Baileyyeah commited on
Commit
00ab8a4
·
verified ·
1 Parent(s): 9b64a48

Add files using upload-large-folder tool

Browse files
Files changed (50) hide show
  1. Blood/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json +0 -0
  2. Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt +0 -0
  3. Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json +0 -0
  4. Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json +0 -0
  5. Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/special_tokens_map.json +31 -0
  6. Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/trainer_state.json +2331 -0
  7. Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/training_args.bin +3 -0
  8. Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/vocab.json +0 -0
  9. Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/added_tokens.json +24 -0
  10. Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/chat_template.jinja +54 -0
  11. Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/config.json +66 -0
  12. Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt +0 -0
  13. Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json +0 -0
  14. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/merges.txt +0 -0
  15. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/special_tokens_map.json +31 -0
  16. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/tokenizer_config.json +208 -0
  17. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/trainer_state.json +2331 -0
  18. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/training_args.bin +3 -0
  19. Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/vocab.json +0 -0
  20. Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt +0 -0
  21. Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/tokenizer_config.json +208 -0
  22. Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json +0 -0
  23. Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/training_args.bin +3 -0
  24. Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json +0 -0
  25. Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt +0 -0
  26. Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json +0 -0
  27. Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/training_args.bin +3 -0
  28. Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json +0 -0
  29. Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/special_tokens_map.json +31 -0
  30. Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/tokenizer_config.json +208 -0
  31. Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/trainer_state.json +0 -0
  32. Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/training_args.bin +3 -0
  33. Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/vocab.json +0 -0
  34. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja +54 -0
  35. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json +66 -0
  36. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/generation_config.json +14 -0
  37. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/merges.txt +0 -0
  38. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/model.safetensors.index.json +443 -0
  39. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/runs/Dec27_04-03-58_csce-yang-s2.engr.tamu.edu/events.out.tfevents.1766829852.csce-yang-s2.engr.tamu.edu.3772169.0 +3 -0
  40. Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/special_tokens_map.json +31 -0
  41. Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json +24 -0
  42. Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja +54 -0
  43. Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json +66 -0
  44. Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/generation_config.json +14 -0
  45. Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/runs/Dec26_20-47-41_csce-yang-s2.engr.tamu.edu/events.out.tfevents.1766803674.csce-yang-s2.engr.tamu.edu.3519012.0 +3 -0
  46. Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json +24 -0
  47. Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja +54 -0
  48. Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json +66 -0
  49. Creditg/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json +24 -0
  50. Creditg/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja +54 -0
Blood/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/trainer_state.json ADDED
@@ -0,0 +1,2331 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 3200,
3
+ "best_metric": 0.8,
4
+ "best_model_checkpoint": "/data/chenlang/Med3DVLM/llm/Blood/seeds/new/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/checkpoint-3200",
5
+ "epoch": 3.0,
6
+ "eval_steps": 100,
7
+ "global_step": 3588,
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,
14
+ "eval_completions/clipped_ratio": 0.0,
15
+ "eval_completions/max_length": 431.5,
16
+ "eval_completions/max_terminated_length": 431.5,
17
+ "eval_completions/mean_length": 362.40625,
18
+ "eval_completions/mean_terminated_length": 362.40625,
19
+ "eval_completions/min_length": 265.0,
20
+ "eval_completions/min_terminated_length": 265.0,
21
+ "eval_frac_reward_zero_std": 0.0,
22
+ "eval_loss": 0.0,
23
+ "eval_num_tokens": 0.0,
24
+ "eval_reward": 0.78125,
25
+ "eval_reward_std": 0.0,
26
+ "eval_rewards/accuracy_reward/mean": 0.78125,
27
+ "eval_rewards/accuracy_reward/std": 0.4166666865348816,
28
+ "eval_runtime": 48.9785,
29
+ "eval_samples_per_second": 1.531,
30
+ "eval_steps_per_second": 0.041,
31
+ "step": 0
32
+ },
33
+ {
34
+ "clip_ratio/high_max": 5.57239654881414e-05,
35
+ "clip_ratio/high_mean": 2.78619827440707e-05,
36
+ "clip_ratio/low_mean": 4.826992335438263e-05,
37
+ "clip_ratio/low_min": 0.0,
38
+ "clip_ratio/region_mean": 7.613190609845333e-05,
39
+ "completions/clipped_ratio": 0.0,
40
+ "completions/max_length": 413.8125,
41
+ "completions/max_terminated_length": 413.8125,
42
+ "completions/mean_length": 370.19140625,
43
+ "completions/mean_terminated_length": 370.19140625,
44
+ "completions/min_length": 331.9375,
45
+ "completions/min_terminated_length": 331.9375,
46
+ "entropy": 0.018870638399675954,
47
+ "epoch": 0.05351170568561873,
48
+ "frac_reward_zero_std": 0.96875,
49
+ "grad_norm": 0.1725948452949524,
50
+ "kl": 0.00010102036628722999,
51
+ "learning_rate": 1.9648829431438127e-06,
52
+ "loss": -0.0002,
53
+ "num_tokens": 323666.0,
54
+ "reward": 0.822265625,
55
+ "reward_std": 0.014959799125790596,
56
+ "rewards/accuracy_reward/mean": 0.822265625,
57
+ "rewards/accuracy_reward/std": 0.18891007266938686,
58
+ "step": 64
59
+ },
60
+ {
61
+ "epoch": 0.08361204013377926,
62
+ "eval_completions/clipped_ratio": 0.0,
63
+ "eval_completions/max_length": 464.0,
64
+ "eval_completions/max_terminated_length": 464.0,
65
+ "eval_completions/mean_length": 364.6484375,
66
+ "eval_completions/mean_terminated_length": 364.6484375,
67
+ "eval_completions/min_length": 270.5,
68
+ "eval_completions/min_terminated_length": 270.5,
69
+ "eval_frac_reward_zero_std": 0.0,
70
+ "eval_loss": 0.0,
71
+ "eval_num_tokens": 501456.0,
72
+ "eval_reward": 0.765625,
73
+ "eval_reward_std": 0.0,
74
+ "eval_rewards/accuracy_reward/mean": 0.765625,
75
+ "eval_rewards/accuracy_reward/std": 0.4265512377023697,
76
+ "eval_runtime": 52.9229,
77
+ "eval_samples_per_second": 1.417,
78
+ "eval_steps_per_second": 0.038,
79
+ "step": 100
80
+ },
81
+ {
82
+ "clip_ratio/high_max": 0.0,
83
+ "clip_ratio/high_mean": 0.0,
84
+ "clip_ratio/low_mean": 0.0,
85
+ "clip_ratio/low_min": 0.0,
86
+ "clip_ratio/region_mean": 0.0,
87
+ "completions/clipped_ratio": 0.0,
88
+ "completions/max_length": 424.0,
89
+ "completions/max_terminated_length": 424.0,
90
+ "completions/mean_length": 380.08482142857144,
91
+ "completions/mean_terminated_length": 380.08482142857144,
92
+ "completions/min_length": 349.57142857142856,
93
+ "completions/min_terminated_length": 349.57142857142856,
94
+ "entropy": 0.024323281067024385,
95
+ "epoch": 0.10702341137123746,
96
+ "frac_reward_zero_std": 1.0,
97
+ "grad_norm": 0.020470470190048218,
98
+ "kl": 0.006877838764921762,
99
+ "learning_rate": 1.9292084726867336e-06,
100
+ "loss": 0.002,
101
+ "num_tokens": 645139.0,
102
+ "reward": 0.8571428571428571,
103
+ "reward_std": 0.0,
104
+ "rewards/accuracy_reward/mean": 0.8571428571428571,
105
+ "rewards/accuracy_reward/std": 0.0,
106
+ "step": 128
107
+ },
108
+ {
109
+ "clip_ratio/high_max": 0.0007680538783461088,
110
+ "clip_ratio/high_mean": 0.0003840269391730544,
111
+ "clip_ratio/low_mean": 0.0009337622955172264,
112
+ "clip_ratio/low_min": 0.0,
113
+ "clip_ratio/region_mean": 0.001317789236054523,
114
+ "completions/clipped_ratio": 0.0,
115
+ "completions/max_length": 419.6875,
116
+ "completions/max_terminated_length": 419.6875,
117
+ "completions/mean_length": 362.955078125,
118
+ "completions/mean_terminated_length": 362.955078125,
119
+ "completions/min_length": 319.5625,
120
+ "completions/min_terminated_length": 319.5625,
121
+ "entropy": 0.027590649275225587,
122
+ "epoch": 0.1605351170568562,
123
+ "frac_reward_zero_std": 0.78125,
124
+ "grad_norm": 2.1669669151306152,
125
+ "kl": 0.02729937141020855,
126
+ "learning_rate": 1.8935340022296544e-06,
127
+ "loss": -0.0048,
128
+ "num_tokens": 965100.0,
129
+ "reward": 0.775390625,
130
+ "reward_std": 0.08694327063858509,
131
+ "rewards/accuracy_reward/mean": 0.775390625,
132
+ "rewards/accuracy_reward/std": 0.25361916795372963,
133
+ "step": 192
134
+ },
135
+ {
136
+ "epoch": 0.16722408026755853,
137
+ "eval_completions/clipped_ratio": 0.0,
138
+ "eval_completions/max_length": 455.0,
139
+ "eval_completions/max_terminated_length": 455.0,
140
+ "eval_completions/mean_length": 367.75,
141
+ "eval_completions/mean_terminated_length": 367.75,
142
+ "eval_completions/min_length": 267.0,
143
+ "eval_completions/min_terminated_length": 267.0,
144
+ "eval_frac_reward_zero_std": 0.0,
145
+ "eval_loss": 0.0,
146
+ "eval_num_tokens": 1005338.0,
147
+ "eval_reward": 0.7265625,
148
+ "eval_reward_std": 0.0,
149
+ "eval_rewards/accuracy_reward/mean": 0.7265625,
150
+ "eval_rewards/accuracy_reward/std": 0.4484642744064331,
151
+ "eval_runtime": 52.6165,
152
+ "eval_samples_per_second": 1.425,
153
+ "eval_steps_per_second": 0.038,
154
+ "step": 200
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.00028142759609701376,
158
+ "clip_ratio/high_mean": 0.00014071379804850688,
159
+ "clip_ratio/low_mean": 4.14181558880955e-05,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.00018213195393660238,
162
+ "completions/clipped_ratio": 0.0,
163
+ "completions/max_length": 411.7142857142857,
164
+ "completions/max_terminated_length": 411.7142857142857,
165
+ "completions/mean_length": 358.44419642857144,
166
+ "completions/mean_terminated_length": 358.44419642857144,
167
+ "completions/min_length": 318.57142857142856,
168
+ "completions/min_terminated_length": 318.57142857142856,
169
+ "entropy": 0.026801060918452486,
170
+ "epoch": 0.2140468227424749,
171
+ "frac_reward_zero_std": 0.9642857142857143,
172
+ "grad_norm": 0.024173056706786156,
173
+ "kl": 0.02256160867204926,
174
+ "learning_rate": 1.8578595317725752e-06,
175
+ "loss": 0.0029,
176
+ "num_tokens": 1283345.0,
177
+ "reward": 0.78125,
178
+ "reward_std": 0.012198750461850847,
179
+ "rewards/accuracy_reward/mean": 0.78125,
180
+ "rewards/accuracy_reward/std": 0.1627097608787673,
181
+ "step": 256
182
+ },
183
+ {
184
+ "epoch": 0.2508361204013378,
185
+ "eval_completions/clipped_ratio": 0.0,
186
+ "eval_completions/max_length": 457.5,
187
+ "eval_completions/max_terminated_length": 457.5,
188
+ "eval_completions/mean_length": 363.546875,
189
+ "eval_completions/mean_terminated_length": 363.546875,
190
+ "eval_completions/min_length": 255.0,
191
+ "eval_completions/min_terminated_length": 255.0,
192
+ "eval_frac_reward_zero_std": 0.0,
193
+ "eval_loss": 0.0,
194
+ "eval_num_tokens": 1504266.0,
195
+ "eval_reward": 0.7578125,
196
+ "eval_reward_std": 0.0,
197
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
198
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
199
+ "eval_runtime": 52.1,
200
+ "eval_samples_per_second": 1.44,
201
+ "eval_steps_per_second": 0.038,
202
+ "step": 300
203
+ },
204
+ {
205
+ "clip_ratio/high_max": 0.0013878081343136727,
206
+ "clip_ratio/high_mean": 0.0006939040671568364,
207
+ "clip_ratio/low_mean": 0.0,
208
+ "clip_ratio/low_min": 0.0,
209
+ "clip_ratio/region_mean": 0.0006939040671568364,
210
+ "completions/clipped_ratio": 0.0,
211
+ "completions/max_length": 440.2,
212
+ "completions/max_terminated_length": 440.2,
213
+ "completions/mean_length": 378.7,
214
+ "completions/mean_terminated_length": 378.7,
215
+ "completions/min_length": 336.8,
216
+ "completions/min_terminated_length": 336.8,
217
+ "entropy": 0.037679997738450766,
218
+ "epoch": 0.26755852842809363,
219
+ "frac_reward_zero_std": 0.8,
220
+ "grad_norm": 0.1935475468635559,
221
+ "kl": 0.5320995265152305,
222
+ "learning_rate": 1.822185061315496e-06,
223
+ "loss": 0.0181,
224
+ "num_tokens": 1606730.0,
225
+ "reward": 0.8875,
226
+ "reward_std": 0.05,
227
+ "rewards/accuracy_reward/mean": 0.8875,
228
+ "rewards/accuracy_reward/std": 0.17231077551841736,
229
+ "step": 320
230
+ },
231
+ {
232
+ "clip_ratio/high_max": 0.0002988548785651801,
233
+ "clip_ratio/high_mean": 0.00014942743928259006,
234
+ "clip_ratio/low_mean": 1.5427591733896406e-05,
235
+ "clip_ratio/low_min": 0.0,
236
+ "clip_ratio/region_mean": 0.00016485502828800236,
237
+ "completions/clipped_ratio": 0.0,
238
+ "completions/max_length": 440.5625,
239
+ "completions/max_terminated_length": 440.5625,
240
+ "completions/mean_length": 376.814453125,
241
+ "completions/mean_terminated_length": 376.814453125,
242
+ "completions/min_length": 319.4375,
243
+ "completions/min_terminated_length": 319.4375,
244
+ "entropy": 0.06541132956044748,
245
+ "epoch": 0.3210702341137124,
246
+ "frac_reward_zero_std": 0.90625,
247
+ "grad_norm": 1.5393034219741821,
248
+ "kl": 0.12934908317402005,
249
+ "learning_rate": 1.786510590858417e-06,
250
+ "loss": 0.0147,
251
+ "num_tokens": 1933899.0,
252
+ "reward": 0.736328125,
253
+ "reward_std": 0.030584799125790596,
254
+ "rewards/accuracy_reward/mean": 0.736328125,
255
+ "rewards/accuracy_reward/std": 0.19989654514938593,
256
+ "step": 384
257
+ },
258
+ {
259
+ "epoch": 0.33444816053511706,
260
+ "eval_completions/clipped_ratio": 0.0,
261
+ "eval_completions/max_length": 472.5,
262
+ "eval_completions/max_terminated_length": 472.5,
263
+ "eval_completions/mean_length": 369.375,
264
+ "eval_completions/mean_terminated_length": 369.375,
265
+ "eval_completions/min_length": 267.0,
266
+ "eval_completions/min_terminated_length": 267.0,
267
+ "eval_frac_reward_zero_std": 0.0,
268
+ "eval_loss": 0.0,
269
+ "eval_num_tokens": 2016137.0,
270
+ "eval_reward": 0.7421875,
271
+ "eval_reward_std": 0.0,
272
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
273
+ "eval_rewards/accuracy_reward/std": 0.4400599002838135,
274
+ "eval_runtime": 53.733,
275
+ "eval_samples_per_second": 1.396,
276
+ "eval_steps_per_second": 0.037,
277
+ "step": 400
278
+ },
279
+ {
280
+ "clip_ratio/high_max": 0.00030761394494523603,
281
+ "clip_ratio/high_mean": 0.00015380697247261801,
282
+ "clip_ratio/low_mean": 7.599783687813517e-05,
283
+ "clip_ratio/low_min": 2.567262223844106e-05,
284
+ "clip_ratio/region_mean": 0.000229804812988732,
285
+ "completions/clipped_ratio": 0.0,
286
+ "completions/max_length": 449.25,
287
+ "completions/max_terminated_length": 449.25,
288
+ "completions/mean_length": 372.3984375,
289
+ "completions/mean_terminated_length": 372.3984375,
290
+ "completions/min_length": 303.3333333333333,
291
+ "completions/min_terminated_length": 303.3333333333333,
292
+ "entropy": 0.06088081297154228,
293
+ "epoch": 0.3745819397993311,
294
+ "frac_reward_zero_std": 0.875,
295
+ "grad_norm": 0.27695611119270325,
296
+ "kl": 0.10456895603177448,
297
+ "learning_rate": 1.7508361204013377e-06,
298
+ "loss": 0.0087,
299
+ "num_tokens": 2259762.0,
300
+ "reward": 0.8151041666666666,
301
+ "reward_std": 0.04548187553882599,
302
+ "rewards/accuracy_reward/mean": 0.8151041666666666,
303
+ "rewards/accuracy_reward/std": 0.20229721814393997,
304
+ "step": 448
305
+ },
306
+ {
307
+ "epoch": 0.4180602006688963,
308
+ "eval_completions/clipped_ratio": 0.0,
309
+ "eval_completions/max_length": 449.5,
310
+ "eval_completions/max_terminated_length": 449.5,
311
+ "eval_completions/mean_length": 367.234375,
312
+ "eval_completions/mean_terminated_length": 367.234375,
313
+ "eval_completions/min_length": 264.0,
314
+ "eval_completions/min_terminated_length": 264.0,
315
+ "eval_frac_reward_zero_std": 0.0,
316
+ "eval_loss": 0.0,
317
+ "eval_num_tokens": 2526718.0,
318
+ "eval_reward": 0.75,
319
+ "eval_reward_std": 0.0,
320
+ "eval_rewards/accuracy_reward/mean": 0.75,
321
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
322
+ "eval_runtime": 50.9559,
323
+ "eval_samples_per_second": 1.472,
324
+ "eval_steps_per_second": 0.039,
325
+ "step": 500
326
+ },
327
+ {
328
+ "clip_ratio/high_max": 0.00046076797298155725,
329
+ "clip_ratio/high_mean": 0.00023038398649077863,
330
+ "clip_ratio/low_mean": 0.0,
331
+ "clip_ratio/low_min": 0.0,
332
+ "clip_ratio/region_mean": 0.00023038398649077863,
333
+ "completions/clipped_ratio": 0.0,
334
+ "completions/max_length": 457.0,
335
+ "completions/max_terminated_length": 457.0,
336
+ "completions/mean_length": 376.90625,
337
+ "completions/mean_terminated_length": 376.90625,
338
+ "completions/min_length": 320.6666666666667,
339
+ "completions/min_terminated_length": 320.6666666666667,
340
+ "entropy": 0.055613856141765915,
341
+ "epoch": 0.4280936454849498,
342
+ "frac_reward_zero_std": 0.8333333333333334,
343
+ "grad_norm": 0.21719853579998016,
344
+ "kl": 0.09068247520675261,
345
+ "learning_rate": 1.7151616499442586e-06,
346
+ "loss": 0.0055,
347
+ "num_tokens": 2588085.0,
348
+ "reward": 0.9791666666666666,
349
+ "reward_std": 0.056927502155303955,
350
+ "rewards/accuracy_reward/mean": 0.9791666666666666,
351
+ "rewards/accuracy_reward/std": 0.08197822670141856,
352
+ "step": 512
353
+ },
354
+ {
355
+ "clip_ratio/high_max": 0.0003454105735727353,
356
+ "clip_ratio/high_mean": 0.00017754574582795613,
357
+ "clip_ratio/low_mean": 2.208435944339726e-05,
358
+ "clip_ratio/low_min": 0.0,
359
+ "clip_ratio/region_mean": 0.00019963010390711133,
360
+ "completions/clipped_ratio": 0.0,
361
+ "completions/max_length": 435.5,
362
+ "completions/max_terminated_length": 435.5,
363
+ "completions/mean_length": 377.986328125,
364
+ "completions/mean_terminated_length": 377.986328125,
365
+ "completions/min_length": 322.0,
366
+ "completions/min_terminated_length": 322.0,
367
+ "entropy": 0.058247681823559105,
368
+ "epoch": 0.4816053511705686,
369
+ "frac_reward_zero_std": 0.875,
370
+ "grad_norm": 0.20317593216896057,
371
+ "kl": 0.09756386690423824,
372
+ "learning_rate": 1.6794871794871794e-06,
373
+ "loss": 0.0125,
374
+ "num_tokens": 2915838.0,
375
+ "reward": 0.767578125,
376
+ "reward_std": 0.04478531330823898,
377
+ "rewards/accuracy_reward/mean": 0.767578125,
378
+ "rewards/accuracy_reward/std": 0.17305080499500036,
379
+ "step": 576
380
+ },
381
+ {
382
+ "epoch": 0.5016722408026756,
383
+ "eval_completions/clipped_ratio": 0.0,
384
+ "eval_completions/max_length": 470.0,
385
+ "eval_completions/max_terminated_length": 470.0,
386
+ "eval_completions/mean_length": 368.40625,
387
+ "eval_completions/mean_terminated_length": 368.40625,
388
+ "eval_completions/min_length": 265.5,
389
+ "eval_completions/min_terminated_length": 265.5,
390
+ "eval_frac_reward_zero_std": 0.0,
391
+ "eval_loss": 0.0,
392
+ "eval_num_tokens": 3034241.0,
393
+ "eval_reward": 0.75,
394
+ "eval_reward_std": 0.0,
395
+ "eval_rewards/accuracy_reward/mean": 0.75,
396
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
397
+ "eval_runtime": 53.2723,
398
+ "eval_samples_per_second": 1.408,
399
+ "eval_steps_per_second": 0.038,
400
+ "step": 600
401
+ },
402
+ {
403
+ "clip_ratio/high_max": 0.00014167100307531654,
404
+ "clip_ratio/high_mean": 7.083550153765827e-05,
405
+ "clip_ratio/low_mean": 8.90313385752961e-06,
406
+ "clip_ratio/low_min": 0.0,
407
+ "clip_ratio/region_mean": 7.973863685037941e-05,
408
+ "completions/clipped_ratio": 0.0,
409
+ "completions/max_length": 430.1,
410
+ "completions/max_terminated_length": 430.1,
411
+ "completions/mean_length": 379.065625,
412
+ "completions/mean_terminated_length": 379.065625,
413
+ "completions/min_length": 333.6,
414
+ "completions/min_terminated_length": 333.6,
415
+ "entropy": 0.04927852246910334,
416
+ "epoch": 0.5351170568561873,
417
+ "frac_reward_zero_std": 0.95,
418
+ "grad_norm": 0.3349158465862274,
419
+ "kl": 0.06723399083130062,
420
+ "learning_rate": 1.6438127090301002e-06,
421
+ "loss": 0.0096,
422
+ "num_tokens": 3239334.0,
423
+ "reward": 0.896875,
424
+ "reward_std": 0.0125,
425
+ "rewards/accuracy_reward/mean": 0.896875,
426
+ "rewards/accuracy_reward/std": 0.11927776783704758,
427
+ "step": 640
428
+ },
429
+ {
430
+ "epoch": 0.5852842809364549,
431
+ "eval_completions/clipped_ratio": 0.0,
432
+ "eval_completions/max_length": 466.0,
433
+ "eval_completions/max_terminated_length": 466.0,
434
+ "eval_completions/mean_length": 368.328125,
435
+ "eval_completions/mean_terminated_length": 368.328125,
436
+ "eval_completions/min_length": 267.0,
437
+ "eval_completions/min_terminated_length": 267.0,
438
+ "eval_frac_reward_zero_std": 0.0,
439
+ "eval_loss": 0.0,
440
+ "eval_num_tokens": 3548491.0,
441
+ "eval_reward": 0.75,
442
+ "eval_reward_std": 0.0,
443
+ "eval_rewards/accuracy_reward/mean": 0.75,
444
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
445
+ "eval_runtime": 52.9481,
446
+ "eval_samples_per_second": 1.416,
447
+ "eval_steps_per_second": 0.038,
448
+ "step": 700
449
+ },
450
+ {
451
+ "clip_ratio/high_max": 0.0,
452
+ "clip_ratio/high_mean": 0.0,
453
+ "clip_ratio/low_mean": 0.0,
454
+ "clip_ratio/low_min": 0.0,
455
+ "clip_ratio/region_mean": 0.0,
456
+ "completions/clipped_ratio": 0.0,
457
+ "completions/max_length": 453.0,
458
+ "completions/max_terminated_length": 453.0,
459
+ "completions/mean_length": 413.65625,
460
+ "completions/mean_terminated_length": 413.65625,
461
+ "completions/min_length": 378.0,
462
+ "completions/min_terminated_length": 378.0,
463
+ "entropy": 0.05146836768835783,
464
+ "epoch": 0.5886287625418061,
465
+ "frac_reward_zero_std": 1.0,
466
+ "grad_norm": 0.23372304439544678,
467
+ "kl": 0.082071078941226,
468
+ "learning_rate": 1.608138238573021e-06,
469
+ "loss": 0.0073,
470
+ "num_tokens": 3570128.0,
471
+ "reward": 0.5,
472
+ "reward_std": 0.0,
473
+ "rewards/accuracy_reward/mean": 0.5,
474
+ "rewards/accuracy_reward/std": 0.5080004930496216,
475
+ "step": 704
476
+ },
477
+ {
478
+ "clip_ratio/high_max": 0.00030011342369107297,
479
+ "clip_ratio/high_mean": 0.00015005671184553648,
480
+ "clip_ratio/low_mean": 0.00011306512169539928,
481
+ "clip_ratio/low_min": 0.0,
482
+ "clip_ratio/region_mean": 0.00026312183535992517,
483
+ "completions/clipped_ratio": 0.0,
484
+ "completions/max_length": 424.0625,
485
+ "completions/max_terminated_length": 424.0625,
486
+ "completions/mean_length": 364.509765625,
487
+ "completions/mean_terminated_length": 364.509765625,
488
+ "completions/min_length": 314.1875,
489
+ "completions/min_terminated_length": 314.1875,
490
+ "entropy": 0.04728981165681034,
491
+ "epoch": 0.6421404682274248,
492
+ "frac_reward_zero_std": 0.84375,
493
+ "grad_norm": 0.14771811664104462,
494
+ "kl": 0.05484659186549834,
495
+ "learning_rate": 1.5724637681159421e-06,
496
+ "loss": 0.0061,
497
+ "num_tokens": 3890933.0,
498
+ "reward": 0.841796875,
499
+ "reward_std": 0.04957009106874466,
500
+ "rewards/accuracy_reward/mean": 0.841796875,
501
+ "rewards/accuracy_reward/std": 0.19991536159068346,
502
+ "step": 768
503
+ },
504
+ {
505
+ "epoch": 0.6688963210702341,
506
+ "eval_completions/clipped_ratio": 0.0,
507
+ "eval_completions/max_length": 438.0,
508
+ "eval_completions/max_terminated_length": 438.0,
509
+ "eval_completions/mean_length": 365.59375,
510
+ "eval_completions/mean_terminated_length": 365.59375,
511
+ "eval_completions/min_length": 270.5,
512
+ "eval_completions/min_terminated_length": 270.5,
513
+ "eval_frac_reward_zero_std": 0.0,
514
+ "eval_loss": 0.0,
515
+ "eval_num_tokens": 4052061.0,
516
+ "eval_reward": 0.75,
517
+ "eval_reward_std": 0.0,
518
+ "eval_rewards/accuracy_reward/mean": 0.75,
519
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
520
+ "eval_runtime": 49.6177,
521
+ "eval_samples_per_second": 1.512,
522
+ "eval_steps_per_second": 0.04,
523
+ "step": 800
524
+ },
525
+ {
526
+ "clip_ratio/high_max": 0.0008106039349513594,
527
+ "clip_ratio/high_mean": 0.0004053019674756797,
528
+ "clip_ratio/low_mean": 0.00016369877903343877,
529
+ "clip_ratio/low_min": 0.0,
530
+ "clip_ratio/region_mean": 0.0005690007437806344,
531
+ "completions/clipped_ratio": 0.0,
532
+ "completions/max_length": 439.25,
533
+ "completions/max_terminated_length": 439.25,
534
+ "completions/mean_length": 382.9375,
535
+ "completions/mean_terminated_length": 382.9375,
536
+ "completions/min_length": 335.875,
537
+ "completions/min_terminated_length": 335.875,
538
+ "entropy": 0.052825499267783016,
539
+ "epoch": 0.6956521739130435,
540
+ "frac_reward_zero_std": 0.6875,
541
+ "grad_norm": 4.8682661056518555,
542
+ "kl": 0.06570573046337813,
543
+ "learning_rate": 1.5367892976588628e-06,
544
+ "loss": 0.007,
545
+ "num_tokens": 4217117.0,
546
+ "reward": 0.7890625,
547
+ "reward_std": 0.09341736882925034,
548
+ "rewards/accuracy_reward/mean": 0.7890625,
549
+ "rewards/accuracy_reward/std": 0.2336982600390911,
550
+ "step": 832
551
+ },
552
+ {
553
+ "clip_ratio/high_max": 0.00012973563389095943,
554
+ "clip_ratio/high_mean": 6.486781694547972e-05,
555
+ "clip_ratio/low_mean": 5.695884419765207e-05,
556
+ "clip_ratio/low_min": 0.0,
557
+ "clip_ratio/region_mean": 0.00012182666205262649,
558
+ "completions/clipped_ratio": 0.0,
559
+ "completions/max_length": 426.75,
560
+ "completions/max_terminated_length": 426.75,
561
+ "completions/mean_length": 370.6953125,
562
+ "completions/mean_terminated_length": 370.6953125,
563
+ "completions/min_length": 323.25,
564
+ "completions/min_terminated_length": 323.25,
565
+ "entropy": 0.0439882168720942,
566
+ "epoch": 0.7491638795986622,
567
+ "frac_reward_zero_std": 0.9375,
568
+ "grad_norm": 0.1999923586845398,
569
+ "kl": 0.05729101897304645,
570
+ "learning_rate": 1.5011148272017838e-06,
571
+ "loss": 0.0087,
572
+ "num_tokens": 4541185.0,
573
+ "reward": 0.869140625,
574
+ "reward_std": 0.02998628467321396,
575
+ "rewards/accuracy_reward/mean": 0.869140625,
576
+ "rewards/accuracy_reward/std": 0.15224276669323444,
577
+ "step": 896
578
+ },
579
+ {
580
+ "epoch": 0.7525083612040134,
581
+ "eval_completions/clipped_ratio": 0.0,
582
+ "eval_completions/max_length": 465.5,
583
+ "eval_completions/max_terminated_length": 465.5,
584
+ "eval_completions/mean_length": 370.953125,
585
+ "eval_completions/mean_terminated_length": 370.953125,
586
+ "eval_completions/min_length": 266.5,
587
+ "eval_completions/min_terminated_length": 266.5,
588
+ "eval_frac_reward_zero_std": 0.0,
589
+ "eval_loss": 0.0,
590
+ "eval_num_tokens": 4560908.0,
591
+ "eval_reward": 0.75,
592
+ "eval_reward_std": 0.0,
593
+ "eval_rewards/accuracy_reward/mean": 0.75,
594
+ "eval_rewards/accuracy_reward/std": 0.43491509556770325,
595
+ "eval_runtime": 52.6548,
596
+ "eval_samples_per_second": 1.424,
597
+ "eval_steps_per_second": 0.038,
598
+ "step": 900
599
+ },
600
+ {
601
+ "clip_ratio/high_max": 0.0002714004241473352,
602
+ "clip_ratio/high_mean": 0.0001357002120736676,
603
+ "clip_ratio/low_mean": 5.78916457016021e-05,
604
+ "clip_ratio/low_min": 0.0,
605
+ "clip_ratio/region_mean": 0.00019359185243956746,
606
+ "completions/clipped_ratio": 0.0,
607
+ "completions/max_length": 420.3333333333333,
608
+ "completions/max_terminated_length": 420.3333333333333,
609
+ "completions/mean_length": 372.46458333333334,
610
+ "completions/mean_terminated_length": 372.46458333333334,
611
+ "completions/min_length": 334.6666666666667,
612
+ "completions/min_terminated_length": 334.6666666666667,
613
+ "entropy": 0.04479737759878238,
614
+ "epoch": 0.802675585284281,
615
+ "frac_reward_zero_std": 0.9,
616
+ "grad_norm": 0.34151706099510193,
617
+ "kl": 0.049933125358074905,
618
+ "learning_rate": 1.4654403567447044e-06,
619
+ "loss": 0.0063,
620
+ "num_tokens": 4865467.0,
621
+ "reward": 0.71875,
622
+ "reward_std": 0.03814757267634074,
623
+ "rewards/accuracy_reward/mean": 0.71875,
624
+ "rewards/accuracy_reward/std": 0.1920533448457718,
625
+ "step": 960
626
+ },
627
+ {
628
+ "epoch": 0.8361204013377926,
629
+ "eval_completions/clipped_ratio": 0.0,
630
+ "eval_completions/max_length": 447.0,
631
+ "eval_completions/max_terminated_length": 447.0,
632
+ "eval_completions/mean_length": 368.5078125,
633
+ "eval_completions/mean_terminated_length": 368.5078125,
634
+ "eval_completions/min_length": 266.0,
635
+ "eval_completions/min_terminated_length": 266.0,
636
+ "eval_frac_reward_zero_std": 0.0,
637
+ "eval_loss": 0.0,
638
+ "eval_num_tokens": 5065573.0,
639
+ "eval_reward": 0.765625,
640
+ "eval_reward_std": 0.0,
641
+ "eval_rewards/accuracy_reward/mean": 0.765625,
642
+ "eval_rewards/accuracy_reward/std": 0.42695629596710205,
643
+ "eval_runtime": 50.7532,
644
+ "eval_samples_per_second": 1.478,
645
+ "eval_steps_per_second": 0.039,
646
+ "step": 1000
647
+ },
648
+ {
649
+ "clip_ratio/high_max": 0.00022080155273821825,
650
+ "clip_ratio/high_mean": 0.00011040077636910912,
651
+ "clip_ratio/low_mean": 0.0002929158811942519,
652
+ "clip_ratio/low_min": 0.0,
653
+ "clip_ratio/region_mean": 0.00040331665756336105,
654
+ "completions/clipped_ratio": 0.0,
655
+ "completions/max_length": 409.6666666666667,
656
+ "completions/max_terminated_length": 409.6666666666667,
657
+ "completions/mean_length": 347.8958333333333,
658
+ "completions/mean_terminated_length": 347.8958333333333,
659
+ "completions/min_length": 302.8333333333333,
660
+ "completions/min_terminated_length": 302.8333333333333,
661
+ "entropy": 0.04160856680634121,
662
+ "epoch": 0.8561872909698997,
663
+ "frac_reward_zero_std": 0.75,
664
+ "grad_norm": 0.6067283749580383,
665
+ "kl": 0.046552122434756406,
666
+ "learning_rate": 1.4297658862876255e-06,
667
+ "loss": 0.004,
668
+ "num_tokens": 5182625.0,
669
+ "reward": 0.7604166666666666,
670
+ "reward_std": 0.07013041774431865,
671
+ "rewards/accuracy_reward/mean": 0.7604166666666666,
672
+ "rewards/accuracy_reward/std": 0.14292824765046439,
673
+ "step": 1024
674
+ },
675
+ {
676
+ "clip_ratio/high_max": 0.0002381671583862044,
677
+ "clip_ratio/high_mean": 0.00012541975047497544,
678
+ "clip_ratio/low_mean": 2.7038077860197518e-05,
679
+ "clip_ratio/low_min": 0.0,
680
+ "clip_ratio/region_mean": 0.0001524578278804256,
681
+ "completions/clipped_ratio": 0.0,
682
+ "completions/max_length": 414.375,
683
+ "completions/max_terminated_length": 414.375,
684
+ "completions/mean_length": 359.837890625,
685
+ "completions/mean_terminated_length": 359.837890625,
686
+ "completions/min_length": 318.75,
687
+ "completions/min_terminated_length": 318.75,
688
+ "entropy": 0.0410496371332556,
689
+ "epoch": 0.9096989966555183,
690
+ "frac_reward_zero_std": 0.875,
691
+ "grad_norm": 2.2683029174804688,
692
+ "kl": 0.04194089516022359,
693
+ "learning_rate": 1.394091415830546e-06,
694
+ "loss": 0.0057,
695
+ "num_tokens": 5500830.0,
696
+ "reward": 0.822265625,
697
+ "reward_std": 0.04230976663529873,
698
+ "rewards/accuracy_reward/mean": 0.822265625,
699
+ "rewards/accuracy_reward/std": 0.14916588738560677,
700
+ "step": 1088
701
+ },
702
+ {
703
+ "epoch": 0.919732441471572,
704
+ "eval_completions/clipped_ratio": 0.0,
705
+ "eval_completions/max_length": 450.5,
706
+ "eval_completions/max_terminated_length": 450.5,
707
+ "eval_completions/mean_length": 368.5625,
708
+ "eval_completions/mean_terminated_length": 368.5625,
709
+ "eval_completions/min_length": 265.5,
710
+ "eval_completions/min_terminated_length": 265.5,
711
+ "eval_frac_reward_zero_std": 0.0,
712
+ "eval_loss": 0.0,
713
+ "eval_num_tokens": 5561185.0,
714
+ "eval_reward": 0.765625,
715
+ "eval_reward_std": 0.0,
716
+ "eval_rewards/accuracy_reward/mean": 0.765625,
717
+ "eval_rewards/accuracy_reward/std": 0.42695629596710205,
718
+ "eval_runtime": 52.0329,
719
+ "eval_samples_per_second": 1.441,
720
+ "eval_steps_per_second": 0.038,
721
+ "step": 1100
722
+ },
723
+ {
724
+ "clip_ratio/high_max": 0.0002082675133491508,
725
+ "clip_ratio/high_mean": 0.00013098904533902757,
726
+ "clip_ratio/low_mean": 8.385539303372543e-05,
727
+ "clip_ratio/low_min": 0.0,
728
+ "clip_ratio/region_mean": 0.00021484444396964347,
729
+ "completions/clipped_ratio": 0.0,
730
+ "completions/max_length": 424.84615384615387,
731
+ "completions/max_terminated_length": 424.84615384615387,
732
+ "completions/mean_length": 366.24278846153845,
733
+ "completions/mean_terminated_length": 366.24278846153845,
734
+ "completions/min_length": 326.53846153846155,
735
+ "completions/min_terminated_length": 326.53846153846155,
736
+ "entropy": 0.04085898066226106,
737
+ "epoch": 0.9632107023411371,
738
+ "frac_reward_zero_std": 0.8076923076923077,
739
+ "grad_norm": 2.3284337520599365,
740
+ "kl": 0.04895740425070891,
741
+ "learning_rate": 1.3584169453734671e-06,
742
+ "loss": 0.0052,
743
+ "num_tokens": 5822518.0,
744
+ "reward": 0.7764423076923077,
745
+ "reward_std": 0.056873598924049966,
746
+ "rewards/accuracy_reward/mean": 0.7764423076923077,
747
+ "rewards/accuracy_reward/std": 0.18679064741501442,
748
+ "step": 1152
749
+ },
750
+ {
751
+ "epoch": 1.0033444816053512,
752
+ "eval_completions/clipped_ratio": 0.0,
753
+ "eval_completions/max_length": 456.0,
754
+ "eval_completions/max_terminated_length": 456.0,
755
+ "eval_completions/mean_length": 366.6328125,
756
+ "eval_completions/mean_terminated_length": 366.6328125,
757
+ "eval_completions/min_length": 264.5,
758
+ "eval_completions/min_terminated_length": 264.5,
759
+ "eval_frac_reward_zero_std": 0.0,
760
+ "eval_loss": 0.0,
761
+ "eval_num_tokens": 6066300.0,
762
+ "eval_reward": 0.7578125,
763
+ "eval_reward_std": 0.0,
764
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
765
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
766
+ "eval_runtime": 51.5856,
767
+ "eval_samples_per_second": 1.454,
768
+ "eval_steps_per_second": 0.039,
769
+ "step": 1200
770
+ },
771
+ {
772
+ "clip_ratio/high_max": 0.0,
773
+ "clip_ratio/high_mean": 0.0,
774
+ "clip_ratio/low_mean": 6.308695265033748e-05,
775
+ "clip_ratio/low_min": 0.0,
776
+ "clip_ratio/region_mean": 6.308695265033748e-05,
777
+ "completions/clipped_ratio": 0.0,
778
+ "completions/max_length": 425.0,
779
+ "completions/max_terminated_length": 425.0,
780
+ "completions/mean_length": 374.2578125,
781
+ "completions/mean_terminated_length": 374.2578125,
782
+ "completions/min_length": 339.5,
783
+ "completions/min_terminated_length": 339.5,
784
+ "entropy": 0.04189286706969142,
785
+ "epoch": 1.0167224080267558,
786
+ "frac_reward_zero_std": 0.875,
787
+ "grad_norm": 0.22261899709701538,
788
+ "kl": 0.04722271178616211,
789
+ "learning_rate": 1.3227424749163878e-06,
790
+ "loss": 0.0023,
791
+ "num_tokens": 6147757.0,
792
+ "reward": 0.7578125,
793
+ "reward_std": 0.03125,
794
+ "rewards/accuracy_reward/mean": 0.7578125,
795
+ "rewards/accuracy_reward/std": 0.2537519633769989,
796
+ "step": 1216
797
+ },
798
+ {
799
+ "clip_ratio/high_max": 0.00024013888196350308,
800
+ "clip_ratio/high_mean": 0.00012006944098175154,
801
+ "clip_ratio/low_mean": 4.1882908590196166e-05,
802
+ "clip_ratio/low_min": 0.0,
803
+ "clip_ratio/region_mean": 0.00016195235002669506,
804
+ "completions/clipped_ratio": 0.0,
805
+ "completions/max_length": 415.25,
806
+ "completions/max_terminated_length": 415.25,
807
+ "completions/mean_length": 357.83984375,
808
+ "completions/mean_terminated_length": 357.83984375,
809
+ "completions/min_length": 317.75,
810
+ "completions/min_terminated_length": 317.75,
811
+ "entropy": 0.03656601076363586,
812
+ "epoch": 1.0702341137123745,
813
+ "frac_reward_zero_std": 0.9375,
814
+ "grad_norm": 0.15465405583381653,
815
+ "kl": 0.03583414322019962,
816
+ "learning_rate": 1.2870680044593088e-06,
817
+ "loss": 0.0001,
818
+ "num_tokens": 6464971.0,
819
+ "reward": 0.873046875,
820
+ "reward_std": 0.030584799125790596,
821
+ "rewards/accuracy_reward/mean": 0.873046875,
822
+ "rewards/accuracy_reward/std": 0.15019487217068672,
823
+ "step": 1280
824
+ },
825
+ {
826
+ "epoch": 1.0869565217391304,
827
+ "eval_completions/clipped_ratio": 0.0,
828
+ "eval_completions/max_length": 474.5,
829
+ "eval_completions/max_terminated_length": 474.5,
830
+ "eval_completions/mean_length": 369.21875,
831
+ "eval_completions/mean_terminated_length": 369.21875,
832
+ "eval_completions/min_length": 260.0,
833
+ "eval_completions/min_terminated_length": 260.0,
834
+ "eval_frac_reward_zero_std": 0.0,
835
+ "eval_loss": 0.0,
836
+ "eval_num_tokens": 6562408.0,
837
+ "eval_reward": 0.765625,
838
+ "eval_reward_std": 0.0,
839
+ "eval_rewards/accuracy_reward/mean": 0.765625,
840
+ "eval_rewards/accuracy_reward/std": 0.4265512377023697,
841
+ "eval_runtime": 53.2759,
842
+ "eval_samples_per_second": 1.408,
843
+ "eval_steps_per_second": 0.038,
844
+ "step": 1300
845
+ },
846
+ {
847
+ "clip_ratio/high_max": 0.0,
848
+ "clip_ratio/high_mean": 0.0,
849
+ "clip_ratio/low_mean": 5.375383294780146e-05,
850
+ "clip_ratio/low_min": 0.0,
851
+ "clip_ratio/region_mean": 5.375383294780146e-05,
852
+ "completions/clipped_ratio": 0.0,
853
+ "completions/max_length": 428.8181818181818,
854
+ "completions/max_terminated_length": 428.8181818181818,
855
+ "completions/mean_length": 372.14204545454544,
856
+ "completions/mean_terminated_length": 372.14204545454544,
857
+ "completions/min_length": 326.8181818181818,
858
+ "completions/min_terminated_length": 326.8181818181818,
859
+ "entropy": 0.03957836308770559,
860
+ "epoch": 1.1237458193979932,
861
+ "frac_reward_zero_std": 0.9545454545454546,
862
+ "grad_norm": 0.2289930284023285,
863
+ "kl": 0.04065175803208893,
864
+ "learning_rate": 1.2513935340022296e-06,
865
+ "loss": 0.0034,
866
+ "num_tokens": 6785642.0,
867
+ "reward": 0.8693181818181818,
868
+ "reward_std": 0.015525682405991987,
869
+ "rewards/accuracy_reward/mean": 0.8693181818181818,
870
+ "rewards/accuracy_reward/std": 0.13818337158723312,
871
+ "step": 1344
872
+ },
873
+ {
874
+ "epoch": 1.1705685618729098,
875
+ "eval_completions/clipped_ratio": 0.0,
876
+ "eval_completions/max_length": 464.0,
877
+ "eval_completions/max_terminated_length": 464.0,
878
+ "eval_completions/mean_length": 366.890625,
879
+ "eval_completions/mean_terminated_length": 366.890625,
880
+ "eval_completions/min_length": 268.0,
881
+ "eval_completions/min_terminated_length": 268.0,
882
+ "eval_frac_reward_zero_std": 0.0,
883
+ "eval_loss": 0.0,
884
+ "eval_num_tokens": 7066640.0,
885
+ "eval_reward": 0.765625,
886
+ "eval_reward_std": 0.0,
887
+ "eval_rewards/accuracy_reward/mean": 0.765625,
888
+ "eval_rewards/accuracy_reward/std": 0.4265512377023697,
889
+ "eval_runtime": 52.2108,
890
+ "eval_samples_per_second": 1.436,
891
+ "eval_steps_per_second": 0.038,
892
+ "step": 1400
893
+ },
894
+ {
895
+ "clip_ratio/high_max": 0.0,
896
+ "clip_ratio/high_mean": 0.0,
897
+ "clip_ratio/low_mean": 8.331963908858597e-05,
898
+ "clip_ratio/low_min": 0.0,
899
+ "clip_ratio/region_mean": 8.331963908858597e-05,
900
+ "completions/clipped_ratio": 0.0,
901
+ "completions/max_length": 403.5,
902
+ "completions/max_terminated_length": 403.5,
903
+ "completions/mean_length": 348.8125,
904
+ "completions/mean_terminated_length": 348.8125,
905
+ "completions/min_length": 298.5,
906
+ "completions/min_terminated_length": 298.5,
907
+ "entropy": 0.03495068661868572,
908
+ "epoch": 1.1772575250836121,
909
+ "frac_reward_zero_std": 0.75,
910
+ "grad_norm": 4.837547302246094,
911
+ "kl": 0.01356734142791538,
912
+ "learning_rate": 1.2157190635451505e-06,
913
+ "loss": 0.0048,
914
+ "num_tokens": 7105684.0,
915
+ "reward": 0.265625,
916
+ "reward_std": 0.0625,
917
+ "rewards/accuracy_reward/mean": 0.265625,
918
+ "rewards/accuracy_reward/std": 0.253503680229187,
919
+ "step": 1408
920
+ },
921
+ {
922
+ "clip_ratio/high_max": 0.00030065215287322644,
923
+ "clip_ratio/high_mean": 0.00015032607643661322,
924
+ "clip_ratio/low_mean": 0.0002008761889555899,
925
+ "clip_ratio/low_min": 0.0,
926
+ "clip_ratio/region_mean": 0.0003512022635732137,
927
+ "completions/clipped_ratio": 0.0,
928
+ "completions/max_length": 431.9375,
929
+ "completions/max_terminated_length": 431.9375,
930
+ "completions/mean_length": 371.126953125,
931
+ "completions/mean_terminated_length": 371.126953125,
932
+ "completions/min_length": 330.5,
933
+ "completions/min_terminated_length": 330.5,
934
+ "entropy": 0.04257212101947516,
935
+ "epoch": 1.2307692307692308,
936
+ "frac_reward_zero_std": 0.78125,
937
+ "grad_norm": 0.13189245760440826,
938
+ "kl": 0.043384918705669406,
939
+ "learning_rate": 1.1800445930880713e-06,
940
+ "loss": 0.0066,
941
+ "num_tokens": 7429781.0,
942
+ "reward": 0.751953125,
943
+ "reward_std": 0.07564390823245049,
944
+ "rewards/accuracy_reward/mean": 0.751953125,
945
+ "rewards/accuracy_reward/std": 0.1867057103663683,
946
+ "step": 1472
947
+ },
948
+ {
949
+ "epoch": 1.254180602006689,
950
+ "eval_completions/clipped_ratio": 0.0,
951
+ "eval_completions/max_length": 461.0,
952
+ "eval_completions/max_terminated_length": 461.0,
953
+ "eval_completions/mean_length": 366.9140625,
954
+ "eval_completions/mean_terminated_length": 366.9140625,
955
+ "eval_completions/min_length": 270.0,
956
+ "eval_completions/min_terminated_length": 270.0,
957
+ "eval_frac_reward_zero_std": 0.0,
958
+ "eval_loss": 0.0,
959
+ "eval_num_tokens": 7570336.0,
960
+ "eval_reward": 0.7421875,
961
+ "eval_reward_std": 0.0,
962
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
963
+ "eval_rewards/accuracy_reward/std": 0.440796360373497,
964
+ "eval_runtime": 52.2808,
965
+ "eval_samples_per_second": 1.435,
966
+ "eval_steps_per_second": 0.038,
967
+ "step": 1500
968
+ },
969
+ {
970
+ "clip_ratio/high_max": 5.763024738472369e-05,
971
+ "clip_ratio/high_mean": 2.8815123692361845e-05,
972
+ "clip_ratio/low_mean": 5.96528205076336e-05,
973
+ "clip_ratio/low_min": 0.0,
974
+ "clip_ratio/region_mean": 8.846794339155572e-05,
975
+ "completions/clipped_ratio": 0.0,
976
+ "completions/max_length": 427.6666666666667,
977
+ "completions/max_terminated_length": 427.6666666666667,
978
+ "completions/mean_length": 371.97569444444446,
979
+ "completions/mean_terminated_length": 371.97569444444446,
980
+ "completions/min_length": 325.77777777777777,
981
+ "completions/min_terminated_length": 325.77777777777777,
982
+ "entropy": 0.03820225586079889,
983
+ "epoch": 1.2842809364548495,
984
+ "frac_reward_zero_std": 0.9444444444444444,
985
+ "grad_norm": 0.10654129832983017,
986
+ "kl": 0.04006648767325613,
987
+ "learning_rate": 1.1443701226309922e-06,
988
+ "loss": 0.004,
989
+ "num_tokens": 7752969.0,
990
+ "reward": 0.7916666666666666,
991
+ "reward_std": 0.02484519945250617,
992
+ "rewards/accuracy_reward/mean": 0.7916666666666666,
993
+ "rewards/accuracy_reward/std": 0.22398564881748623,
994
+ "step": 1536
995
+ },
996
+ {
997
+ "clip_ratio/high_max": 0.0005374772144932649,
998
+ "clip_ratio/high_mean": 0.00027859043575517717,
999
+ "clip_ratio/low_mean": 6.284001574385911e-05,
1000
+ "clip_ratio/low_min": 0.0,
1001
+ "clip_ratio/region_mean": 0.0003414304496800469,
1002
+ "completions/clipped_ratio": 0.0,
1003
+ "completions/max_length": 412.6875,
1004
+ "completions/max_terminated_length": 412.6875,
1005
+ "completions/mean_length": 366.66015625,
1006
+ "completions/mean_terminated_length": 366.66015625,
1007
+ "completions/min_length": 330.625,
1008
+ "completions/min_terminated_length": 330.625,
1009
+ "entropy": 0.037426625698572025,
1010
+ "epoch": 1.3377926421404682,
1011
+ "frac_reward_zero_std": 0.78125,
1012
+ "grad_norm": 0.09183552861213684,
1013
+ "kl": 0.037651331345841754,
1014
+ "learning_rate": 1.108695652173913e-06,
1015
+ "loss": 0.005,
1016
+ "num_tokens": 8074795.0,
1017
+ "reward": 0.865234375,
1018
+ "reward_std": 0.07917051576077938,
1019
+ "rewards/accuracy_reward/mean": 0.865234375,
1020
+ "rewards/accuracy_reward/std": 0.18074841145426035,
1021
+ "step": 1600
1022
+ },
1023
+ {
1024
+ "epoch": 1.3377926421404682,
1025
+ "eval_completions/clipped_ratio": 0.0,
1026
+ "eval_completions/max_length": 469.0,
1027
+ "eval_completions/max_terminated_length": 469.0,
1028
+ "eval_completions/mean_length": 364.515625,
1029
+ "eval_completions/mean_terminated_length": 364.515625,
1030
+ "eval_completions/min_length": 262.0,
1031
+ "eval_completions/min_terminated_length": 262.0,
1032
+ "eval_frac_reward_zero_std": 0.0,
1033
+ "eval_loss": 0.0,
1034
+ "eval_num_tokens": 8074795.0,
1035
+ "eval_reward": 0.75,
1036
+ "eval_reward_std": 0.0,
1037
+ "eval_rewards/accuracy_reward/mean": 0.75,
1038
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
1039
+ "eval_runtime": 53.0535,
1040
+ "eval_samples_per_second": 1.414,
1041
+ "eval_steps_per_second": 0.038,
1042
+ "step": 1600
1043
+ },
1044
+ {
1045
+ "clip_ratio/high_max": 0.00030433517986239167,
1046
+ "clip_ratio/high_mean": 0.0001576538966219232,
1047
+ "clip_ratio/low_mean": 0.00012775148616128718,
1048
+ "clip_ratio/low_min": 1.0768435458885506e-05,
1049
+ "clip_ratio/region_mean": 0.0002854053846021998,
1050
+ "completions/clipped_ratio": 0.0,
1051
+ "completions/max_length": 420.1875,
1052
+ "completions/max_terminated_length": 420.1875,
1053
+ "completions/mean_length": 360.537109375,
1054
+ "completions/mean_terminated_length": 360.537109375,
1055
+ "completions/min_length": 314.875,
1056
+ "completions/min_terminated_length": 314.875,
1057
+ "entropy": 0.04038277090876363,
1058
+ "epoch": 1.391304347826087,
1059
+ "frac_reward_zero_std": 0.84375,
1060
+ "grad_norm": 0.7070130705833435,
1061
+ "kl": 0.05217625901423162,
1062
+ "learning_rate": 1.0730211817168338e-06,
1063
+ "loss": 0.0026,
1064
+ "num_tokens": 8393710.0,
1065
+ "reward": 0.904296875,
1066
+ "reward_std": 0.06716620735824108,
1067
+ "rewards/accuracy_reward/mean": 0.904296875,
1068
+ "rewards/accuracy_reward/std": 0.1349763683974743,
1069
+ "step": 1664
1070
+ },
1071
+ {
1072
+ "epoch": 1.4214046822742474,
1073
+ "eval_completions/clipped_ratio": 0.0,
1074
+ "eval_completions/max_length": 451.0,
1075
+ "eval_completions/max_terminated_length": 451.0,
1076
+ "eval_completions/mean_length": 365.5234375,
1077
+ "eval_completions/mean_terminated_length": 365.5234375,
1078
+ "eval_completions/min_length": 266.0,
1079
+ "eval_completions/min_terminated_length": 266.0,
1080
+ "eval_frac_reward_zero_std": 0.0,
1081
+ "eval_loss": 0.0,
1082
+ "eval_num_tokens": 8574597.0,
1083
+ "eval_reward": 0.7265625,
1084
+ "eval_reward_std": 0.0,
1085
+ "eval_rewards/accuracy_reward/mean": 0.7265625,
1086
+ "eval_rewards/accuracy_reward/std": 0.44916021823883057,
1087
+ "eval_runtime": 51.329,
1088
+ "eval_samples_per_second": 1.461,
1089
+ "eval_steps_per_second": 0.039,
1090
+ "step": 1700
1091
+ },
1092
+ {
1093
+ "clip_ratio/high_max": 0.0,
1094
+ "clip_ratio/high_mean": 0.0,
1095
+ "clip_ratio/low_mean": 0.0,
1096
+ "clip_ratio/low_min": 0.0,
1097
+ "clip_ratio/region_mean": 0.0,
1098
+ "completions/clipped_ratio": 0.0,
1099
+ "completions/max_length": 399.42857142857144,
1100
+ "completions/max_terminated_length": 399.42857142857144,
1101
+ "completions/mean_length": 372.0892857142857,
1102
+ "completions/mean_terminated_length": 372.0892857142857,
1103
+ "completions/min_length": 343.85714285714283,
1104
+ "completions/min_terminated_length": 343.85714285714283,
1105
+ "entropy": 0.029882594078247036,
1106
+ "epoch": 1.4448160535117056,
1107
+ "frac_reward_zero_std": 1.0,
1108
+ "grad_norm": 0.08598916977643967,
1109
+ "kl": 0.025003603040074398,
1110
+ "learning_rate": 1.0373467112597547e-06,
1111
+ "loss": -0.0,
1112
+ "num_tokens": 8716537.0,
1113
+ "reward": 0.8571428571428571,
1114
+ "reward_std": 0.0,
1115
+ "rewards/accuracy_reward/mean": 0.8571428571428571,
1116
+ "rewards/accuracy_reward/std": 0.1451429980141776,
1117
+ "step": 1728
1118
+ },
1119
+ {
1120
+ "clip_ratio/high_max": 0.00011557655489013996,
1121
+ "clip_ratio/high_mean": 5.778827744506998e-05,
1122
+ "clip_ratio/low_mean": 0.00013030926675128285,
1123
+ "clip_ratio/low_min": 0.0,
1124
+ "clip_ratio/region_mean": 0.00018809754510584753,
1125
+ "completions/clipped_ratio": 0.0,
1126
+ "completions/max_length": 425.625,
1127
+ "completions/max_terminated_length": 425.625,
1128
+ "completions/mean_length": 365.990234375,
1129
+ "completions/mean_terminated_length": 365.990234375,
1130
+ "completions/min_length": 315.9375,
1131
+ "completions/min_terminated_length": 315.9375,
1132
+ "entropy": 0.03568679823365528,
1133
+ "epoch": 1.4983277591973243,
1134
+ "frac_reward_zero_std": 0.90625,
1135
+ "grad_norm": 0.3338335156440735,
1136
+ "kl": 0.0331722792113851,
1137
+ "learning_rate": 1.0016722408026757e-06,
1138
+ "loss": 0.0045,
1139
+ "num_tokens": 9038004.0,
1140
+ "reward": 0.888671875,
1141
+ "reward_std": 0.03960913047194481,
1142
+ "rewards/accuracy_reward/mean": 0.888671875,
1143
+ "rewards/accuracy_reward/std": 0.13977273274213076,
1144
+ "step": 1792
1145
+ },
1146
+ {
1147
+ "epoch": 1.5050167224080266,
1148
+ "eval_completions/clipped_ratio": 0.0,
1149
+ "eval_completions/max_length": 470.5,
1150
+ "eval_completions/max_terminated_length": 470.5,
1151
+ "eval_completions/mean_length": 366.8203125,
1152
+ "eval_completions/mean_terminated_length": 366.8203125,
1153
+ "eval_completions/min_length": 265.5,
1154
+ "eval_completions/min_terminated_length": 265.5,
1155
+ "eval_frac_reward_zero_std": 0.0,
1156
+ "eval_loss": 0.0,
1157
+ "eval_num_tokens": 9078781.0,
1158
+ "eval_reward": 0.7421875,
1159
+ "eval_reward_std": 0.0,
1160
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
1161
+ "eval_rewards/accuracy_reward/std": 0.440796360373497,
1162
+ "eval_runtime": 53.1667,
1163
+ "eval_samples_per_second": 1.411,
1164
+ "eval_steps_per_second": 0.038,
1165
+ "step": 1800
1166
+ },
1167
+ {
1168
+ "clip_ratio/high_max": 0.00013944351563363204,
1169
+ "clip_ratio/high_mean": 0.00010243838707018378,
1170
+ "clip_ratio/low_mean": 9.132678574782663e-05,
1171
+ "clip_ratio/low_min": 0.0,
1172
+ "clip_ratio/region_mean": 0.00019376517125887664,
1173
+ "completions/clipped_ratio": 0.0,
1174
+ "completions/max_length": 405.0,
1175
+ "completions/max_terminated_length": 405.0,
1176
+ "completions/mean_length": 359.12723214285717,
1177
+ "completions/mean_terminated_length": 359.12723214285717,
1178
+ "completions/min_length": 317.7857142857143,
1179
+ "completions/min_terminated_length": 317.7857142857143,
1180
+ "entropy": 0.038793727689023526,
1181
+ "epoch": 1.551839464882943,
1182
+ "frac_reward_zero_std": 0.8571428571428571,
1183
+ "grad_norm": 0.17789730429649353,
1184
+ "kl": 0.03541430658203483,
1185
+ "learning_rate": 9.659977703455963e-07,
1186
+ "loss": 0.0068,
1187
+ "num_tokens": 9357078.0,
1188
+ "reward": 0.7388392857142857,
1189
+ "reward_std": 0.06432593294552394,
1190
+ "rewards/accuracy_reward/mean": 0.7388392857142857,
1191
+ "rewards/accuracy_reward/std": 0.23442292426313674,
1192
+ "step": 1856
1193
+ },
1194
+ {
1195
+ "epoch": 1.588628762541806,
1196
+ "eval_completions/clipped_ratio": 0.0,
1197
+ "eval_completions/max_length": 441.0,
1198
+ "eval_completions/max_terminated_length": 441.0,
1199
+ "eval_completions/mean_length": 366.5625,
1200
+ "eval_completions/mean_terminated_length": 366.5625,
1201
+ "eval_completions/min_length": 263.0,
1202
+ "eval_completions/min_terminated_length": 263.0,
1203
+ "eval_frac_reward_zero_std": 0.0,
1204
+ "eval_loss": 0.0,
1205
+ "eval_num_tokens": 9582765.0,
1206
+ "eval_reward": 0.7578125,
1207
+ "eval_reward_std": 0.0,
1208
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1209
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
1210
+ "eval_runtime": 49.6242,
1211
+ "eval_samples_per_second": 1.511,
1212
+ "eval_steps_per_second": 0.04,
1213
+ "step": 1900
1214
+ },
1215
+ {
1216
+ "clip_ratio/high_max": 3.42465762514621e-05,
1217
+ "clip_ratio/high_mean": 1.712328812573105e-05,
1218
+ "clip_ratio/low_mean": 0.00013308838533703237,
1219
+ "clip_ratio/low_min": 0.0,
1220
+ "clip_ratio/region_mean": 0.0001502116705523804,
1221
+ "completions/clipped_ratio": 0.0,
1222
+ "completions/max_length": 402.4,
1223
+ "completions/max_terminated_length": 402.4,
1224
+ "completions/mean_length": 365.99375,
1225
+ "completions/mean_terminated_length": 365.99375,
1226
+ "completions/min_length": 327.2,
1227
+ "completions/min_terminated_length": 327.2,
1228
+ "entropy": 0.04094527019187808,
1229
+ "epoch": 1.605351170568562,
1230
+ "frac_reward_zero_std": 0.9,
1231
+ "grad_norm": 1.3623830080032349,
1232
+ "kl": 0.04199032899923623,
1233
+ "learning_rate": 9.303232998885172e-07,
1234
+ "loss": 0.0033,
1235
+ "num_tokens": 9683324.0,
1236
+ "reward": 0.94375,
1237
+ "reward_std": 0.05123475193977356,
1238
+ "rewards/accuracy_reward/mean": 0.94375,
1239
+ "rewards/accuracy_reward/std": 0.09136068224906921,
1240
+ "step": 1920
1241
+ },
1242
+ {
1243
+ "clip_ratio/high_max": 0.00038922205567359924,
1244
+ "clip_ratio/high_mean": 0.0002362860759603791,
1245
+ "clip_ratio/low_mean": 0.0002494512982593733,
1246
+ "clip_ratio/low_min": 7.466646729881177e-05,
1247
+ "clip_ratio/region_mean": 0.00048573737376500503,
1248
+ "completions/clipped_ratio": 0.0,
1249
+ "completions/max_length": 423.1875,
1250
+ "completions/max_terminated_length": 423.1875,
1251
+ "completions/mean_length": 367.5,
1252
+ "completions/mean_terminated_length": 367.5,
1253
+ "completions/min_length": 324.5625,
1254
+ "completions/min_terminated_length": 324.5625,
1255
+ "entropy": 0.03955537982983515,
1256
+ "epoch": 1.6588628762541806,
1257
+ "frac_reward_zero_std": 0.6875,
1258
+ "grad_norm": 0.08120284974575043,
1259
+ "kl": 0.03543006930885895,
1260
+ "learning_rate": 8.94648829431438e-07,
1261
+ "loss": 0.007,
1262
+ "num_tokens": 10005740.0,
1263
+ "reward": 0.689453125,
1264
+ "reward_std": 0.11965099535882473,
1265
+ "rewards/accuracy_reward/mean": 0.689453125,
1266
+ "rewards/accuracy_reward/std": 0.3270665444433689,
1267
+ "step": 1984
1268
+ },
1269
+ {
1270
+ "epoch": 1.6722408026755853,
1271
+ "eval_completions/clipped_ratio": 0.0,
1272
+ "eval_completions/max_length": 440.0,
1273
+ "eval_completions/max_terminated_length": 440.0,
1274
+ "eval_completions/mean_length": 365.8125,
1275
+ "eval_completions/mean_terminated_length": 365.8125,
1276
+ "eval_completions/min_length": 268.5,
1277
+ "eval_completions/min_terminated_length": 268.5,
1278
+ "eval_frac_reward_zero_std": 0.0,
1279
+ "eval_loss": 0.0,
1280
+ "eval_num_tokens": 10085229.0,
1281
+ "eval_reward": 0.7578125,
1282
+ "eval_reward_std": 0.0,
1283
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1284
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
1285
+ "eval_runtime": 49.643,
1286
+ "eval_samples_per_second": 1.511,
1287
+ "eval_steps_per_second": 0.04,
1288
+ "step": 2000
1289
+ },
1290
+ {
1291
+ "clip_ratio/high_max": 0.00018998604597679028,
1292
+ "clip_ratio/high_mean": 9.499302298839514e-05,
1293
+ "clip_ratio/low_mean": 4.345102024672087e-05,
1294
+ "clip_ratio/low_min": 0.0,
1295
+ "clip_ratio/region_mean": 0.0001384440426287862,
1296
+ "completions/clipped_ratio": 0.0,
1297
+ "completions/max_length": 410.5,
1298
+ "completions/max_terminated_length": 410.5,
1299
+ "completions/mean_length": 365.359375,
1300
+ "completions/mean_terminated_length": 365.359375,
1301
+ "completions/min_length": 325.6666666666667,
1302
+ "completions/min_terminated_length": 325.6666666666667,
1303
+ "entropy": 0.0357375051244162,
1304
+ "epoch": 1.7123745819397993,
1305
+ "frac_reward_zero_std": 0.9166666666666666,
1306
+ "grad_norm": 0.16435562074184418,
1307
+ "kl": 0.031716163145271516,
1308
+ "learning_rate": 8.589743589743588e-07,
1309
+ "loss": 0.0036,
1310
+ "num_tokens": 10326135.0,
1311
+ "reward": 0.7552083333333334,
1312
+ "reward_std": 0.03286577512820562,
1313
+ "rewards/accuracy_reward/mean": 0.7552083333333334,
1314
+ "rewards/accuracy_reward/std": 0.19700236121813455,
1315
+ "step": 2048
1316
+ },
1317
+ {
1318
+ "epoch": 1.7558528428093645,
1319
+ "eval_completions/clipped_ratio": 0.0,
1320
+ "eval_completions/max_length": 470.5,
1321
+ "eval_completions/max_terminated_length": 470.5,
1322
+ "eval_completions/mean_length": 364.2578125,
1323
+ "eval_completions/mean_terminated_length": 364.2578125,
1324
+ "eval_completions/min_length": 270.0,
1325
+ "eval_completions/min_terminated_length": 270.0,
1326
+ "eval_frac_reward_zero_std": 0.0,
1327
+ "eval_loss": 0.0,
1328
+ "eval_num_tokens": 10587665.0,
1329
+ "eval_reward": 0.75,
1330
+ "eval_reward_std": 0.0,
1331
+ "eval_rewards/accuracy_reward/mean": 0.75,
1332
+ "eval_rewards/accuracy_reward/std": 0.43491509556770325,
1333
+ "eval_runtime": 52.8003,
1334
+ "eval_samples_per_second": 1.42,
1335
+ "eval_steps_per_second": 0.038,
1336
+ "step": 2100
1337
+ },
1338
+ {
1339
+ "clip_ratio/high_max": 0.0005627853630964333,
1340
+ "clip_ratio/high_mean": 0.00028139268154821667,
1341
+ "clip_ratio/low_mean": 5.2444093550244965e-05,
1342
+ "clip_ratio/low_min": 0.0,
1343
+ "clip_ratio/region_mean": 0.00033383678237441927,
1344
+ "completions/clipped_ratio": 0.0,
1345
+ "completions/max_length": 384.0,
1346
+ "completions/max_terminated_length": 384.0,
1347
+ "completions/mean_length": 342.6354166666667,
1348
+ "completions/mean_terminated_length": 342.6354166666667,
1349
+ "completions/min_length": 297.0,
1350
+ "completions/min_terminated_length": 297.0,
1351
+ "entropy": 0.04007397824898362,
1352
+ "epoch": 1.7658862876254182,
1353
+ "frac_reward_zero_std": 0.8333333333333334,
1354
+ "grad_norm": 1.784582495689392,
1355
+ "kl": 0.028852593871609617,
1356
+ "learning_rate": 8.232998885172798e-07,
1357
+ "loss": 0.0001,
1358
+ "num_tokens": 10645710.0,
1359
+ "reward": 0.7604166666666666,
1360
+ "reward_std": 0.08539125323295593,
1361
+ "rewards/accuracy_reward/mean": 0.7604166666666666,
1362
+ "rewards/accuracy_reward/std": 0.1522678037484487,
1363
+ "step": 2112
1364
+ },
1365
+ {
1366
+ "clip_ratio/high_max": 5.1616162636491936e-05,
1367
+ "clip_ratio/high_mean": 2.5808081318245968e-05,
1368
+ "clip_ratio/low_mean": 6.810115883126855e-05,
1369
+ "clip_ratio/low_min": 0.0,
1370
+ "clip_ratio/region_mean": 9.390924014951452e-05,
1371
+ "completions/clipped_ratio": 0.0,
1372
+ "completions/max_length": 416.125,
1373
+ "completions/max_terminated_length": 416.125,
1374
+ "completions/mean_length": 371.861328125,
1375
+ "completions/mean_terminated_length": 371.861328125,
1376
+ "completions/min_length": 327.9375,
1377
+ "completions/min_terminated_length": 327.9375,
1378
+ "entropy": 0.033375646657077596,
1379
+ "epoch": 1.819397993311037,
1380
+ "frac_reward_zero_std": 0.90625,
1381
+ "grad_norm": 0.11332476139068604,
1382
+ "kl": 0.024399012820822463,
1383
+ "learning_rate": 7.876254180602006e-07,
1384
+ "loss": 0.0021,
1385
+ "num_tokens": 10970199.0,
1386
+ "reward": 0.794921875,
1387
+ "reward_std": 0.03300705552101135,
1388
+ "rewards/accuracy_reward/mean": 0.794921875,
1389
+ "rewards/accuracy_reward/std": 0.15756188333034515,
1390
+ "step": 2176
1391
+ },
1392
+ {
1393
+ "epoch": 1.839464882943144,
1394
+ "eval_completions/clipped_ratio": 0.0,
1395
+ "eval_completions/max_length": 440.5,
1396
+ "eval_completions/max_terminated_length": 440.5,
1397
+ "eval_completions/mean_length": 366.1171875,
1398
+ "eval_completions/mean_terminated_length": 366.1171875,
1399
+ "eval_completions/min_length": 265.0,
1400
+ "eval_completions/min_terminated_length": 265.0,
1401
+ "eval_frac_reward_zero_std": 0.0,
1402
+ "eval_loss": 0.0,
1403
+ "eval_num_tokens": 11089407.0,
1404
+ "eval_reward": 0.75,
1405
+ "eval_reward_std": 0.0,
1406
+ "eval_rewards/accuracy_reward/mean": 0.75,
1407
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
1408
+ "eval_runtime": 50.0612,
1409
+ "eval_samples_per_second": 1.498,
1410
+ "eval_steps_per_second": 0.04,
1411
+ "step": 2200
1412
+ },
1413
+ {
1414
+ "clip_ratio/high_max": 0.00023123159917304292,
1415
+ "clip_ratio/high_mean": 0.00011561579958652146,
1416
+ "clip_ratio/low_mean": 0.00028777941915905105,
1417
+ "clip_ratio/low_min": 0.0,
1418
+ "clip_ratio/region_mean": 0.00040339521365240216,
1419
+ "completions/clipped_ratio": 0.0,
1420
+ "completions/max_length": 410.2,
1421
+ "completions/max_terminated_length": 410.2,
1422
+ "completions/mean_length": 367.084375,
1423
+ "completions/mean_terminated_length": 367.084375,
1424
+ "completions/min_length": 331.6,
1425
+ "completions/min_terminated_length": 331.6,
1426
+ "entropy": 0.04230717392638326,
1427
+ "epoch": 1.8729096989966556,
1428
+ "frac_reward_zero_std": 0.75,
1429
+ "grad_norm": 0.10948460549116135,
1430
+ "kl": 0.04778804142260924,
1431
+ "learning_rate": 7.519509476031214e-07,
1432
+ "loss": 0.0034,
1433
+ "num_tokens": 11290714.0,
1434
+ "reward": 0.80625,
1435
+ "reward_std": 0.09691373407840728,
1436
+ "rewards/accuracy_reward/mean": 0.80625,
1437
+ "rewards/accuracy_reward/std": 0.2130134329199791,
1438
+ "step": 2240
1439
+ },
1440
+ {
1441
+ "epoch": 1.9230769230769231,
1442
+ "eval_completions/clipped_ratio": 0.0,
1443
+ "eval_completions/max_length": 463.5,
1444
+ "eval_completions/max_terminated_length": 463.5,
1445
+ "eval_completions/mean_length": 367.1015625,
1446
+ "eval_completions/mean_terminated_length": 367.1015625,
1447
+ "eval_completions/min_length": 276.5,
1448
+ "eval_completions/min_terminated_length": 276.5,
1449
+ "eval_frac_reward_zero_std": 0.0,
1450
+ "eval_loss": 0.0,
1451
+ "eval_num_tokens": 11593515.0,
1452
+ "eval_reward": 0.75,
1453
+ "eval_reward_std": 0.0,
1454
+ "eval_rewards/accuracy_reward/mean": 0.75,
1455
+ "eval_rewards/accuracy_reward/std": 0.4364357888698578,
1456
+ "eval_runtime": 52.3654,
1457
+ "eval_samples_per_second": 1.432,
1458
+ "eval_steps_per_second": 0.038,
1459
+ "step": 2300
1460
+ },
1461
+ {
1462
+ "clip_ratio/high_max": 0.0,
1463
+ "clip_ratio/high_mean": 0.0,
1464
+ "clip_ratio/low_mean": 0.0,
1465
+ "clip_ratio/low_min": 0.0,
1466
+ "clip_ratio/region_mean": 0.0,
1467
+ "completions/clipped_ratio": 0.0,
1468
+ "completions/max_length": 421.0,
1469
+ "completions/max_terminated_length": 421.0,
1470
+ "completions/mean_length": 402.28125,
1471
+ "completions/mean_terminated_length": 402.28125,
1472
+ "completions/min_length": 373.0,
1473
+ "completions/min_terminated_length": 373.0,
1474
+ "entropy": 0.042067136615514755,
1475
+ "epoch": 1.9264214046822743,
1476
+ "frac_reward_zero_std": 1.0,
1477
+ "grad_norm": 0.24160102009773254,
1478
+ "kl": 0.039941526018083096,
1479
+ "learning_rate": 7.162764771460424e-07,
1480
+ "loss": 0.0014,
1481
+ "num_tokens": 11614788.0,
1482
+ "reward": 1.0,
1483
+ "reward_std": 0.0,
1484
+ "rewards/accuracy_reward/mean": 1.0,
1485
+ "rewards/accuracy_reward/std": 0.0,
1486
+ "step": 2304
1487
+ },
1488
+ {
1489
+ "clip_ratio/high_max": 0.0001278845456909039,
1490
+ "clip_ratio/high_mean": 6.394227284545195e-05,
1491
+ "clip_ratio/low_mean": 0.00010635303306116839,
1492
+ "clip_ratio/low_min": 0.0,
1493
+ "clip_ratio/region_mean": 0.00017029530681611504,
1494
+ "completions/clipped_ratio": 0.0,
1495
+ "completions/max_length": 422.125,
1496
+ "completions/max_terminated_length": 422.125,
1497
+ "completions/mean_length": 369.134765625,
1498
+ "completions/mean_terminated_length": 369.134765625,
1499
+ "completions/min_length": 324.6875,
1500
+ "completions/min_terminated_length": 324.6875,
1501
+ "entropy": 0.034209970210213214,
1502
+ "epoch": 1.979933110367893,
1503
+ "frac_reward_zero_std": 0.90625,
1504
+ "grad_norm": 0.1534082293510437,
1505
+ "kl": 0.022877497935724023,
1506
+ "learning_rate": 6.806020066889632e-07,
1507
+ "loss": 0.005,
1508
+ "num_tokens": 11937993.0,
1509
+ "reward": 0.845703125,
1510
+ "reward_std": 0.04271013289690018,
1511
+ "rewards/accuracy_reward/mean": 0.845703125,
1512
+ "rewards/accuracy_reward/std": 0.1940061654895544,
1513
+ "step": 2368
1514
+ },
1515
+ {
1516
+ "epoch": 2.0066889632107023,
1517
+ "eval_completions/clipped_ratio": 0.0,
1518
+ "eval_completions/max_length": 461.0,
1519
+ "eval_completions/max_terminated_length": 461.0,
1520
+ "eval_completions/mean_length": 367.28125,
1521
+ "eval_completions/mean_terminated_length": 367.28125,
1522
+ "eval_completions/min_length": 264.5,
1523
+ "eval_completions/min_terminated_length": 264.5,
1524
+ "eval_frac_reward_zero_std": 0.0,
1525
+ "eval_loss": 0.0,
1526
+ "eval_num_tokens": 12101741.0,
1527
+ "eval_reward": 0.7578125,
1528
+ "eval_reward_std": 0.0,
1529
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1530
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
1531
+ "eval_runtime": 51.9127,
1532
+ "eval_samples_per_second": 1.445,
1533
+ "eval_steps_per_second": 0.039,
1534
+ "step": 2400
1535
+ },
1536
+ {
1537
+ "clip_ratio/high_max": 9.742790280142799e-05,
1538
+ "clip_ratio/high_mean": 4.8713951400713995e-05,
1539
+ "clip_ratio/low_mean": 8.524941222276539e-05,
1540
+ "clip_ratio/low_min": 0.0,
1541
+ "clip_ratio/region_mean": 0.00013396335998550057,
1542
+ "completions/clipped_ratio": 0.0,
1543
+ "completions/max_length": 397.625,
1544
+ "completions/max_terminated_length": 397.625,
1545
+ "completions/mean_length": 346.6640625,
1546
+ "completions/mean_terminated_length": 346.6640625,
1547
+ "completions/min_length": 302.75,
1548
+ "completions/min_terminated_length": 302.75,
1549
+ "entropy": 0.03151974832871929,
1550
+ "epoch": 2.0334448160535117,
1551
+ "frac_reward_zero_std": 0.9375,
1552
+ "grad_norm": 0.025068683549761772,
1553
+ "kl": 0.017846403601652128,
1554
+ "learning_rate": 6.449275362318841e-07,
1555
+ "loss": 0.0036,
1556
+ "num_tokens": 12257511.0,
1557
+ "reward": 0.98828125,
1558
+ "reward_std": 0.025194555521011353,
1559
+ "rewards/accuracy_reward/mean": 0.98828125,
1560
+ "rewards/accuracy_reward/std": 0.037018071860075,
1561
+ "step": 2432
1562
+ },
1563
+ {
1564
+ "clip_ratio/high_max": 0.000281647706287913,
1565
+ "clip_ratio/high_mean": 0.0001408238531439565,
1566
+ "clip_ratio/low_mean": 4.2170279357378604e-05,
1567
+ "clip_ratio/low_min": 0.0,
1568
+ "clip_ratio/region_mean": 0.0001829941306823457,
1569
+ "completions/clipped_ratio": 0.0,
1570
+ "completions/max_length": 422.1875,
1571
+ "completions/max_terminated_length": 422.1875,
1572
+ "completions/mean_length": 371.1328125,
1573
+ "completions/mean_terminated_length": 371.1328125,
1574
+ "completions/min_length": 332.0625,
1575
+ "completions/min_terminated_length": 332.0625,
1576
+ "entropy": 0.033946564319194295,
1577
+ "epoch": 2.0869565217391304,
1578
+ "frac_reward_zero_std": 0.875,
1579
+ "grad_norm": 2.1868627071380615,
1580
+ "kl": 0.028850825376423472,
1581
+ "learning_rate": 6.092530657748049e-07,
1582
+ "loss": 0.0016,
1583
+ "num_tokens": 12581787.0,
1584
+ "reward": 0.82421875,
1585
+ "reward_std": 0.04742163047194481,
1586
+ "rewards/accuracy_reward/mean": 0.82421875,
1587
+ "rewards/accuracy_reward/std": 0.21811604965478182,
1588
+ "step": 2496
1589
+ },
1590
+ {
1591
+ "epoch": 2.0903010033444818,
1592
+ "eval_completions/clipped_ratio": 0.0,
1593
+ "eval_completions/max_length": 433.0,
1594
+ "eval_completions/max_terminated_length": 433.0,
1595
+ "eval_completions/mean_length": 362.4140625,
1596
+ "eval_completions/mean_terminated_length": 362.4140625,
1597
+ "eval_completions/min_length": 273.0,
1598
+ "eval_completions/min_terminated_length": 273.0,
1599
+ "eval_frac_reward_zero_std": 0.0,
1600
+ "eval_loss": 0.0,
1601
+ "eval_num_tokens": 12601846.0,
1602
+ "eval_reward": 0.765625,
1603
+ "eval_reward_std": 0.0,
1604
+ "eval_rewards/accuracy_reward/mean": 0.765625,
1605
+ "eval_rewards/accuracy_reward/std": 0.42533111572265625,
1606
+ "eval_runtime": 49.4436,
1607
+ "eval_samples_per_second": 1.517,
1608
+ "eval_steps_per_second": 0.04,
1609
+ "step": 2500
1610
+ },
1611
+ {
1612
+ "clip_ratio/high_max": 0.0001637306500924751,
1613
+ "clip_ratio/high_mean": 8.186532504623756e-05,
1614
+ "clip_ratio/low_mean": 0.0001552711376765122,
1615
+ "clip_ratio/low_min": 1.0738831285076837e-05,
1616
+ "clip_ratio/region_mean": 0.00023713646126755824,
1617
+ "completions/clipped_ratio": 0.0,
1618
+ "completions/max_length": 402.93333333333334,
1619
+ "completions/max_terminated_length": 402.93333333333334,
1620
+ "completions/mean_length": 360.23333333333335,
1621
+ "completions/mean_terminated_length": 360.23333333333335,
1622
+ "completions/min_length": 321.26666666666665,
1623
+ "completions/min_terminated_length": 321.26666666666665,
1624
+ "entropy": 0.035695391427725556,
1625
+ "epoch": 2.140468227424749,
1626
+ "frac_reward_zero_std": 0.8333333333333334,
1627
+ "grad_norm": 0.17780426144599915,
1628
+ "kl": 0.026257768408443856,
1629
+ "learning_rate": 5.735785953177257e-07,
1630
+ "loss": 0.0007,
1631
+ "num_tokens": 12900518.0,
1632
+ "reward": 0.8229166666666666,
1633
+ "reward_std": 0.06161968111991882,
1634
+ "rewards/accuracy_reward/mean": 0.8229166666666666,
1635
+ "rewards/accuracy_reward/std": 0.19885458350181578,
1636
+ "step": 2560
1637
+ },
1638
+ {
1639
+ "epoch": 2.1739130434782608,
1640
+ "eval_completions/clipped_ratio": 0.0,
1641
+ "eval_completions/max_length": 433.5,
1642
+ "eval_completions/max_terminated_length": 433.5,
1643
+ "eval_completions/mean_length": 366.2890625,
1644
+ "eval_completions/mean_terminated_length": 366.2890625,
1645
+ "eval_completions/min_length": 268.5,
1646
+ "eval_completions/min_terminated_length": 268.5,
1647
+ "eval_frac_reward_zero_std": 0.0,
1648
+ "eval_loss": 0.0,
1649
+ "eval_num_tokens": 13104242.0,
1650
+ "eval_reward": 0.75,
1651
+ "eval_reward_std": 0.0,
1652
+ "eval_rewards/accuracy_reward/mean": 0.75,
1653
+ "eval_rewards/accuracy_reward/std": 0.43491509556770325,
1654
+ "eval_runtime": 48.5867,
1655
+ "eval_samples_per_second": 1.544,
1656
+ "eval_steps_per_second": 0.041,
1657
+ "step": 2600
1658
+ },
1659
+ {
1660
+ "clip_ratio/high_max": 0.00015852885553613305,
1661
+ "clip_ratio/high_mean": 7.926442776806653e-05,
1662
+ "clip_ratio/low_mean": 0.0,
1663
+ "clip_ratio/low_min": 0.0,
1664
+ "clip_ratio/region_mean": 7.926442776806653e-05,
1665
+ "completions/clipped_ratio": 0.0,
1666
+ "completions/max_length": 411.3333333333333,
1667
+ "completions/max_terminated_length": 411.3333333333333,
1668
+ "completions/mean_length": 362.515625,
1669
+ "completions/mean_terminated_length": 362.515625,
1670
+ "completions/min_length": 318.5,
1671
+ "completions/min_terminated_length": 318.5,
1672
+ "entropy": 0.03652742908646663,
1673
+ "epoch": 2.1939799331103678,
1674
+ "frac_reward_zero_std": 0.9166666666666666,
1675
+ "grad_norm": 2.8967080116271973,
1676
+ "kl": 0.03786919802466097,
1677
+ "learning_rate": 5.379041248606467e-07,
1678
+ "loss": 0.0026,
1679
+ "num_tokens": 13224117.0,
1680
+ "reward": 0.9947916666666666,
1681
+ "reward_std": 0.020833333333333332,
1682
+ "rewards/accuracy_reward/mean": 0.9947916666666666,
1683
+ "rewards/accuracy_reward/std": 0.029462782045205433,
1684
+ "step": 2624
1685
+ },
1686
+ {
1687
+ "clip_ratio/high_max": 0.0001364857480439241,
1688
+ "clip_ratio/high_mean": 6.824287402196205e-05,
1689
+ "clip_ratio/low_mean": 3.928913429263048e-05,
1690
+ "clip_ratio/low_min": 0.0,
1691
+ "clip_ratio/region_mean": 0.00010753200831459253,
1692
+ "completions/clipped_ratio": 0.0,
1693
+ "completions/max_length": 413.3125,
1694
+ "completions/max_terminated_length": 413.3125,
1695
+ "completions/mean_length": 359.306640625,
1696
+ "completions/mean_terminated_length": 359.306640625,
1697
+ "completions/min_length": 312.6875,
1698
+ "completions/min_terminated_length": 312.6875,
1699
+ "entropy": 0.030745981523068622,
1700
+ "epoch": 2.2474916387959865,
1701
+ "frac_reward_zero_std": 0.875,
1702
+ "grad_norm": 0.49525848031044006,
1703
+ "kl": 0.02309208803171714,
1704
+ "learning_rate": 5.022296544035675e-07,
1705
+ "loss": 0.0007,
1706
+ "num_tokens": 13542306.0,
1707
+ "reward": 0.859375,
1708
+ "reward_std": 0.04318207688629627,
1709
+ "rewards/accuracy_reward/mean": 0.859375,
1710
+ "rewards/accuracy_reward/std": 0.12716256082057953,
1711
+ "step": 2688
1712
+ },
1713
+ {
1714
+ "epoch": 2.25752508361204,
1715
+ "eval_completions/clipped_ratio": 0.0,
1716
+ "eval_completions/max_length": 452.0,
1717
+ "eval_completions/max_terminated_length": 452.0,
1718
+ "eval_completions/mean_length": 367.140625,
1719
+ "eval_completions/mean_terminated_length": 367.140625,
1720
+ "eval_completions/min_length": 268.5,
1721
+ "eval_completions/min_terminated_length": 268.5,
1722
+ "eval_frac_reward_zero_std": 0.0,
1723
+ "eval_loss": 0.0,
1724
+ "eval_num_tokens": 13602666.0,
1725
+ "eval_reward": 0.75,
1726
+ "eval_reward_std": 0.0,
1727
+ "eval_rewards/accuracy_reward/mean": 0.75,
1728
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
1729
+ "eval_runtime": 50.6555,
1730
+ "eval_samples_per_second": 1.481,
1731
+ "eval_steps_per_second": 0.039,
1732
+ "step": 2700
1733
+ },
1734
+ {
1735
+ "clip_ratio/high_max": 0.0001540342379415121,
1736
+ "clip_ratio/high_mean": 7.701711897075606e-05,
1737
+ "clip_ratio/low_mean": 0.00010309644074788174,
1738
+ "clip_ratio/low_min": 0.0,
1739
+ "clip_ratio/region_mean": 0.00018011355971863779,
1740
+ "completions/clipped_ratio": 0.0,
1741
+ "completions/max_length": 406.15384615384613,
1742
+ "completions/max_terminated_length": 406.15384615384613,
1743
+ "completions/mean_length": 361.2932692307692,
1744
+ "completions/mean_terminated_length": 361.2932692307692,
1745
+ "completions/min_length": 329.53846153846155,
1746
+ "completions/min_terminated_length": 329.53846153846155,
1747
+ "entropy": 0.03364891480081356,
1748
+ "epoch": 2.3010033444816056,
1749
+ "frac_reward_zero_std": 0.8461538461538461,
1750
+ "grad_norm": 0.16395771503448486,
1751
+ "kl": 0.024445113113660436,
1752
+ "learning_rate": 4.665551839464883e-07,
1753
+ "loss": -0.0,
1754
+ "num_tokens": 13861908.0,
1755
+ "reward": 0.7067307692307693,
1756
+ "reward_std": 0.05965869701825655,
1757
+ "rewards/accuracy_reward/mean": 0.7067307692307693,
1758
+ "rewards/accuracy_reward/std": 0.30705370811315685,
1759
+ "step": 2752
1760
+ },
1761
+ {
1762
+ "epoch": 2.3411371237458196,
1763
+ "eval_completions/clipped_ratio": 0.0,
1764
+ "eval_completions/max_length": 464.0,
1765
+ "eval_completions/max_terminated_length": 464.0,
1766
+ "eval_completions/mean_length": 363.5078125,
1767
+ "eval_completions/mean_terminated_length": 363.5078125,
1768
+ "eval_completions/min_length": 263.0,
1769
+ "eval_completions/min_terminated_length": 263.0,
1770
+ "eval_frac_reward_zero_std": 0.0,
1771
+ "eval_loss": 0.0,
1772
+ "eval_num_tokens": 14100795.0,
1773
+ "eval_reward": 0.7578125,
1774
+ "eval_reward_std": 0.0,
1775
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1776
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
1777
+ "eval_runtime": 52.222,
1778
+ "eval_samples_per_second": 1.436,
1779
+ "eval_steps_per_second": 0.038,
1780
+ "step": 2800
1781
+ },
1782
+ {
1783
+ "clip_ratio/high_max": 0.00022265735606197268,
1784
+ "clip_ratio/high_mean": 0.00011132867803098634,
1785
+ "clip_ratio/low_mean": 0.0003003920064656995,
1786
+ "clip_ratio/low_min": 0.0,
1787
+ "clip_ratio/region_mean": 0.00041172068449668586,
1788
+ "completions/clipped_ratio": 0.0,
1789
+ "completions/max_length": 402.0,
1790
+ "completions/max_terminated_length": 402.0,
1791
+ "completions/mean_length": 361.625,
1792
+ "completions/mean_terminated_length": 361.625,
1793
+ "completions/min_length": 323.0,
1794
+ "completions/min_terminated_length": 323.0,
1795
+ "entropy": 0.03735521179623902,
1796
+ "epoch": 2.3545150501672243,
1797
+ "frac_reward_zero_std": 0.75,
1798
+ "grad_norm": 0.0680985152721405,
1799
+ "kl": 0.03388472790538799,
1800
+ "learning_rate": 4.308807134894091e-07,
1801
+ "loss": 0.0023,
1802
+ "num_tokens": 14180619.0,
1803
+ "reward": 0.7890625,
1804
+ "reward_std": 0.09308473765850067,
1805
+ "rewards/accuracy_reward/mean": 0.7890625,
1806
+ "rewards/accuracy_reward/std": 0.2507517486810684,
1807
+ "step": 2816
1808
+ },
1809
+ {
1810
+ "clip_ratio/high_max": 0.00022784904467698652,
1811
+ "clip_ratio/high_mean": 0.00011392452233849326,
1812
+ "clip_ratio/low_mean": 2.761044834187487e-05,
1813
+ "clip_ratio/low_min": 0.0,
1814
+ "clip_ratio/region_mean": 0.00014153497068036813,
1815
+ "completions/clipped_ratio": 0.0,
1816
+ "completions/max_length": 423.0,
1817
+ "completions/max_terminated_length": 423.0,
1818
+ "completions/mean_length": 372.39453125,
1819
+ "completions/mean_terminated_length": 372.39453125,
1820
+ "completions/min_length": 323.75,
1821
+ "completions/min_terminated_length": 323.75,
1822
+ "entropy": 0.03146844085131306,
1823
+ "epoch": 2.408026755852843,
1824
+ "frac_reward_zero_std": 0.90625,
1825
+ "grad_norm": 0.2218427062034607,
1826
+ "kl": 0.02288907259719508,
1827
+ "learning_rate": 3.9520624303232996e-07,
1828
+ "loss": 0.006,
1829
+ "num_tokens": 14505509.0,
1830
+ "reward": 0.841796875,
1831
+ "reward_std": 0.0450726542621851,
1832
+ "rewards/accuracy_reward/mean": 0.841796875,
1833
+ "rewards/accuracy_reward/std": 0.19855350628495216,
1834
+ "step": 2880
1835
+ },
1836
+ {
1837
+ "epoch": 2.4247491638795986,
1838
+ "eval_completions/clipped_ratio": 0.0,
1839
+ "eval_completions/max_length": 454.5,
1840
+ "eval_completions/max_terminated_length": 454.5,
1841
+ "eval_completions/mean_length": 363.7421875,
1842
+ "eval_completions/mean_terminated_length": 363.7421875,
1843
+ "eval_completions/min_length": 271.0,
1844
+ "eval_completions/min_terminated_length": 271.0,
1845
+ "eval_frac_reward_zero_std": 0.0,
1846
+ "eval_loss": 0.0,
1847
+ "eval_num_tokens": 14605224.0,
1848
+ "eval_reward": 0.7734375,
1849
+ "eval_reward_std": 0.0,
1850
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1851
+ "eval_rewards/accuracy_reward/std": 0.4218114912509918,
1852
+ "eval_runtime": 51.4851,
1853
+ "eval_samples_per_second": 1.457,
1854
+ "eval_steps_per_second": 0.039,
1855
+ "step": 2900
1856
+ },
1857
+ {
1858
+ "clip_ratio/high_max": 0.0002011400329965082,
1859
+ "clip_ratio/high_mean": 0.0001005700164982541,
1860
+ "clip_ratio/low_mean": 1.566317884928801e-05,
1861
+ "clip_ratio/low_min": 0.0,
1862
+ "clip_ratio/region_mean": 0.00011623319534754211,
1863
+ "completions/clipped_ratio": 0.0,
1864
+ "completions/max_length": 418.72727272727275,
1865
+ "completions/max_terminated_length": 418.72727272727275,
1866
+ "completions/mean_length": 366.46022727272725,
1867
+ "completions/mean_terminated_length": 366.46022727272725,
1868
+ "completions/min_length": 318.6363636363636,
1869
+ "completions/min_terminated_length": 318.6363636363636,
1870
+ "entropy": 0.03805397533472966,
1871
+ "epoch": 2.4615384615384617,
1872
+ "frac_reward_zero_std": 0.9090909090909091,
1873
+ "grad_norm": 0.09983903169631958,
1874
+ "kl": 0.03686384881306863,
1875
+ "learning_rate": 3.595317725752508e-07,
1876
+ "loss": 0.0005,
1877
+ "num_tokens": 14826522.0,
1878
+ "reward": 0.8636363636363636,
1879
+ "reward_std": 0.04065578092228283,
1880
+ "rewards/accuracy_reward/mean": 0.8636363636363636,
1881
+ "rewards/accuracy_reward/std": 0.1676255545832894,
1882
+ "step": 2944
1883
+ },
1884
+ {
1885
+ "epoch": 2.508361204013378,
1886
+ "eval_completions/clipped_ratio": 0.0,
1887
+ "eval_completions/max_length": 462.5,
1888
+ "eval_completions/max_terminated_length": 462.5,
1889
+ "eval_completions/mean_length": 366.421875,
1890
+ "eval_completions/mean_terminated_length": 366.421875,
1891
+ "eval_completions/min_length": 266.5,
1892
+ "eval_completions/min_terminated_length": 266.5,
1893
+ "eval_frac_reward_zero_std": 0.0,
1894
+ "eval_loss": 0.0,
1895
+ "eval_num_tokens": 15108038.0,
1896
+ "eval_reward": 0.7734375,
1897
+ "eval_reward_std": 0.0,
1898
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1899
+ "eval_rewards/accuracy_reward/std": 0.42097054421901703,
1900
+ "eval_runtime": 52.7107,
1901
+ "eval_samples_per_second": 1.423,
1902
+ "eval_steps_per_second": 0.038,
1903
+ "step": 3000
1904
+ },
1905
+ {
1906
+ "clip_ratio/high_max": 0.0,
1907
+ "clip_ratio/high_mean": 0.0,
1908
+ "clip_ratio/low_mean": 0.0,
1909
+ "clip_ratio/low_min": 0.0,
1910
+ "clip_ratio/region_mean": 0.0,
1911
+ "completions/clipped_ratio": 0.0,
1912
+ "completions/max_length": 400.5,
1913
+ "completions/max_terminated_length": 400.5,
1914
+ "completions/mean_length": 366.59375,
1915
+ "completions/mean_terminated_length": 366.59375,
1916
+ "completions/min_length": 343.5,
1917
+ "completions/min_terminated_length": 343.5,
1918
+ "entropy": 0.03544338117353618,
1919
+ "epoch": 2.5150501672240804,
1920
+ "frac_reward_zero_std": 1.0,
1921
+ "grad_norm": 0.025996780022978783,
1922
+ "kl": 0.02990366171434289,
1923
+ "learning_rate": 3.238573021181717e-07,
1924
+ "loss": 0.0043,
1925
+ "num_tokens": 15148284.0,
1926
+ "reward": 0.75,
1927
+ "reward_std": 0.0,
1928
+ "rewards/accuracy_reward/mean": 0.75,
1929
+ "rewards/accuracy_reward/std": 0.2540002465248108,
1930
+ "step": 3008
1931
+ },
1932
+ {
1933
+ "clip_ratio/high_max": 0.00011581025501072872,
1934
+ "clip_ratio/high_mean": 5.790512750536436e-05,
1935
+ "clip_ratio/low_mean": 4.125472969462862e-05,
1936
+ "clip_ratio/low_min": 0.0,
1937
+ "clip_ratio/region_mean": 9.915985674524563e-05,
1938
+ "completions/clipped_ratio": 0.0,
1939
+ "completions/max_length": 417.3125,
1940
+ "completions/max_terminated_length": 417.3125,
1941
+ "completions/mean_length": 361.859375,
1942
+ "completions/mean_terminated_length": 361.859375,
1943
+ "completions/min_length": 316.625,
1944
+ "completions/min_terminated_length": 316.625,
1945
+ "entropy": 0.03599540321738459,
1946
+ "epoch": 2.568561872909699,
1947
+ "frac_reward_zero_std": 0.875,
1948
+ "grad_norm": 0.09647044539451599,
1949
+ "kl": 0.027639411630161703,
1950
+ "learning_rate": 2.881828316610925e-07,
1951
+ "loss": 0.0014,
1952
+ "num_tokens": 15467604.0,
1953
+ "reward": 0.744140625,
1954
+ "reward_std": 0.06018522381782532,
1955
+ "rewards/accuracy_reward/mean": 0.744140625,
1956
+ "rewards/accuracy_reward/std": 0.19861802645027637,
1957
+ "step": 3072
1958
+ },
1959
+ {
1960
+ "epoch": 2.591973244147157,
1961
+ "eval_completions/clipped_ratio": 0.0,
1962
+ "eval_completions/max_length": 441.5,
1963
+ "eval_completions/max_terminated_length": 441.5,
1964
+ "eval_completions/mean_length": 363.015625,
1965
+ "eval_completions/mean_terminated_length": 363.015625,
1966
+ "eval_completions/min_length": 260.0,
1967
+ "eval_completions/min_terminated_length": 260.0,
1968
+ "eval_frac_reward_zero_std": 0.0,
1969
+ "eval_loss": 0.0,
1970
+ "eval_num_tokens": 15606449.0,
1971
+ "eval_reward": 0.7421875,
1972
+ "eval_reward_std": 0.0,
1973
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
1974
+ "eval_rewards/accuracy_reward/std": 0.4400599002838135,
1975
+ "eval_runtime": 49.2034,
1976
+ "eval_samples_per_second": 1.524,
1977
+ "eval_steps_per_second": 0.041,
1978
+ "step": 3100
1979
+ },
1980
+ {
1981
+ "clip_ratio/high_max": 0.0,
1982
+ "clip_ratio/high_mean": 0.0,
1983
+ "clip_ratio/low_mean": 0.0,
1984
+ "clip_ratio/low_min": 0.0,
1985
+ "clip_ratio/region_mean": 0.0,
1986
+ "completions/clipped_ratio": 0.0,
1987
+ "completions/max_length": 422.44444444444446,
1988
+ "completions/max_terminated_length": 422.44444444444446,
1989
+ "completions/mean_length": 376.46527777777777,
1990
+ "completions/mean_terminated_length": 376.46527777777777,
1991
+ "completions/min_length": 340.0,
1992
+ "completions/min_terminated_length": 340.0,
1993
+ "entropy": 0.027494923961866233,
1994
+ "epoch": 2.6220735785953178,
1995
+ "frac_reward_zero_std": 1.0,
1996
+ "grad_norm": 0.137000173330307,
1997
+ "kl": 0.014413958009552315,
1998
+ "learning_rate": 2.5250836120401335e-07,
1999
+ "loss": 0.0073,
2000
+ "num_tokens": 15790375.0,
2001
+ "reward": 0.9444444444444444,
2002
+ "reward_std": 0.0,
2003
+ "rewards/accuracy_reward/mean": 0.9444444444444444,
2004
+ "rewards/accuracy_reward/std": 0.05644449922773573,
2005
+ "step": 3136
2006
+ },
2007
+ {
2008
+ "clip_ratio/high_max": 0.00017072596619982505,
2009
+ "clip_ratio/high_mean": 8.536298309991253e-05,
2010
+ "clip_ratio/low_mean": 0.00011088756036770064,
2011
+ "clip_ratio/low_min": 0.0,
2012
+ "clip_ratio/region_mean": 0.00019625054346761317,
2013
+ "completions/clipped_ratio": 0.0,
2014
+ "completions/max_length": 416.9375,
2015
+ "completions/max_terminated_length": 416.9375,
2016
+ "completions/mean_length": 364.75390625,
2017
+ "completions/mean_terminated_length": 364.75390625,
2018
+ "completions/min_length": 317.125,
2019
+ "completions/min_terminated_length": 317.125,
2020
+ "entropy": 0.033622848975937814,
2021
+ "epoch": 2.6755852842809364,
2022
+ "frac_reward_zero_std": 0.8125,
2023
+ "grad_norm": 0.20958314836025238,
2024
+ "kl": 0.028272411605030356,
2025
+ "learning_rate": 2.1683389074693424e-07,
2026
+ "loss": 0.0041,
2027
+ "num_tokens": 16111225.0,
2028
+ "reward": 0.751953125,
2029
+ "reward_std": 0.0578227024525404,
2030
+ "rewards/accuracy_reward/mean": 0.751953125,
2031
+ "rewards/accuracy_reward/std": 0.23291070200502872,
2032
+ "step": 3200
2033
+ },
2034
+ {
2035
+ "epoch": 2.6755852842809364,
2036
+ "eval_completions/clipped_ratio": 0.0,
2037
+ "eval_completions/max_length": 454.0,
2038
+ "eval_completions/max_terminated_length": 454.0,
2039
+ "eval_completions/mean_length": 364.9765625,
2040
+ "eval_completions/mean_terminated_length": 364.9765625,
2041
+ "eval_completions/min_length": 265.5,
2042
+ "eval_completions/min_terminated_length": 265.5,
2043
+ "eval_frac_reward_zero_std": 0.0,
2044
+ "eval_loss": 0.0,
2045
+ "eval_num_tokens": 16111225.0,
2046
+ "eval_reward": 0.7734375,
2047
+ "eval_reward_std": 0.0,
2048
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
2049
+ "eval_rewards/accuracy_reward/std": 0.4218114912509918,
2050
+ "eval_runtime": 51.4302,
2051
+ "eval_samples_per_second": 1.458,
2052
+ "eval_steps_per_second": 0.039,
2053
+ "step": 3200
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.00018844604164769407,
2057
+ "clip_ratio/high_mean": 9.422302082384704e-05,
2058
+ "clip_ratio/low_mean": 3.695105260703713e-05,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.00013117407343088416,
2061
+ "completions/clipped_ratio": 0.0,
2062
+ "completions/max_length": 423.5,
2063
+ "completions/max_terminated_length": 423.5,
2064
+ "completions/mean_length": 375.486328125,
2065
+ "completions/mean_terminated_length": 375.486328125,
2066
+ "completions/min_length": 332.5625,
2067
+ "completions/min_terminated_length": 332.5625,
2068
+ "entropy": 0.03422654085443355,
2069
+ "epoch": 2.729096989966555,
2070
+ "frac_reward_zero_std": 0.875,
2071
+ "grad_norm": 0.615790843963623,
2072
+ "kl": 0.03400245206444197,
2073
+ "learning_rate": 1.8115942028985507e-07,
2074
+ "loss": 0.0017,
2075
+ "num_tokens": 16437810.0,
2076
+ "reward": 0.826171875,
2077
+ "reward_std": 0.04914448596537113,
2078
+ "rewards/accuracy_reward/mean": 0.826171875,
2079
+ "rewards/accuracy_reward/std": 0.15970242768526077,
2080
+ "step": 3264
2081
+ },
2082
+ {
2083
+ "epoch": 2.759197324414716,
2084
+ "eval_completions/clipped_ratio": 0.0,
2085
+ "eval_completions/max_length": 438.0,
2086
+ "eval_completions/max_terminated_length": 438.0,
2087
+ "eval_completions/mean_length": 364.484375,
2088
+ "eval_completions/mean_terminated_length": 364.484375,
2089
+ "eval_completions/min_length": 265.0,
2090
+ "eval_completions/min_terminated_length": 265.0,
2091
+ "eval_frac_reward_zero_std": 0.0,
2092
+ "eval_loss": 0.0,
2093
+ "eval_num_tokens": 16618024.0,
2094
+ "eval_reward": 0.75,
2095
+ "eval_reward_std": 0.0,
2096
+ "eval_rewards/accuracy_reward/mean": 0.75,
2097
+ "eval_rewards/accuracy_reward/std": 0.4364357888698578,
2098
+ "eval_runtime": 49.5831,
2099
+ "eval_samples_per_second": 1.513,
2100
+ "eval_steps_per_second": 0.04,
2101
+ "step": 3300
2102
+ },
2103
+ {
2104
+ "clip_ratio/high_max": 0.0,
2105
+ "clip_ratio/high_mean": 0.0,
2106
+ "clip_ratio/low_mean": 0.0,
2107
+ "clip_ratio/low_min": 0.0,
2108
+ "clip_ratio/region_mean": 0.0,
2109
+ "completions/clipped_ratio": 0.0,
2110
+ "completions/max_length": 405.2857142857143,
2111
+ "completions/max_terminated_length": 405.2857142857143,
2112
+ "completions/mean_length": 369.0982142857143,
2113
+ "completions/mean_terminated_length": 369.0982142857143,
2114
+ "completions/min_length": 336.2857142857143,
2115
+ "completions/min_terminated_length": 336.2857142857143,
2116
+ "entropy": 0.03487457235210708,
2117
+ "epoch": 2.782608695652174,
2118
+ "frac_reward_zero_std": 0.9285714285714286,
2119
+ "grad_norm": 0.22031717002391815,
2120
+ "kl": 0.027312338524747508,
2121
+ "learning_rate": 1.454849498327759e-07,
2122
+ "loss": 0.0031,
2123
+ "num_tokens": 16759374.0,
2124
+ "reward": 0.6473214285714286,
2125
+ "reward_std": 0.017857142857142856,
2126
+ "rewards/accuracy_reward/mean": 0.6473214285714286,
2127
+ "rewards/accuracy_reward/std": 0.36271561895098003,
2128
+ "step": 3328
2129
+ },
2130
+ {
2131
+ "clip_ratio/high_max": 9.045138540386688e-05,
2132
+ "clip_ratio/high_mean": 4.522569270193344e-05,
2133
+ "clip_ratio/low_mean": 0.0,
2134
+ "clip_ratio/low_min": 0.0,
2135
+ "clip_ratio/region_mean": 4.522569270193344e-05,
2136
+ "completions/clipped_ratio": 0.0,
2137
+ "completions/max_length": 410.0,
2138
+ "completions/max_terminated_length": 410.0,
2139
+ "completions/mean_length": 373.52734375,
2140
+ "completions/mean_terminated_length": 373.52734375,
2141
+ "completions/min_length": 339.0625,
2142
+ "completions/min_terminated_length": 339.0625,
2143
+ "entropy": 0.028402078489307314,
2144
+ "epoch": 2.8361204013377925,
2145
+ "frac_reward_zero_std": 0.96875,
2146
+ "grad_norm": 0.012958893552422523,
2147
+ "kl": 0.012670181547264292,
2148
+ "learning_rate": 1.0981047937569676e-07,
2149
+ "loss": 0.0014,
2150
+ "num_tokens": 17084732.0,
2151
+ "reward": 0.779296875,
2152
+ "reward_std": 0.0078125,
2153
+ "rewards/accuracy_reward/mean": 0.779296875,
2154
+ "rewards/accuracy_reward/std": 0.16979869734495878,
2155
+ "step": 3392
2156
+ },
2157
+ {
2158
+ "epoch": 2.842809364548495,
2159
+ "eval_completions/clipped_ratio": 0.0,
2160
+ "eval_completions/max_length": 461.5,
2161
+ "eval_completions/max_terminated_length": 461.5,
2162
+ "eval_completions/mean_length": 365.59375,
2163
+ "eval_completions/mean_terminated_length": 365.59375,
2164
+ "eval_completions/min_length": 266.5,
2165
+ "eval_completions/min_terminated_length": 266.5,
2166
+ "eval_frac_reward_zero_std": 0.0,
2167
+ "eval_loss": 0.0,
2168
+ "eval_num_tokens": 17122985.0,
2169
+ "eval_reward": 0.7578125,
2170
+ "eval_reward_std": 0.0,
2171
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
2172
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
2173
+ "eval_runtime": 52.0405,
2174
+ "eval_samples_per_second": 1.441,
2175
+ "eval_steps_per_second": 0.038,
2176
+ "step": 3400
2177
+ },
2178
+ {
2179
+ "clip_ratio/high_max": 0.0002831326554379692,
2180
+ "clip_ratio/high_mean": 0.0001415663277189846,
2181
+ "clip_ratio/low_mean": 3.088332752148355e-05,
2182
+ "clip_ratio/low_min": 0.0,
2183
+ "clip_ratio/region_mean": 0.00017244965420104563,
2184
+ "completions/clipped_ratio": 0.0,
2185
+ "completions/max_length": 413.35714285714283,
2186
+ "completions/max_terminated_length": 413.35714285714283,
2187
+ "completions/mean_length": 371.0267857142857,
2188
+ "completions/mean_terminated_length": 371.0267857142857,
2189
+ "completions/min_length": 333.85714285714283,
2190
+ "completions/min_terminated_length": 333.85714285714283,
2191
+ "entropy": 0.035724993396018236,
2192
+ "epoch": 2.8896321070234112,
2193
+ "frac_reward_zero_std": 0.7857142857142857,
2194
+ "grad_norm": 0.18393680453300476,
2195
+ "kl": 0.03556311088842839,
2196
+ "learning_rate": 7.41360089186176e-08,
2197
+ "loss": 0.0008,
2198
+ "num_tokens": 17406757.0,
2199
+ "reward": 0.7991071428571429,
2200
+ "reward_std": 0.07387761771678925,
2201
+ "rewards/accuracy_reward/mean": 0.7991071428571429,
2202
+ "rewards/accuracy_reward/std": 0.22847512364387512,
2203
+ "step": 3456
2204
+ },
2205
+ {
2206
+ "epoch": 2.9264214046822743,
2207
+ "eval_completions/clipped_ratio": 0.0,
2208
+ "eval_completions/max_length": 445.0,
2209
+ "eval_completions/max_terminated_length": 445.0,
2210
+ "eval_completions/mean_length": 364.71875,
2211
+ "eval_completions/mean_terminated_length": 364.71875,
2212
+ "eval_completions/min_length": 269.5,
2213
+ "eval_completions/min_terminated_length": 269.5,
2214
+ "eval_frac_reward_zero_std": 0.0,
2215
+ "eval_loss": 0.0,
2216
+ "eval_num_tokens": 17632843.0,
2217
+ "eval_reward": 0.7578125,
2218
+ "eval_reward_std": 0.0,
2219
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
2220
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
2221
+ "eval_runtime": 49.7391,
2222
+ "eval_samples_per_second": 1.508,
2223
+ "eval_steps_per_second": 0.04,
2224
+ "step": 3500
2225
+ },
2226
+ {
2227
+ "clip_ratio/high_max": 3.3898305264301595e-05,
2228
+ "clip_ratio/high_mean": 1.6949152632150798e-05,
2229
+ "clip_ratio/low_mean": 5.1591800001915544e-05,
2230
+ "clip_ratio/low_min": 0.0,
2231
+ "clip_ratio/region_mean": 6.854095263406635e-05,
2232
+ "completions/clipped_ratio": 0.0,
2233
+ "completions/max_length": 402.8,
2234
+ "completions/max_terminated_length": 402.8,
2235
+ "completions/mean_length": 370.7625,
2236
+ "completions/mean_terminated_length": 370.7625,
2237
+ "completions/min_length": 338.4,
2238
+ "completions/min_terminated_length": 338.4,
2239
+ "entropy": 0.03341134488582611,
2240
+ "epoch": 2.94314381270903,
2241
+ "frac_reward_zero_std": 0.9,
2242
+ "grad_norm": 0.08761726319789886,
2243
+ "kl": 0.012898083019535989,
2244
+ "learning_rate": 3.846153846153846e-08,
2245
+ "loss": -0.0003,
2246
+ "num_tokens": 17733989.0,
2247
+ "reward": 0.61875,
2248
+ "reward_std": 0.04031128883361816,
2249
+ "rewards/accuracy_reward/mean": 0.61875,
2250
+ "rewards/accuracy_reward/std": 0.20139827728271484,
2251
+ "step": 3520
2252
+ },
2253
+ {
2254
+ "clip_ratio/high_max": 9.24700652831234e-05,
2255
+ "clip_ratio/high_mean": 4.62350326415617e-05,
2256
+ "clip_ratio/low_mean": 1.0218824172625318e-05,
2257
+ "clip_ratio/low_min": 0.0,
2258
+ "clip_ratio/region_mean": 5.645385681418702e-05,
2259
+ "completions/clipped_ratio": 0.0,
2260
+ "completions/max_length": 413.9375,
2261
+ "completions/max_terminated_length": 413.9375,
2262
+ "completions/mean_length": 365.89453125,
2263
+ "completions/mean_terminated_length": 365.89453125,
2264
+ "completions/min_length": 322.625,
2265
+ "completions/min_terminated_length": 322.625,
2266
+ "entropy": 0.03428677894407883,
2267
+ "epoch": 2.9966555183946486,
2268
+ "frac_reward_zero_std": 0.9375,
2269
+ "grad_norm": 0.1562715768814087,
2270
+ "kl": 0.022461608245976095,
2271
+ "learning_rate": 2.787068004459309e-09,
2272
+ "loss": 0.0023,
2273
+ "num_tokens": 18055679.0,
2274
+ "reward": 0.861328125,
2275
+ "reward_std": 0.025633705779910088,
2276
+ "rewards/accuracy_reward/mean": 0.861328125,
2277
+ "rewards/accuracy_reward/std": 0.10192735586315393,
2278
+ "step": 3584
2279
+ },
2280
+ {
2281
+ "clip_ratio/high_max": 0.0,
2282
+ "clip_ratio/high_mean": 0.0,
2283
+ "clip_ratio/low_mean": 0.0006547821831190959,
2284
+ "clip_ratio/low_min": 0.0,
2285
+ "clip_ratio/region_mean": 0.0006547821831190959,
2286
+ "completions/clipped_ratio": 0.0,
2287
+ "completions/max_length": 399.0,
2288
+ "completions/max_terminated_length": 399.0,
2289
+ "completions/mean_length": 354.875,
2290
+ "completions/mean_terminated_length": 354.875,
2291
+ "completions/min_length": 311.0,
2292
+ "completions/min_terminated_length": 311.0,
2293
+ "entropy": 0.03880858235061169,
2294
+ "epoch": 3.0,
2295
+ "frac_reward_zero_std": 0.5,
2296
+ "kl": 0.017735540866851807,
2297
+ "num_tokens": 18075371.0,
2298
+ "reward": 0.53125,
2299
+ "reward_std": 0.125,
2300
+ "rewards/accuracy_reward/mean": 0.53125,
2301
+ "rewards/accuracy_reward/std": 0.507007360458374,
2302
+ "step": 3588,
2303
+ "total_flos": 0.0,
2304
+ "train_loss": 0.004108291035237719,
2305
+ "train_runtime": 30603.4675,
2306
+ "train_samples_per_second": 0.059,
2307
+ "train_steps_per_second": 0.117
2308
+ }
2309
+ ],
2310
+ "logging_steps": 64,
2311
+ "max_steps": 3588,
2312
+ "num_input_tokens_seen": 18075371,
2313
+ "num_train_epochs": 3,
2314
+ "save_steps": 100,
2315
+ "stateful_callbacks": {
2316
+ "TrainerControl": {
2317
+ "args": {
2318
+ "should_epoch_stop": false,
2319
+ "should_evaluate": false,
2320
+ "should_log": false,
2321
+ "should_save": true,
2322
+ "should_training_stop": true
2323
+ },
2324
+ "attributes": {}
2325
+ }
2326
+ },
2327
+ "total_flos": 0.0,
2328
+ "train_batch_size": 4,
2329
+ "trial_name": null,
2330
+ "trial_params": null
2331
+ }
Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdb56ccff88d174aa43a5931acc1d93ce10dc8bb303dd2029218ca8cddd08d3d
3
+ size 8440
Blood/seed_2/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151645,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 2048,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 11008,
12
+ "layer_types": [
13
+ "full_attention",
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention"
49
+ ],
50
+ "max_position_embeddings": 32768,
51
+ "max_window_layers": 70,
52
+ "model_type": "qwen2",
53
+ "num_attention_heads": 16,
54
+ "num_hidden_layers": 36,
55
+ "num_key_value_heads": 2,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": true,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.55.2",
63
+ "use_cache": true,
64
+ "use_sliding_window": false,
65
+ "vocab_size": 151936
66
+ }
Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/tokenizer_config.json ADDED
@@ -0,0 +1,208 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ }
181
+ },
182
+ "additional_special_tokens": [
183
+ "<|im_start|>",
184
+ "<|im_end|>",
185
+ "<|object_ref_start|>",
186
+ "<|object_ref_end|>",
187
+ "<|box_start|>",
188
+ "<|box_end|>",
189
+ "<|quad_start|>",
190
+ "<|quad_end|>",
191
+ "<|vision_start|>",
192
+ "<|vision_end|>",
193
+ "<|vision_pad|>",
194
+ "<|image_pad|>",
195
+ "<|video_pad|>"
196
+ ],
197
+ "bos_token": null,
198
+ "clean_up_tokenization_spaces": false,
199
+ "eos_token": "<|im_end|>",
200
+ "errors": "replace",
201
+ "extra_special_tokens": {},
202
+ "model_max_length": 2048,
203
+ "pad_token": "<|endoftext|>",
204
+ "padding_side": "right",
205
+ "split_special_tokens": false,
206
+ "tokenizer_class": "Qwen2Tokenizer",
207
+ "unk_token": null
208
+ }
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/trainer_state.json ADDED
@@ -0,0 +1,2331 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 700,
3
+ "best_metric": 0.8,
4
+ "best_model_checkpoint": "/data/chenlang/Med3DVLM/llm/Blood/seeds/new/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/checkpoint-700",
5
+ "epoch": 3.0,
6
+ "eval_steps": 100,
7
+ "global_step": 3588,
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,
14
+ "eval_completions/clipped_ratio": 0.0,
15
+ "eval_completions/max_length": 500.0,
16
+ "eval_completions/max_terminated_length": 500.0,
17
+ "eval_completions/mean_length": 406.5546875,
18
+ "eval_completions/mean_terminated_length": 406.5546875,
19
+ "eval_completions/min_length": 253.0,
20
+ "eval_completions/min_terminated_length": 253.0,
21
+ "eval_frac_reward_zero_std": 0.0,
22
+ "eval_loss": 0.0,
23
+ "eval_num_tokens": 0.0,
24
+ "eval_reward": 0.7734375,
25
+ "eval_reward_std": 0.0,
26
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
27
+ "eval_rewards/accuracy_reward/std": 0.4192774146795273,
28
+ "eval_runtime": 56.5762,
29
+ "eval_samples_per_second": 1.326,
30
+ "eval_steps_per_second": 0.035,
31
+ "step": 0
32
+ },
33
+ {
34
+ "clip_ratio/high_max": 0.0005737788524129428,
35
+ "clip_ratio/high_mean": 0.0002868894262064714,
36
+ "clip_ratio/low_mean": 6.672977997368434e-05,
37
+ "clip_ratio/low_min": 0.0,
38
+ "clip_ratio/region_mean": 0.0003536192048159137,
39
+ "completions/clipped_ratio": 0.0,
40
+ "completions/max_length": 484.5,
41
+ "completions/max_terminated_length": 484.5,
42
+ "completions/mean_length": 388.58203125,
43
+ "completions/mean_terminated_length": 388.58203125,
44
+ "completions/min_length": 301.9375,
45
+ "completions/min_terminated_length": 301.9375,
46
+ "entropy": 0.05455621759756468,
47
+ "epoch": 0.05351170568561873,
48
+ "frac_reward_zero_std": 0.875,
49
+ "grad_norm": 0.04455799236893654,
50
+ "kl": 0.0009309545080213866,
51
+ "learning_rate": 1.9648829431438127e-06,
52
+ "loss": 0.0002,
53
+ "num_tokens": 333050.0,
54
+ "reward": 0.890625,
55
+ "reward_std": 0.04175759106874466,
56
+ "rewards/accuracy_reward/mean": 0.890625,
57
+ "rewards/accuracy_reward/std": 0.15554950665682554,
58
+ "step": 64
59
+ },
60
+ {
61
+ "epoch": 0.08361204013377926,
62
+ "eval_completions/clipped_ratio": 0.0,
63
+ "eval_completions/max_length": 486.0,
64
+ "eval_completions/max_terminated_length": 486.0,
65
+ "eval_completions/mean_length": 405.34375,
66
+ "eval_completions/mean_terminated_length": 405.34375,
67
+ "eval_completions/min_length": 248.5,
68
+ "eval_completions/min_terminated_length": 248.5,
69
+ "eval_frac_reward_zero_std": 0.0,
70
+ "eval_loss": 0.0,
71
+ "eval_num_tokens": 525875.0,
72
+ "eval_reward": 0.75,
73
+ "eval_reward_std": 0.0,
74
+ "eval_rewards/accuracy_reward/mean": 0.75,
75
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
76
+ "eval_runtime": 54.9627,
77
+ "eval_samples_per_second": 1.365,
78
+ "eval_steps_per_second": 0.036,
79
+ "step": 100
80
+ },
81
+ {
82
+ "clip_ratio/high_max": 0.0009298518229375727,
83
+ "clip_ratio/high_mean": 0.00046492591146878634,
84
+ "clip_ratio/low_mean": 0.0005951209833645928,
85
+ "clip_ratio/low_min": 0.0,
86
+ "clip_ratio/region_mean": 0.0010600468865179988,
87
+ "completions/clipped_ratio": 0.0,
88
+ "completions/max_length": 480.0,
89
+ "completions/max_terminated_length": 480.0,
90
+ "completions/mean_length": 411.0625,
91
+ "completions/mean_terminated_length": 411.0625,
92
+ "completions/min_length": 340.42857142857144,
93
+ "completions/min_terminated_length": 340.42857142857144,
94
+ "entropy": 0.05500567252082484,
95
+ "epoch": 0.10702341137123746,
96
+ "frac_reward_zero_std": 0.7857142857142857,
97
+ "grad_norm": 0.20202167332172394,
98
+ "kl": 0.01307384398699339,
99
+ "learning_rate": 1.9292084726867336e-06,
100
+ "loss": 0.0036,
101
+ "num_tokens": 676641.0,
102
+ "reward": 0.9151785714285714,
103
+ "reward_std": 0.10542563455445426,
104
+ "rewards/accuracy_reward/mean": 0.9151785714285714,
105
+ "rewards/accuracy_reward/std": 0.1708522183554513,
106
+ "step": 128
107
+ },
108
+ {
109
+ "clip_ratio/high_max": 0.0007363232725765556,
110
+ "clip_ratio/high_mean": 0.0004593329576891847,
111
+ "clip_ratio/low_mean": 0.0005956716659056838,
112
+ "clip_ratio/low_min": 0.0,
113
+ "clip_ratio/region_mean": 0.0010550046185926476,
114
+ "completions/clipped_ratio": 0.0,
115
+ "completions/max_length": 472.375,
116
+ "completions/max_terminated_length": 472.375,
117
+ "completions/mean_length": 394.740234375,
118
+ "completions/mean_terminated_length": 394.740234375,
119
+ "completions/min_length": 325.0,
120
+ "completions/min_terminated_length": 325.0,
121
+ "entropy": 0.06305783963762224,
122
+ "epoch": 0.1605351170568562,
123
+ "frac_reward_zero_std": 0.75,
124
+ "grad_norm": 0.21355415880680084,
125
+ "kl": 0.10797693437052658,
126
+ "learning_rate": 1.8935340022296544e-06,
127
+ "loss": 0.0099,
128
+ "num_tokens": 1012812.0,
129
+ "reward": 0.767578125,
130
+ "reward_std": 0.09855521284043789,
131
+ "rewards/accuracy_reward/mean": 0.767578125,
132
+ "rewards/accuracy_reward/std": 0.2924370802938938,
133
+ "step": 192
134
+ },
135
+ {
136
+ "epoch": 0.16722408026755853,
137
+ "eval_completions/clipped_ratio": 0.0,
138
+ "eval_completions/max_length": 512.5,
139
+ "eval_completions/max_terminated_length": 512.5,
140
+ "eval_completions/mean_length": 410.9296875,
141
+ "eval_completions/mean_terminated_length": 410.9296875,
142
+ "eval_completions/min_length": 256.5,
143
+ "eval_completions/min_terminated_length": 256.5,
144
+ "eval_frac_reward_zero_std": 0.0,
145
+ "eval_loss": 0.0,
146
+ "eval_num_tokens": 1056424.0,
147
+ "eval_reward": 0.7578125,
148
+ "eval_reward_std": 0.0,
149
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
150
+ "eval_rewards/accuracy_reward/std": 0.4293344020843506,
151
+ "eval_runtime": 57.214,
152
+ "eval_samples_per_second": 1.311,
153
+ "eval_steps_per_second": 0.035,
154
+ "step": 200
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.000557512412862187,
158
+ "clip_ratio/high_mean": 0.00028394421990794526,
159
+ "clip_ratio/low_mean": 0.00011464747876743786,
160
+ "clip_ratio/low_min": 3.0672824193191316e-05,
161
+ "clip_ratio/region_mean": 0.0003985916934782706,
162
+ "completions/clipped_ratio": 0.0,
163
+ "completions/max_length": 490.5,
164
+ "completions/max_terminated_length": 490.5,
165
+ "completions/mean_length": 409.39732142857144,
166
+ "completions/mean_terminated_length": 409.39732142857144,
167
+ "completions/min_length": 330.7857142857143,
168
+ "completions/min_terminated_length": 330.7857142857143,
169
+ "entropy": 0.08145251737109252,
170
+ "epoch": 0.2140468227424749,
171
+ "frac_reward_zero_std": 0.7142857142857143,
172
+ "grad_norm": 1.6099765300750732,
173
+ "kl": 0.06665297087082374,
174
+ "learning_rate": 1.8578595317725752e-06,
175
+ "loss": 0.0123,
176
+ "num_tokens": 1357162.0,
177
+ "reward": 0.8236607142857143,
178
+ "reward_std": 0.11614774167537689,
179
+ "rewards/accuracy_reward/mean": 0.8236607142857143,
180
+ "rewards/accuracy_reward/std": 0.24122382700443268,
181
+ "step": 256
182
+ },
183
+ {
184
+ "epoch": 0.2508361204013378,
185
+ "eval_completions/clipped_ratio": 0.0,
186
+ "eval_completions/max_length": 498.5,
187
+ "eval_completions/max_terminated_length": 498.5,
188
+ "eval_completions/mean_length": 412.2578125,
189
+ "eval_completions/mean_terminated_length": 412.2578125,
190
+ "eval_completions/min_length": 254.0,
191
+ "eval_completions/min_terminated_length": 254.0,
192
+ "eval_frac_reward_zero_std": 0.0,
193
+ "eval_loss": 0.0,
194
+ "eval_num_tokens": 1593123.0,
195
+ "eval_reward": 0.71875,
196
+ "eval_reward_std": 0.0,
197
+ "eval_rewards/accuracy_reward/mean": 0.71875,
198
+ "eval_rewards/accuracy_reward/std": 0.4518062025308609,
199
+ "eval_runtime": 56.2376,
200
+ "eval_samples_per_second": 1.334,
201
+ "eval_steps_per_second": 0.036,
202
+ "step": 300
203
+ },
204
+ {
205
+ "clip_ratio/high_max": 0.0,
206
+ "clip_ratio/high_mean": 0.0,
207
+ "clip_ratio/low_mean": 0.0,
208
+ "clip_ratio/low_min": 0.0,
209
+ "clip_ratio/region_mean": 0.0,
210
+ "completions/clipped_ratio": 0.0,
211
+ "completions/max_length": 477.6,
212
+ "completions/max_terminated_length": 477.6,
213
+ "completions/mean_length": 388.625,
214
+ "completions/mean_terminated_length": 388.625,
215
+ "completions/min_length": 303.0,
216
+ "completions/min_terminated_length": 303.0,
217
+ "entropy": 0.07273879945278168,
218
+ "epoch": 0.26755852842809363,
219
+ "frac_reward_zero_std": 1.0,
220
+ "grad_norm": 0.1347026526927948,
221
+ "kl": 0.039967592991888526,
222
+ "learning_rate": 1.822185061315496e-06,
223
+ "loss": 0.0026,
224
+ "num_tokens": 1697159.0,
225
+ "reward": 0.8,
226
+ "reward_std": 0.0,
227
+ "rewards/accuracy_reward/mean": 0.8,
228
+ "rewards/accuracy_reward/std": 0.20320019721984864,
229
+ "step": 320
230
+ },
231
+ {
232
+ "clip_ratio/high_max": 0.00035602805201051524,
233
+ "clip_ratio/high_mean": 0.00017801402600525762,
234
+ "clip_ratio/low_mean": 2.308020066266181e-05,
235
+ "clip_ratio/low_min": 0.0,
236
+ "clip_ratio/region_mean": 0.00020109422803216148,
237
+ "completions/clipped_ratio": 0.0,
238
+ "completions/max_length": 465.4375,
239
+ "completions/max_terminated_length": 465.4375,
240
+ "completions/mean_length": 390.41796875,
241
+ "completions/mean_terminated_length": 390.41796875,
242
+ "completions/min_length": 311.9375,
243
+ "completions/min_terminated_length": 311.9375,
244
+ "entropy": 0.07640599430305883,
245
+ "epoch": 0.3210702341137124,
246
+ "frac_reward_zero_std": 0.84375,
247
+ "grad_norm": 2.0845470428466797,
248
+ "kl": 0.0484608694241615,
249
+ "learning_rate": 1.786510590858417e-06,
250
+ "loss": 0.0024,
251
+ "num_tokens": 2031133.0,
252
+ "reward": 0.818359375,
253
+ "reward_std": 0.05523413047194481,
254
+ "rewards/accuracy_reward/mean": 0.818359375,
255
+ "rewards/accuracy_reward/std": 0.24027520697563887,
256
+ "step": 384
257
+ },
258
+ {
259
+ "epoch": 0.33444816053511706,
260
+ "eval_completions/clipped_ratio": 0.0,
261
+ "eval_completions/max_length": 569.5,
262
+ "eval_completions/max_terminated_length": 569.5,
263
+ "eval_completions/mean_length": 407.1875,
264
+ "eval_completions/mean_terminated_length": 407.1875,
265
+ "eval_completions/min_length": 262.5,
266
+ "eval_completions/min_terminated_length": 262.5,
267
+ "eval_frac_reward_zero_std": 0.0,
268
+ "eval_loss": 0.0,
269
+ "eval_num_tokens": 2120978.0,
270
+ "eval_reward": 0.7109375,
271
+ "eval_reward_std": 0.0,
272
+ "eval_rewards/accuracy_reward/mean": 0.7109375,
273
+ "eval_rewards/accuracy_reward/std": 0.4561667740345001,
274
+ "eval_runtime": 63.5869,
275
+ "eval_samples_per_second": 1.179,
276
+ "eval_steps_per_second": 0.031,
277
+ "step": 400
278
+ },
279
+ {
280
+ "clip_ratio/high_max": 0.00033333520574766834,
281
+ "clip_ratio/high_mean": 0.00016666760287383417,
282
+ "clip_ratio/low_mean": 5.0458998278675914e-05,
283
+ "clip_ratio/low_min": 0.0,
284
+ "clip_ratio/region_mean": 0.00021712660115251006,
285
+ "completions/clipped_ratio": 0.0,
286
+ "completions/max_length": 449.4166666666667,
287
+ "completions/max_terminated_length": 449.4166666666667,
288
+ "completions/mean_length": 384.1979166666667,
289
+ "completions/mean_terminated_length": 384.1979166666667,
290
+ "completions/min_length": 315.9166666666667,
291
+ "completions/min_terminated_length": 315.9166666666667,
292
+ "entropy": 0.07296963362023234,
293
+ "epoch": 0.3745819397993311,
294
+ "frac_reward_zero_std": 0.8333333333333334,
295
+ "grad_norm": 1.0998990535736084,
296
+ "kl": 0.04535827544653633,
297
+ "learning_rate": 1.7508361204013377e-06,
298
+ "loss": 0.0057,
299
+ "num_tokens": 2369022.0,
300
+ "reward": 0.765625,
301
+ "reward_std": 0.08216285953919093,
302
+ "rewards/accuracy_reward/mean": 0.765625,
303
+ "rewards/accuracy_reward/std": 0.2741037880380948,
304
+ "step": 448
305
+ },
306
+ {
307
+ "epoch": 0.4180602006688963,
308
+ "eval_completions/clipped_ratio": 0.0,
309
+ "eval_completions/max_length": 499.5,
310
+ "eval_completions/max_terminated_length": 499.5,
311
+ "eval_completions/mean_length": 404.6796875,
312
+ "eval_completions/mean_terminated_length": 404.6796875,
313
+ "eval_completions/min_length": 250.5,
314
+ "eval_completions/min_terminated_length": 250.5,
315
+ "eval_frac_reward_zero_std": 0.0,
316
+ "eval_loss": 0.0,
317
+ "eval_num_tokens": 2652857.0,
318
+ "eval_reward": 0.7578125,
319
+ "eval_reward_std": 0.0,
320
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
321
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
322
+ "eval_runtime": 55.605,
323
+ "eval_samples_per_second": 1.349,
324
+ "eval_steps_per_second": 0.036,
325
+ "step": 500
326
+ },
327
+ {
328
+ "clip_ratio/high_max": 0.0,
329
+ "clip_ratio/high_mean": 0.0,
330
+ "clip_ratio/low_mean": 0.0,
331
+ "clip_ratio/low_min": 0.0,
332
+ "clip_ratio/region_mean": 0.0,
333
+ "completions/clipped_ratio": 0.0,
334
+ "completions/max_length": 496.6666666666667,
335
+ "completions/max_terminated_length": 496.6666666666667,
336
+ "completions/mean_length": 439.2083333333333,
337
+ "completions/mean_terminated_length": 439.2083333333333,
338
+ "completions/min_length": 393.6666666666667,
339
+ "completions/min_terminated_length": 393.6666666666667,
340
+ "entropy": 0.08005397270123164,
341
+ "epoch": 0.4280936454849498,
342
+ "frac_reward_zero_std": 1.0,
343
+ "grad_norm": 0.31341928243637085,
344
+ "kl": 0.049614113134642444,
345
+ "learning_rate": 1.7151616499442586e-06,
346
+ "loss": 0.0047,
347
+ "num_tokens": 2720205.0,
348
+ "reward": 1.0,
349
+ "reward_std": 0.0,
350
+ "rewards/accuracy_reward/mean": 1.0,
351
+ "rewards/accuracy_reward/std": 0.0,
352
+ "step": 512
353
+ },
354
+ {
355
+ "clip_ratio/high_max": 0.0001360287678835448,
356
+ "clip_ratio/high_mean": 6.80143839417724e-05,
357
+ "clip_ratio/low_mean": 5.609181516774697e-05,
358
+ "clip_ratio/low_min": 0.0,
359
+ "clip_ratio/region_mean": 0.00012410619820002466,
360
+ "completions/clipped_ratio": 0.0,
361
+ "completions/max_length": 477.9375,
362
+ "completions/max_terminated_length": 477.9375,
363
+ "completions/mean_length": 398.828125,
364
+ "completions/mean_terminated_length": 398.828125,
365
+ "completions/min_length": 336.25,
366
+ "completions/min_terminated_length": 336.25,
367
+ "entropy": 0.07518683368107304,
368
+ "epoch": 0.4816053511705686,
369
+ "frac_reward_zero_std": 0.875,
370
+ "grad_norm": 0.19274647533893585,
371
+ "kl": 0.045726225442194846,
372
+ "learning_rate": 1.6794871794871794e-06,
373
+ "loss": 0.007,
374
+ "num_tokens": 3058629.0,
375
+ "reward": 0.81640625,
376
+ "reward_std": 0.04836062714457512,
377
+ "rewards/accuracy_reward/mean": 0.81640625,
378
+ "rewards/accuracy_reward/std": 0.22097810450941324,
379
+ "step": 576
380
+ },
381
+ {
382
+ "epoch": 0.5016722408026756,
383
+ "eval_completions/clipped_ratio": 0.0,
384
+ "eval_completions/max_length": 554.0,
385
+ "eval_completions/max_terminated_length": 554.0,
386
+ "eval_completions/mean_length": 409.75,
387
+ "eval_completions/mean_terminated_length": 409.75,
388
+ "eval_completions/min_length": 258.5,
389
+ "eval_completions/min_terminated_length": 258.5,
390
+ "eval_frac_reward_zero_std": 0.0,
391
+ "eval_loss": 0.0,
392
+ "eval_num_tokens": 3190091.0,
393
+ "eval_reward": 0.71875,
394
+ "eval_reward_std": 0.0,
395
+ "eval_rewards/accuracy_reward/mean": 0.71875,
396
+ "eval_rewards/accuracy_reward/std": 0.4528248459100723,
397
+ "eval_runtime": 61.6003,
398
+ "eval_samples_per_second": 1.218,
399
+ "eval_steps_per_second": 0.032,
400
+ "step": 600
401
+ },
402
+ {
403
+ "clip_ratio/high_max": 0.0004203029937343672,
404
+ "clip_ratio/high_mean": 0.0002619155522552319,
405
+ "clip_ratio/low_mean": 3.8457466143881905e-05,
406
+ "clip_ratio/low_min": 0.0,
407
+ "clip_ratio/region_mean": 0.00030037301839911377,
408
+ "completions/clipped_ratio": 0.0,
409
+ "completions/max_length": 477.6,
410
+ "completions/max_terminated_length": 477.6,
411
+ "completions/mean_length": 401.790625,
412
+ "completions/mean_terminated_length": 401.790625,
413
+ "completions/min_length": 326.8,
414
+ "completions/min_terminated_length": 326.8,
415
+ "entropy": 0.07270417772233487,
416
+ "epoch": 0.5351170568561873,
417
+ "frac_reward_zero_std": 0.75,
418
+ "grad_norm": 0.29571664333343506,
419
+ "kl": 0.04063929356634617,
420
+ "learning_rate": 1.6438127090301002e-06,
421
+ "loss": 0.0078,
422
+ "num_tokens": 3402504.0,
423
+ "reward": 0.875,
424
+ "reward_std": 0.09910764694213867,
425
+ "rewards/accuracy_reward/mean": 0.875,
426
+ "rewards/accuracy_reward/std": 0.19127324670553209,
427
+ "step": 640
428
+ },
429
+ {
430
+ "epoch": 0.5852842809364549,
431
+ "eval_completions/clipped_ratio": 0.0,
432
+ "eval_completions/max_length": 481.5,
433
+ "eval_completions/max_terminated_length": 481.5,
434
+ "eval_completions/mean_length": 406.0390625,
435
+ "eval_completions/mean_terminated_length": 406.0390625,
436
+ "eval_completions/min_length": 250.0,
437
+ "eval_completions/min_terminated_length": 250.0,
438
+ "eval_frac_reward_zero_std": 0.0,
439
+ "eval_loss": 0.0,
440
+ "eval_num_tokens": 3716985.0,
441
+ "eval_reward": 0.75,
442
+ "eval_reward_std": 0.0,
443
+ "eval_rewards/accuracy_reward/mean": 0.75,
444
+ "eval_rewards/accuracy_reward/std": 0.4364357888698578,
445
+ "eval_runtime": 54.6683,
446
+ "eval_samples_per_second": 1.372,
447
+ "eval_steps_per_second": 0.037,
448
+ "step": 700
449
+ },
450
+ {
451
+ "clip_ratio/high_max": 0.0,
452
+ "clip_ratio/high_mean": 0.0,
453
+ "clip_ratio/low_mean": 0.0,
454
+ "clip_ratio/low_min": 0.0,
455
+ "clip_ratio/region_mean": 0.0,
456
+ "completions/clipped_ratio": 0.0,
457
+ "completions/max_length": 489.0,
458
+ "completions/max_terminated_length": 489.0,
459
+ "completions/mean_length": 423.3125,
460
+ "completions/mean_terminated_length": 423.3125,
461
+ "completions/min_length": 362.0,
462
+ "completions/min_terminated_length": 362.0,
463
+ "entropy": 0.08161978796124458,
464
+ "epoch": 0.5886287625418061,
465
+ "frac_reward_zero_std": 1.0,
466
+ "grad_norm": 0.20454226434230804,
467
+ "kl": 0.025429074652493,
468
+ "learning_rate": 1.608138238573021e-06,
469
+ "loss": 0.0024,
470
+ "num_tokens": 3738931.0,
471
+ "reward": 1.0,
472
+ "reward_std": 0.0,
473
+ "rewards/accuracy_reward/mean": 1.0,
474
+ "rewards/accuracy_reward/std": 0.0,
475
+ "step": 704
476
+ },
477
+ {
478
+ "clip_ratio/high_max": 0.00013286507055454422,
479
+ "clip_ratio/high_mean": 6.643253527727211e-05,
480
+ "clip_ratio/low_mean": 9.813576070882846e-06,
481
+ "clip_ratio/low_min": 0.0,
482
+ "clip_ratio/region_mean": 7.624611134815495e-05,
483
+ "completions/clipped_ratio": 0.0,
484
+ "completions/max_length": 487.6875,
485
+ "completions/max_terminated_length": 487.6875,
486
+ "completions/mean_length": 411.509765625,
487
+ "completions/mean_terminated_length": 411.509765625,
488
+ "completions/min_length": 345.5,
489
+ "completions/min_terminated_length": 345.5,
490
+ "entropy": 0.07385398226324469,
491
+ "epoch": 0.6421404682274248,
492
+ "frac_reward_zero_std": 0.875,
493
+ "grad_norm": 0.29667434096336365,
494
+ "kl": 0.03678021153609734,
495
+ "learning_rate": 1.5724637681159421e-06,
496
+ "loss": 0.0047,
497
+ "num_tokens": 4083880.0,
498
+ "reward": 0.7421875,
499
+ "reward_std": 0.036034777760505676,
500
+ "rewards/accuracy_reward/mean": 0.7421875,
501
+ "rewards/accuracy_reward/std": 0.2421548506245017,
502
+ "step": 768
503
+ },
504
+ {
505
+ "epoch": 0.6688963210702341,
506
+ "eval_completions/clipped_ratio": 0.0,
507
+ "eval_completions/max_length": 477.0,
508
+ "eval_completions/max_terminated_length": 477.0,
509
+ "eval_completions/mean_length": 408.3671875,
510
+ "eval_completions/mean_terminated_length": 408.3671875,
511
+ "eval_completions/min_length": 260.5,
512
+ "eval_completions/min_terminated_length": 260.5,
513
+ "eval_frac_reward_zero_std": 0.0,
514
+ "eval_loss": 0.0,
515
+ "eval_num_tokens": 4258863.0,
516
+ "eval_reward": 0.75,
517
+ "eval_reward_std": 0.0,
518
+ "eval_rewards/accuracy_reward/mean": 0.75,
519
+ "eval_rewards/accuracy_reward/std": 0.4364357888698578,
520
+ "eval_runtime": 53.8893,
521
+ "eval_samples_per_second": 1.392,
522
+ "eval_steps_per_second": 0.037,
523
+ "step": 800
524
+ },
525
+ {
526
+ "clip_ratio/high_max": 0.00016690819029463455,
527
+ "clip_ratio/high_mean": 8.345409514731728e-05,
528
+ "clip_ratio/low_mean": 6.137437776487786e-05,
529
+ "clip_ratio/low_min": 0.0,
530
+ "clip_ratio/region_mean": 0.00014482847291219514,
531
+ "completions/clipped_ratio": 0.0,
532
+ "completions/max_length": 482.875,
533
+ "completions/max_terminated_length": 482.875,
534
+ "completions/mean_length": 402.09375,
535
+ "completions/mean_terminated_length": 402.09375,
536
+ "completions/min_length": 325.625,
537
+ "completions/min_terminated_length": 325.625,
538
+ "entropy": 0.07018477574456483,
539
+ "epoch": 0.6956521739130435,
540
+ "frac_reward_zero_std": 0.875,
541
+ "grad_norm": 0.2572539150714874,
542
+ "kl": 0.028143660005298443,
543
+ "learning_rate": 1.5367892976588628e-06,
544
+ "loss": 0.0114,
545
+ "num_tokens": 4428839.0,
546
+ "reward": 0.8359375,
547
+ "reward_std": 0.05721627548336983,
548
+ "rewards/accuracy_reward/mean": 0.8359375,
549
+ "rewards/accuracy_reward/std": 0.08951975032687187,
550
+ "step": 832
551
+ },
552
+ {
553
+ "clip_ratio/high_max": 0.00012594104646268534,
554
+ "clip_ratio/high_mean": 7.248635802170611e-05,
555
+ "clip_ratio/low_mean": 2.5105027361860266e-05,
556
+ "clip_ratio/low_min": 0.0,
557
+ "clip_ratio/region_mean": 9.759138629306108e-05,
558
+ "completions/clipped_ratio": 0.0,
559
+ "completions/max_length": 461.0,
560
+ "completions/max_terminated_length": 461.0,
561
+ "completions/mean_length": 393.013671875,
562
+ "completions/mean_terminated_length": 393.013671875,
563
+ "completions/min_length": 330.625,
564
+ "completions/min_terminated_length": 330.625,
565
+ "entropy": 0.07270828442415223,
566
+ "epoch": 0.7491638795986622,
567
+ "frac_reward_zero_std": 0.84375,
568
+ "grad_norm": 0.2998928725719452,
569
+ "kl": 0.03330114685377339,
570
+ "learning_rate": 1.5011148272017838e-06,
571
+ "loss": 0.002,
572
+ "num_tokens": 4764254.0,
573
+ "reward": 0.7890625,
574
+ "reward_std": 0.05880707688629627,
575
+ "rewards/accuracy_reward/mean": 0.7890625,
576
+ "rewards/accuracy_reward/std": 0.2213432164862752,
577
+ "step": 896
578
+ },
579
+ {
580
+ "epoch": 0.7525083612040134,
581
+ "eval_completions/clipped_ratio": 0.0,
582
+ "eval_completions/max_length": 495.5,
583
+ "eval_completions/max_terminated_length": 495.5,
584
+ "eval_completions/mean_length": 404.3046875,
585
+ "eval_completions/mean_terminated_length": 404.3046875,
586
+ "eval_completions/min_length": 250.5,
587
+ "eval_completions/min_terminated_length": 250.5,
588
+ "eval_frac_reward_zero_std": 0.0,
589
+ "eval_loss": 0.0,
590
+ "eval_num_tokens": 4784677.0,
591
+ "eval_reward": 0.7578125,
592
+ "eval_reward_std": 0.0,
593
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
594
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
595
+ "eval_runtime": 55.9911,
596
+ "eval_samples_per_second": 1.339,
597
+ "eval_steps_per_second": 0.036,
598
+ "step": 900
599
+ },
600
+ {
601
+ "clip_ratio/high_max": 0.0001594289894759034,
602
+ "clip_ratio/high_mean": 7.97144947379517e-05,
603
+ "clip_ratio/low_mean": 5.4793119003685814e-05,
604
+ "clip_ratio/low_min": 0.0,
605
+ "clip_ratio/region_mean": 0.00013450761374163753,
606
+ "completions/clipped_ratio": 0.0,
607
+ "completions/max_length": 479.3333333333333,
608
+ "completions/max_terminated_length": 479.3333333333333,
609
+ "completions/mean_length": 413.57708333333335,
610
+ "completions/mean_terminated_length": 413.57708333333335,
611
+ "completions/min_length": 336.1333333333333,
612
+ "completions/min_terminated_length": 336.1333333333333,
613
+ "entropy": 0.07614361097415288,
614
+ "epoch": 0.802675585284281,
615
+ "frac_reward_zero_std": 0.8333333333333334,
616
+ "grad_norm": 0.17494894564151764,
617
+ "kl": 0.03481740249941746,
618
+ "learning_rate": 1.4654403567447044e-06,
619
+ "loss": 0.0094,
620
+ "num_tokens": 5109066.0,
621
+ "reward": 0.5916666666666667,
622
+ "reward_std": 0.07507966558138529,
623
+ "rewards/accuracy_reward/mean": 0.5916666666666667,
624
+ "rewards/accuracy_reward/std": 0.392855371038119,
625
+ "step": 960
626
+ },
627
+ {
628
+ "epoch": 0.8361204013377926,
629
+ "eval_completions/clipped_ratio": 0.0,
630
+ "eval_completions/max_length": 528.0,
631
+ "eval_completions/max_terminated_length": 528.0,
632
+ "eval_completions/mean_length": 403.3515625,
633
+ "eval_completions/mean_terminated_length": 403.3515625,
634
+ "eval_completions/min_length": 247.5,
635
+ "eval_completions/min_terminated_length": 247.5,
636
+ "eval_frac_reward_zero_std": 0.0,
637
+ "eval_loss": 0.0,
638
+ "eval_num_tokens": 5326799.0,
639
+ "eval_reward": 0.765625,
640
+ "eval_reward_std": 0.0,
641
+ "eval_rewards/accuracy_reward/mean": 0.765625,
642
+ "eval_rewards/accuracy_reward/std": 0.4265512377023697,
643
+ "eval_runtime": 59.4013,
644
+ "eval_samples_per_second": 1.263,
645
+ "eval_steps_per_second": 0.034,
646
+ "step": 1000
647
+ },
648
+ {
649
+ "clip_ratio/high_max": 0.00015891849276764938,
650
+ "clip_ratio/high_mean": 7.945924638382469e-05,
651
+ "clip_ratio/low_mean": 9.069825197608831e-05,
652
+ "clip_ratio/low_min": 0.0,
653
+ "clip_ratio/region_mean": 0.0001701574971472534,
654
+ "completions/clipped_ratio": 0.0,
655
+ "completions/max_length": 499.1666666666667,
656
+ "completions/max_terminated_length": 499.1666666666667,
657
+ "completions/mean_length": 409.9635416666667,
658
+ "completions/mean_terminated_length": 409.9635416666667,
659
+ "completions/min_length": 343.1666666666667,
660
+ "completions/min_terminated_length": 343.1666666666667,
661
+ "entropy": 0.0759583196292321,
662
+ "epoch": 0.8561872909698997,
663
+ "frac_reward_zero_std": 0.8333333333333334,
664
+ "grad_norm": 2.6942520141601562,
665
+ "kl": 0.03244620930248251,
666
+ "learning_rate": 1.4297658862876255e-06,
667
+ "loss": 0.0043,
668
+ "num_tokens": 5455880.0,
669
+ "reward": 0.84375,
670
+ "reward_std": 0.05442607402801514,
671
+ "rewards/accuracy_reward/mean": 0.84375,
672
+ "rewards/accuracy_reward/std": 0.16766637563705444,
673
+ "step": 1024
674
+ },
675
+ {
676
+ "clip_ratio/high_max": 0.000129164556710748,
677
+ "clip_ratio/high_mean": 6.4582278355374e-05,
678
+ "clip_ratio/low_mean": 5.761430657003075e-06,
679
+ "clip_ratio/low_min": 0.0,
680
+ "clip_ratio/region_mean": 7.034370901237708e-05,
681
+ "completions/clipped_ratio": 0.0,
682
+ "completions/max_length": 488.875,
683
+ "completions/max_terminated_length": 488.875,
684
+ "completions/mean_length": 404.009765625,
685
+ "completions/mean_terminated_length": 404.009765625,
686
+ "completions/min_length": 334.6875,
687
+ "completions/min_terminated_length": 334.6875,
688
+ "entropy": 0.06897921918425709,
689
+ "epoch": 0.9096989966555183,
690
+ "frac_reward_zero_std": 0.9375,
691
+ "grad_norm": 0.1773137003183365,
692
+ "kl": 0.024513703643606277,
693
+ "learning_rate": 1.394091415830546e-06,
694
+ "loss": 0.0075,
695
+ "num_tokens": 5797005.0,
696
+ "reward": 0.9609375,
697
+ "reward_std": 0.021347813308238983,
698
+ "rewards/accuracy_reward/mean": 0.9609375,
699
+ "rewards/accuracy_reward/std": 0.06249186582863331,
700
+ "step": 1088
701
+ },
702
+ {
703
+ "epoch": 0.919732441471572,
704
+ "eval_completions/clipped_ratio": 0.0,
705
+ "eval_completions/max_length": 519.0,
706
+ "eval_completions/max_terminated_length": 519.0,
707
+ "eval_completions/mean_length": 405.5390625,
708
+ "eval_completions/mean_terminated_length": 405.5390625,
709
+ "eval_completions/min_length": 255.0,
710
+ "eval_completions/min_terminated_length": 255.0,
711
+ "eval_frac_reward_zero_std": 0.0,
712
+ "eval_loss": 0.0,
713
+ "eval_num_tokens": 5856780.0,
714
+ "eval_reward": 0.7890625,
715
+ "eval_reward_std": 0.0,
716
+ "eval_rewards/accuracy_reward/mean": 0.7890625,
717
+ "eval_rewards/accuracy_reward/std": 0.41108599305152893,
718
+ "eval_runtime": 58.2782,
719
+ "eval_samples_per_second": 1.287,
720
+ "eval_steps_per_second": 0.034,
721
+ "step": 1100
722
+ },
723
+ {
724
+ "clip_ratio/high_max": 0.00017278441541398375,
725
+ "clip_ratio/high_mean": 0.00010357623978052288,
726
+ "clip_ratio/low_mean": 6.92867117602593e-05,
727
+ "clip_ratio/low_min": 2.2228219537422636e-05,
728
+ "clip_ratio/region_mean": 0.00017286295154078218,
729
+ "completions/clipped_ratio": 0.0,
730
+ "completions/max_length": 477.53846153846155,
731
+ "completions/max_terminated_length": 477.53846153846155,
732
+ "completions/mean_length": 406.42788461538464,
733
+ "completions/mean_terminated_length": 406.42788461538464,
734
+ "completions/min_length": 329.7692307692308,
735
+ "completions/min_terminated_length": 329.7692307692308,
736
+ "entropy": 0.07567127815519388,
737
+ "epoch": 0.9632107023411371,
738
+ "frac_reward_zero_std": 0.8076923076923077,
739
+ "grad_norm": 0.16391460597515106,
740
+ "kl": 0.026717175675842624,
741
+ "learning_rate": 1.3584169453734671e-06,
742
+ "loss": -0.0006,
743
+ "num_tokens": 6134910.0,
744
+ "reward": 0.7043269230769231,
745
+ "reward_std": 0.07569497594466576,
746
+ "rewards/accuracy_reward/mean": 0.7043269230769231,
747
+ "rewards/accuracy_reward/std": 0.28925886291723985,
748
+ "step": 1152
749
+ },
750
+ {
751
+ "epoch": 1.0033444816053512,
752
+ "eval_completions/clipped_ratio": 0.0,
753
+ "eval_completions/max_length": 513.5,
754
+ "eval_completions/max_terminated_length": 513.5,
755
+ "eval_completions/mean_length": 403.953125,
756
+ "eval_completions/mean_terminated_length": 403.953125,
757
+ "eval_completions/min_length": 244.5,
758
+ "eval_completions/min_terminated_length": 244.5,
759
+ "eval_frac_reward_zero_std": 0.0,
760
+ "eval_loss": 0.0,
761
+ "eval_num_tokens": 6392167.0,
762
+ "eval_reward": 0.7421875,
763
+ "eval_reward_std": 0.0,
764
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
765
+ "eval_rewards/accuracy_reward/std": 0.4400599002838135,
766
+ "eval_runtime": 58.9215,
767
+ "eval_samples_per_second": 1.273,
768
+ "eval_steps_per_second": 0.034,
769
+ "step": 1200
770
+ },
771
+ {
772
+ "clip_ratio/high_max": 0.00014671361714135855,
773
+ "clip_ratio/high_mean": 7.335680857067928e-05,
774
+ "clip_ratio/low_mean": 0.0,
775
+ "clip_ratio/low_min": 0.0,
776
+ "clip_ratio/region_mean": 7.335680857067928e-05,
777
+ "completions/clipped_ratio": 0.0,
778
+ "completions/max_length": 477.75,
779
+ "completions/max_terminated_length": 477.75,
780
+ "completions/mean_length": 390.328125,
781
+ "completions/mean_terminated_length": 390.328125,
782
+ "completions/min_length": 323.25,
783
+ "completions/min_terminated_length": 323.25,
784
+ "entropy": 0.06390998419374228,
785
+ "epoch": 1.0167224080267558,
786
+ "frac_reward_zero_std": 0.875,
787
+ "grad_norm": 0.46081018447875977,
788
+ "kl": 0.022215171542484313,
789
+ "learning_rate": 1.3227424749163878e-06,
790
+ "loss": 0.0039,
791
+ "num_tokens": 6475585.0,
792
+ "reward": 0.515625,
793
+ "reward_std": 0.042695626616477966,
794
+ "rewards/accuracy_reward/mean": 0.515625,
795
+ "rewards/accuracy_reward/std": 0.3154839165508747,
796
+ "step": 1216
797
+ },
798
+ {
799
+ "clip_ratio/high_max": 0.0001303492090301006,
800
+ "clip_ratio/high_mean": 6.51746045150503e-05,
801
+ "clip_ratio/low_mean": 3.732054119609529e-05,
802
+ "clip_ratio/low_min": 0.0,
803
+ "clip_ratio/region_mean": 0.00010249514525639825,
804
+ "completions/clipped_ratio": 0.0,
805
+ "completions/max_length": 475.6875,
806
+ "completions/max_terminated_length": 475.6875,
807
+ "completions/mean_length": 398.3046875,
808
+ "completions/mean_terminated_length": 398.3046875,
809
+ "completions/min_length": 336.4375,
810
+ "completions/min_terminated_length": 336.4375,
811
+ "entropy": 0.06825923337601125,
812
+ "epoch": 1.0702341137123745,
813
+ "frac_reward_zero_std": 0.90625,
814
+ "grad_norm": 0.09459348767995834,
815
+ "kl": 0.02288869075709954,
816
+ "learning_rate": 1.2870680044593088e-06,
817
+ "loss": 0.0012,
818
+ "num_tokens": 6813581.0,
819
+ "reward": 0.83203125,
820
+ "reward_std": 0.039282044395804405,
821
+ "rewards/accuracy_reward/mean": 0.83203125,
822
+ "rewards/accuracy_reward/std": 0.19980881083756685,
823
+ "step": 1280
824
+ },
825
+ {
826
+ "epoch": 1.0869565217391304,
827
+ "eval_completions/clipped_ratio": 0.0,
828
+ "eval_completions/max_length": 489.5,
829
+ "eval_completions/max_terminated_length": 489.5,
830
+ "eval_completions/mean_length": 403.1953125,
831
+ "eval_completions/mean_terminated_length": 403.1953125,
832
+ "eval_completions/min_length": 250.5,
833
+ "eval_completions/min_terminated_length": 250.5,
834
+ "eval_frac_reward_zero_std": 0.0,
835
+ "eval_loss": 0.0,
836
+ "eval_num_tokens": 6918687.0,
837
+ "eval_reward": 0.75,
838
+ "eval_reward_std": 0.0,
839
+ "eval_rewards/accuracy_reward/mean": 0.75,
840
+ "eval_rewards/accuracy_reward/std": 0.4364357888698578,
841
+ "eval_runtime": 55.4236,
842
+ "eval_samples_per_second": 1.353,
843
+ "eval_steps_per_second": 0.036,
844
+ "step": 1300
845
+ },
846
+ {
847
+ "clip_ratio/high_max": 0.00025820941928858787,
848
+ "clip_ratio/high_mean": 0.00012910470964429393,
849
+ "clip_ratio/low_mean": 4.9222709177146584e-05,
850
+ "clip_ratio/low_min": 0.0,
851
+ "clip_ratio/region_mean": 0.0001783274194828912,
852
+ "completions/clipped_ratio": 0.0,
853
+ "completions/max_length": 458.3636363636364,
854
+ "completions/max_terminated_length": 458.3636363636364,
855
+ "completions/mean_length": 396.36647727272725,
856
+ "completions/mean_terminated_length": 396.36647727272725,
857
+ "completions/min_length": 336.09090909090907,
858
+ "completions/min_terminated_length": 336.09090909090907,
859
+ "entropy": 0.07279148037460717,
860
+ "epoch": 1.1237458193979932,
861
+ "frac_reward_zero_std": 0.8636363636363636,
862
+ "grad_norm": 0.20247255265712738,
863
+ "kl": 0.026983414967121047,
864
+ "learning_rate": 1.2513935340022296e-06,
865
+ "loss": 0.0055,
866
+ "num_tokens": 7150464.0,
867
+ "reward": 0.8579545454545454,
868
+ "reward_std": 0.05345123464410955,
869
+ "rewards/accuracy_reward/mean": 0.8579545454545454,
870
+ "rewards/accuracy_reward/std": 0.1866862787441774,
871
+ "step": 1344
872
+ },
873
+ {
874
+ "epoch": 1.1705685618729098,
875
+ "eval_completions/clipped_ratio": 0.0,
876
+ "eval_completions/max_length": 482.0,
877
+ "eval_completions/max_terminated_length": 482.0,
878
+ "eval_completions/mean_length": 402.9296875,
879
+ "eval_completions/mean_terminated_length": 402.9296875,
880
+ "eval_completions/min_length": 260.0,
881
+ "eval_completions/min_terminated_length": 260.0,
882
+ "eval_frac_reward_zero_std": 0.0,
883
+ "eval_loss": 0.0,
884
+ "eval_num_tokens": 7449414.0,
885
+ "eval_reward": 0.78125,
886
+ "eval_reward_std": 0.0,
887
+ "eval_rewards/accuracy_reward/mean": 0.78125,
888
+ "eval_rewards/accuracy_reward/std": 0.41270557045936584,
889
+ "eval_runtime": 54.4896,
890
+ "eval_samples_per_second": 1.376,
891
+ "eval_steps_per_second": 0.037,
892
+ "step": 1400
893
+ },
894
+ {
895
+ "clip_ratio/high_max": 0.0,
896
+ "clip_ratio/high_mean": 0.0,
897
+ "clip_ratio/low_mean": 0.0,
898
+ "clip_ratio/low_min": 0.0,
899
+ "clip_ratio/region_mean": 0.0,
900
+ "completions/clipped_ratio": 0.0,
901
+ "completions/max_length": 514.5,
902
+ "completions/max_terminated_length": 514.5,
903
+ "completions/mean_length": 438.859375,
904
+ "completions/mean_terminated_length": 438.859375,
905
+ "completions/min_length": 364.5,
906
+ "completions/min_terminated_length": 364.5,
907
+ "entropy": 0.08545570261776447,
908
+ "epoch": 1.1772575250836121,
909
+ "frac_reward_zero_std": 1.0,
910
+ "grad_norm": 0.280918151140213,
911
+ "kl": 0.03301918879151344,
912
+ "learning_rate": 1.2157190635451505e-06,
913
+ "loss": 0.0048,
914
+ "num_tokens": 7494285.0,
915
+ "reward": 1.0,
916
+ "reward_std": 0.0,
917
+ "rewards/accuracy_reward/mean": 1.0,
918
+ "rewards/accuracy_reward/std": 0.0,
919
+ "step": 1408
920
+ },
921
+ {
922
+ "clip_ratio/high_max": 0.00033302290194114903,
923
+ "clip_ratio/high_mean": 0.00016651145097057451,
924
+ "clip_ratio/low_mean": 8.066673490247922e-05,
925
+ "clip_ratio/low_min": 0.0,
926
+ "clip_ratio/region_mean": 0.0002471781854183064,
927
+ "completions/clipped_ratio": 0.0,
928
+ "completions/max_length": 474.5,
929
+ "completions/max_terminated_length": 474.5,
930
+ "completions/mean_length": 393.86328125,
931
+ "completions/mean_terminated_length": 393.86328125,
932
+ "completions/min_length": 322.75,
933
+ "completions/min_terminated_length": 322.75,
934
+ "entropy": 0.0690437750890851,
935
+ "epoch": 1.2307692307692308,
936
+ "frac_reward_zero_std": 0.8125,
937
+ "grad_norm": 0.7645003795623779,
938
+ "kl": 0.022957888922974234,
939
+ "learning_rate": 1.1800445930880713e-06,
940
+ "loss": 0.0,
941
+ "num_tokens": 7830087.0,
942
+ "reward": 0.77734375,
943
+ "reward_std": 0.07537247985601425,
944
+ "rewards/accuracy_reward/mean": 0.77734375,
945
+ "rewards/accuracy_reward/std": 0.27959971968084574,
946
+ "step": 1472
947
+ },
948
+ {
949
+ "epoch": 1.254180602006689,
950
+ "eval_completions/clipped_ratio": 0.0,
951
+ "eval_completions/max_length": 500.0,
952
+ "eval_completions/max_terminated_length": 500.0,
953
+ "eval_completions/mean_length": 404.375,
954
+ "eval_completions/mean_terminated_length": 404.375,
955
+ "eval_completions/min_length": 253.0,
956
+ "eval_completions/min_terminated_length": 253.0,
957
+ "eval_frac_reward_zero_std": 0.0,
958
+ "eval_loss": 0.0,
959
+ "eval_num_tokens": 7979615.0,
960
+ "eval_reward": 0.7578125,
961
+ "eval_reward_std": 0.0,
962
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
963
+ "eval_rewards/accuracy_reward/std": 0.4293344020843506,
964
+ "eval_runtime": 56.577,
965
+ "eval_samples_per_second": 1.326,
966
+ "eval_steps_per_second": 0.035,
967
+ "step": 1500
968
+ },
969
+ {
970
+ "clip_ratio/high_max": 5.1156129403453735e-05,
971
+ "clip_ratio/high_mean": 2.5578064701726867e-05,
972
+ "clip_ratio/low_mean": 5.29933167854324e-05,
973
+ "clip_ratio/low_min": 0.0,
974
+ "clip_ratio/region_mean": 7.857138148715926e-05,
975
+ "completions/clipped_ratio": 0.0,
976
+ "completions/max_length": 470.8888888888889,
977
+ "completions/max_terminated_length": 470.8888888888889,
978
+ "completions/mean_length": 382.21875,
979
+ "completions/mean_terminated_length": 382.21875,
980
+ "completions/min_length": 294.8888888888889,
981
+ "completions/min_terminated_length": 294.8888888888889,
982
+ "entropy": 0.06619931716057989,
983
+ "epoch": 1.2842809364548495,
984
+ "frac_reward_zero_std": 0.9444444444444444,
985
+ "grad_norm": 0.1408282369375229,
986
+ "kl": 0.015102360274694446,
987
+ "learning_rate": 1.1443701226309922e-06,
988
+ "loss": 0.0011,
989
+ "num_tokens": 8165182.0,
990
+ "reward": 0.7326388888888888,
991
+ "reward_std": 0.022395160463121202,
992
+ "rewards/accuracy_reward/mean": 0.7326388888888888,
993
+ "rewards/accuracy_reward/std": 0.2812214295069377,
994
+ "step": 1536
995
+ },
996
+ {
997
+ "clip_ratio/high_max": 6.820588987466181e-05,
998
+ "clip_ratio/high_mean": 3.879795667671715e-05,
999
+ "clip_ratio/low_mean": 1.4458280929829925e-05,
1000
+ "clip_ratio/low_min": 0.0,
1001
+ "clip_ratio/region_mean": 5.325623669705237e-05,
1002
+ "completions/clipped_ratio": 0.0,
1003
+ "completions/max_length": 459.75,
1004
+ "completions/max_terminated_length": 459.75,
1005
+ "completions/mean_length": 387.818359375,
1006
+ "completions/mean_terminated_length": 387.818359375,
1007
+ "completions/min_length": 320.75,
1008
+ "completions/min_terminated_length": 320.75,
1009
+ "entropy": 0.06727623427286744,
1010
+ "epoch": 1.3377926421404682,
1011
+ "frac_reward_zero_std": 0.90625,
1012
+ "grad_norm": 0.13344120979309082,
1013
+ "kl": 0.019887725342186968,
1014
+ "learning_rate": 1.108695652173913e-06,
1015
+ "loss": 0.0039,
1016
+ "num_tokens": 8497873.0,
1017
+ "reward": 0.87109375,
1018
+ "reward_std": 0.038397299125790596,
1019
+ "rewards/accuracy_reward/mean": 0.87109375,
1020
+ "rewards/accuracy_reward/std": 0.12367496453225613,
1021
+ "step": 1600
1022
+ },
1023
+ {
1024
+ "epoch": 1.3377926421404682,
1025
+ "eval_completions/clipped_ratio": 0.0,
1026
+ "eval_completions/max_length": 501.0,
1027
+ "eval_completions/max_terminated_length": 501.0,
1028
+ "eval_completions/mean_length": 398.4453125,
1029
+ "eval_completions/mean_terminated_length": 398.4453125,
1030
+ "eval_completions/min_length": 258.5,
1031
+ "eval_completions/min_terminated_length": 258.5,
1032
+ "eval_frac_reward_zero_std": 0.0,
1033
+ "eval_loss": 0.0,
1034
+ "eval_num_tokens": 8497873.0,
1035
+ "eval_reward": 0.78125,
1036
+ "eval_reward_std": 0.0,
1037
+ "eval_rewards/accuracy_reward/mean": 0.78125,
1038
+ "eval_rewards/accuracy_reward/std": 0.4166666865348816,
1039
+ "eval_runtime": 57.0247,
1040
+ "eval_samples_per_second": 1.315,
1041
+ "eval_steps_per_second": 0.035,
1042
+ "step": 1600
1043
+ },
1044
+ {
1045
+ "clip_ratio/high_max": 0.0002794734828057699,
1046
+ "clip_ratio/high_mean": 0.00013973674140288495,
1047
+ "clip_ratio/low_mean": 8.053069359448273e-05,
1048
+ "clip_ratio/low_min": 0.0,
1049
+ "clip_ratio/region_mean": 0.00022026743636160973,
1050
+ "completions/clipped_ratio": 0.0,
1051
+ "completions/max_length": 474.1875,
1052
+ "completions/max_terminated_length": 474.1875,
1053
+ "completions/mean_length": 400.814453125,
1054
+ "completions/mean_terminated_length": 400.814453125,
1055
+ "completions/min_length": 339.875,
1056
+ "completions/min_terminated_length": 339.875,
1057
+ "entropy": 0.06888031965354457,
1058
+ "epoch": 1.391304347826087,
1059
+ "frac_reward_zero_std": 0.78125,
1060
+ "grad_norm": 0.9663128852844238,
1061
+ "kl": 0.021658355797626427,
1062
+ "learning_rate": 1.0730211817168338e-06,
1063
+ "loss": 0.0007,
1064
+ "num_tokens": 8837250.0,
1065
+ "reward": 0.904296875,
1066
+ "reward_std": 0.08930579200387001,
1067
+ "rewards/accuracy_reward/mean": 0.904296875,
1068
+ "rewards/accuracy_reward/std": 0.16615330334752798,
1069
+ "step": 1664
1070
+ },
1071
+ {
1072
+ "epoch": 1.4214046822742474,
1073
+ "eval_completions/clipped_ratio": 0.0,
1074
+ "eval_completions/max_length": 494.0,
1075
+ "eval_completions/max_terminated_length": 494.0,
1076
+ "eval_completions/mean_length": 404.3203125,
1077
+ "eval_completions/mean_terminated_length": 404.3203125,
1078
+ "eval_completions/min_length": 253.5,
1079
+ "eval_completions/min_terminated_length": 253.5,
1080
+ "eval_frac_reward_zero_std": 0.0,
1081
+ "eval_loss": 0.0,
1082
+ "eval_num_tokens": 9031483.0,
1083
+ "eval_reward": 0.7734375,
1084
+ "eval_reward_std": 0.0,
1085
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1086
+ "eval_rewards/accuracy_reward/std": 0.4218114912509918,
1087
+ "eval_runtime": 56.1823,
1088
+ "eval_samples_per_second": 1.335,
1089
+ "eval_steps_per_second": 0.036,
1090
+ "step": 1700
1091
+ },
1092
+ {
1093
+ "clip_ratio/high_max": 0.0005283370930036264,
1094
+ "clip_ratio/high_mean": 0.0002641685465018132,
1095
+ "clip_ratio/low_mean": 0.00010849346810053768,
1096
+ "clip_ratio/low_min": 0.0,
1097
+ "clip_ratio/region_mean": 0.0003726620166811959,
1098
+ "completions/clipped_ratio": 0.0,
1099
+ "completions/max_length": 477.42857142857144,
1100
+ "completions/max_terminated_length": 477.42857142857144,
1101
+ "completions/mean_length": 406.51785714285717,
1102
+ "completions/mean_terminated_length": 406.51785714285717,
1103
+ "completions/min_length": 335.85714285714283,
1104
+ "completions/min_terminated_length": 335.85714285714283,
1105
+ "entropy": 0.07962273646678243,
1106
+ "epoch": 1.4448160535117056,
1107
+ "frac_reward_zero_std": 0.6428571428571429,
1108
+ "grad_norm": 2.3605217933654785,
1109
+ "kl": 0.03000039840117097,
1110
+ "learning_rate": 1.0373467112597547e-06,
1111
+ "loss": 0.0022,
1112
+ "num_tokens": 9181295.0,
1113
+ "reward": 0.7991071428571429,
1114
+ "reward_std": 0.14185658523014613,
1115
+ "rewards/accuracy_reward/mean": 0.7991071428571429,
1116
+ "rewards/accuracy_reward/std": 0.3185030094214848,
1117
+ "step": 1728
1118
+ },
1119
+ {
1120
+ "clip_ratio/high_max": 6.881475019326899e-05,
1121
+ "clip_ratio/high_mean": 3.4407375096634496e-05,
1122
+ "clip_ratio/low_mean": 3.38522177116829e-05,
1123
+ "clip_ratio/low_min": 0.0,
1124
+ "clip_ratio/region_mean": 6.82595937178121e-05,
1125
+ "completions/clipped_ratio": 0.0,
1126
+ "completions/max_length": 482.0,
1127
+ "completions/max_terminated_length": 482.0,
1128
+ "completions/mean_length": 412.345703125,
1129
+ "completions/mean_terminated_length": 412.345703125,
1130
+ "completions/min_length": 338.9375,
1131
+ "completions/min_terminated_length": 338.9375,
1132
+ "entropy": 0.07128582720179111,
1133
+ "epoch": 1.4983277591973243,
1134
+ "frac_reward_zero_std": 0.90625,
1135
+ "grad_norm": 0.141985222697258,
1136
+ "kl": 0.02132676024484681,
1137
+ "learning_rate": 1.0016722408026757e-06,
1138
+ "loss": 0.0083,
1139
+ "num_tokens": 9526640.0,
1140
+ "reward": 0.78515625,
1141
+ "reward_std": 0.038909729570150375,
1142
+ "rewards/accuracy_reward/mean": 0.78515625,
1143
+ "rewards/accuracy_reward/std": 0.1886013774201274,
1144
+ "step": 1792
1145
+ },
1146
+ {
1147
+ "epoch": 1.5050167224080266,
1148
+ "eval_completions/clipped_ratio": 0.0,
1149
+ "eval_completions/max_length": 479.5,
1150
+ "eval_completions/max_terminated_length": 479.5,
1151
+ "eval_completions/mean_length": 400.7265625,
1152
+ "eval_completions/mean_terminated_length": 400.7265625,
1153
+ "eval_completions/min_length": 249.5,
1154
+ "eval_completions/min_terminated_length": 249.5,
1155
+ "eval_frac_reward_zero_std": 0.0,
1156
+ "eval_loss": 0.0,
1157
+ "eval_num_tokens": 9570330.0,
1158
+ "eval_reward": 0.765625,
1159
+ "eval_reward_std": 0.0,
1160
+ "eval_rewards/accuracy_reward/mean": 0.765625,
1161
+ "eval_rewards/accuracy_reward/std": 0.4265512377023697,
1162
+ "eval_runtime": 53.566,
1163
+ "eval_samples_per_second": 1.4,
1164
+ "eval_steps_per_second": 0.037,
1165
+ "step": 1800
1166
+ },
1167
+ {
1168
+ "clip_ratio/high_max": 0.0001362620886149151,
1169
+ "clip_ratio/high_mean": 7.850707126116114e-05,
1170
+ "clip_ratio/low_mean": 2.5352627354940133e-05,
1171
+ "clip_ratio/low_min": 0.0,
1172
+ "clip_ratio/region_mean": 0.00010385969861610127,
1173
+ "completions/clipped_ratio": 0.0,
1174
+ "completions/max_length": 460.7142857142857,
1175
+ "completions/max_terminated_length": 460.7142857142857,
1176
+ "completions/mean_length": 387.95535714285717,
1177
+ "completions/mean_terminated_length": 387.95535714285717,
1178
+ "completions/min_length": 318.2142857142857,
1179
+ "completions/min_terminated_length": 318.2142857142857,
1180
+ "entropy": 0.06739772630057164,
1181
+ "epoch": 1.551839464882943,
1182
+ "frac_reward_zero_std": 0.8928571428571429,
1183
+ "grad_norm": 0.3352620601654053,
1184
+ "kl": 0.0183140971857938,
1185
+ "learning_rate": 9.659977703455963e-07,
1186
+ "loss": 0.0004,
1187
+ "num_tokens": 9861430.0,
1188
+ "reward": 0.8638392857142857,
1189
+ "reward_std": 0.03898446474756513,
1190
+ "rewards/accuracy_reward/mean": 0.8638392857142857,
1191
+ "rewards/accuracy_reward/std": 0.07134833719049181,
1192
+ "step": 1856
1193
+ },
1194
+ {
1195
+ "epoch": 1.588628762541806,
1196
+ "eval_completions/clipped_ratio": 0.0,
1197
+ "eval_completions/max_length": 482.5,
1198
+ "eval_completions/max_terminated_length": 482.5,
1199
+ "eval_completions/mean_length": 401.71875,
1200
+ "eval_completions/mean_terminated_length": 401.71875,
1201
+ "eval_completions/min_length": 260.0,
1202
+ "eval_completions/min_terminated_length": 260.0,
1203
+ "eval_frac_reward_zero_std": 0.0,
1204
+ "eval_loss": 0.0,
1205
+ "eval_num_tokens": 10096758.0,
1206
+ "eval_reward": 0.7578125,
1207
+ "eval_reward_std": 0.0,
1208
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1209
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
1210
+ "eval_runtime": 53.9573,
1211
+ "eval_samples_per_second": 1.39,
1212
+ "eval_steps_per_second": 0.037,
1213
+ "step": 1900
1214
+ },
1215
+ {
1216
+ "clip_ratio/high_max": 0.0,
1217
+ "clip_ratio/high_mean": 0.0,
1218
+ "clip_ratio/low_mean": 0.0,
1219
+ "clip_ratio/low_min": 0.0,
1220
+ "clip_ratio/region_mean": 0.0,
1221
+ "completions/clipped_ratio": 0.0,
1222
+ "completions/max_length": 444.6,
1223
+ "completions/max_terminated_length": 444.6,
1224
+ "completions/mean_length": 399.85625,
1225
+ "completions/mean_terminated_length": 399.85625,
1226
+ "completions/min_length": 321.6,
1227
+ "completions/min_terminated_length": 321.6,
1228
+ "entropy": 0.06419210489839315,
1229
+ "epoch": 1.605351170568562,
1230
+ "frac_reward_zero_std": 1.0,
1231
+ "grad_norm": 0.09829365462064743,
1232
+ "kl": 0.012354878752375953,
1233
+ "learning_rate": 9.303232998885172e-07,
1234
+ "loss": 0.0017,
1235
+ "num_tokens": 10202687.0,
1236
+ "reward": 0.7,
1237
+ "reward_std": 0.0,
1238
+ "rewards/accuracy_reward/mean": 0.7,
1239
+ "rewards/accuracy_reward/std": 0.30480029582977297,
1240
+ "step": 1920
1241
+ },
1242
+ {
1243
+ "clip_ratio/high_max": 0.00022835662821307778,
1244
+ "clip_ratio/high_mean": 0.00011417831410653889,
1245
+ "clip_ratio/low_mean": 0.0,
1246
+ "clip_ratio/low_min": 0.0,
1247
+ "clip_ratio/region_mean": 0.00011417831410653889,
1248
+ "completions/clipped_ratio": 0.0,
1249
+ "completions/max_length": 481.375,
1250
+ "completions/max_terminated_length": 481.375,
1251
+ "completions/mean_length": 408.984375,
1252
+ "completions/mean_terminated_length": 408.984375,
1253
+ "completions/min_length": 338.625,
1254
+ "completions/min_terminated_length": 338.625,
1255
+ "entropy": 0.07211820350494236,
1256
+ "epoch": 1.6588628762541806,
1257
+ "frac_reward_zero_std": 0.875,
1258
+ "grad_norm": 0.12647898495197296,
1259
+ "kl": 0.02006025089212926,
1260
+ "learning_rate": 8.94648829431438e-07,
1261
+ "loss": -0.0016,
1262
+ "num_tokens": 10546247.0,
1263
+ "reward": 0.76171875,
1264
+ "reward_std": 0.04412011243402958,
1265
+ "rewards/accuracy_reward/mean": 0.76171875,
1266
+ "rewards/accuracy_reward/std": 0.223596909083426,
1267
+ "step": 1984
1268
+ },
1269
+ {
1270
+ "epoch": 1.6722408026755853,
1271
+ "eval_completions/clipped_ratio": 0.0,
1272
+ "eval_completions/max_length": 497.0,
1273
+ "eval_completions/max_terminated_length": 497.0,
1274
+ "eval_completions/mean_length": 399.8671875,
1275
+ "eval_completions/mean_terminated_length": 399.8671875,
1276
+ "eval_completions/min_length": 252.5,
1277
+ "eval_completions/min_terminated_length": 252.5,
1278
+ "eval_frac_reward_zero_std": 0.0,
1279
+ "eval_loss": 0.0,
1280
+ "eval_num_tokens": 10629155.0,
1281
+ "eval_reward": 0.734375,
1282
+ "eval_reward_std": 0.0,
1283
+ "eval_rewards/accuracy_reward/mean": 0.734375,
1284
+ "eval_rewards/accuracy_reward/std": 0.44372452795505524,
1285
+ "eval_runtime": 55.3514,
1286
+ "eval_samples_per_second": 1.355,
1287
+ "eval_steps_per_second": 0.036,
1288
+ "step": 2000
1289
+ },
1290
+ {
1291
+ "clip_ratio/high_max": 0.00010111791198141873,
1292
+ "clip_ratio/high_mean": 5.0558955990709364e-05,
1293
+ "clip_ratio/low_mean": 6.405972696181077e-05,
1294
+ "clip_ratio/low_min": 0.0,
1295
+ "clip_ratio/region_mean": 0.00011461868173986052,
1296
+ "completions/clipped_ratio": 0.0,
1297
+ "completions/max_length": 469.5833333333333,
1298
+ "completions/max_terminated_length": 469.5833333333333,
1299
+ "completions/mean_length": 398.1744791666667,
1300
+ "completions/mean_terminated_length": 398.1744791666667,
1301
+ "completions/min_length": 322.8333333333333,
1302
+ "completions/min_terminated_length": 322.8333333333333,
1303
+ "entropy": 0.06620154102953772,
1304
+ "epoch": 1.7123745819397993,
1305
+ "frac_reward_zero_std": 0.9583333333333334,
1306
+ "grad_norm": 0.10976500064134598,
1307
+ "kl": 0.011338657544304928,
1308
+ "learning_rate": 8.589743589743588e-07,
1309
+ "loss": 0.0002,
1310
+ "num_tokens": 10882678.0,
1311
+ "reward": 0.8854166666666666,
1312
+ "reward_std": 0.018633899589379627,
1313
+ "rewards/accuracy_reward/mean": 0.8854166666666666,
1314
+ "rewards/accuracy_reward/std": 0.12565586219231287,
1315
+ "step": 2048
1316
+ },
1317
+ {
1318
+ "epoch": 1.7558528428093645,
1319
+ "eval_completions/clipped_ratio": 0.0,
1320
+ "eval_completions/max_length": 493.5,
1321
+ "eval_completions/max_terminated_length": 493.5,
1322
+ "eval_completions/mean_length": 403.921875,
1323
+ "eval_completions/mean_terminated_length": 403.921875,
1324
+ "eval_completions/min_length": 257.0,
1325
+ "eval_completions/min_terminated_length": 257.0,
1326
+ "eval_frac_reward_zero_std": 0.0,
1327
+ "eval_loss": 0.0,
1328
+ "eval_num_tokens": 11156763.0,
1329
+ "eval_reward": 0.796875,
1330
+ "eval_reward_std": 0.0,
1331
+ "eval_rewards/accuracy_reward/mean": 0.796875,
1332
+ "eval_rewards/accuracy_reward/std": 0.40503229200839996,
1333
+ "eval_runtime": 55.1933,
1334
+ "eval_samples_per_second": 1.359,
1335
+ "eval_steps_per_second": 0.036,
1336
+ "step": 2100
1337
+ },
1338
+ {
1339
+ "clip_ratio/high_max": 0.0003107721859123558,
1340
+ "clip_ratio/high_mean": 0.0001553860929561779,
1341
+ "clip_ratio/low_mean": 2.6172529032919556e-05,
1342
+ "clip_ratio/low_min": 0.0,
1343
+ "clip_ratio/region_mean": 0.00018155862198909745,
1344
+ "completions/clipped_ratio": 0.0,
1345
+ "completions/max_length": 480.6666666666667,
1346
+ "completions/max_terminated_length": 480.6666666666667,
1347
+ "completions/mean_length": 413.375,
1348
+ "completions/mean_terminated_length": 413.375,
1349
+ "completions/min_length": 345.6666666666667,
1350
+ "completions/min_terminated_length": 345.6666666666667,
1351
+ "entropy": 0.07046312031646569,
1352
+ "epoch": 1.7658862876254182,
1353
+ "frac_reward_zero_std": 0.8333333333333334,
1354
+ "grad_norm": 1.1858398914337158,
1355
+ "kl": 0.028363421481723588,
1356
+ "learning_rate": 8.232998885172798e-07,
1357
+ "loss": 0.0004,
1358
+ "num_tokens": 11221679.0,
1359
+ "reward": 0.7708333333333334,
1360
+ "reward_std": 0.08333333333333333,
1361
+ "rewards/accuracy_reward/mean": 0.7708333333333334,
1362
+ "rewards/accuracy_reward/std": 0.3015194237232208,
1363
+ "step": 2112
1364
+ },
1365
+ {
1366
+ "clip_ratio/high_max": 0.0001258936235899455,
1367
+ "clip_ratio/high_mean": 6.755052481821622e-05,
1368
+ "clip_ratio/low_mean": 8.187777348211966e-05,
1369
+ "clip_ratio/low_min": 9.603564649296459e-06,
1370
+ "clip_ratio/region_mean": 0.00014942829784558853,
1371
+ "completions/clipped_ratio": 0.0,
1372
+ "completions/max_length": 480.9375,
1373
+ "completions/max_terminated_length": 480.9375,
1374
+ "completions/mean_length": 406.427734375,
1375
+ "completions/mean_terminated_length": 406.427734375,
1376
+ "completions/min_length": 338.5625,
1377
+ "completions/min_terminated_length": 338.5625,
1378
+ "entropy": 0.0692291718441993,
1379
+ "epoch": 1.819397993311037,
1380
+ "frac_reward_zero_std": 0.78125,
1381
+ "grad_norm": 2.778951406478882,
1382
+ "kl": 0.01803175114764599,
1383
+ "learning_rate": 7.876254180602006e-07,
1384
+ "loss": 0.0106,
1385
+ "num_tokens": 11563930.0,
1386
+ "reward": 0.724609375,
1387
+ "reward_std": 0.08298446424305439,
1388
+ "rewards/accuracy_reward/mean": 0.724609375,
1389
+ "rewards/accuracy_reward/std": 0.2212205920368433,
1390
+ "step": 2176
1391
+ },
1392
+ {
1393
+ "epoch": 1.839464882943144,
1394
+ "eval_completions/clipped_ratio": 0.0,
1395
+ "eval_completions/max_length": 485.5,
1396
+ "eval_completions/max_terminated_length": 485.5,
1397
+ "eval_completions/mean_length": 399.2578125,
1398
+ "eval_completions/mean_terminated_length": 399.2578125,
1399
+ "eval_completions/min_length": 257.5,
1400
+ "eval_completions/min_terminated_length": 257.5,
1401
+ "eval_frac_reward_zero_std": 0.0,
1402
+ "eval_loss": 0.0,
1403
+ "eval_num_tokens": 11693002.0,
1404
+ "eval_reward": 0.75,
1405
+ "eval_reward_std": 0.0,
1406
+ "eval_rewards/accuracy_reward/mean": 0.75,
1407
+ "eval_rewards/accuracy_reward/std": 0.43605661392211914,
1408
+ "eval_runtime": 54.222,
1409
+ "eval_samples_per_second": 1.383,
1410
+ "eval_steps_per_second": 0.037,
1411
+ "step": 2200
1412
+ },
1413
+ {
1414
+ "clip_ratio/high_max": 0.00011596212425502017,
1415
+ "clip_ratio/high_mean": 5.7981062127510086e-05,
1416
+ "clip_ratio/low_mean": 4.922251537209377e-05,
1417
+ "clip_ratio/low_min": 0.0,
1418
+ "clip_ratio/region_mean": 0.00010720357749960385,
1419
+ "completions/clipped_ratio": 0.0,
1420
+ "completions/max_length": 449.5,
1421
+ "completions/max_terminated_length": 449.5,
1422
+ "completions/mean_length": 388.690625,
1423
+ "completions/mean_terminated_length": 388.690625,
1424
+ "completions/min_length": 337.8,
1425
+ "completions/min_terminated_length": 337.8,
1426
+ "entropy": 0.06797689041122794,
1427
+ "epoch": 1.8729096989966556,
1428
+ "frac_reward_zero_std": 0.9,
1429
+ "grad_norm": 0.057320088148117065,
1430
+ "kl": 0.016946042542986105,
1431
+ "learning_rate": 7.519509476031214e-07,
1432
+ "loss": 0.0025,
1433
+ "num_tokens": 11901255.0,
1434
+ "reward": 0.815625,
1435
+ "reward_std": 0.04797805547714233,
1436
+ "rewards/accuracy_reward/mean": 0.815625,
1437
+ "rewards/accuracy_reward/std": 0.22800256609916686,
1438
+ "step": 2240
1439
+ },
1440
+ {
1441
+ "epoch": 1.9230769230769231,
1442
+ "eval_completions/clipped_ratio": 0.0,
1443
+ "eval_completions/max_length": 498.5,
1444
+ "eval_completions/max_terminated_length": 498.5,
1445
+ "eval_completions/mean_length": 403.8515625,
1446
+ "eval_completions/mean_terminated_length": 403.8515625,
1447
+ "eval_completions/min_length": 252.0,
1448
+ "eval_completions/min_terminated_length": 252.0,
1449
+ "eval_frac_reward_zero_std": 0.0,
1450
+ "eval_loss": 0.0,
1451
+ "eval_num_tokens": 12225880.0,
1452
+ "eval_reward": 0.7578125,
1453
+ "eval_reward_std": 0.0,
1454
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1455
+ "eval_rewards/accuracy_reward/std": 0.4293344020843506,
1456
+ "eval_runtime": 55.7697,
1457
+ "eval_samples_per_second": 1.345,
1458
+ "eval_steps_per_second": 0.036,
1459
+ "step": 2300
1460
+ },
1461
+ {
1462
+ "clip_ratio/high_max": 0.0,
1463
+ "clip_ratio/high_mean": 0.0,
1464
+ "clip_ratio/low_mean": 0.0,
1465
+ "clip_ratio/low_min": 0.0,
1466
+ "clip_ratio/region_mean": 0.0,
1467
+ "completions/clipped_ratio": 0.0,
1468
+ "completions/max_length": 498.0,
1469
+ "completions/max_terminated_length": 498.0,
1470
+ "completions/mean_length": 437.8125,
1471
+ "completions/mean_terminated_length": 437.8125,
1472
+ "completions/min_length": 389.0,
1473
+ "completions/min_terminated_length": 389.0,
1474
+ "entropy": 0.07395693473517895,
1475
+ "epoch": 1.9264214046822743,
1476
+ "frac_reward_zero_std": 1.0,
1477
+ "grad_norm": 0.1640491783618927,
1478
+ "kl": 0.018250879365950823,
1479
+ "learning_rate": 7.162764771460424e-07,
1480
+ "loss": 0.0057,
1481
+ "num_tokens": 12248242.0,
1482
+ "reward": 0.5,
1483
+ "reward_std": 0.0,
1484
+ "rewards/accuracy_reward/mean": 0.5,
1485
+ "rewards/accuracy_reward/std": 0.5080004930496216,
1486
+ "step": 2304
1487
+ },
1488
+ {
1489
+ "clip_ratio/high_max": 8.477509709337028e-05,
1490
+ "clip_ratio/high_mean": 4.238754854668514e-05,
1491
+ "clip_ratio/low_mean": 5.9837275784957455e-05,
1492
+ "clip_ratio/low_min": 0.0,
1493
+ "clip_ratio/region_mean": 0.00010222482342214789,
1494
+ "completions/clipped_ratio": 0.0,
1495
+ "completions/max_length": 482.5625,
1496
+ "completions/max_terminated_length": 482.5625,
1497
+ "completions/mean_length": 422.08984375,
1498
+ "completions/mean_terminated_length": 422.08984375,
1499
+ "completions/min_length": 365.75,
1500
+ "completions/min_terminated_length": 365.75,
1501
+ "entropy": 0.06926613615360111,
1502
+ "epoch": 1.979933110367893,
1503
+ "frac_reward_zero_std": 0.8125,
1504
+ "grad_norm": 0.22166195511817932,
1505
+ "kl": 0.01740988768506213,
1506
+ "learning_rate": 6.806020066889632e-07,
1507
+ "loss": 0.0104,
1508
+ "num_tokens": 12598512.0,
1509
+ "reward": 0.8359375,
1510
+ "reward_std": 0.07015972957015038,
1511
+ "rewards/accuracy_reward/mean": 0.8359375,
1512
+ "rewards/accuracy_reward/std": 0.2143780468031764,
1513
+ "step": 2368
1514
+ },
1515
+ {
1516
+ "epoch": 2.0066889632107023,
1517
+ "eval_completions/clipped_ratio": 0.0,
1518
+ "eval_completions/max_length": 496.5,
1519
+ "eval_completions/max_terminated_length": 496.5,
1520
+ "eval_completions/mean_length": 401.453125,
1521
+ "eval_completions/mean_terminated_length": 401.453125,
1522
+ "eval_completions/min_length": 257.5,
1523
+ "eval_completions/min_terminated_length": 257.5,
1524
+ "eval_frac_reward_zero_std": 0.0,
1525
+ "eval_loss": 0.0,
1526
+ "eval_num_tokens": 12769080.0,
1527
+ "eval_reward": 0.7421875,
1528
+ "eval_reward_std": 0.0,
1529
+ "eval_rewards/accuracy_reward/mean": 0.7421875,
1530
+ "eval_rewards/accuracy_reward/std": 0.4400599002838135,
1531
+ "eval_runtime": 55.3171,
1532
+ "eval_samples_per_second": 1.356,
1533
+ "eval_steps_per_second": 0.036,
1534
+ "step": 2400
1535
+ },
1536
+ {
1537
+ "clip_ratio/high_max": 0.00021754772023996338,
1538
+ "clip_ratio/high_mean": 0.00010877386011998169,
1539
+ "clip_ratio/low_mean": 5.1196228923799936e-05,
1540
+ "clip_ratio/low_min": 0.0,
1541
+ "clip_ratio/region_mean": 0.00015997009177226573,
1542
+ "completions/clipped_ratio": 0.0,
1543
+ "completions/max_length": 461.25,
1544
+ "completions/max_terminated_length": 461.25,
1545
+ "completions/mean_length": 383.29296875,
1546
+ "completions/mean_terminated_length": 383.29296875,
1547
+ "completions/min_length": 308.125,
1548
+ "completions/min_terminated_length": 308.125,
1549
+ "entropy": 0.060442831134423614,
1550
+ "epoch": 2.0334448160535117,
1551
+ "frac_reward_zero_std": 0.8125,
1552
+ "grad_norm": 1.918314814567566,
1553
+ "kl": 0.01564921912358841,
1554
+ "learning_rate": 6.449275362318841e-07,
1555
+ "loss": 0.0024,
1556
+ "num_tokens": 12934419.0,
1557
+ "reward": 0.8046875,
1558
+ "reward_std": 0.07889671996235847,
1559
+ "rewards/accuracy_reward/mean": 0.8046875,
1560
+ "rewards/accuracy_reward/std": 0.24284254387021065,
1561
+ "step": 2432
1562
+ },
1563
+ {
1564
+ "clip_ratio/high_max": 0.00015874328892095946,
1565
+ "clip_ratio/high_mean": 7.937164446047973e-05,
1566
+ "clip_ratio/low_mean": 0.0,
1567
+ "clip_ratio/low_min": 0.0,
1568
+ "clip_ratio/region_mean": 7.937164446047973e-05,
1569
+ "completions/clipped_ratio": 0.0,
1570
+ "completions/max_length": 464.0625,
1571
+ "completions/max_terminated_length": 464.0625,
1572
+ "completions/mean_length": 390.00390625,
1573
+ "completions/mean_terminated_length": 390.00390625,
1574
+ "completions/min_length": 312.875,
1575
+ "completions/min_terminated_length": 312.875,
1576
+ "entropy": 0.06245889712590724,
1577
+ "epoch": 2.0869565217391304,
1578
+ "frac_reward_zero_std": 0.84375,
1579
+ "grad_norm": 0.18844355642795563,
1580
+ "kl": 0.01606141249249049,
1581
+ "learning_rate": 6.092530657748049e-07,
1582
+ "loss": 0.004,
1583
+ "num_tokens": 13268181.0,
1584
+ "reward": 0.9140625,
1585
+ "reward_std": 0.04738743044435978,
1586
+ "rewards/accuracy_reward/mean": 0.9140625,
1587
+ "rewards/accuracy_reward/std": 0.13519056886434555,
1588
+ "step": 2496
1589
+ },
1590
+ {
1591
+ "epoch": 2.0903010033444818,
1592
+ "eval_completions/clipped_ratio": 0.0,
1593
+ "eval_completions/max_length": 492.0,
1594
+ "eval_completions/max_terminated_length": 492.0,
1595
+ "eval_completions/mean_length": 399.9609375,
1596
+ "eval_completions/mean_terminated_length": 399.9609375,
1597
+ "eval_completions/min_length": 242.0,
1598
+ "eval_completions/min_terminated_length": 242.0,
1599
+ "eval_frac_reward_zero_std": 0.0,
1600
+ "eval_loss": 0.0,
1601
+ "eval_num_tokens": 13288161.0,
1602
+ "eval_reward": 0.7734375,
1603
+ "eval_reward_std": 0.0,
1604
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1605
+ "eval_rewards/accuracy_reward/std": 0.4218114912509918,
1606
+ "eval_runtime": 54.5548,
1607
+ "eval_samples_per_second": 1.375,
1608
+ "eval_steps_per_second": 0.037,
1609
+ "step": 2500
1610
+ },
1611
+ {
1612
+ "clip_ratio/high_max": 0.00021818012367778769,
1613
+ "clip_ratio/high_mean": 0.00011889975430676714,
1614
+ "clip_ratio/low_mean": 4.738835559692234e-05,
1615
+ "clip_ratio/low_min": 0.0,
1616
+ "clip_ratio/region_mean": 0.0001662881089335618,
1617
+ "completions/clipped_ratio": 0.0,
1618
+ "completions/max_length": 496.1333333333333,
1619
+ "completions/max_terminated_length": 496.1333333333333,
1620
+ "completions/mean_length": 400.3125,
1621
+ "completions/mean_terminated_length": 400.3125,
1622
+ "completions/min_length": 329.06666666666666,
1623
+ "completions/min_terminated_length": 329.06666666666666,
1624
+ "entropy": 0.07029800980041424,
1625
+ "epoch": 2.140468227424749,
1626
+ "frac_reward_zero_std": 0.7666666666666667,
1627
+ "grad_norm": 0.12676171958446503,
1628
+ "kl": 0.017872429029860845,
1629
+ "learning_rate": 5.735785953177257e-07,
1630
+ "loss": -0.0005,
1631
+ "num_tokens": 13606071.0,
1632
+ "reward": 0.7895833333333333,
1633
+ "reward_std": 0.08755014538764953,
1634
+ "rewards/accuracy_reward/mean": 0.7895833333333333,
1635
+ "rewards/accuracy_reward/std": 0.2574740966161092,
1636
+ "step": 2560
1637
+ },
1638
+ {
1639
+ "epoch": 2.1739130434782608,
1640
+ "eval_completions/clipped_ratio": 0.0,
1641
+ "eval_completions/max_length": 496.5,
1642
+ "eval_completions/max_terminated_length": 496.5,
1643
+ "eval_completions/mean_length": 402.9921875,
1644
+ "eval_completions/mean_terminated_length": 402.9921875,
1645
+ "eval_completions/min_length": 267.5,
1646
+ "eval_completions/min_terminated_length": 267.5,
1647
+ "eval_frac_reward_zero_std": 0.0,
1648
+ "eval_loss": 0.0,
1649
+ "eval_num_tokens": 13815070.0,
1650
+ "eval_reward": 0.75,
1651
+ "eval_reward_std": 0.0,
1652
+ "eval_rewards/accuracy_reward/mean": 0.75,
1653
+ "eval_rewards/accuracy_reward/std": 0.43491509556770325,
1654
+ "eval_runtime": 55.3195,
1655
+ "eval_samples_per_second": 1.356,
1656
+ "eval_steps_per_second": 0.036,
1657
+ "step": 2600
1658
+ },
1659
+ {
1660
+ "clip_ratio/high_max": 0.0,
1661
+ "clip_ratio/high_mean": 0.0,
1662
+ "clip_ratio/low_mean": 0.0,
1663
+ "clip_ratio/low_min": 0.0,
1664
+ "clip_ratio/region_mean": 0.0,
1665
+ "completions/clipped_ratio": 0.0,
1666
+ "completions/max_length": 452.6666666666667,
1667
+ "completions/max_terminated_length": 452.6666666666667,
1668
+ "completions/mean_length": 378.0729166666667,
1669
+ "completions/mean_terminated_length": 378.0729166666667,
1670
+ "completions/min_length": 310.8333333333333,
1671
+ "completions/min_terminated_length": 310.8333333333333,
1672
+ "entropy": 0.06803068161631624,
1673
+ "epoch": 2.1939799331103678,
1674
+ "frac_reward_zero_std": 1.0,
1675
+ "grad_norm": 0.2740255296230316,
1676
+ "kl": 0.01596026936507163,
1677
+ "learning_rate": 5.379041248606467e-07,
1678
+ "loss": 0.0024,
1679
+ "num_tokens": 13937980.0,
1680
+ "reward": 1.0,
1681
+ "reward_std": 0.0,
1682
+ "rewards/accuracy_reward/mean": 1.0,
1683
+ "rewards/accuracy_reward/std": 0.0,
1684
+ "step": 2624
1685
+ },
1686
+ {
1687
+ "clip_ratio/high_max": 0.0001891011461339076,
1688
+ "clip_ratio/high_mean": 9.45505730669538e-05,
1689
+ "clip_ratio/low_mean": 4.221822200634051e-05,
1690
+ "clip_ratio/low_min": 0.0,
1691
+ "clip_ratio/region_mean": 0.0001367687941637996,
1692
+ "completions/clipped_ratio": 0.0,
1693
+ "completions/max_length": 474.625,
1694
+ "completions/max_terminated_length": 474.625,
1695
+ "completions/mean_length": 410.6328125,
1696
+ "completions/mean_terminated_length": 410.6328125,
1697
+ "completions/min_length": 351.3125,
1698
+ "completions/min_terminated_length": 351.3125,
1699
+ "entropy": 0.0657982278498821,
1700
+ "epoch": 2.2474916387959865,
1701
+ "frac_reward_zero_std": 0.78125,
1702
+ "grad_norm": 0.4276233911514282,
1703
+ "kl": 0.016005878359464987,
1704
+ "learning_rate": 5.022296544035675e-07,
1705
+ "loss": 0.0034,
1706
+ "num_tokens": 14282400.0,
1707
+ "reward": 0.783203125,
1708
+ "reward_std": 0.07250870577991009,
1709
+ "rewards/accuracy_reward/mean": 0.783203125,
1710
+ "rewards/accuracy_reward/std": 0.20767589565366507,
1711
+ "step": 2688
1712
+ },
1713
+ {
1714
+ "epoch": 2.25752508361204,
1715
+ "eval_completions/clipped_ratio": 0.0,
1716
+ "eval_completions/max_length": 492.0,
1717
+ "eval_completions/max_terminated_length": 492.0,
1718
+ "eval_completions/mean_length": 399.5078125,
1719
+ "eval_completions/mean_terminated_length": 399.5078125,
1720
+ "eval_completions/min_length": 246.5,
1721
+ "eval_completions/min_terminated_length": 246.5,
1722
+ "eval_frac_reward_zero_std": 0.0,
1723
+ "eval_loss": 0.0,
1724
+ "eval_num_tokens": 14347182.0,
1725
+ "eval_reward": 0.7578125,
1726
+ "eval_reward_std": 0.0,
1727
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1728
+ "eval_rewards/accuracy_reward/std": 0.4293344020843506,
1729
+ "eval_runtime": 55.0226,
1730
+ "eval_samples_per_second": 1.363,
1731
+ "eval_steps_per_second": 0.036,
1732
+ "step": 2700
1733
+ },
1734
+ {
1735
+ "clip_ratio/high_max": 7.66884936288429e-05,
1736
+ "clip_ratio/high_mean": 5.171135405991943e-05,
1737
+ "clip_ratio/low_mean": 3.0614767008675976e-05,
1738
+ "clip_ratio/low_min": 0.0,
1739
+ "clip_ratio/region_mean": 8.232612162828445e-05,
1740
+ "completions/clipped_ratio": 0.0,
1741
+ "completions/max_length": 474.38461538461536,
1742
+ "completions/max_terminated_length": 474.38461538461536,
1743
+ "completions/mean_length": 403.1394230769231,
1744
+ "completions/mean_terminated_length": 403.1394230769231,
1745
+ "completions/min_length": 346.0769230769231,
1746
+ "completions/min_terminated_length": 346.0769230769231,
1747
+ "entropy": 0.07397414020334299,
1748
+ "epoch": 2.3010033444816056,
1749
+ "frac_reward_zero_std": 0.8846153846153846,
1750
+ "grad_norm": 0.23086923360824585,
1751
+ "kl": 0.019218805854996808,
1752
+ "learning_rate": 4.665551839464883e-07,
1753
+ "loss": 0.0077,
1754
+ "num_tokens": 14623944.0,
1755
+ "reward": 0.7331730769230769,
1756
+ "reward_std": 0.039953023195266724,
1757
+ "rewards/accuracy_reward/mean": 0.7331730769230769,
1758
+ "rewards/accuracy_reward/std": 0.21692430056058443,
1759
+ "step": 2752
1760
+ },
1761
+ {
1762
+ "epoch": 2.3411371237458196,
1763
+ "eval_completions/clipped_ratio": 0.0,
1764
+ "eval_completions/max_length": 496.5,
1765
+ "eval_completions/max_terminated_length": 496.5,
1766
+ "eval_completions/mean_length": 403.1953125,
1767
+ "eval_completions/mean_terminated_length": 403.1953125,
1768
+ "eval_completions/min_length": 247.0,
1769
+ "eval_completions/min_terminated_length": 247.0,
1770
+ "eval_frac_reward_zero_std": 0.0,
1771
+ "eval_loss": 0.0,
1772
+ "eval_num_tokens": 14876012.0,
1773
+ "eval_reward": 0.7734375,
1774
+ "eval_reward_std": 0.0,
1775
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1776
+ "eval_rewards/accuracy_reward/std": 0.42097054421901703,
1777
+ "eval_runtime": 55.4157,
1778
+ "eval_samples_per_second": 1.353,
1779
+ "eval_steps_per_second": 0.036,
1780
+ "step": 2800
1781
+ },
1782
+ {
1783
+ "clip_ratio/high_max": 0.0,
1784
+ "clip_ratio/high_mean": 0.0,
1785
+ "clip_ratio/low_mean": 0.0,
1786
+ "clip_ratio/low_min": 0.0,
1787
+ "clip_ratio/region_mean": 0.0,
1788
+ "completions/clipped_ratio": 0.0,
1789
+ "completions/max_length": 472.5,
1790
+ "completions/max_terminated_length": 472.5,
1791
+ "completions/mean_length": 443.96875,
1792
+ "completions/mean_terminated_length": 443.96875,
1793
+ "completions/min_length": 406.25,
1794
+ "completions/min_terminated_length": 406.25,
1795
+ "entropy": 0.0675924657844007,
1796
+ "epoch": 2.3545150501672243,
1797
+ "frac_reward_zero_std": 1.0,
1798
+ "grad_norm": 0.1234281063079834,
1799
+ "kl": 0.009361097356304526,
1800
+ "learning_rate": 4.308807134894091e-07,
1801
+ "loss": 0.0064,
1802
+ "num_tokens": 14966392.0,
1803
+ "reward": 0.75,
1804
+ "reward_std": 0.0,
1805
+ "rewards/accuracy_reward/mean": 0.75,
1806
+ "rewards/accuracy_reward/std": 0.2540002465248108,
1807
+ "step": 2816
1808
+ },
1809
+ {
1810
+ "clip_ratio/high_max": 0.0002307154645677656,
1811
+ "clip_ratio/high_mean": 0.00012432216226443416,
1812
+ "clip_ratio/low_mean": 2.3828054509067442e-05,
1813
+ "clip_ratio/low_min": 0.0,
1814
+ "clip_ratio/region_mean": 0.00014815021540925954,
1815
+ "completions/clipped_ratio": 0.0,
1816
+ "completions/max_length": 468.375,
1817
+ "completions/max_terminated_length": 468.375,
1818
+ "completions/mean_length": 402.505859375,
1819
+ "completions/mean_terminated_length": 402.505859375,
1820
+ "completions/min_length": 342.0625,
1821
+ "completions/min_terminated_length": 342.0625,
1822
+ "entropy": 0.06295629125088453,
1823
+ "epoch": 2.408026755852843,
1824
+ "frac_reward_zero_std": 0.78125,
1825
+ "grad_norm": 0.45806559920310974,
1826
+ "kl": 0.014739077185367933,
1827
+ "learning_rate": 3.9520624303232996e-07,
1828
+ "loss": 0.001,
1829
+ "num_tokens": 15306555.0,
1830
+ "reward": 0.73828125,
1831
+ "reward_std": 0.07284127548336983,
1832
+ "rewards/accuracy_reward/mean": 0.73828125,
1833
+ "rewards/accuracy_reward/std": 0.23197946790605783,
1834
+ "step": 2880
1835
+ },
1836
+ {
1837
+ "epoch": 2.4247491638795986,
1838
+ "eval_completions/clipped_ratio": 0.0,
1839
+ "eval_completions/max_length": 488.5,
1840
+ "eval_completions/max_terminated_length": 488.5,
1841
+ "eval_completions/mean_length": 400.0234375,
1842
+ "eval_completions/mean_terminated_length": 400.0234375,
1843
+ "eval_completions/min_length": 252.5,
1844
+ "eval_completions/min_terminated_length": 252.5,
1845
+ "eval_frac_reward_zero_std": 0.0,
1846
+ "eval_loss": 0.0,
1847
+ "eval_num_tokens": 15416775.0,
1848
+ "eval_reward": 0.734375,
1849
+ "eval_reward_std": 0.0,
1850
+ "eval_rewards/accuracy_reward/mean": 0.734375,
1851
+ "eval_rewards/accuracy_reward/std": 0.4419216513633728,
1852
+ "eval_runtime": 54.9601,
1853
+ "eval_samples_per_second": 1.365,
1854
+ "eval_steps_per_second": 0.036,
1855
+ "step": 2900
1856
+ },
1857
+ {
1858
+ "clip_ratio/high_max": 0.00010461304654282603,
1859
+ "clip_ratio/high_mean": 5.230652327141301e-05,
1860
+ "clip_ratio/low_mean": 2.0310278738510203e-05,
1861
+ "clip_ratio/low_min": 0.0,
1862
+ "clip_ratio/region_mean": 7.261680200992322e-05,
1863
+ "completions/clipped_ratio": 0.0,
1864
+ "completions/max_length": 468.1818181818182,
1865
+ "completions/max_terminated_length": 468.1818181818182,
1866
+ "completions/mean_length": 392.5681818181818,
1867
+ "completions/mean_terminated_length": 392.5681818181818,
1868
+ "completions/min_length": 327.3636363636364,
1869
+ "completions/min_terminated_length": 327.3636363636364,
1870
+ "entropy": 0.06535574548285115,
1871
+ "epoch": 2.4615384615384617,
1872
+ "frac_reward_zero_std": 0.8181818181818182,
1873
+ "grad_norm": 3.3948781490325928,
1874
+ "kl": 0.017127034026336994,
1875
+ "learning_rate": 3.595317725752508e-07,
1876
+ "loss": -0.0017,
1877
+ "num_tokens": 15647103.0,
1878
+ "reward": 0.7784090909090909,
1879
+ "reward_std": 0.05937389893965288,
1880
+ "rewards/accuracy_reward/mean": 0.7784090909090909,
1881
+ "rewards/accuracy_reward/std": 0.2568319250236858,
1882
+ "step": 2944
1883
+ },
1884
+ {
1885
+ "epoch": 2.508361204013378,
1886
+ "eval_completions/clipped_ratio": 0.0,
1887
+ "eval_completions/max_length": 477.5,
1888
+ "eval_completions/max_terminated_length": 477.5,
1889
+ "eval_completions/mean_length": 404.0078125,
1890
+ "eval_completions/mean_terminated_length": 404.0078125,
1891
+ "eval_completions/min_length": 251.0,
1892
+ "eval_completions/min_terminated_length": 251.0,
1893
+ "eval_frac_reward_zero_std": 0.0,
1894
+ "eval_loss": 0.0,
1895
+ "eval_num_tokens": 15947417.0,
1896
+ "eval_reward": 0.7578125,
1897
+ "eval_reward_std": 0.0,
1898
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
1899
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
1900
+ "eval_runtime": 53.5131,
1901
+ "eval_samples_per_second": 1.402,
1902
+ "eval_steps_per_second": 0.037,
1903
+ "step": 3000
1904
+ },
1905
+ {
1906
+ "clip_ratio/high_max": 0.0,
1907
+ "clip_ratio/high_mean": 0.0,
1908
+ "clip_ratio/low_mean": 0.0,
1909
+ "clip_ratio/low_min": 0.0,
1910
+ "clip_ratio/region_mean": 0.0,
1911
+ "completions/clipped_ratio": 0.0,
1912
+ "completions/max_length": 502.0,
1913
+ "completions/max_terminated_length": 502.0,
1914
+ "completions/mean_length": 403.9375,
1915
+ "completions/mean_terminated_length": 403.9375,
1916
+ "completions/min_length": 337.0,
1917
+ "completions/min_terminated_length": 337.0,
1918
+ "entropy": 0.07481803093105555,
1919
+ "epoch": 2.5150501672240804,
1920
+ "frac_reward_zero_std": 1.0,
1921
+ "grad_norm": 0.20490163564682007,
1922
+ "kl": 0.015758564579300582,
1923
+ "learning_rate": 3.238573021181717e-07,
1924
+ "loss": 0.0052,
1925
+ "num_tokens": 15990069.0,
1926
+ "reward": 1.0,
1927
+ "reward_std": 0.0,
1928
+ "rewards/accuracy_reward/mean": 1.0,
1929
+ "rewards/accuracy_reward/std": 0.0,
1930
+ "step": 3008
1931
+ },
1932
+ {
1933
+ "clip_ratio/high_max": 7.721042129560374e-05,
1934
+ "clip_ratio/high_mean": 3.860521064780187e-05,
1935
+ "clip_ratio/low_mean": 2.3580541892442852e-05,
1936
+ "clip_ratio/low_min": 0.0,
1937
+ "clip_ratio/region_mean": 6.218575299499207e-05,
1938
+ "completions/clipped_ratio": 0.0,
1939
+ "completions/max_length": 456.375,
1940
+ "completions/max_terminated_length": 456.375,
1941
+ "completions/mean_length": 397.22265625,
1942
+ "completions/mean_terminated_length": 397.22265625,
1943
+ "completions/min_length": 335.5,
1944
+ "completions/min_terminated_length": 335.5,
1945
+ "entropy": 0.06511772476369515,
1946
+ "epoch": 2.568561872909699,
1947
+ "frac_reward_zero_std": 0.90625,
1948
+ "grad_norm": 0.22943955659866333,
1949
+ "kl": 0.011846352602333354,
1950
+ "learning_rate": 2.881828316610925e-07,
1951
+ "loss": -0.0014,
1952
+ "num_tokens": 16327623.0,
1953
+ "reward": 0.677734375,
1954
+ "reward_std": 0.04015435464680195,
1955
+ "rewards/accuracy_reward/mean": 0.677734375,
1956
+ "rewards/accuracy_reward/std": 0.2386060394346714,
1957
+ "step": 3072
1958
+ },
1959
+ {
1960
+ "epoch": 2.591973244147157,
1961
+ "eval_completions/clipped_ratio": 0.0,
1962
+ "eval_completions/max_length": 510.5,
1963
+ "eval_completions/max_terminated_length": 510.5,
1964
+ "eval_completions/mean_length": 404.40625,
1965
+ "eval_completions/mean_terminated_length": 404.40625,
1966
+ "eval_completions/min_length": 255.5,
1967
+ "eval_completions/min_terminated_length": 255.5,
1968
+ "eval_frac_reward_zero_std": 0.0,
1969
+ "eval_loss": 0.0,
1970
+ "eval_num_tokens": 16482544.0,
1971
+ "eval_reward": 0.7734375,
1972
+ "eval_reward_std": 0.0,
1973
+ "eval_rewards/accuracy_reward/mean": 0.7734375,
1974
+ "eval_rewards/accuracy_reward/std": 0.4218114912509918,
1975
+ "eval_runtime": 57.7157,
1976
+ "eval_samples_per_second": 1.299,
1977
+ "eval_steps_per_second": 0.035,
1978
+ "step": 3100
1979
+ },
1980
+ {
1981
+ "clip_ratio/high_max": 0.00018287577970315598,
1982
+ "clip_ratio/high_mean": 9.143788985157799e-05,
1983
+ "clip_ratio/low_mean": 6.674907328690299e-05,
1984
+ "clip_ratio/low_min": 0.0,
1985
+ "clip_ratio/region_mean": 0.0001581869639469207,
1986
+ "completions/clipped_ratio": 0.0,
1987
+ "completions/max_length": 477.77777777777777,
1988
+ "completions/max_terminated_length": 477.77777777777777,
1989
+ "completions/mean_length": 413.81597222222223,
1990
+ "completions/mean_terminated_length": 413.81597222222223,
1991
+ "completions/min_length": 366.55555555555554,
1992
+ "completions/min_terminated_length": 366.55555555555554,
1993
+ "entropy": 0.06471226343678103,
1994
+ "epoch": 2.6220735785953178,
1995
+ "frac_reward_zero_std": 0.6666666666666666,
1996
+ "grad_norm": 0.1000482365489006,
1997
+ "kl": 0.020034029368414648,
1998
+ "learning_rate": 2.5250836120401335e-07,
1999
+ "loss": -0.0036,
2000
+ "num_tokens": 16677259.0,
2001
+ "reward": 0.7881944444444444,
2002
+ "reward_std": 0.13741063740518358,
2003
+ "rewards/accuracy_reward/mean": 0.7881944444444444,
2004
+ "rewards/accuracy_reward/std": 0.2844773613744312,
2005
+ "step": 3136
2006
+ },
2007
+ {
2008
+ "clip_ratio/high_max": 0.00013618932462122757,
2009
+ "clip_ratio/high_mean": 6.809466231061378e-05,
2010
+ "clip_ratio/low_mean": 2.2072627871239092e-05,
2011
+ "clip_ratio/low_min": 0.0,
2012
+ "clip_ratio/region_mean": 9.016729109134758e-05,
2013
+ "completions/clipped_ratio": 0.0,
2014
+ "completions/max_length": 468.25,
2015
+ "completions/max_terminated_length": 468.25,
2016
+ "completions/mean_length": 407.998046875,
2017
+ "completions/mean_terminated_length": 407.998046875,
2018
+ "completions/min_length": 342.375,
2019
+ "completions/min_terminated_length": 342.375,
2020
+ "entropy": 0.06463818287011236,
2021
+ "epoch": 2.6755852842809364,
2022
+ "frac_reward_zero_std": 0.875,
2023
+ "grad_norm": 0.7343602776527405,
2024
+ "kl": 0.013379381600543638,
2025
+ "learning_rate": 2.1683389074693424e-07,
2026
+ "loss": 0.0021,
2027
+ "num_tokens": 17020314.0,
2028
+ "reward": 0.74609375,
2029
+ "reward_std": 0.04175759106874466,
2030
+ "rewards/accuracy_reward/mean": 0.74609375,
2031
+ "rewards/accuracy_reward/std": 0.15711678750813007,
2032
+ "step": 3200
2033
+ },
2034
+ {
2035
+ "epoch": 2.6755852842809364,
2036
+ "eval_completions/clipped_ratio": 0.0,
2037
+ "eval_completions/max_length": 491.0,
2038
+ "eval_completions/max_terminated_length": 491.0,
2039
+ "eval_completions/mean_length": 400.1171875,
2040
+ "eval_completions/mean_terminated_length": 400.1171875,
2041
+ "eval_completions/min_length": 256.0,
2042
+ "eval_completions/min_terminated_length": 256.0,
2043
+ "eval_frac_reward_zero_std": 0.0,
2044
+ "eval_loss": 0.0,
2045
+ "eval_num_tokens": 17020314.0,
2046
+ "eval_reward": 0.71875,
2047
+ "eval_reward_std": 0.0,
2048
+ "eval_rewards/accuracy_reward/mean": 0.71875,
2049
+ "eval_rewards/accuracy_reward/std": 0.4528248459100723,
2050
+ "eval_runtime": 55.4159,
2051
+ "eval_samples_per_second": 1.353,
2052
+ "eval_steps_per_second": 0.036,
2053
+ "step": 3200
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.00017083798411476891,
2057
+ "clip_ratio/high_mean": 8.541899205738446e-05,
2058
+ "clip_ratio/low_mean": 4.446499588084407e-06,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 8.986549164546886e-05,
2061
+ "completions/clipped_ratio": 0.0,
2062
+ "completions/max_length": 469.6875,
2063
+ "completions/max_terminated_length": 469.6875,
2064
+ "completions/mean_length": 410.0,
2065
+ "completions/mean_terminated_length": 410.0,
2066
+ "completions/min_length": 347.5625,
2067
+ "completions/min_terminated_length": 347.5625,
2068
+ "entropy": 0.0666909699793905,
2069
+ "epoch": 2.729096989966555,
2070
+ "frac_reward_zero_std": 0.875,
2071
+ "grad_norm": 0.291899174451828,
2072
+ "kl": 0.01309701916488848,
2073
+ "learning_rate": 1.8115942028985507e-07,
2074
+ "loss": 0.004,
2075
+ "num_tokens": 17364426.0,
2076
+ "reward": 0.939453125,
2077
+ "reward_std": 0.05133409798145294,
2078
+ "rewards/accuracy_reward/mean": 0.939453125,
2079
+ "rewards/accuracy_reward/std": 0.10974337719380856,
2080
+ "step": 3264
2081
+ },
2082
+ {
2083
+ "epoch": 2.759197324414716,
2084
+ "eval_completions/clipped_ratio": 0.0,
2085
+ "eval_completions/max_length": 497.5,
2086
+ "eval_completions/max_terminated_length": 497.5,
2087
+ "eval_completions/mean_length": 402.3671875,
2088
+ "eval_completions/mean_terminated_length": 402.3671875,
2089
+ "eval_completions/min_length": 249.5,
2090
+ "eval_completions/min_terminated_length": 249.5,
2091
+ "eval_frac_reward_zero_std": 0.0,
2092
+ "eval_loss": 0.0,
2093
+ "eval_num_tokens": 17554649.0,
2094
+ "eval_reward": 0.7578125,
2095
+ "eval_reward_std": 0.0,
2096
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
2097
+ "eval_rewards/accuracy_reward/std": 0.4316960424184799,
2098
+ "eval_runtime": 55.6271,
2099
+ "eval_samples_per_second": 1.348,
2100
+ "eval_steps_per_second": 0.036,
2101
+ "step": 3300
2102
+ },
2103
+ {
2104
+ "clip_ratio/high_max": 8.781038715304541e-05,
2105
+ "clip_ratio/high_mean": 4.3905193576522705e-05,
2106
+ "clip_ratio/low_mean": 0.0,
2107
+ "clip_ratio/low_min": 0.0,
2108
+ "clip_ratio/region_mean": 4.3905193576522705e-05,
2109
+ "completions/clipped_ratio": 0.0,
2110
+ "completions/max_length": 477.0,
2111
+ "completions/max_terminated_length": 477.0,
2112
+ "completions/mean_length": 399.60714285714283,
2113
+ "completions/mean_terminated_length": 399.60714285714283,
2114
+ "completions/min_length": 324.42857142857144,
2115
+ "completions/min_terminated_length": 324.42857142857144,
2116
+ "entropy": 0.06731948908418417,
2117
+ "epoch": 2.782608695652174,
2118
+ "frac_reward_zero_std": 0.9285714285714286,
2119
+ "grad_norm": 0.080546535551548,
2120
+ "kl": 0.010967594387433824,
2121
+ "learning_rate": 1.454849498327759e-07,
2122
+ "loss": 0.0032,
2123
+ "num_tokens": 17702785.0,
2124
+ "reward": 0.78125,
2125
+ "reward_std": 0.017857142857142856,
2126
+ "rewards/accuracy_reward/mean": 0.78125,
2127
+ "rewards/accuracy_reward/std": 0.24296831020287105,
2128
+ "step": 3328
2129
+ },
2130
+ {
2131
+ "clip_ratio/high_max": 1.8371545593254268e-05,
2132
+ "clip_ratio/high_mean": 9.185772796627134e-06,
2133
+ "clip_ratio/low_mean": 0.0,
2134
+ "clip_ratio/low_min": 0.0,
2135
+ "clip_ratio/region_mean": 9.185772796627134e-06,
2136
+ "completions/clipped_ratio": 0.0,
2137
+ "completions/max_length": 476.4375,
2138
+ "completions/max_terminated_length": 476.4375,
2139
+ "completions/mean_length": 416.22265625,
2140
+ "completions/mean_terminated_length": 416.22265625,
2141
+ "completions/min_length": 358.375,
2142
+ "completions/min_terminated_length": 358.375,
2143
+ "entropy": 0.06470249680569395,
2144
+ "epoch": 2.8361204013377925,
2145
+ "frac_reward_zero_std": 0.96875,
2146
+ "grad_norm": 0.08876697719097137,
2147
+ "kl": 0.012055652209639334,
2148
+ "learning_rate": 1.0981047937569676e-07,
2149
+ "loss": 0.0017,
2150
+ "num_tokens": 18050083.0,
2151
+ "reward": 0.8515625,
2152
+ "reward_std": 0.013975424692034721,
2153
+ "rewards/accuracy_reward/mean": 0.8515625,
2154
+ "rewards/accuracy_reward/std": 0.11625076457858086,
2155
+ "step": 3392
2156
+ },
2157
+ {
2158
+ "epoch": 2.842809364548495,
2159
+ "eval_completions/clipped_ratio": 0.0,
2160
+ "eval_completions/max_length": 492.0,
2161
+ "eval_completions/max_terminated_length": 492.0,
2162
+ "eval_completions/mean_length": 399.5234375,
2163
+ "eval_completions/mean_terminated_length": 399.5234375,
2164
+ "eval_completions/min_length": 251.0,
2165
+ "eval_completions/min_terminated_length": 251.0,
2166
+ "eval_frac_reward_zero_std": 0.0,
2167
+ "eval_loss": 0.0,
2168
+ "eval_num_tokens": 18091884.0,
2169
+ "eval_reward": 0.7578125,
2170
+ "eval_reward_std": 0.0,
2171
+ "eval_rewards/accuracy_reward/mean": 0.7578125,
2172
+ "eval_rewards/accuracy_reward/std": 0.4309118092060089,
2173
+ "eval_runtime": 55.1048,
2174
+ "eval_samples_per_second": 1.361,
2175
+ "eval_steps_per_second": 0.036,
2176
+ "step": 3400
2177
+ },
2178
+ {
2179
+ "clip_ratio/high_max": 0.00021469751144260435,
2180
+ "clip_ratio/high_mean": 0.00010734875572130218,
2181
+ "clip_ratio/low_mean": 2.356954012481895e-05,
2182
+ "clip_ratio/low_min": 0.0,
2183
+ "clip_ratio/region_mean": 0.00013091829532640986,
2184
+ "completions/clipped_ratio": 0.0,
2185
+ "completions/max_length": 439.2857142857143,
2186
+ "completions/max_terminated_length": 439.2857142857143,
2187
+ "completions/mean_length": 381.5424107142857,
2188
+ "completions/mean_terminated_length": 381.5424107142857,
2189
+ "completions/min_length": 314.07142857142856,
2190
+ "completions/min_terminated_length": 314.07142857142856,
2191
+ "entropy": 0.05965407398928489,
2192
+ "epoch": 2.8896321070234112,
2193
+ "frac_reward_zero_std": 0.7857142857142857,
2194
+ "grad_norm": 0.1913478970527649,
2195
+ "kl": 0.014099701430365843,
2196
+ "learning_rate": 7.41360089186176e-08,
2197
+ "loss": 0.0007,
2198
+ "num_tokens": 18380207.0,
2199
+ "reward": 0.8415178571428571,
2200
+ "reward_std": 0.07720869992460523,
2201
+ "rewards/accuracy_reward/mean": 0.8415178571428571,
2202
+ "rewards/accuracy_reward/std": 0.23203862139156886,
2203
+ "step": 3456
2204
+ },
2205
+ {
2206
+ "epoch": 2.9264214046822743,
2207
+ "eval_completions/clipped_ratio": 0.0,
2208
+ "eval_completions/max_length": 565.0,
2209
+ "eval_completions/max_terminated_length": 565.0,
2210
+ "eval_completions/mean_length": 401.8046875,
2211
+ "eval_completions/mean_terminated_length": 401.8046875,
2212
+ "eval_completions/min_length": 248.5,
2213
+ "eval_completions/min_terminated_length": 248.5,
2214
+ "eval_frac_reward_zero_std": 0.0,
2215
+ "eval_loss": 0.0,
2216
+ "eval_num_tokens": 18618071.0,
2217
+ "eval_reward": 0.734375,
2218
+ "eval_reward_std": 0.0,
2219
+ "eval_rewards/accuracy_reward/mean": 0.734375,
2220
+ "eval_rewards/accuracy_reward/std": 0.44479964673519135,
2221
+ "eval_runtime": 62.8493,
2222
+ "eval_samples_per_second": 1.193,
2223
+ "eval_steps_per_second": 0.032,
2224
+ "step": 3500
2225
+ },
2226
+ {
2227
+ "clip_ratio/high_max": 0.0002402667363639921,
2228
+ "clip_ratio/high_mean": 0.00012013336818199605,
2229
+ "clip_ratio/low_mean": 1.6688919276930392e-05,
2230
+ "clip_ratio/low_min": 0.0,
2231
+ "clip_ratio/region_mean": 0.00013682228745892644,
2232
+ "completions/clipped_ratio": 0.0,
2233
+ "completions/max_length": 415.0,
2234
+ "completions/max_terminated_length": 415.0,
2235
+ "completions/mean_length": 341.05625,
2236
+ "completions/mean_terminated_length": 341.05625,
2237
+ "completions/min_length": 284.4,
2238
+ "completions/min_terminated_length": 284.4,
2239
+ "entropy": 0.05576919969171286,
2240
+ "epoch": 2.94314381270903,
2241
+ "frac_reward_zero_std": 0.8,
2242
+ "grad_norm": 0.03020313009619713,
2243
+ "kl": 0.007373337184253615,
2244
+ "learning_rate": 3.846153846153846e-08,
2245
+ "loss": 0.0075,
2246
+ "num_tokens": 18714528.0,
2247
+ "reward": 0.8875,
2248
+ "reward_std": 0.05,
2249
+ "rewards/accuracy_reward/mean": 0.8875,
2250
+ "rewards/accuracy_reward/std": 0.17231077551841736,
2251
+ "step": 3520
2252
+ },
2253
+ {
2254
+ "clip_ratio/high_max": 4.61889658254222e-05,
2255
+ "clip_ratio/high_mean": 2.30944829127111e-05,
2256
+ "clip_ratio/low_mean": 1.4195853509590961e-05,
2257
+ "clip_ratio/low_min": 0.0,
2258
+ "clip_ratio/region_mean": 3.729033642230206e-05,
2259
+ "completions/clipped_ratio": 0.0,
2260
+ "completions/max_length": 477.3125,
2261
+ "completions/max_terminated_length": 477.3125,
2262
+ "completions/mean_length": 402.041015625,
2263
+ "completions/mean_terminated_length": 402.041015625,
2264
+ "completions/min_length": 324.8125,
2265
+ "completions/min_terminated_length": 324.8125,
2266
+ "entropy": 0.06624187622219324,
2267
+ "epoch": 2.9966555183946486,
2268
+ "frac_reward_zero_std": 0.90625,
2269
+ "grad_norm": 0.13859905302524567,
2270
+ "kl": 0.010573866235063178,
2271
+ "learning_rate": 2.787068004459309e-09,
2272
+ "loss": 0.0006,
2273
+ "num_tokens": 19054453.0,
2274
+ "reward": 0.80078125,
2275
+ "reward_std": 0.04357584938406944,
2276
+ "rewards/accuracy_reward/mean": 0.80078125,
2277
+ "rewards/accuracy_reward/std": 0.23527752235531807,
2278
+ "step": 3584
2279
+ },
2280
+ {
2281
+ "clip_ratio/high_max": 0.0,
2282
+ "clip_ratio/high_mean": 0.0,
2283
+ "clip_ratio/low_mean": 0.0,
2284
+ "clip_ratio/low_min": 0.0,
2285
+ "clip_ratio/region_mean": 0.0,
2286
+ "completions/clipped_ratio": 0.0,
2287
+ "completions/max_length": 474.0,
2288
+ "completions/max_terminated_length": 474.0,
2289
+ "completions/mean_length": 436.0,
2290
+ "completions/mean_terminated_length": 436.0,
2291
+ "completions/min_length": 404.0,
2292
+ "completions/min_terminated_length": 404.0,
2293
+ "entropy": 0.057431732304394245,
2294
+ "epoch": 3.0,
2295
+ "frac_reward_zero_std": 1.0,
2296
+ "kl": 0.0017359714256599545,
2297
+ "num_tokens": 19076757.0,
2298
+ "reward": 0.5,
2299
+ "reward_std": 0.0,
2300
+ "rewards/accuracy_reward/mean": 0.5,
2301
+ "rewards/accuracy_reward/std": 0.5080004930496216,
2302
+ "step": 3588,
2303
+ "total_flos": 0.0,
2304
+ "train_loss": 0.0036438228976481916,
2305
+ "train_runtime": 33932.84,
2306
+ "train_samples_per_second": 0.053,
2307
+ "train_steps_per_second": 0.106
2308
+ }
2309
+ ],
2310
+ "logging_steps": 64,
2311
+ "max_steps": 3588,
2312
+ "num_input_tokens_seen": 19076757,
2313
+ "num_train_epochs": 3,
2314
+ "save_steps": 100,
2315
+ "stateful_callbacks": {
2316
+ "TrainerControl": {
2317
+ "args": {
2318
+ "should_epoch_stop": false,
2319
+ "should_evaluate": false,
2320
+ "should_log": false,
2321
+ "should_save": true,
2322
+ "should_training_stop": true
2323
+ },
2324
+ "attributes": {}
2325
+ }
2326
+ },
2327
+ "total_flos": 0.0,
2328
+ "train_batch_size": 4,
2329
+ "trial_name": null,
2330
+ "trial_params": null
2331
+ }
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60566d17cc6a771231c47b0222598f237d07dd9ccb7ebb24570e10a1ce48df26
3
+ size 8440
Blood/seed_4/GRPO/LLM-Qwen-2.5-3B-GRPO-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/tokenizer_config.json ADDED
@@ -0,0 +1,208 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ }
181
+ },
182
+ "additional_special_tokens": [
183
+ "<|im_start|>",
184
+ "<|im_end|>",
185
+ "<|object_ref_start|>",
186
+ "<|object_ref_end|>",
187
+ "<|box_start|>",
188
+ "<|box_end|>",
189
+ "<|quad_start|>",
190
+ "<|quad_end|>",
191
+ "<|vision_start|>",
192
+ "<|vision_end|>",
193
+ "<|vision_pad|>",
194
+ "<|image_pad|>",
195
+ "<|video_pad|>"
196
+ ],
197
+ "bos_token": null,
198
+ "clean_up_tokenization_spaces": false,
199
+ "eos_token": "<|im_end|>",
200
+ "errors": "replace",
201
+ "extra_special_tokens": {},
202
+ "model_max_length": 2048,
203
+ "pad_token": "<|endoftext|>",
204
+ "padding_side": "right",
205
+ "split_special_tokens": false,
206
+ "tokenizer_class": "Qwen2Tokenizer",
207
+ "unk_token": null
208
+ }
Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6905fd3c472196526867cc09e4f992af48ddcf0f4c1a1e397648834a2b94e7be
3
+ size 7224
Blood/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1f8748df38c7b102c499ccbb8a0ac775ca63c4f57906979ab16bddcc98a60bd
3
+ size 7224
Blood/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Blood-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/tokenizer_config.json ADDED
@@ -0,0 +1,208 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ }
181
+ },
182
+ "additional_special_tokens": [
183
+ "<|im_start|>",
184
+ "<|im_end|>",
185
+ "<|object_ref_start|>",
186
+ "<|object_ref_end|>",
187
+ "<|box_start|>",
188
+ "<|box_end|>",
189
+ "<|quad_start|>",
190
+ "<|quad_end|>",
191
+ "<|vision_start|>",
192
+ "<|vision_end|>",
193
+ "<|vision_pad|>",
194
+ "<|image_pad|>",
195
+ "<|video_pad|>"
196
+ ],
197
+ "bos_token": null,
198
+ "clean_up_tokenization_spaces": false,
199
+ "eos_token": "<|im_end|>",
200
+ "errors": "replace",
201
+ "extra_special_tokens": {},
202
+ "model_max_length": 2048,
203
+ "pad_token": "<|endoftext|>",
204
+ "padding_side": "right",
205
+ "split_special_tokens": false,
206
+ "tokenizer_class": "Qwen2Tokenizer",
207
+ "unk_token": null
208
+ }
Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:131da205e9a7b27af61c7dc65bcdfc32dca1c21f1fd15d85e80f5be9890ca4b6
3
+ size 7224
Creditg/seed_0/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151645,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 2048,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 11008,
12
+ "layer_types": [
13
+ "full_attention",
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention"
49
+ ],
50
+ "max_position_embeddings": 32768,
51
+ "max_window_layers": 70,
52
+ "model_type": "qwen2",
53
+ "num_attention_heads": 16,
54
+ "num_hidden_layers": 36,
55
+ "num_key_value_heads": 2,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": true,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.55.2",
63
+ "use_cache": true,
64
+ "use_sliding_window": false,
65
+ "vocab_size": 151936
66
+ }
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 151645,
6
+ 151643
7
+ ],
8
+ "pad_token_id": 151643,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.7,
11
+ "top_k": 20,
12
+ "top_p": 0.8,
13
+ "transformers_version": "4.55.2"
14
+ }
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/model.safetensors.index.json ADDED
@@ -0,0 +1,443 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 3085938688,
4
+ "total_size": 6794207232
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00002-of-00002.safetensors",
8
+ "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
9
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
10
+ "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
11
+ "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
12
+ "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
13
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
14
+ "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
15
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
16
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
17
+ "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
18
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
19
+ "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
20
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
21
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
22
+ "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
23
+ "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
24
+ "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
25
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
26
+ "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
27
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
28
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
29
+ "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
30
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
31
+ "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
32
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
33
+ "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
34
+ "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
35
+ "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
36
+ "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
37
+ "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
38
+ "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
39
+ "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
40
+ "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
41
+ "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
42
+ "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
43
+ "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
44
+ "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
45
+ "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
46
+ "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
47
+ "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
48
+ "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
49
+ "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
50
+ "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
51
+ "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
52
+ "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
53
+ "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
54
+ "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
55
+ "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
56
+ "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
57
+ "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
58
+ "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
59
+ "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
60
+ "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
61
+ "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
62
+ "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
63
+ "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
64
+ "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
65
+ "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
66
+ "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
67
+ "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
68
+ "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
69
+ "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
70
+ "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
71
+ "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
72
+ "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
73
+ "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
74
+ "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
75
+ "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
76
+ "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
77
+ "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
78
+ "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
79
+ "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
80
+ "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
81
+ "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
82
+ "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
83
+ "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
84
+ "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
85
+ "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
86
+ "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
87
+ "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
88
+ "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
89
+ "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
90
+ "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
91
+ "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
92
+ "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
93
+ "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
94
+ "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
95
+ "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
96
+ "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
97
+ "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
98
+ "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
99
+ "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
100
+ "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
101
+ "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
102
+ "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
103
+ "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
104
+ "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
105
+ "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
106
+ "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
107
+ "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
108
+ "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
109
+ "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
110
+ "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
111
+ "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
112
+ "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
113
+ "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
114
+ "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
115
+ "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
116
+ "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
117
+ "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
118
+ "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
119
+ "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
120
+ "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
121
+ "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
122
+ "model.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
123
+ "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
124
+ "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
125
+ "model.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
126
+ "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
127
+ "model.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
128
+ "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
129
+ "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
130
+ "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
131
+ "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
132
+ "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
133
+ "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
134
+ "model.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
135
+ "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
136
+ "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
137
+ "model.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
138
+ "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
139
+ "model.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
140
+ "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
141
+ "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
142
+ "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
143
+ "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
144
+ "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
145
+ "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
146
+ "model.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
147
+ "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
148
+ "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
149
+ "model.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
150
+ "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
151
+ "model.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
152
+ "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
153
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
154
+ "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
155
+ "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
156
+ "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
157
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
158
+ "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
159
+ "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
160
+ "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
161
+ "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
162
+ "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
163
+ "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
164
+ "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
165
+ "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
166
+ "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
167
+ "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
168
+ "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
169
+ "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
170
+ "model.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
171
+ "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
172
+ "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
173
+ "model.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
174
+ "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
175
+ "model.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
176
+ "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
177
+ "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
178
+ "model.layers.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
179
+ "model.layers.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
180
+ "model.layers.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
181
+ "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
182
+ "model.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
183
+ "model.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
184
+ "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
185
+ "model.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
186
+ "model.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
187
+ "model.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
188
+ "model.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
189
+ "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
190
+ "model.layers.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
191
+ "model.layers.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
192
+ "model.layers.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
193
+ "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
194
+ "model.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
195
+ "model.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
196
+ "model.layers.22.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
197
+ "model.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
198
+ "model.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
199
+ "model.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
200
+ "model.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
201
+ "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
202
+ "model.layers.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
203
+ "model.layers.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
204
+ "model.layers.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
205
+ "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
206
+ "model.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
207
+ "model.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
208
+ "model.layers.23.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
209
+ "model.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
210
+ "model.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
211
+ "model.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
212
+ "model.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
213
+ "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
214
+ "model.layers.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
215
+ "model.layers.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
216
+ "model.layers.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
217
+ "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
218
+ "model.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
219
+ "model.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
220
+ "model.layers.24.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
221
+ "model.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
222
+ "model.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
223
+ "model.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
224
+ "model.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
225
+ "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
226
+ "model.layers.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
227
+ "model.layers.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
228
+ "model.layers.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
229
+ "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
230
+ "model.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
231
+ "model.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
232
+ "model.layers.25.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
233
+ "model.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
234
+ "model.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
235
+ "model.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
236
+ "model.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
237
+ "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
238
+ "model.layers.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
239
+ "model.layers.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
240
+ "model.layers.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
241
+ "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
242
+ "model.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
243
+ "model.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
244
+ "model.layers.26.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
245
+ "model.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
246
+ "model.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
247
+ "model.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
248
+ "model.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
249
+ "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
250
+ "model.layers.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
251
+ "model.layers.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
252
+ "model.layers.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
253
+ "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
254
+ "model.layers.27.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
255
+ "model.layers.27.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
256
+ "model.layers.27.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
257
+ "model.layers.27.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
258
+ "model.layers.27.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
259
+ "model.layers.27.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
260
+ "model.layers.27.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
261
+ "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
262
+ "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
263
+ "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
264
+ "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
265
+ "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
266
+ "model.layers.28.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
267
+ "model.layers.28.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
268
+ "model.layers.28.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
269
+ "model.layers.28.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
270
+ "model.layers.28.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
271
+ "model.layers.28.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
272
+ "model.layers.28.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
273
+ "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
274
+ "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
275
+ "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
276
+ "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
277
+ "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
278
+ "model.layers.29.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
279
+ "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
280
+ "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
281
+ "model.layers.29.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
282
+ "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
283
+ "model.layers.29.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
284
+ "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
285
+ "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
286
+ "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
287
+ "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
288
+ "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
289
+ "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
290
+ "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
291
+ "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
292
+ "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
293
+ "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
294
+ "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
295
+ "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
296
+ "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
297
+ "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
298
+ "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
299
+ "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
300
+ "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
301
+ "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
302
+ "model.layers.30.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
303
+ "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
304
+ "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
305
+ "model.layers.30.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
306
+ "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
307
+ "model.layers.30.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
308
+ "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
309
+ "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
310
+ "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
311
+ "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
312
+ "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
313
+ "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
314
+ "model.layers.31.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
315
+ "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
316
+ "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
317
+ "model.layers.31.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
318
+ "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
319
+ "model.layers.31.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
320
+ "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
321
+ "model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
322
+ "model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
323
+ "model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
324
+ "model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
325
+ "model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
326
+ "model.layers.32.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
327
+ "model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
328
+ "model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
329
+ "model.layers.32.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
330
+ "model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
331
+ "model.layers.32.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
332
+ "model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
333
+ "model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
334
+ "model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
335
+ "model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
336
+ "model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
337
+ "model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
338
+ "model.layers.33.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
339
+ "model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
340
+ "model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
341
+ "model.layers.33.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
342
+ "model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
343
+ "model.layers.33.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
344
+ "model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
345
+ "model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
346
+ "model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
347
+ "model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
348
+ "model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
349
+ "model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
350
+ "model.layers.34.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
351
+ "model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
352
+ "model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
353
+ "model.layers.34.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
354
+ "model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
355
+ "model.layers.34.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
356
+ "model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
357
+ "model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
358
+ "model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
359
+ "model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
360
+ "model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
361
+ "model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
362
+ "model.layers.35.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
363
+ "model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
364
+ "model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
365
+ "model.layers.35.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
366
+ "model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
367
+ "model.layers.35.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
368
+ "model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
369
+ "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
370
+ "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
371
+ "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
372
+ "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
373
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
374
+ "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
375
+ "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
376
+ "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
377
+ "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
378
+ "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
379
+ "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
380
+ "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
381
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
382
+ "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
383
+ "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
384
+ "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
385
+ "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
386
+ "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
387
+ "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
388
+ "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
389
+ "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
390
+ "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
391
+ "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
392
+ "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
393
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
394
+ "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
395
+ "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
396
+ "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
397
+ "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
398
+ "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
399
+ "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
400
+ "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
401
+ "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
402
+ "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
403
+ "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
404
+ "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
405
+ "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
406
+ "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
407
+ "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
408
+ "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
409
+ "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
410
+ "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
411
+ "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
412
+ "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
413
+ "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
414
+ "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
415
+ "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
416
+ "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
417
+ "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
418
+ "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
419
+ "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
420
+ "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
421
+ "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
422
+ "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
423
+ "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
424
+ "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
425
+ "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
426
+ "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
427
+ "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
428
+ "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
429
+ "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
430
+ "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
431
+ "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
432
+ "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
433
+ "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
434
+ "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
435
+ "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
436
+ "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
437
+ "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
438
+ "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
439
+ "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
440
+ "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
441
+ "model.norm.weight": "model-00002-of-00002.safetensors"
442
+ }
443
+ }
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/runs/Dec27_04-03-58_csce-yang-s2.engr.tamu.edu/events.out.tfevents.1766829852.csce-yang-s2.engr.tamu.edu.3772169.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:497304e7d4692743d2b00ef6d187d2d7f19b02d917dba75bc65064a7d85d8b87
3
+ size 230533
Creditg/seed_15/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151645,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 2048,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 11008,
12
+ "layer_types": [
13
+ "full_attention",
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention"
49
+ ],
50
+ "max_position_embeddings": 32768,
51
+ "max_window_layers": 70,
52
+ "model_type": "qwen2",
53
+ "num_attention_heads": 16,
54
+ "num_hidden_layers": 36,
55
+ "num_key_value_heads": 2,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": true,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.55.2",
63
+ "use_cache": true,
64
+ "use_sliding_window": false,
65
+ "vocab_size": 151936
66
+ }
Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/generation_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 151645,
6
+ 151643
7
+ ],
8
+ "pad_token_id": 151643,
9
+ "repetition_penalty": 1.05,
10
+ "temperature": 0.7,
11
+ "top_k": 20,
12
+ "top_p": 0.8,
13
+ "transformers_version": "4.55.2"
14
+ }
Creditg/seed_2/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/runs/Dec26_20-47-41_csce-yang-s2.engr.tamu.edu/events.out.tfevents.1766803674.csce-yang-s2.engr.tamu.edu.3519012.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb4f1b3ad53582e30e446e152bec6698b30a9db255b003b8a7d2a305e5f5bf14
3
+ size 236873
Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
Creditg/seed_4/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151645,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 2048,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 11008,
12
+ "layer_types": [
13
+ "full_attention",
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention"
49
+ ],
50
+ "max_position_embeddings": 32768,
51
+ "max_window_layers": 70,
52
+ "model_type": "qwen2",
53
+ "num_attention_heads": 16,
54
+ "num_hidden_layers": 36,
55
+ "num_key_value_heads": 2,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": true,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.55.2",
63
+ "use_cache": true,
64
+ "use_sliding_window": false,
65
+ "vocab_size": 151936
66
+ }
Creditg/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
Creditg/seed_50/SFT/LLM-Qwen-2.5-3B-SFT-decision-tree-Creditg-serialized/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}