.43896804004620715,
805
+ "grad_norm": 0.1828005611896515,
806
+ "learning_rate": 6.225200208443936e-06,
807
+ "loss": 0.7205,
808
+ "step": 1140
809
+ },
810
+ {
811
+ "epoch": 0.44281863688871775,
812
+ "grad_norm": 0.17308096587657928,
813
+ "learning_rate": 6.16464966886616e-06,
814
+ "loss": 0.7043,
815
+ "step": 1150
816
+ },
817
+ {
818
+ "epoch": 0.44666923373122835,
819
+ "grad_norm": 0.19987842440605164,
820
+ "learning_rate": 6.103917981478065e-06,
821
+ "loss": 0.7347,
822
+ "step": 1160
823
+ },
824
+ {
825
+ "epoch": 0.45051983057373896,
826
+ "grad_norm": 0.16362804174423218,
827
+ "learning_rate": 6.043014592392489e-06,
828
+ "loss": 0.7406,
829
+ "step": 1170
830
+ },
831
+ {
832
+ "epoch": 0.4543704274162495,
833
+ "grad_norm": 0.16938254237174988,
834
+ "learning_rate": 5.981948974428485e-06,
835
+ "loss": 0.722,
836
+ "step": 1180
837
+ },
838
+ {
839
+ "epoch": 0.4582210242587601,
840
+ "grad_norm": 0.1641385704278946,
841
+ "learning_rate": 5.920730625637934e-06,
842
+ "loss": 0.6955,
843
+ "step": 1190
844
+ },
845
+ {
846
+ "epoch": 0.4620716211012707,
847
+ "grad_norm": 0.17055128514766693,
848
+ "learning_rate": 5.8593690678282355e-06,
849
+ "loss": 0.7076,
850
+ "step": 1200
851
+ },
852
+ {
853
+ "epoch": 0.4659222179437813,
854
+ "grad_norm": 0.16328445076942444,
855
+ "learning_rate": 5.797873845081301e-06,
856
+ "loss": 0.7047,
857
+ "step": 1210
858
+ },
859
+ {
860
+ "epoch": 0.46977281478629185,
861
+ "grad_norm": 0.17706719040870667,
862
+ "learning_rate": 5.736254522269075e-06,
863
+ "loss": 0.7289,
864
+ "step": 1220
865
+ },
866
+ {
867
+ "epoch": 0.47362341162880245,
868
+ "grad_norm": 0.16769511997699738,
869
+ "learning_rate": 5.6745206835658386e-06,
870
+ "loss": 0.7267,
871
+ "step": 1230
872
+ },
873
+ {
874
+ "epoch": 0.47747400847131305,
875
+ "grad_norm": 0.16450582444667816,
876
+ "learning_rate": 5.612681930957495e-06,
877
+ "loss": 0.7057,
878
+ "step": 1240
879
+ },
880
+ {
881
+ "epoch": 0.48132460531382365,
882
+ "grad_norm": 0.16470426321029663,
883
+ "learning_rate": 5.550747882748095e-06,
884
+ "loss": 0.6952,
885
+ "step": 1250
886
+ },
887
+ {
888
+ "epoch": 0.48517520215633425,
889
+ "grad_norm": 0.18448762595653534,
890
+ "learning_rate": 5.4887281720638195e-06,
891
+ "loss": 0.7123,
892
+ "step": 1260
893
+ },
894
+ {
895
+ "epoch": 0.4890257989988448,
896
+ "grad_norm": 0.17835403978824615,
897
+ "learning_rate": 5.42663244535466e-06,
898
+ "loss": 0.6869,
899
+ "step": 1270
900
+ },
901
+ {
902
+ "epoch": 0.4928763958413554,
903
+ "grad_norm": 0.17476056516170502,
904
+ "learning_rate": 5.3644703608940205e-06,
905
+ "loss": 0.7053,
906
+ "step": 1280
907
+ },
908
+ {
909
+ "epoch": 0.496726992683866,
910
+ "grad_norm": 0.16679257154464722,
911
+ "learning_rate": 5.302251587276491e-06,
912
+ "loss": 0.7163,
913
+ "step": 1290
914
+ },
915
+ {
916
+ "epoch": 0.5005775895263765,
917
+ "grad_norm": 0.1622665822505951,
918
+ "learning_rate": 5.2399858019140005e-06,
919
+ "loss": 0.7067,
920
+ "step": 1300
921
+ },
922
+ {
923
+ "epoch": 0.5044281863688872,
924
+ "grad_norm": 0.17537492513656616,
925
+ "learning_rate": 5.177682689530615e-06,
926
+ "loss": 0.7295,
927
+ "step": 1310
928
+ },
929
+ {
930
+ "epoch": 0.5082787832113977,
931
+ "grad_norm": 0.18007946014404297,
932
+ "learning_rate": 5.11535194065619e-06,
933
+ "loss": 0.6955,
934
+ "step": 1320
935
+ },
936
+ {
937
+ "epoch": 0.5121293800539084,
938
+ "grad_norm": 0.1903083175420761,
939
+ "learning_rate": 5.053003250119113e-06,
940
+ "loss": 0.7026,
941
+ "step": 1330
942
+ },
943
+ {
944
+ "epoch": 0.515979976896419,
945
+ "grad_norm": 0.17099252343177795,
946
+ "learning_rate": 4.9906463155383925e-06,
947
+ "loss": 0.7103,
948
+ "step": 1340
949
+ },
950
+ {
951
+ "epoch": 0.5198305737389295,
952
+ "grad_norm": 0.17034003138542175,
953
+ "learning_rate": 4.9282908358153046e-06,
954
+ "loss": 0.71,
955
+ "step": 1350
956
+ },
957
+ {
958
+ "epoch": 0.5236811705814401,
959
+ "grad_norm": 0.18668518960475922,
960
+ "learning_rate": 4.8659465096248345e-06,
961
+ "loss": 0.7229,
962
+ "step": 1360
963
+ },
964
+ {
965
+ "epoch": 0.5275317674239507,
966
+ "grad_norm": 0.18048568069934845,
967
+ "learning_rate": 4.803623033907169e-06,
968
+ "loss": 0.7133,
969
+ "step": 1370
970
+ },
971
+ {
972
+ "epoch": 0.5313823642664613,
973
+ "grad_norm": 0.16927576065063477,
974
+ "learning_rate": 4.741330102359441e-06,
975
+ "loss": 0.7286,
976
+ "step": 1380
977
+ },
978
+ {
979
+ "epoch": 0.5352329611089719,
980
+ "grad_norm": 0.17502231895923615,
981
+ "learning_rate": 4.679077403927996e-06,
982
+ "loss": 0.7034,
983
+ "step": 1390
984
+ },
985
+ {
986
+ "epoch": 0.5390835579514824,
987
+ "grad_norm": 0.16677217185497284,
988
+ "learning_rate": 4.616874621301373e-06,
989
+ "loss": 0.7055,
990
+ "step": 1400
991
+ },
992
+ {
993
+ "epoch": 0.5429341547939931,
994
+ "grad_norm": 0.1794423907995224,
995
+ "learning_rate": 4.554731429404293e-06,
996
+ "loss": 0.7262,
997
+ "step": 1410
998
+ },
999
+ {
1000
+ "epoch": 0.5467847516365036,
1001
+ "grad_norm": 0.17891331017017365,
1002
+ "learning_rate": 4.492657493892822e-06,
1003
+ "loss": 0.705,
1004
+ "step": 1420
1005
+ },
1006
+ {
1007
+ "epoch": 0.5506353484790143,
1008
+ "grad_norm": 0.1732056885957718,
1009
+ "learning_rate": 4.430662469650999e-06,
1010
+ "loss": 0.7193,
1011
+ "step": 1430
1012
+ },
1013
+ {
1014
+ "epoch": 0.5544859453215248,
1015
+ "grad_norm": 0.17596466839313507,
1016
+ "learning_rate": 4.368755999289118e-06,
1017
+ "loss": 0.7219,
1018
+ "step": 1440
1019
+ },
1020
+ {
1021
+ "epoch": 0.5583365421640354,
1022
+ "grad_norm": 0.18150314688682556,
1023
+ "learning_rate": 4.306947711643949e-06,
1024
+ "loss": 0.7053,
1025
+ "step": 1450
1026
+ },
1027
+ {
1028
+ "epoch": 0.562187139006546,
1029
+ "grad_norm": 0.17256776988506317,
1030
+ "learning_rate": 4.245247220281068e-06,
1031
+ "loss": 0.7101,
1032
+ "step": 1460
1033
+ },
1034
+ {
1035
+ "epoch": 0.5660377358490566,
1036
+ "grad_norm": 0.1705535352230072,
1037
+ "learning_rate": 4.183664121999587e-06,
1038
+ "loss": 0.7034,
1039
+ "step": 1470
1040
+ },
1041
+ {
1042
+ "epoch": 0.5698883326915672,
1043
+ "grad_norm": 0.17244026064872742,
1044
+ "learning_rate": 4.122207995339474e-06,
1045
+ "loss": 0.6995,
1046
+ "step": 1480
1047
+ },
1048
+ {
1049
+ "epoch": 0.5737389295340778,
1050
+ "grad_norm": 0.16721037030220032,
1051
+ "learning_rate": 4.060888399091734e-06,
1052
+ "loss": 0.7079,
1053
+ "step": 1490
1054
+ },
1055
+ {
1056
+ "epoch": 0.5775895263765883,
1057
+ "grad_norm": 0.17194902896881104,
1058
+ "learning_rate": 3.999714870811639e-06,
1059
+ "loss": 0.7204,
1060
+ "step": 1500
1061
+ },
1062
+ {
1063
+ "epoch": 0.581440123219099,
1064
+ "grad_norm": 0.1716313362121582,
1065
+ "learning_rate": 3.938696925335275e-06,
1066
+ "loss": 0.7132,
1067
+ "step": 1510
1068
+ },
1069
+ {
1070
+ "epoch": 0.5852907200616095,
1071
+ "grad_norm": 0.16842377185821533,
1072
+ "learning_rate": 3.87784405329962e-06,
1073
+ "loss": 0.7212,
1074
+ "step": 1520
1075
+ },
1076
+ {
1077
+ "epoch": 0.5891413169041202,
1078
+ "grad_norm": 0.16351418197155,
1079
+ "learning_rate": 3.817165719666382e-06,
1080
+ "loss": 0.7196,
1081
+ "step": 1530
1082
+ },
1083
+ {
1084
+ "epoch": 0.5929919137466307,
1085
+ "grad_norm": 0.17018556594848633,
1086
+ "learning_rate": 3.756671362249841e-06,
1087
+ "loss": 0.7249,
1088
+ "step": 1540
1089
+ },
1090
+ {
1091
+ "epoch": 0.5968425105891413,
1092
+ "grad_norm": 0.16127856075763702,
1093
+ "learning_rate": 3.6963703902488907e-06,
1094
+ "loss": 0.7108,
1095
+ "step": 1550
1096
+ },
1097
+ {
1098
+ "epoch": 0.6006931074316519,
1099
+ "grad_norm": 0.1657448709011078,
1100
+ "learning_rate": 3.6362721827835655e-06,
1101
+ "loss": 0.7187,
1102
+ "step": 1560
1103
+ },
1104
+ {
1105
+ "epoch": 0.6045437042741625,
1106
+ "grad_norm": 0.18213684856891632,
1107
+ "learning_rate": 3.576386087436212e-06,
1108
+ "loss": 0.7275,
1109
+ "step": 1570
1110
+ },
1111
+ {
1112
+ "epoch": 0.608394301116673,
1113
+ "grad_norm": 0.17878413200378418,
1114
+ "learning_rate": 3.5167214187975913e-06,
1115
+ "loss": 0.7189,
1116
+ "step": 1580
1117
+ },
1118
+ {
1119
+ "epoch": 0.6122448979591837,
1120
+ "grad_norm": 0.15755130350589752,
1121
+ "learning_rate": 3.457287457018086e-06,
1122
+ "loss": 0.7143,
1123
+ "step": 1590
1124
+ },
1125
+ {
1126
+ "epoch": 0.6160954948016942,
1127
+ "grad_norm": 0.17290335893630981,
1128
+ "learning_rate": 3.3980934463643003e-06,
1129
+ "loss": 0.7116,
1130
+ "step": 1600
1131
+ },
1132
+ {
1133
+ "epoch": 0.6199460916442049,
1134
+ "grad_norm": 0.15915852785110474,
1135
+ "learning_rate": 3.3391485937812053e-06,
1136
+ "loss": 0.6885,
1137
+ "step": 1610
1138
+ },
1139
+ {
1140
+ "epoch": 0.6237966884867154,
1141
+ "grad_norm": 0.1713958978652954,
1142
+ "learning_rate": 3.280462067460115e-06,
1143
+ "loss": 0.6893,
1144
+ "step": 1620
1145
+ },
1146
+ {
1147
+ "epoch": 0.627647285329226,
1148
+ "grad_norm": 0.17184418439865112,
1149
+ "learning_rate": 3.222042995412669e-06,
1150
+ "loss": 0.7144,
1151
+ "step": 1630
1152
+ },
1153
+ {
1154
+ "epoch": 0.6314978821717366,
1155
+ "grad_norm": 0.1632545292377472,
1156
+ "learning_rate": 3.163900464051089e-06,
1157
+ "loss": 0.7054,
1158
+ "step": 1640
1159
+ },
1160
+ {
1161
+ "epoch": 0.6353484790142472,
1162
+ "grad_norm": 0.1811217963695526,
1163
+ "learning_rate": 3.1060435167748826e-06,
1164
+ "loss": 0.7208,
1165
+ "step": 1650
1166
+ },
1167
+ {
1168
+ "epoch": 0.6391990758567578,
1169
+ "grad_norm": 0.16855701804161072,
1170
+ "learning_rate": 3.048481152564253e-06,
1171
+ "loss": 0.6988,
1172
+ "step": 1660
1173
+ },
1174
+ {
1175
+ "epoch": 0.6430496726992684,
1176
+ "grad_norm": 0.1804233342409134,
1177
+ "learning_rate": 2.9912223245804017e-06,
1178
+ "loss": 0.7164,
1179
+ "step": 1670
1180
+ },
1181
+ {
1182
+ "epoch": 0.6469002695417789,
1183
+ "grad_norm": 0.1720559000968933,
1184
+ "learning_rate": 2.934275938772981e-06,
1185
+ "loss": 0.7064,
1186
+ "step": 1680
1187
+ },
1188
+ {
1189
+ "epoch": 0.6507508663842896,
1190
+ "grad_norm": 0.17344790697097778,
1191
+ "learning_rate": 2.8776508524948676e-06,
1192
+ "loss": 0.7014,
1193
+ "step": 1690
1194
+ },
1195
+ {
1196
+ "epoch": 0.6546014632268001,
1197
+ "grad_norm": 0.16272534430027008,
1198
+ "learning_rate": 2.821355873124503e-06,
1199
+ "loss": 0.7037,
1200
+ "step": 1700
1201
+ },
1202
+ {
1203
+ "epoch": 0.6584520600693108,
1204
+ "grad_norm": 0.17318332195281982,
1205
+ "learning_rate": 2.7653997566960144e-06,
1206
+ "loss": 0.7123,
1207
+ "step": 1710
1208
+ },
1209
+ {
1210
+ "epoch": 0.6623026569118213,
1211
+ "grad_norm": 0.18791598081588745,
1212
+ "learning_rate": 2.709791206537309e-06,
1213
+ "loss": 0.7052,
1214
+ "step": 1720
1215
+ },
1216
+ {
1217
+ "epoch": 0.6661532537543319,
1218
+ "grad_norm": 0.18518894910812378,
1219
+ "learning_rate": 2.6545388719163747e-06,
1220
+ "loss": 0.706,
1221
+ "step": 1730
1222
+ },
1223
+ {
1224
+ "epoch": 0.6700038505968425,
1225
+ "grad_norm": 0.16961555182933807,
1226
+ "learning_rate": 2.599651346695979e-06,
1227
+ "loss": 0.7159,
1228
+ "step": 1740
1229
+ },
1230
+ {
1231
+ "epoch": 0.6738544474393531,
1232
+ "grad_norm": 0.1653808206319809,
1233
+ "learning_rate": 2.545137167997007e-06,
1234
+ "loss": 0.7138,
1235
+ "step": 1750
1236
+ },
1237
+ {
1238
+ "epoch": 0.6777050442818637,
1239
+ "grad_norm": 0.1633240133523941,
1240
+ "learning_rate": 2.4910048148705972e-06,
1241
+ "loss": 0.7021,
1242
+ "step": 1760
1243
+ },
1244
+ {
1245
+ "epoch": 0.6815556411243743,
1246
+ "grad_norm": 0.17832452058792114,
1247
+ "learning_rate": 2.4372627069793327e-06,
1248
+ "loss": 0.7401,
1249
+ "step": 1770
1250
+ },
1251
+ {
1252
+ "epoch": 0.6854062379668848,
1253
+ "grad_norm": 0.16505220532417297,
1254
+ "learning_rate": 2.3839192032876486e-06,
1255
+ "loss": 0.7156,
1256
+ "step": 1780
1257
+ },
1258
+ {
1259
+ "epoch": 0.6892568348093955,
1260
+ "grad_norm": 0.17864443361759186,
1261
+ "learning_rate": 2.3309826007617047e-06,
1262
+ "loss": 0.7316,
1263
+ "step": 1790
1264
+ },
1265
+ {
1266
+ "epoch": 0.693107431651906,
1267
+ "grad_norm": 0.1664564609527588,
1268
+ "learning_rate": 2.278461133078881e-06,
1269
+ "loss": 0.7002,
1270
+ "step": 1800
1271
+ },
1272
+ {
1273
+ "epoch": 0.6969580284944167,
1274
+ "grad_norm": 0.17273657023906708,
1275
+ "learning_rate": 2.2263629693471258e-06,
1276
+ "loss": 0.693,
1277
+ "step": 1810
1278
+ },
1279
+ {
1280
+ "epoch": 0.7008086253369272,
1281
+ "grad_norm": 0.16442269086837769,
1282
+ "learning_rate": 2.1746962128343414e-06,
1283
+ "loss": 0.7151,
1284
+ "step": 1820
1285
+ },
1286
+ {
1287
+ "epoch": 0.7046592221794378,
1288
+ "grad_norm": 0.1783774495124817,
1289
+ "learning_rate": 2.1234688997080183e-06,
1290
+ "loss": 0.6999,
1291
+ "step": 1830
1292
+ },
1293
+ {
1294
+ "epoch": 0.7085098190219484,
1295
+ "grad_norm": 0.16475073993206024,
1296
+ "learning_rate": 2.0726889977853053e-06,
1297
+ "loss": 0.7055,
1298
+ "step": 1840
1299
+ },
1300
+ {
1301
+ "epoch": 0.712360415864459,
1302
+ "grad_norm": 0.17759501934051514,
1303
+ "learning_rate": 2.022364405293703e-06,
1304
+ "loss": 0.7208,
1305
+ "step": 1850
1306
+ },
1307
+ {
1308
+ "epoch": 0.7162110127069696,
1309
+ "grad_norm": 0.16667591035366058,
1310
+ "learning_rate": 1.9725029496425814e-06,
1311
+ "loss": 0.7175,
1312
+ "step": 1860
1313
+ },
1314
+ {
1315
+ "epoch": 0.7200616095494802,
1316
+ "grad_norm": 0.16606342792510986,
1317
+ "learning_rate": 1.923112386205727e-06,
1318
+ "loss": 0.7089,
1319
+ "step": 1870
1320
+ },
1321
+ {
1322
+ "epoch": 0.7239122063919907,
1323
+ "grad_norm": 0.17656585574150085,
1324
+ "learning_rate": 1.874200397115079e-06,
1325
+ "loss": 0.7191,
1326
+ "step": 1880
1327
+ },
1328
+ {
1329
+ "epoch": 0.7277628032345014,
1330
+ "grad_norm": 0.17283694446086884,
1331
+ "learning_rate": 1.825774590065857e-06,
1332
+ "loss": 0.7042,
1333
+ "step": 1890
1334
+ },
1335
+ {
1336
+ "epoch": 0.7316134000770119,
1337
+ "grad_norm": 0.17800167202949524,
1338
+ "learning_rate": 1.7778424971332813e-06,
1339
+ "loss": 0.7151,
1340
+ "step": 1900
1341
+ },
1342
+ {
1343
+ "epoch": 0.7354639969195226,
1344
+ "grad_norm": 0.16654133796691895,
1345
+ "learning_rate": 1.7304115736010407e-06,
1346
+ "loss": 0.7107,
1347
+ "step": 1910
1348
+ },
1349
+ {
1350
+ "epoch": 0.7393145937620331,
1351
+ "grad_norm": 0.16922228038311005,
1352
+ "learning_rate": 1.6834891968017108e-06,
1353
+ "loss": 0.7028,
1354
+ "step": 1920
1355
+ },
1356
+ {
1357
+ "epoch": 0.7431651906045437,
1358
+ "grad_norm": 0.1779944896697998,
1359
+ "learning_rate": 1.6370826649692884e-06,
1360
+ "loss": 0.7176,
1361
+ "step": 1930
1362
+ },
1363
+ {
1364
+ "epoch": 0.7470157874470543,
1365
+ "grad_norm": 0.1670387238264084,
1366
+ "learning_rate": 1.5911991961040463e-06,
1367
+ "loss": 0.7209,
1368
+ "step": 1940
1369
+ },
1370
+ {
1371
+ "epoch": 0.7508663842895649,
1372
+ "grad_norm": 0.1781718134880066,
1373
+ "learning_rate": 1.545845926849851e-06,
1374
+ "loss": 0.7051,
1375
+ "step": 1950
1376
+ },
1377
+ {
1378
+ "epoch": 0.7547169811320755,
1379
+ "grad_norm": 0.18440252542495728,
1380
+ "learning_rate": 1.5010299113841397e-06,
1381
+ "loss": 0.7286,
1382
+ "step": 1960
1383
+ },
1384
+ {
1385
+ "epoch": 0.7585675779745861,
1386
+ "grad_norm": 0.17158637940883636,
1387
+ "learning_rate": 1.4567581203207238e-06,
1388
+ "loss": 0.7181,
1389
+ "step": 1970
1390
+ },
1391
+ {
1392
+ "epoch": 0.7624181748170966,
1393
+ "grad_norm": 0.1843189299106598,
1394
+ "learning_rate": 1.413037439625593e-06,
1395
+ "loss": 0.7178,
1396
+ "step": 1980
1397
+ },
1398
+ {
1399
+ "epoch": 0.7662687716596073,
1400
+ "grad_norm": 0.18571186065673828,
1401
+ "learning_rate": 1.3698746695458847e-06,
1402
+ "loss": 0.707,
1403
+ "step": 1990
1404
+ },
1405
+ {
1406
+ "epoch": 0.7701193685021178,
1407
+ "grad_norm": 0.17674745619297028,
1408
+ "learning_rate": 1.3272765235521745e-06,
1409
+ "loss": 0.7079,
1410
+ "step": 2000
1411
+ },
1412
+ {
1413
+ "epoch": 0.7739699653446284,
1414
+ "grad_norm": 0.16598600149154663,
1415
+ "learning_rate": 1.2852496272942877e-06,
1416
+ "loss": 0.7153,
1417
+ "step": 2010
1418
+ },
1419
+ {
1420
+ "epoch": 0.777820562187139,
1421
+ "grad_norm": 0.16544374823570251,
1422
+ "learning_rate": 1.2438005175707396e-06,
1423
+ "loss": 0.6933,
1424
+ "step": 2020
1425
+ },
1426
+ {
1427
+ "epoch": 0.7816711590296496,
1428
+ "grad_norm": 0.17070452868938446,
1429
+ "learning_rate": 1.2029356413120297e-06,
1430
+ "loss": 0.689,
1431
+ "step": 2030
1432
+ },
1433
+ {
1434
+ "epoch": 0.7855217558721602,
1435
+ "grad_norm": 0.17231673002243042,
1436
+ "learning_rate": 1.1626613545778786e-06,
1437
+ "loss": 0.7045,
1438
+ "step": 2040
1439
+ },
1440
+ {
1441
+ "epoch": 0.7893723527146708,
1442
+ "grad_norm": 0.16082847118377686,
1443
+ "learning_rate": 1.1229839215686294e-06,
1444
+ "loss": 0.7075,
1445
+ "step": 2050
1446
+ },
1447
+ {
1448
+ "epoch": 0.7932229495571813,
1449
+ "grad_norm": 0.15411065518856049,
1450
+ "learning_rate": 1.0839095136509188e-06,
1451
+ "loss": 0.6918,
1452
+ "step": 2060
1453
+ },
1454
+ {
1455
+ "epoch": 0.797073546399692,
1456
+ "grad_norm": 0.17241427302360535,
1457
+ "learning_rate": 1.045444208397791e-06,
1458
+ "loss": 0.7066,
1459
+ "step": 2070
1460
+ },
1461
+ {
1462
+ "epoch": 0.8009241432422025,
1463
+ "grad_norm": 0.16257914900779724,
1464
+ "learning_rate": 1.007593988643402e-06,
1465
+ "loss": 0.6852,
1466
+ "step": 2080
1467
+ },
1468
+ {
1469
+ "epoch": 0.8047747400847132,
1470
+ "grad_norm": 0.1578209400177002,
1471
+ "learning_rate": 9.703647415524613e-07,
1472
+ "loss": 0.7129,
1473
+ "step": 2090
1474
+ },
1475
+ {
1476
+ "epoch": 0.8086253369272237,
1477
+ "grad_norm": 0.15883062779903412,
1478
+ "learning_rate": 9.337622577045485e-07,
1479
+ "loss": 0.7021,
1480
+ "step": 2100
1481
+ },
1482
+ {
1483
+ "epoch": 0.8124759337697343,
1484
+ "grad_norm": 0.16800457239151,
1485
+ "learning_rate": 8.977922301934588e-07,
1486
+ "loss": 0.7107,
1487
+ "step": 2110
1488
+ },
1489
+ {
1490
+ "epoch": 0.8163265306122449,
1491
+ "grad_norm": 0.1827409416437149,
1492
+ "learning_rate": 8.624602537417009e-07,
1493
+ "loss": 0.7245,
1494
+ "step": 2120
1495
+ },
1496
+ {
1497
+ "epoch": 0.8201771274547555,
1498
+ "grad_norm": 0.17340503633022308,
1499
+ "learning_rate": 8.277718238303123e-07,
1500
+ "loss": 0.705,
1501
+ "step": 2130
1502
+ },
1503
+ {
1504
+ "epoch": 0.8240277242972661,
1505
+ "grad_norm": 0.17089290916919708,
1506
+ "learning_rate": 7.937323358440935e-07,
1507
+ "loss": 0.7078,
1508
+ "step": 2140
1509
+ },
1510
+ {
1511
+ "epoch": 0.8278783211397767,
1512
+ "grad_norm": 0.16699570417404175,
1513
+ "learning_rate": 7.603470842324223e-07,
1514
+ "loss": 0.6941,
1515
+ "step": 2150
1516
+ },
1517
+ {
1518
+ "epoch": 0.8317289179822872,
1519
+ "grad_norm": 0.16421230137348175,
1520
+ "learning_rate": 7.276212616857575e-07,
1521
+ "loss": 0.7165,
1522
+ "step": 2160
1523
+ },
1524
+ {
1525
+ "epoch": 0.8355795148247979,
1526
+ "grad_norm": 0.17130818963050842,
1527
+ "learning_rate": 6.955599583279837e-07,
1528
+ "loss": 0.7096,
1529
+ "step": 2170
1530
+ },
1531
+ {
1532
+ "epoch": 0.8394301116673084,
1533
+ "grad_norm": 0.16459695994853973,
1534
+ "learning_rate": 6.641681609246981e-07,
1535
+ "loss": 0.7072,
1536
+ "step": 2180
1537
+ },
1538
+ {
1539
+ "epoch": 0.8432807085098191,
1540
+ "grad_norm": 0.1756402552127838,
1541
+ "learning_rate": 6.334507521075711e-07,
1542
+ "loss": 0.7049,
1543
+ "step": 2190
1544
+ },
1545
+ {
1546
+ "epoch": 0.8471313053523296,
1547
+ "grad_norm": 0.174482524394989,
1548
+ "learning_rate": 6.034125096149179e-07,
1549
+ "loss": 0.7085,
1550
+ "step": 2200
1551
+ },
1552
+ {
1553
+ "epoch": 0.8509819021948402,
1554
+ "grad_norm": 0.159109964966774,
1555
+ "learning_rate": 5.740581055485722e-07,
1556
+ "loss": 0.6795,
1557
+ "step": 2210
1558
+ },
1559
+ {
1560
+ "epoch": 0.8548324990373508,
1561
+ "grad_norm": 0.1627909243106842,
1562
+ "learning_rate": 5.453921056471961e-07,
1563
+ "loss": 0.7124,
1564
+ "step": 2220
1565
+ },
1566
+ {
1567
+ "epoch": 0.8586830958798614,
1568
+ "grad_norm": 0.18585991859436035,
1569
+ "learning_rate": 5.174189685761288e-07,
1570
+ "loss": 0.7066,
1571
+ "step": 2230
1572
+ },
1573
+ {
1574
+ "epoch": 0.862533692722372,
1575
+ "grad_norm": 0.16287393867969513,
1576
+ "learning_rate": 4.901430452338979e-07,
1577
+ "loss": 0.7192,
1578
+ "step": 2240
1579
+ },
1580
+ {
1581
+ "epoch": 0.8663842895648826,
1582
+ "grad_norm": 0.17171747982501984,
1583
+ "learning_rate": 4.6356857807548286e-07,
1584
+ "loss": 0.7054,
1585
+ "step": 2250
1586
+ },
1587
+ {
1588
+ "epoch": 0.8702348864073931,
1589
+ "grad_norm": 0.19086970388889313,
1590
+ "learning_rate": 4.376997004524541e-07,
1591
+ "loss": 0.6882,
1592
+ "step": 2260
1593
+ },
1594
+ {
1595
+ "epoch": 0.8740854832499038,
1596
+ "grad_norm": 0.17048843204975128,
1597
+ "learning_rate": 4.12540435970073e-07,
1598
+ "loss": 0.705,
1599
+ "step": 2270
1600
+ },
1601
+ {
1602
+ "epoch": 0.8779360800924143,
1603
+ "grad_norm": 0.16426725685596466,
1604
+ "learning_rate": 3.880946978614703e-07,
1605
+ "loss": 0.7152,
1606
+ "step": 2280
1607
+ },
1608
+ {
1609
+ "epoch": 0.881786676934925,
1610
+ "grad_norm": 0.16695010662078857,
1611
+ "learning_rate": 3.643662883789878e-07,
1612
+ "loss": 0.7099,
1613
+ "step": 2290
1614
+ },
1615
+ {
1616
+ "epoch": 0.8856372737774355,
1617
+ "grad_norm": 0.1634049415588379,
1618
+ "learning_rate": 3.41358898202781e-07,
1619
+ "loss": 0.6955,
1620
+ "step": 2300
1621
+ },
1622
+ {
1623
+ "epoch": 0.889487870619946,
1624
+ "grad_norm": 0.16256767511367798,
1625
+ "learning_rate": 3.190761058667741e-07,
1626
+ "loss": 0.7078,
1627
+ "step": 2310
1628
+ },
1629
+ {
1630
+ "epoch": 0.8933384674624567,
1631
+ "grad_norm": 0.15762171149253845,
1632
+ "learning_rate": 2.9752137720206463e-07,
1633
+ "loss": 0.7082,
1634
+ "step": 2320
1635
+ },
1636
+ {
1637
+ "epoch": 0.8971890643049673,
1638
+ "grad_norm": 0.17422233521938324,
1639
+ "learning_rate": 2.7669806479785067e-07,
1640
+ "loss": 0.7193,
1641
+ "step": 2330
1642
+ },
1643
+ {
1644
+ "epoch": 0.9010396611474779,
1645
+ "grad_norm": 0.16270925104618073,
1646
+ "learning_rate": 2.566094074799763e-07,
1647
+ "loss": 0.709,
1648
+ "step": 2340
1649
+ },
1650
+ {
1651
+ "epoch": 0.9048902579899885,
1652
+ "grad_norm": 0.16947312653064728,
1653
+ "learning_rate": 2.3725852980716557e-07,
1654
+ "loss": 0.6931,
1655
+ "step": 2350
1656
+ },
1657
+ {
1658
+ "epoch": 0.908740854832499,
1659
+ "grad_norm": 0.16585607826709747,
1660
+ "learning_rate": 2.1864844158503983e-07,
1661
+ "loss": 0.7062,
1662
+ "step": 2360
1663
+ },
1664
+ {
1665
+ "epoch": 0.9125914516750097,
1666
+ "grad_norm": 0.18474096059799194,
1667
+ "learning_rate": 2.007820373979702e-07,
1668
+ "loss": 0.6987,
1669
+ "step": 2370
1670
+ },
1671
+ {
1672
+ "epoch": 0.9164420485175202,
1673
+ "grad_norm": 0.16417932510375977,
1674
+ "learning_rate": 1.8366209615886532e-07,
1675
+ "loss": 0.7119,
1676
+ "step": 2380
1677
+ },
1678
+ {
1679
+ "epoch": 0.9202926453600309,
1680
+ "grad_norm": 0.16731320321559906,
1681
+ "learning_rate": 1.6729128067693613e-07,
1682
+ "loss": 0.725,
1683
+ "step": 2390
1684
+ },
1685
+ {
1686
+ "epoch": 0.9241432422025414,
1687
+ "grad_norm": 0.18339282274246216,
1688
+ "learning_rate": 1.5167213724353426e-07,
1689
+ "loss": 0.7181,
1690
+ "step": 2400
1691
+ },
1692
+ {
1693
+ "epoch": 0.927993839045052,
1694
+ "grad_norm": 0.16345539689064026,
1695
+ "learning_rate": 1.3680709523610058e-07,
1696
+ "loss": 0.6945,
1697
+ "step": 2410
1698
+ },
1699
+ {
1700
+ "epoch": 0.9318444358875626,
1701
+ "grad_norm": 0.16204354166984558,
1702
+ "learning_rate": 1.2269846674030418e-07,
1703
+ "loss": 0.7231,
1704
+ "step": 2420
1705
+ },
1706
+ {
1707
+ "epoch": 0.9356950327300732,
1708
+ "grad_norm": 0.1662166863679886,
1709
+ "learning_rate": 1.0934844619042595e-07,
1710
+ "loss": 0.7183,
1711
+ "step": 2430
1712
+ },
1713
+ {
1714
+ "epoch": 0.9395456295725837,
1715
+ "grad_norm": 0.16929581761360168,
1716
+ "learning_rate": 9.675911002803961e-08,
1717
+ "loss": 0.7055,
1718
+ "step": 2440
1719
+ },
1720
+ {
1721
+ "epoch": 0.9433962264150944,
1722
+ "grad_norm": 0.17421160638332367,
1723
+ "learning_rate": 8.493241637904381e-08,
1724
+ "loss": 0.6945,
1725
+ "step": 2450
1726
+ },
1727
+ {
1728
+ "epoch": 0.9472468232576049,
1729
+ "grad_norm": 0.18213938176631927,
1730
+ "learning_rate": 7.387020474909689e-08,
1731
+ "loss": 0.7217,
1732
+ "step": 2460
1733
+ },
1734
+ {
1735
+ "epoch": 0.9510974201001156,
1736
+ "grad_norm": 0.17582881450653076,
1737
+ "learning_rate": 6.357419573750856e-08,
1738
+ "loss": 0.7169,
1739
+ "step": 2470
1740
+ },
1741
+ {
1742
+ "epoch": 0.9549480169426261,
1743
+ "grad_norm": 0.1662108153104782,
1744
+ "learning_rate": 5.4045990769613365e-08,
1745
+ "loss": 0.7031,
1746
+ "step": 2480
1747
+ },
1748
+ {
1749
+ "epoch": 0.9587986137851366,
1750
+ "grad_norm": 0.16325069963932037,
1751
+ "learning_rate": 4.5287071847694434e-08,
1752
+ "loss": 0.7039,
1753
+ "step": 2490
1754
+ },
1755
+ {
1756
+ "epoch": 0.9626492106276473,
1757
+ "grad_norm": 0.16816158592700958,
1758
+ "learning_rate": 3.7298801320467817e-08,
1759
+ "loss": 0.7088,
1760
+ "step": 2500
1761
+ },
1762
+ {
1763
+ "epoch": 0.9664998074701578,
1764
+ "grad_norm": 0.16526630520820618,
1765
+ "learning_rate": 3.0082421671192576e-08,
1766
+ "loss": 0.7065,
1767
+ "step": 2510
1768
+ },
1769
+ {
1770
+ "epoch": 0.9703504043126685,
1771
+ "grad_norm": 0.16447442770004272,
1772
+ "learning_rate": 2.3639055324410355e-08,
1773
+ "loss": 0.7223,
1774
+ "step": 2520
1775
+ },
1776
+ {
1777
+ "epoch": 0.974201001155179,
1778
+ "grad_norm": 0.17489105463027954,
1779
+ "learning_rate": 1.7969704471369477e-08,
1780
+ "loss": 0.7005,
1781
+ "step": 2530
1782
+ },
1783
+ {
1784
+ "epoch": 0.9780515979976896,
1785
+ "grad_norm": 0.1632070392370224,
1786
+ "learning_rate": 1.3075250914142413e-08,
1787
+ "loss": 0.703,
1788
+ "step": 2540
1789
+ },
1790
+ {
1791
+ "epoch": 0.9819021948402002,
1792
+ "grad_norm": 0.1564241349697113,
1793
+ "learning_rate": 8.956455928476049e-09,
1794
+ "loss": 0.7166,
1795
+ "step": 2550
1796
+ },
1797
+ {
1798
+ "epoch": 0.9857527916827108,
1799
+ "grad_norm": 0.16202977299690247,
1800
+ "learning_rate": 5.6139601453786275e-09,
1801
+ "loss": 0.7,
1802
+ "step": 2560
1803
+ },
1804
+ {
1805
+ "epoch": 0.9896033885252215,
1806
+ "grad_norm": 0.16260544955730438,
1807
+ "learning_rate": 3.0482834514805602e-09,
1808
+ "loss": 0.6983,
1809
+ "step": 2570
1810
+ },
1811
+ {
1812
+ "epoch": 0.993453985367732,
1813
+ "grad_norm": 0.16054078936576843,
1814
+ "learning_rate": 1.2598249081713366e-09,
1815
+ "loss": 0.6869,
1816
+ "step": 2580
1817
+ },
1818
+ {
1819
+ "epoch": 0.9973045822102425,
1820
+ "grad_norm": 0.17293651401996613,
1821
+ "learning_rate": 2.488626895302826e-10,
1822
+ "loss": 0.7228,
1823
+ "step": 2590
1824
+ },
1825
+ {
1826
+ "epoch": 1.0,
1827
+ "step": 2597,
1828
+ "total_flos": 7.170128148946747e+18,
1829
+ "train_loss": 0.7246010998464062,
1830
+ "train_runtime": 15924.5491,
1831
+ "train_samples_per_second": 41.753,
1832
+ "train_steps_per_second": 0.163
1833
+ }
1834
+ ],
1835
+ "logging_steps": 10,
1836
+ "max_steps": 2597,
1837
+ "num_input_tokens_seen": 584317136,
1838
+ "num_train_epochs": 1,
1839
+ "save_steps": 500,
1840
+ "stateful_callbacks": {
1841
+ "TrainerControl": {
1842
+ "args": {
1843
+ "should_epoch_stop": false,
1844
+ "should_evaluate": false,
1845
+ "should_log": false,
1846
+ "should_save": true,
1847
+ "should_training_stop": true
1848
+ },
1849
+ "attributes": {}
1850
+ }
1851
+ },
1852
+ "total_flos": 7.170128148946747e+18,
1853
+ "train_batch_size": 16,
1854
+ "trial_name": null,
1855
+ "trial_params": null
1856
+ }
continuous/smollm2/siglip2_sm16_256/finetune/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dfe6329e4b02bd8d23b36d8805ad0ecd4a3aa894e54500ec8c6c0f98114e36c9
3
+ size 6264
continuous/smollm2/siglip2_sm16_256/finetune/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
continuous/smollm2/siglip2_sm16_256/pretrain/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": null,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-16-SigLIP2-256",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm16_256.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm16_256/pretrain/mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e23e4c84feef7c4f0d758baac0a8cefbe6db3bc623aff9e73d6d6c79cad39cd
3
+ size 26232888
continuous/smollm2/siglip2_sm16_256/pretrain/trainer_state.json ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 545,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.01834862385321101,
14
+ "grad_norm": 3.445356607437134,
15
+ "learning_rate": 0.00032142857142857147,
16
+ "loss": 3.8146,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 0.03669724770642202,
21
+ "grad_norm": 0.13813941180706024,
22
+ "learning_rate": 0.0006785714285714287,
23
+ "loss": 2.8488,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 0.05504587155963303,
28
+ "grad_norm": 0.062422581017017365,
29
+ "learning_rate": 0.0009999907688176173,
30
+ "loss": 2.7692,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 0.07339449541284404,
35
+ "grad_norm": 0.1660098284482956,
36
+ "learning_rate": 0.0009988834393115766,
37
+ "loss": 2.7283,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 0.09174311926605505,
42
+ "grad_norm": 0.12383703887462616,
43
+ "learning_rate": 0.0009959345572940772,
44
+ "loss": 2.7045,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 0.11009174311926606,
49
+ "grad_norm": 0.11411943286657333,
50
+ "learning_rate": 0.0009911550081155983,
51
+ "loss": 2.6278,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 0.12844036697247707,
56
+ "grad_norm": 0.09219418466091156,
57
+ "learning_rate": 0.0009845624347566433,
58
+ "loss": 2.5431,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 0.14678899082568808,
63
+ "grad_norm": 0.13013620674610138,
64
+ "learning_rate": 0.0009761811727013547,
65
+ "loss": 2.4595,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 0.1651376146788991,
70
+ "grad_norm": 0.1488083004951477,
71
+ "learning_rate": 0.0009660421601067666,
72
+ "loss": 2.3985,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 0.1834862385321101,
77
+ "grad_norm": 0.11741623282432556,
78
+ "learning_rate": 0.0009541828235992926,
79
+ "loss": 2.3586,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 0.2018348623853211,
84
+ "grad_norm": 0.11350785940885544,
85
+ "learning_rate": 0.0009406469401200151,
86
+ "loss": 2.3116,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 0.22018348623853212,
91
+ "grad_norm": 0.10228355973958969,
92
+ "learning_rate": 0.0009254844753287492,
93
+ "loss": 2.2849,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 0.23853211009174313,
98
+ "grad_norm": 0.14226453006267548,
99
+ "learning_rate": 0.0009087513991633924,
100
+ "loss": 2.2514,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 0.25688073394495414,
105
+ "grad_norm": 0.11405840516090393,
106
+ "learning_rate": 0.0008905094792353917,
107
+ "loss": 2.226,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 0.27522935779816515,
112
+ "grad_norm": 0.12276880443096161,
113
+ "learning_rate": 0.0008708260528239789,
114
+ "loss": 2.2077,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 0.29357798165137616,
119
+ "grad_norm": 0.12694914638996124,
120
+ "learning_rate": 0.0008497737783108237,
121
+ "loss": 2.163,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.3119266055045872,
126
+ "grad_norm": 0.10896015167236328,
127
+ "learning_rate": 0.0008274303669726426,
128
+ "loss": 2.1379,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 0.3302752293577982,
133
+ "grad_norm": 0.10039632767438889,
134
+ "learning_rate": 0.0008038782961218137,
135
+ "loss": 2.1144,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 0.3486238532110092,
140
+ "grad_norm": 0.10828958451747894,
141
+ "learning_rate": 0.000779204504653894,
142
+ "loss": 2.099,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 0.3669724770642202,
147
+ "grad_norm": 0.1023016944527626,
148
+ "learning_rate": 0.0007535000721258791,
149
+ "loss": 2.0717,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 0.3853211009174312,
154
+ "grad_norm": 0.11902939528226852,
155
+ "learning_rate": 0.00072685988254984,
156
+ "loss": 2.067,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 0.4036697247706422,
161
+ "grad_norm": 0.09173259139060974,
162
+ "learning_rate": 0.0006993822741429906,
163
+ "loss": 2.0716,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 0.42201834862385323,
168
+ "grad_norm": 0.09877891093492508,
169
+ "learning_rate": 0.0006711686763270818,
170
+ "loss": 2.0573,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 0.44036697247706424,
175
+ "grad_norm": 0.08655282855033875,
176
+ "learning_rate": 0.0006423232353170831,
177
+ "loss": 2.0524,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 0.45871559633027525,
182
+ "grad_norm": 0.11764185130596161,
183
+ "learning_rate": 0.0006129524296812335,
184
+ "loss": 2.0525,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 0.47706422018348627,
189
+ "grad_norm": 0.10041780024766922,
190
+ "learning_rate": 0.0005831646772915651,
191
+ "loss": 2.0559,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 0.4954128440366973,
196
+ "grad_norm": 0.0936015248298645,
197
+ "learning_rate": 0.0005530699351157809,
198
+ "loss": 2.0332,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 0.5137614678899083,
203
+ "grad_norm": 0.0941554382443428,
204
+ "learning_rate": 0.0005227792933277943,
205
+ "loss": 2.0378,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 0.5321100917431193,
210
+ "grad_norm": 0.0938389003276825,
211
+ "learning_rate": 0.0004924045652352092,
212
+ "loss": 2.027,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 0.5504587155963303,
217
+ "grad_norm": 0.08497871458530426,
218
+ "learning_rate": 0.0004620578745374633,
219
+ "loss": 2.0261,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 0.5688073394495413,
224
+ "grad_norm": 0.10620149970054626,
225
+ "learning_rate": 0.0004318512414382059,
226
+ "loss": 2.0301,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 0.5871559633027523,
231
+ "grad_norm": 0.08616603165864944,
232
+ "learning_rate": 0.0004018961691397148,
233
+ "loss": 2.0301,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 0.6055045871559633,
238
+ "grad_norm": 0.09978330880403519,
239
+ "learning_rate": 0.0003723032322457458,
240
+ "loss": 2.0154,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 0.6238532110091743,
245
+ "grad_norm": 0.11187147349119186,
246
+ "learning_rate": 0.0003431816685921625,
247
+ "loss": 2.0173,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 0.6422018348623854,
252
+ "grad_norm": 0.08981547504663467,
253
+ "learning_rate": 0.0003146389760120416,
254
+ "loss": 2.0153,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 0.6605504587155964,
259
+ "grad_norm": 0.0966462716460228,
260
+ "learning_rate": 0.00028678051552373485,
261
+ "loss": 2.0156,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 0.6788990825688074,
266
+ "grad_norm": 0.08413543552160263,
267
+ "learning_rate": 0.0002597091224066581,
268
+ "loss": 2.0102,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 0.6972477064220184,
273
+ "grad_norm": 0.08554989099502563,
274
+ "learning_rate": 0.0002335247266004629,
275
+ "loss": 2.0151,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 0.7155963302752294,
280
+ "grad_norm": 0.09147220104932785,
281
+ "learning_rate": 0.000208323983828828,
282
+ "loss": 2.0062,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 0.7339449541284404,
287
+ "grad_norm": 0.0871782973408699,
288
+ "learning_rate": 0.0001841999188095224,
289
+ "loss": 2.001,
290
+ "step": 400
291
+ },
292
+ {
293
+ "epoch": 0.7522935779816514,
294
+ "grad_norm": 0.09297121316194534,
295
+ "learning_rate": 0.00016124158186777676,
296
+ "loss": 2.0066,
297
+ "step": 410
298
+ },
299
+ {
300
+ "epoch": 0.7706422018348624,
301
+ "grad_norm": 0.09897100925445557,
302
+ "learning_rate": 0.0001395337202205211,
303
+ "loss": 2.0,
304
+ "step": 420
305
+ },
306
+ {
307
+ "epoch": 0.7889908256880734,
308
+ "grad_norm": 0.08241917937994003,
309
+ "learning_rate": 0.00011915646514489292,
310
+ "loss": 2.0004,
311
+ "step": 430
312
+ },
313
+ {
314
+ "epoch": 0.8073394495412844,
315
+ "grad_norm": 0.08575671166181564,
316
+ "learning_rate": 0.00010018503618578818,
317
+ "loss": 2.0031,
318
+ "step": 440
319
+ },
320
+ {
321
+ "epoch": 0.8256880733944955,
322
+ "grad_norm": 0.08442652225494385,
323
+ "learning_rate": 8.268946349432583e-05,
324
+ "loss": 1.9983,
325
+ "step": 450
326
+ },
327
+ {
328
+ "epoch": 0.8440366972477065,
329
+ "grad_norm": 0.08504480868577957,
330
+ "learning_rate": 6.673432932217117e-05,
331
+ "loss": 1.9935,
332
+ "step": 460
333
+ },
334
+ {
335
+ "epoch": 0.8623853211009175,
336
+ "grad_norm": 0.08285156637430191,
337
+ "learning_rate": 5.2378529625954684e-05,
338
+ "loss": 1.9851,
339
+ "step": 470
340
+ },
341
+ {
342
+ "epoch": 0.8807339449541285,
343
+ "grad_norm": 0.08321565389633179,
344
+ "learning_rate": 3.9675056661785556e-05,
345
+ "loss": 1.9893,
346
+ "step": 480
347
+ },
348
+ {
349
+ "epoch": 0.8990825688073395,
350
+ "grad_norm": 0.08429694175720215,
351
+ "learning_rate": 2.8670803372379538e-05,
352
+ "loss": 1.995,
353
+ "step": 490
354
+ },
355
+ {
356
+ "epoch": 0.9174311926605505,
357
+ "grad_norm": 0.08694891631603241,
358
+ "learning_rate": 1.9406390288876586e-05,
359
+ "loss": 1.9983,
360
+ "step": 500
361
+ },
362
+ {
363
+ "epoch": 0.9357798165137615,
364
+ "grad_norm": 0.08322856575250626,
365
+ "learning_rate": 1.191601558631461e-05,
366
+ "loss": 1.991,
367
+ "step": 510
368
+ },
369
+ {
370
+ "epoch": 0.9541284403669725,
371
+ "grad_norm": 0.08070629835128784,
372
+ "learning_rate": 6.2273288462559305e-06,
373
+ "loss": 1.9854,
374
+ "step": 520
375
+ },
376
+ {
377
+ "epoch": 0.9724770642201835,
378
+ "grad_norm": 0.08326620608568192,
379
+ "learning_rate": 2.361328992552314e-06,
380
+ "loss": 1.9904,
381
+ "step": 530
382
+ },
383
+ {
384
+ "epoch": 0.9908256880733946,
385
+ "grad_norm": 0.08383011072874069,
386
+ "learning_rate": 3.322867770044202e-07,
387
+ "loss": 1.9868,
388
+ "step": 540
389
+ },
390
+ {
391
+ "epoch": 1.0,
392
+ "step": 545,
393
+ "total_flos": 4.252888502821192e+17,
394
+ "train_loss": 2.1778375940585355,
395
+ "train_runtime": 1702.1944,
396
+ "train_samples_per_second": 327.887,
397
+ "train_steps_per_second": 0.32
398
+ }
399
+ ],
400
+ "logging_steps": 10,
401
+ "max_steps": 545,
402
+ "num_input_tokens_seen": 34658176,
403
+ "num_train_epochs": 1,
404
+ "save_steps": 500,
405
+ "stateful_callbacks": {
406
+ "TrainerControl": {
407
+ "args": {
408
+ "should_epoch_stop": false,
409
+ "should_evaluate": false,
410
+ "should_log": false,
411
+ "should_save": false,
412
+ "should_training_stop": false
413
+ },
414
+ "attributes": {}
415
+ }
416
+ },
417
+ "total_flos": 4.252888502821192e+17,
418
+ "train_batch_size": 64,
419
+ "trial_name": null,
420
+ "trial_params": null
421
+ }
continuous/smollm2/siglip2_sm16_384/finetune/chat_template.jinja ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
2
+ You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
3
+ ' }}{% endif %}{{'<|im_start|>' + message['role'] + '
4
+ ' + message['content'] + '<|im_end|>' + '
5
+ '}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
6
+ ' }}{% endif %}
continuous/smollm2/siglip2_sm16_384/finetune/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_faster_video": false,
3
+ "add_time_instruction": false,
4
+ "architectures": [
5
+ "LlavaLlamaForCausalLM"
6
+ ],
7
+ "attention_bias": false,
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 2,
12
+ "faster_token_stride": 10,
13
+ "force_sample": false,
14
+ "head_dim": 64,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 2048,
17
+ "image_aspect_ratio": "square",
18
+ "image_crop_resolution": null,
19
+ "image_grid_pinpoints": null,
20
+ "image_split_resolution": null,
21
+ "initializer_range": 0.02,
22
+ "intermediate_size": 8192,
23
+ "max_position_embeddings": 8192,
24
+ "mlp_bias": false,
25
+ "mm_hidden_size": 1152,
26
+ "mm_newline_position": "grid",
27
+ "mm_patch_merge_type": "flat",
28
+ "mm_projector_lr": 0.0001,
29
+ "mm_projector_type": "mlp2x_gelu",
30
+ "mm_resampler_type": null,
31
+ "mm_spatial_pool_mode": "bilinear",
32
+ "mm_spatial_pool_stride": null,
33
+ "mm_tunable_parts": "mm_mlp_adapter,mm_language_model",
34
+ "mm_use_im_patch_token": false,
35
+ "mm_use_im_start_end": false,
36
+ "mm_vision_select_feature": "patch",
37
+ "mm_vision_select_layer": -2,
38
+ "mm_vision_tower": "open_clip_hub:ViT-SO400M-16-SigLIP2-384",
39
+ "mm_vision_tower_lr": null,
40
+ "model_type": "llama",
41
+ "num_attention_heads": 32,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 32,
44
+ "pad_token_id": 0,
45
+ "pos_skipping_range": 4096,
46
+ "pretraining_tp": 1,
47
+ "rms_norm_eps": 1e-05,
48
+ "rope_scaling": null,
49
+ "rope_theta": 130000,
50
+ "tie_word_embeddings": true,
51
+ "tokenizer_model_max_length": 8192,
52
+ "tokenizer_padding_side": "right",
53
+ "transformers.js_config": {
54
+ "dtype": "q4",
55
+ "kv_cache_dtype": {
56
+ "fp16": "float16",
57
+ "q4f16": "float16"
58
+ },
59
+ "use_external_data_format": {
60
+ "model.onnx": true,
61
+ "model_fp16.onnx": true
62
+ }
63
+ },
64
+ "transformers_version": "4.57.6",
65
+ "use_cache": true,
66
+ "use_mm_proj": true,
67
+ "use_pos_skipping": false,
68
+ "vision_tower_pretrained": "/cache/ckpt/download/tokenizer/continuous/siglip2/siglip2_sm16_384.npz",
69
+ "vocab_size": 49152
70
+ }
continuous/smollm2/siglip2_sm16_384/finetune/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": [
5
+ 2
6
+ ],
7
+ "pad_token_id": 0,
8
+ "transformers_version": "4.57.6"
9
+ }
continuous/smollm2/siglip2_sm16_384/finetune/merges.txt ADDED
The diff for this file is too large to render. See raw diff