Hugging Face's logo

xinpeng
/
auc-sft-batch

Safetensors
Model card Files
xet
Community
xinpeng commited on
Commit
342ed1b
·
verified ·
1 Parent(s): 471372d

Batch upload: 20250922_054148

Browse files
Files changed (28) hide show
  1. 20250922_054148/experiment_log.txt +16 -0
  2. 20250922_054148/filtered_data/filter_info.json +58 -0
  3. 20250922_054148/filtered_data/filtered_top_10pct.parquet +3 -0
  4. 20250922_054148/filtered_data/filtered_top_10pct_auc.csv +474 -0
  5. 20250922_054148/filtered_data/filtered_top_20pct.parquet +3 -0
  6. 20250922_054148/filtered_data/filtered_top_20pct_auc.csv +421 -0
  7. 20250922_054148/filtered_data/filtered_top_50pct.parquet +3 -0
  8. 20250922_054148/filtered_data/filtered_top_50pct_auc.csv +264 -0
  9. 20250922_054148/filtered_data/full_data.parquet +3 -0
  10. 20250922_054148/filtered_data/full_data_auc.csv +526 -0
  11. 20250922_054148/results_summary.txt +16 -0
  12. 20250922_054148/run_analysis.sh +44 -0
  13. 20250922_054148/sft_results/filtered_top_10pct/gpu_lock_wrapper.sh +31 -0
  14. 20250922_054148/sft_results/filtered_top_10pct/run_sft_script.sh +40 -0
  15. 20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet +3 -0
  16. 20250922_054148/sft_results/filtered_top_10pct/training_log.txt +50 -0
  17. 20250922_054148/sft_results/filtered_top_20pct/gpu_lock_wrapper.sh +31 -0
  18. 20250922_054148/sft_results/filtered_top_20pct/run_sft_script.sh +40 -0
  19. 20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet +3 -0
  20. 20250922_054148/sft_results/filtered_top_20pct/training_log.txt +50 -0
  21. 20250922_054148/sft_results/filtered_top_50pct/gpu_lock_wrapper.sh +31 -0
  22. 20250922_054148/sft_results/filtered_top_50pct/run_sft_script.sh +40 -0
  23. 20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet +3 -0
  24. 20250922_054148/sft_results/filtered_top_50pct/training_log.txt +50 -0
  25. 20250922_054148/sft_results/full_data/gpu_lock_wrapper.sh +31 -0
  26. 20250922_054148/sft_results/full_data/run_sft_script.sh +40 -0
  27. 20250922_054148/sft_results/full_data/sft_training_data.parquet +3 -0
  28. 20250922_054148/sft_results/full_data/training_log.txt +50 -0
20250922_054148/experiment_log.txt ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Parallel AUC-Filtered SFT Experiments - 2025-09-22 05:41:49.500540
2
+ Configuration:
3
+ Model: Qwen/Qwen2.5-1.5B
4
+ Filter Percentages: [0, 10, 20, 50]
5
+ SFT Epochs: 2
6
+ SFT Batch Size: 4
7
+ Filtered Files: 4
8
+
9
+ Filter: filtered_top_20pct - Started at Mon Sep 22 05:41:49 UTC 2025
10
+ Filter: filtered_top_10pct - Started at Mon Sep 22 05:41:49 UTC 2025
11
+ Filter: full_data - Started at Mon Sep 22 05:41:49 UTC 2025
12
+ Filter: filtered_top_50pct - Started at Mon Sep 22 05:41:49 UTC 2025
13
+ Filter: filtered_top_10pct - Failed with exit code 1 at Mon Sep 22 06:01:08 UTC 2025
14
+ Filter: filtered_top_20pct - Failed with exit code 1 at Mon Sep 22 06:01:08 UTC 2025
15
+ Filter: full_data - Failed with exit code 1 at Mon Sep 22 06:01:08 UTC 2025
16
+ Filter: filtered_top_50pct - Failed with exit code 1 at Mon Sep 22 06:01:09 UTC 2025
20250922_054148/filtered_data/filter_info.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "full_data": {
3
+ "inference_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet",
4
+ "auc_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data_auc.csv",
5
+ "sample_count": 525,
6
+ "auc_stats": {
7
+ "mean": 44.05714285714286,
8
+ "min": 4.0,
9
+ "max": 90.0,
10
+ "std": 22.1591349743965
11
+ },
12
+ "case_type_distribution": {
13
+ "nonhacking": 525
14
+ }
15
+ },
16
+ "filtered_top_10pct": {
17
+ "inference_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet",
18
+ "auc_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct_auc.csv",
19
+ "sample_count": 473,
20
+ "auc_stats": {
21
+ "mean": 40.16279069767442,
22
+ "min": 4.0,
23
+ "max": 73.0,
24
+ "std": 19.738999570898045
25
+ },
26
+ "case_type_distribution": {
27
+ "nonhacking": 473
28
+ }
29
+ },
30
+ "filtered_top_20pct": {
31
+ "inference_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet",
32
+ "auc_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct_auc.csv",
33
+ "sample_count": 420,
34
+ "auc_stats": {
35
+ "mean": 36.457142857142856,
36
+ "min": 4.0,
37
+ "max": 65.0,
38
+ "std": 17.755044230749405
39
+ },
40
+ "case_type_distribution": {
41
+ "nonhacking": 420
42
+ }
43
+ },
44
+ "filtered_top_50pct": {
45
+ "inference_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet",
46
+ "auc_file_path": "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct_auc.csv",
47
+ "sample_count": 263,
48
+ "auc_stats": {
49
+ "mean": 25.247148288973385,
50
+ "min": 4.0,
51
+ "max": 45.0,
52
+ "std": 12.230754528667553
53
+ },
54
+ "case_type_distribution": {
55
+ "nonhacking": 263
56
+ }
57
+ }
58
+ }
20250922_054148/filtered_data/filtered_top_10pct.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f394e64e591fb75d8b301e76b6068cf15ccd8b3315f4eaff293625594fa89651
3
+ size 233477
20250922_054148/filtered_data/filtered_top_10pct_auc.csv ADDED
@@ -0,0 +1,474 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ sample_id,auc_10_100,num_points,case_type,training_step
2
+ 0,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
3
+ 1,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
4
+ 2,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
5
+ 3,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
6
+ 4,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
7
+ 5,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
8
+ 6,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
9
+ 7,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
10
+ 8,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
11
+ 9,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
12
+ 10,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
13
+ 11,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
14
+ 12,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
15
+ 13,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
16
+ 14,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
17
+ 15,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
18
+ 16,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
19
+ 17,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20
+ 18,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
21
+ 20,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
22
+ 21,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
23
+ 22,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
24
+ 23,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
25
+ 24,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
26
+ 25,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
27
+ 26,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
28
+ 27,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
29
+ 28,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
30
+ 29,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
31
+ 30,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
32
+ 31,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
33
+ 32,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
34
+ 33,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
35
+ 34,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
36
+ 35,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
37
+ 36,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
38
+ 38,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
39
+ 39,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
40
+ 40,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
41
+ 41,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
42
+ 42,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
43
+ 43,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
44
+ 44,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
45
+ 45,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
46
+ 46,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
47
+ 47,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
48
+ 48,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
49
+ 49,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
50
+ 50,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
51
+ 51,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
52
+ 52,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
53
+ 53,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
54
+ 54,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
55
+ 55,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
56
+ 56,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
57
+ 57,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
58
+ 58,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
59
+ 59,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
60
+ 60,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
61
+ 61,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
62
+ 62,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
63
+ 63,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
64
+ 64,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
65
+ 65,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
66
+ 66,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
67
+ 67,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
68
+ 68,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
69
+ 70,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
70
+ 71,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
71
+ 72,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
72
+ 73,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
73
+ 74,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
74
+ 75,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
75
+ 76,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
76
+ 77,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
77
+ 78,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
78
+ 79,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
79
+ 80,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
80
+ 81,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
81
+ 82,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
82
+ 83,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
83
+ 84,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
84
+ 86,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
85
+ 88,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
86
+ 89,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
87
+ 90,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
88
+ 91,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
89
+ 92,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
90
+ 93,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
91
+ 94,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
92
+ 95,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
93
+ 96,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
94
+ 97,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
95
+ 98,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
96
+ 99,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
97
+ 100,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
98
+ 101,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
99
+ 102,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
100
+ 103,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
101
+ 104,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
102
+ 105,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
103
+ 106,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
104
+ 107,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
105
+ 108,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
106
+ 109,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
107
+ 110,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
108
+ 111,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
109
+ 112,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
110
+ 113,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
111
+ 114,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
112
+ 115,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
113
+ 116,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
114
+ 117,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
115
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
116
+ 120,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
117
+ 121,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
118
+ 122,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
119
+ 123,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
120
+ 124,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
121
+ 125,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
122
+ 126,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
123
+ 127,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
124
+ 128,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
125
+ 129,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
126
+ 130,48.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
127
+ 131,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
128
+ 132,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
129
+ 133,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
130
+ 134,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
131
+ 135,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
132
+ 136,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
133
+ 137,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
134
+ 138,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
135
+ 139,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
136
+ 140,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
137
+ 141,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
138
+ 142,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
139
+ 143,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
140
+ 144,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
141
+ 145,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
142
+ 146,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
143
+ 147,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
144
+ 148,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
145
+ 149,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
146
+ 150,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
147
+ 151,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
148
+ 152,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
149
+ 153,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
150
+ 154,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
151
+ 155,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
152
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
153
+ 157,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
154
+ 158,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
155
+ 159,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
156
+ 160,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
157
+ 161,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
158
+ 162,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
159
+ 163,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
160
+ 164,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
161
+ 166,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
162
+ 167,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
163
+ 168,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
164
+ 169,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
165
+ 170,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
166
+ 171,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
167
+ 172,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
168
+ 173,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
169
+ 174,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
170
+ 175,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
171
+ 176,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
172
+ 177,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
173
+ 178,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
174
+ 179,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
175
+ 180,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
176
+ 181,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
177
+ 182,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
178
+ 183,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
179
+ 184,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
180
+ 185,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
181
+ 186,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
182
+ 187,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
183
+ 189,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
184
+ 190,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
185
+ 192,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
186
+ 193,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
187
+ 194,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
188
+ 195,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
189
+ 196,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
190
+ 197,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
191
+ 199,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
192
+ 200,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
193
+ 201,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
194
+ 202,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
195
+ 203,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
196
+ 204,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
197
+ 205,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
198
+ 206,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
199
+ 207,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
200
+ 208,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
201
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
202
+ 210,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
203
+ 211,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
204
+ 212,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
205
+ 213,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
206
+ 214,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
207
+ 215,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
208
+ 216,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
209
+ 217,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
210
+ 219,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
211
+ 221,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
212
+ 222,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
213
+ 223,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
214
+ 224,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
215
+ 0,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
216
+ 2,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
217
+ 3,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
218
+ 5,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
219
+ 6,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
220
+ 7,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
221
+ 8,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
222
+ 9,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
223
+ 10,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
224
+ 12,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
225
+ 13,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
226
+ 14,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
227
+ 15,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
228
+ 16,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
229
+ 17,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
230
+ 18,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
231
+ 19,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
232
+ 21,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
233
+ 22,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
234
+ 23,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
235
+ 24,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
236
+ 25,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
237
+ 26,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
238
+ 27,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
239
+ 28,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
240
+ 29,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
241
+ 30,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
242
+ 31,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
243
+ 32,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
244
+ 33,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
245
+ 34,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
246
+ 35,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
247
+ 36,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
248
+ 37,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
249
+ 39,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
250
+ 40,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
251
+ 41,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
252
+ 42,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
253
+ 43,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
254
+ 44,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
255
+ 45,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
256
+ 46,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
257
+ 48,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
258
+ 50,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
259
+ 51,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
260
+ 52,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
261
+ 53,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
262
+ 54,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
263
+ 56,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
264
+ 57,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
265
+ 58,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
266
+ 60,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
267
+ 63,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
268
+ 64,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
269
+ 65,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
270
+ 66,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
271
+ 68,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
272
+ 69,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
273
+ 70,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
274
+ 71,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
275
+ 73,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
276
+ 74,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
277
+ 75,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
278
+ 76,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
279
+ 77,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
280
+ 78,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
281
+ 79,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
282
+ 80,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
283
+ 81,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
284
+ 82,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
285
+ 83,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
286
+ 84,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
287
+ 85,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
288
+ 86,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
289
+ 87,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
290
+ 88,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
291
+ 90,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
292
+ 91,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
293
+ 92,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
294
+ 93,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
295
+ 94,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
296
+ 95,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
297
+ 96,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
298
+ 97,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
299
+ 98,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
300
+ 99,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
301
+ 100,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
302
+ 101,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
303
+ 102,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
304
+ 103,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
305
+ 104,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
306
+ 105,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
307
+ 106,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
308
+ 107,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
309
+ 108,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
310
+ 109,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
311
+ 110,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
312
+ 111,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
313
+ 113,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
314
+ 115,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
315
+ 116,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
316
+ 117,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
317
+ 118,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
318
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
319
+ 120,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
320
+ 121,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
321
+ 122,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
322
+ 123,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
323
+ 124,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
324
+ 126,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
325
+ 128,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
326
+ 129,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
327
+ 130,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
328
+ 131,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
329
+ 132,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
330
+ 133,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
331
+ 134,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
332
+ 135,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
333
+ 136,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
334
+ 138,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
335
+ 140,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
336
+ 141,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
337
+ 142,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
338
+ 144,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
339
+ 145,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
340
+ 146,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
341
+ 148,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
342
+ 149,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
343
+ 150,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
344
+ 151,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
345
+ 152,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
346
+ 153,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
347
+ 154,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
348
+ 155,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
349
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
350
+ 157,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
351
+ 158,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
352
+ 159,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
353
+ 160,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
354
+ 161,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
355
+ 163,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
356
+ 164,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
357
+ 165,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
358
+ 166,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
359
+ 167,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
360
+ 168,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
361
+ 169,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
362
+ 170,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
363
+ 171,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
364
+ 172,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
365
+ 173,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
366
+ 174,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
367
+ 175,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
368
+ 176,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
369
+ 177,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
370
+ 178,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
371
+ 179,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
372
+ 180,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
373
+ 181,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
374
+ 182,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
375
+ 185,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
376
+ 186,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
377
+ 187,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
378
+ 188,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
379
+ 189,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
380
+ 190,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
381
+ 191,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
382
+ 192,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
383
+ 193,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
384
+ 194,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
385
+ 195,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
386
+ 196,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
387
+ 198,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
388
+ 199,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
389
+ 200,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
390
+ 201,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
391
+ 202,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
392
+ 204,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
393
+ 205,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
394
+ 206,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
395
+ 207,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
396
+ 208,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
397
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
398
+ 210,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
399
+ 211,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
400
+ 212,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
401
+ 213,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
402
+ 214,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
403
+ 216,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
404
+ 217,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
405
+ 218,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
406
+ 219,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
407
+ 220,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
408
+ 221,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
409
+ 222,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
410
+ 224,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
411
+ 226,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
412
+ 227,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
413
+ 228,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
414
+ 229,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
415
+ 230,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
416
+ 231,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
417
+ 232,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
418
+ 234,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
419
+ 235,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
420
+ 236,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
421
+ 237,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
422
+ 239,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
423
+ 240,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
424
+ 241,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
425
+ 242,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
426
+ 243,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
427
+ 244,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
428
+ 245,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
429
+ 246,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
430
+ 247,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
431
+ 248,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
432
+ 250,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
433
+ 252,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
434
+ 253,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
435
+ 254,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
436
+ 255,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
437
+ 256,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
438
+ 257,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
439
+ 258,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
440
+ 259,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
441
+ 260,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
442
+ 262,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
443
+ 263,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
444
+ 264,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
445
+ 265,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
446
+ 267,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
447
+ 268,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
448
+ 269,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
449
+ 270,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
450
+ 272,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
451
+ 273,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
452
+ 274,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
453
+ 275,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
454
+ 276,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
455
+ 277,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
456
+ 278,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
457
+ 279,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
458
+ 280,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
459
+ 281,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
460
+ 282,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
461
+ 284,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
462
+ 285,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
463
+ 286,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
464
+ 287,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
465
+ 290,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
466
+ 291,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
467
+ 292,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
468
+ 293,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
469
+ 294,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
470
+ 295,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
471
+ 296,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
472
+ 297,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
473
+ 298,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
474
+ 299,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20250922_054148/filtered_data/filtered_top_20pct.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:149e7d664133e541bbc1143235b57dbf40ed9d42d88ceaae87006033d28782e4
3
+ size 211384
20250922_054148/filtered_data/filtered_top_20pct_auc.csv ADDED
@@ -0,0 +1,421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ sample_id,auc_10_100,num_points,case_type,training_step
2
+ 0,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
3
+ 1,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
4
+ 2,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
5
+ 3,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
6
+ 4,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
7
+ 6,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
8
+ 7,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
9
+ 8,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
10
+ 9,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
11
+ 10,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
12
+ 11,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
13
+ 12,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
14
+ 13,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
15
+ 15,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
16
+ 16,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
17
+ 17,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
18
+ 21,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
19
+ 22,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20
+ 23,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
21
+ 24,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
22
+ 25,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
23
+ 26,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
24
+ 27,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
25
+ 28,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
26
+ 29,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
27
+ 31,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
28
+ 32,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
29
+ 33,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
30
+ 34,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
31
+ 35,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
32
+ 36,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
33
+ 38,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
34
+ 39,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
35
+ 40,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
36
+ 42,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
37
+ 44,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
38
+ 45,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
39
+ 46,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
40
+ 47,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
41
+ 48,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
42
+ 49,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
43
+ 50,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
44
+ 51,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
45
+ 52,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
46
+ 53,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
47
+ 54,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
48
+ 55,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
49
+ 56,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
50
+ 57,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
51
+ 58,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
52
+ 60,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
53
+ 61,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
54
+ 62,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
55
+ 63,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
56
+ 64,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
57
+ 65,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
58
+ 66,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
59
+ 67,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
60
+ 68,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
61
+ 70,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
62
+ 71,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
63
+ 72,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
64
+ 73,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
65
+ 74,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
66
+ 75,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
67
+ 76,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
68
+ 77,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
69
+ 78,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
70
+ 79,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
71
+ 80,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
72
+ 81,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
73
+ 82,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
74
+ 83,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
75
+ 84,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
76
+ 86,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
77
+ 88,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
78
+ 89,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
79
+ 90,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
80
+ 92,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
81
+ 93,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
82
+ 94,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
83
+ 95,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
84
+ 96,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
85
+ 97,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
86
+ 98,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
87
+ 99,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
88
+ 100,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
89
+ 101,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
90
+ 102,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
91
+ 103,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
92
+ 104,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
93
+ 105,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
94
+ 106,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
95
+ 107,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
96
+ 108,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
97
+ 109,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
98
+ 110,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
99
+ 111,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
100
+ 112,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
101
+ 113,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
102
+ 114,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
103
+ 115,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
104
+ 117,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
105
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
106
+ 120,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
107
+ 121,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
108
+ 122,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
109
+ 123,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
110
+ 124,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
111
+ 125,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
112
+ 126,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
113
+ 127,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
114
+ 128,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
115
+ 130,48.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
116
+ 131,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
117
+ 132,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
118
+ 133,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
119
+ 134,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
120
+ 135,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
121
+ 136,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
122
+ 137,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
123
+ 138,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
124
+ 139,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
125
+ 140,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
126
+ 141,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
127
+ 142,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
128
+ 143,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
129
+ 144,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
130
+ 145,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
131
+ 146,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
132
+ 147,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
133
+ 148,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
134
+ 149,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
135
+ 150,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
136
+ 151,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
137
+ 152,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
138
+ 154,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
139
+ 155,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
140
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
141
+ 157,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
142
+ 158,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
143
+ 159,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
144
+ 160,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
145
+ 161,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
146
+ 162,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
147
+ 163,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
148
+ 164,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
149
+ 166,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
150
+ 167,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
151
+ 168,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
152
+ 169,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
153
+ 171,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
154
+ 172,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
155
+ 173,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
156
+ 174,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
157
+ 175,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
158
+ 176,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
159
+ 177,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
160
+ 178,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
161
+ 179,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
162
+ 180,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
163
+ 181,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
164
+ 182,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
165
+ 183,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
166
+ 185,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
167
+ 186,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
168
+ 187,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
169
+ 189,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
170
+ 190,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
171
+ 192,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
172
+ 193,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
173
+ 194,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
174
+ 195,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
175
+ 196,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
176
+ 197,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
177
+ 199,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
178
+ 200,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
179
+ 201,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
180
+ 202,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
181
+ 203,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
182
+ 204,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
183
+ 205,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
184
+ 206,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
185
+ 207,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
186
+ 208,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
187
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
188
+ 210,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
189
+ 211,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
190
+ 212,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
191
+ 213,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
192
+ 214,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
193
+ 215,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
194
+ 216,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
195
+ 217,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
196
+ 219,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
197
+ 221,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
198
+ 222,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
199
+ 223,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
200
+ 224,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
201
+ 0,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
202
+ 2,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
203
+ 3,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
204
+ 6,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
205
+ 7,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
206
+ 10,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
207
+ 12,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
208
+ 13,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
209
+ 15,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
210
+ 16,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
211
+ 19,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
212
+ 21,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
213
+ 22,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
214
+ 23,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
215
+ 25,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
216
+ 26,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
217
+ 27,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
218
+ 28,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
219
+ 29,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
220
+ 31,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
221
+ 34,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
222
+ 35,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
223
+ 36,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
224
+ 37,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
225
+ 39,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
226
+ 40,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
227
+ 41,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
228
+ 43,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
229
+ 44,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
230
+ 45,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
231
+ 46,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
232
+ 48,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
233
+ 50,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
234
+ 51,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
235
+ 52,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
236
+ 53,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
237
+ 56,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
238
+ 57,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
239
+ 58,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
240
+ 65,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
241
+ 66,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
242
+ 68,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
243
+ 69,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
244
+ 70,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
245
+ 71,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
246
+ 73,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
247
+ 76,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
248
+ 77,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
249
+ 78,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
250
+ 79,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
251
+ 80,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
252
+ 81,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
253
+ 82,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
254
+ 83,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
255
+ 84,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
256
+ 85,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
257
+ 86,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
258
+ 87,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
259
+ 88,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
260
+ 90,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
261
+ 91,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
262
+ 92,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
263
+ 93,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
264
+ 95,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
265
+ 96,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
266
+ 97,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
267
+ 98,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
268
+ 99,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
269
+ 100,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
270
+ 101,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
271
+ 102,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
272
+ 103,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
273
+ 104,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
274
+ 105,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
275
+ 106,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
276
+ 107,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
277
+ 108,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
278
+ 110,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
279
+ 111,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
280
+ 113,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
281
+ 115,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
282
+ 116,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
283
+ 118,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
284
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
285
+ 120,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
286
+ 121,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
287
+ 122,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
288
+ 123,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
289
+ 124,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
290
+ 126,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
291
+ 129,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
292
+ 130,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
293
+ 131,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
294
+ 132,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
295
+ 133,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
296
+ 134,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
297
+ 136,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
298
+ 138,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
299
+ 140,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
300
+ 141,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
301
+ 142,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
302
+ 144,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
303
+ 145,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
304
+ 146,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
305
+ 148,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
306
+ 149,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
307
+ 150,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
308
+ 151,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
309
+ 153,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
310
+ 154,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
311
+ 155,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
312
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
313
+ 157,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
314
+ 158,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
315
+ 159,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
316
+ 160,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
317
+ 161,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
318
+ 163,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
319
+ 165,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
320
+ 167,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
321
+ 168,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
322
+ 169,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
323
+ 170,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
324
+ 171,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
325
+ 172,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
326
+ 173,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
327
+ 175,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
328
+ 176,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
329
+ 177,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
330
+ 178,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
331
+ 179,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
332
+ 180,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
333
+ 181,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
334
+ 182,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
335
+ 185,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
336
+ 188,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
337
+ 189,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
338
+ 190,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
339
+ 191,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
340
+ 192,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
341
+ 193,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
342
+ 194,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
343
+ 196,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
344
+ 198,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
345
+ 199,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
346
+ 200,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
347
+ 202,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
348
+ 204,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
349
+ 205,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
350
+ 206,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
351
+ 207,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
352
+ 208,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
353
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
354
+ 210,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
355
+ 211,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
356
+ 212,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
357
+ 213,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
358
+ 216,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
359
+ 217,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
360
+ 218,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
361
+ 219,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
362
+ 220,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
363
+ 221,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
364
+ 222,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
365
+ 224,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
366
+ 226,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
367
+ 227,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
368
+ 228,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
369
+ 229,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
370
+ 232,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
371
+ 234,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
372
+ 235,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
373
+ 236,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
374
+ 237,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
375
+ 239,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
376
+ 240,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
377
+ 241,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
378
+ 242,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
379
+ 243,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
380
+ 244,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
381
+ 245,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
382
+ 246,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
383
+ 247,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
384
+ 248,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
385
+ 250,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
386
+ 252,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
387
+ 253,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
388
+ 254,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
389
+ 255,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
390
+ 256,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
391
+ 258,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
392
+ 259,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
393
+ 260,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
394
+ 262,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
395
+ 263,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
396
+ 264,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
397
+ 267,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
398
+ 269,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
399
+ 270,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
400
+ 272,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
401
+ 275,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
402
+ 276,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
403
+ 277,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
404
+ 278,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
405
+ 279,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
406
+ 280,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
407
+ 281,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
408
+ 282,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
409
+ 284,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
410
+ 285,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
411
+ 286,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
412
+ 290,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
413
+ 291,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
414
+ 292,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
415
+ 293,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
416
+ 294,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
417
+ 295,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
418
+ 296,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
419
+ 297,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
420
+ 298,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
421
+ 299,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20250922_054148/filtered_data/filtered_top_50pct.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80ee39eb66b6afcf66488d33cd8320658cc5e6c98b262a115b1af8637fb234c0
3
+ size 142735
20250922_054148/filtered_data/filtered_top_50pct_auc.csv ADDED
@@ -0,0 +1,264 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ sample_id,auc_10_100,num_points,case_type,training_step
2
+ 0,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
3
+ 3,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
4
+ 6,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
5
+ 7,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
6
+ 8,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
7
+ 10,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
8
+ 11,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
9
+ 13,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
10
+ 15,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
11
+ 16,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
12
+ 21,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
13
+ 23,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
14
+ 24,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
15
+ 25,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
16
+ 26,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
17
+ 27,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
18
+ 31,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
19
+ 32,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20
+ 34,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
21
+ 35,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
22
+ 36,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
23
+ 39,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
24
+ 42,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
25
+ 45,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
26
+ 46,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
27
+ 48,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
28
+ 49,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
29
+ 50,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
30
+ 51,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
31
+ 52,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
32
+ 53,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
33
+ 54,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
34
+ 55,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
35
+ 58,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
36
+ 60,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
37
+ 62,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
38
+ 63,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
39
+ 64,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
40
+ 65,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
41
+ 66,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
42
+ 67,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
43
+ 72,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
44
+ 74,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
45
+ 75,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
46
+ 76,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
47
+ 77,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
48
+ 79,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
49
+ 80,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
50
+ 81,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
51
+ 82,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
52
+ 83,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
53
+ 84,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
54
+ 86,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
55
+ 88,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
56
+ 89,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
57
+ 90,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
58
+ 93,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
59
+ 96,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
60
+ 97,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
61
+ 98,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
62
+ 99,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
63
+ 100,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
64
+ 101,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
65
+ 102,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
66
+ 104,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
67
+ 105,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
68
+ 106,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
69
+ 107,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
70
+ 108,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
71
+ 109,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
72
+ 110,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
73
+ 111,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
74
+ 112,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
75
+ 113,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
76
+ 114,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
77
+ 115,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
78
+ 117,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
79
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
80
+ 120,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
81
+ 121,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
82
+ 122,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
83
+ 124,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
84
+ 125,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
85
+ 127,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
86
+ 128,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
87
+ 131,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
88
+ 132,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
89
+ 133,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
90
+ 134,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
91
+ 135,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
92
+ 137,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
93
+ 140,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
94
+ 142,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
95
+ 143,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
96
+ 144,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
97
+ 146,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
98
+ 148,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
99
+ 151,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
100
+ 152,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
101
+ 154,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
102
+ 155,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
103
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
104
+ 158,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
105
+ 159,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
106
+ 160,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
107
+ 161,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
108
+ 163,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
109
+ 166,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
110
+ 167,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
111
+ 168,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
112
+ 169,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
113
+ 173,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
114
+ 176,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
115
+ 179,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
116
+ 181,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
117
+ 182,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
118
+ 183,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
119
+ 185,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
120
+ 186,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
121
+ 187,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
122
+ 189,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
123
+ 190,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
124
+ 194,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
125
+ 195,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
126
+ 196,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
127
+ 197,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
128
+ 200,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
129
+ 201,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
130
+ 202,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
131
+ 206,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
132
+ 208,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
133
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
134
+ 210,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
135
+ 211,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
136
+ 212,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
137
+ 213,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
138
+ 214,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
139
+ 215,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
140
+ 216,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
141
+ 217,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
142
+ 219,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
143
+ 221,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
144
+ 222,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
145
+ 223,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
146
+ 224,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
147
+ 3,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
148
+ 6,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
149
+ 7,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
150
+ 13,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
151
+ 16,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
152
+ 22,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
153
+ 25,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
154
+ 27,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
155
+ 28,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
156
+ 31,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
157
+ 39,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
158
+ 44,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
159
+ 46,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
160
+ 48,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
161
+ 50,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
162
+ 53,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
163
+ 56,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
164
+ 57,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
165
+ 65,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
166
+ 68,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
167
+ 69,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
168
+ 70,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
169
+ 71,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
170
+ 77,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
171
+ 79,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
172
+ 80,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
173
+ 81,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
174
+ 84,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
175
+ 86,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
176
+ 87,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
177
+ 88,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
178
+ 90,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
179
+ 91,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
180
+ 92,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
181
+ 95,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
182
+ 99,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
183
+ 100,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
184
+ 102,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
185
+ 104,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
186
+ 107,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
187
+ 116,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
188
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
189
+ 120,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
190
+ 121,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
191
+ 123,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
192
+ 124,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
193
+ 126,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
194
+ 129,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
195
+ 130,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
196
+ 134,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
197
+ 138,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
198
+ 140,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
199
+ 141,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
200
+ 142,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
201
+ 144,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
202
+ 149,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
203
+ 150,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
204
+ 153,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
205
+ 155,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
206
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
207
+ 158,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
208
+ 159,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
209
+ 160,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
210
+ 169,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
211
+ 170,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
212
+ 171,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
213
+ 172,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
214
+ 175,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
215
+ 176,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
216
+ 177,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
217
+ 178,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
218
+ 179,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
219
+ 180,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
220
+ 181,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
221
+ 189,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
222
+ 191,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
223
+ 193,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
224
+ 194,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
225
+ 196,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
226
+ 198,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
227
+ 199,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
228
+ 202,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
229
+ 204,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
230
+ 205,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
231
+ 206,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
232
+ 208,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
233
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
234
+ 211,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
235
+ 212,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
236
+ 216,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
237
+ 217,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
238
+ 220,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
239
+ 224,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
240
+ 226,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
241
+ 234,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
242
+ 235,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
243
+ 239,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
244
+ 241,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
245
+ 242,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
246
+ 245,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
247
+ 246,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
248
+ 247,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
249
+ 250,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
250
+ 253,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
251
+ 255,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
252
+ 259,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
253
+ 262,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
254
+ 270,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
255
+ 275,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
256
+ 276,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
257
+ 281,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
258
+ 282,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
259
+ 286,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
260
+ 291,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
261
+ 295,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
262
+ 296,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
263
+ 297,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
264
+ 298,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20250922_054148/filtered_data/full_data.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88a59e1e90e6ea1155e84670fc7d1a16005d13f97e35e4f3d1df99ebad97c048
3
+ size 254127
20250922_054148/filtered_data/full_data_auc.csv ADDED
@@ -0,0 +1,526 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ sample_id,auc_10_100,num_points,case_type,training_step
2
+ 0,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
3
+ 1,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
4
+ 2,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
5
+ 3,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
6
+ 4,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
7
+ 5,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
8
+ 6,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
9
+ 7,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
10
+ 8,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
11
+ 9,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
12
+ 10,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
13
+ 11,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
14
+ 12,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
15
+ 13,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
16
+ 14,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
17
+ 15,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
18
+ 16,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
19
+ 17,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20
+ 18,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
21
+ 19,84.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
22
+ 20,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
23
+ 21,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
24
+ 22,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
25
+ 23,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
26
+ 24,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
27
+ 25,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
28
+ 26,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
29
+ 27,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
30
+ 28,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
31
+ 29,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
32
+ 30,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
33
+ 31,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
34
+ 32,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
35
+ 33,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
36
+ 34,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
37
+ 35,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
38
+ 36,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
39
+ 37,76.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
40
+ 38,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
41
+ 39,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
42
+ 40,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
43
+ 41,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
44
+ 42,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
45
+ 43,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
46
+ 44,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
47
+ 45,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
48
+ 46,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
49
+ 47,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
50
+ 48,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
51
+ 49,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
52
+ 50,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
53
+ 51,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
54
+ 52,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
55
+ 53,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
56
+ 54,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
57
+ 55,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
58
+ 56,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
59
+ 57,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
60
+ 58,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
61
+ 59,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
62
+ 60,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
63
+ 61,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
64
+ 62,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
65
+ 63,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
66
+ 64,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
67
+ 65,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
68
+ 66,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
69
+ 67,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
70
+ 68,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
71
+ 69,82.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
72
+ 70,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
73
+ 71,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
74
+ 72,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
75
+ 73,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
76
+ 74,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
77
+ 75,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
78
+ 76,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
79
+ 77,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
80
+ 78,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
81
+ 79,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
82
+ 80,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
83
+ 81,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
84
+ 82,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
85
+ 83,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
86
+ 84,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
87
+ 85,78.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
88
+ 86,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
89
+ 87,88.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
90
+ 88,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
91
+ 89,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
92
+ 90,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
93
+ 91,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
94
+ 92,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
95
+ 93,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
96
+ 94,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
97
+ 95,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
98
+ 96,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
99
+ 97,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
100
+ 98,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
101
+ 99,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
102
+ 100,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
103
+ 101,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
104
+ 102,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
105
+ 103,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
106
+ 104,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
107
+ 105,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
108
+ 106,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
109
+ 107,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
110
+ 108,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
111
+ 109,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
112
+ 110,26.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
113
+ 111,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
114
+ 112,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
115
+ 113,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
116
+ 114,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
117
+ 115,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
118
+ 116,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
119
+ 117,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
120
+ 118,76.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
121
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
122
+ 120,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
123
+ 121,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
124
+ 122,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
125
+ 123,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
126
+ 124,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
127
+ 125,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
128
+ 126,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
129
+ 127,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
130
+ 128,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
131
+ 129,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
132
+ 130,48.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
133
+ 131,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
134
+ 132,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
135
+ 133,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
136
+ 134,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
137
+ 135,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
138
+ 136,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
139
+ 137,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
140
+ 138,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
141
+ 139,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
142
+ 140,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
143
+ 141,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
144
+ 142,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
145
+ 143,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
146
+ 144,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
147
+ 145,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
148
+ 146,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
149
+ 147,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
150
+ 148,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
151
+ 149,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
152
+ 150,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
153
+ 151,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
154
+ 152,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
155
+ 153,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
156
+ 154,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
157
+ 155,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
158
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
159
+ 157,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
160
+ 158,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
161
+ 159,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
162
+ 160,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
163
+ 161,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
164
+ 162,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
165
+ 163,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
166
+ 164,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
167
+ 165,88.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
168
+ 166,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
169
+ 167,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
170
+ 168,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
171
+ 169,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
172
+ 170,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
173
+ 171,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
174
+ 172,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
175
+ 173,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
176
+ 174,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
177
+ 175,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
178
+ 176,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
179
+ 177,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
180
+ 178,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
181
+ 179,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
182
+ 180,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
183
+ 181,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
184
+ 182,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
185
+ 183,13.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
186
+ 184,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
187
+ 185,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
188
+ 186,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
189
+ 187,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
190
+ 188,79.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
191
+ 189,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
192
+ 190,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
193
+ 191,81.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
194
+ 192,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
195
+ 193,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
196
+ 194,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
197
+ 195,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
198
+ 196,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
199
+ 197,12.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
200
+ 198,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
201
+ 199,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
202
+ 200,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
203
+ 201,4.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
204
+ 202,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
205
+ 203,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
206
+ 204,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
207
+ 205,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
208
+ 206,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
209
+ 207,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
210
+ 208,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
211
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
212
+ 210,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
213
+ 211,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
214
+ 212,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
215
+ 213,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
216
+ 214,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
217
+ 215,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
218
+ 216,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
219
+ 217,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
220
+ 218,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
221
+ 219,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
222
+ 220,86.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
223
+ 221,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
224
+ 222,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
225
+ 223,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
226
+ 224,44.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
227
+ 0,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
228
+ 1,88.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
229
+ 2,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
230
+ 3,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
231
+ 4,81.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
232
+ 5,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
233
+ 6,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
234
+ 7,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
235
+ 8,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
236
+ 9,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
237
+ 10,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
238
+ 11,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
239
+ 12,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
240
+ 13,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
241
+ 14,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
242
+ 15,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
243
+ 16,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
244
+ 17,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
245
+ 18,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
246
+ 19,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
247
+ 20,81.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
248
+ 21,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
249
+ 22,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
250
+ 23,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
251
+ 24,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
252
+ 25,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
253
+ 26,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
254
+ 27,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
255
+ 28,45.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
256
+ 29,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
257
+ 30,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
258
+ 31,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
259
+ 32,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
260
+ 33,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
261
+ 34,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
262
+ 35,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
263
+ 36,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
264
+ 37,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
265
+ 38,80.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
266
+ 39,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
267
+ 40,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
268
+ 41,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
269
+ 42,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
270
+ 43,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
271
+ 44,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
272
+ 45,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
273
+ 46,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
274
+ 47,82.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
275
+ 48,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
276
+ 49,75.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
277
+ 50,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
278
+ 51,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
279
+ 52,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
280
+ 53,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
281
+ 54,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
282
+ 55,82.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
283
+ 56,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
284
+ 57,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
285
+ 58,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
286
+ 59,84.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
287
+ 60,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
288
+ 61,85.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
289
+ 62,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
290
+ 63,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
291
+ 64,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
292
+ 65,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
293
+ 66,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
294
+ 67,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
295
+ 68,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
296
+ 69,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
297
+ 70,14.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
298
+ 71,11.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
299
+ 72,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
300
+ 73,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
301
+ 74,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
302
+ 75,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
303
+ 76,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
304
+ 77,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
305
+ 78,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
306
+ 79,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
307
+ 80,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
308
+ 81,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
309
+ 82,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
310
+ 83,62.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
311
+ 84,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
312
+ 85,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
313
+ 86,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
314
+ 87,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
315
+ 88,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
316
+ 89,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
317
+ 90,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
318
+ 91,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
319
+ 92,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
320
+ 93,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
321
+ 94,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
322
+ 95,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
323
+ 96,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
324
+ 97,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
325
+ 98,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
326
+ 99,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
327
+ 100,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
328
+ 101,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
329
+ 102,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
330
+ 103,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
331
+ 104,28.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
332
+ 105,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
333
+ 106,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
334
+ 107,36.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
335
+ 108,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
336
+ 109,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
337
+ 110,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
338
+ 111,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
339
+ 112,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
340
+ 113,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
341
+ 114,90.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
342
+ 115,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
343
+ 116,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
344
+ 117,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
345
+ 118,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
346
+ 119,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
347
+ 120,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
348
+ 121,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
349
+ 122,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
350
+ 123,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
351
+ 124,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
352
+ 125,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
353
+ 126,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
354
+ 127,79.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
355
+ 128,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
356
+ 129,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
357
+ 130,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
358
+ 131,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
359
+ 132,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
360
+ 133,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
361
+ 134,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
362
+ 135,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
363
+ 136,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
364
+ 137,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
365
+ 138,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
366
+ 139,76.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
367
+ 140,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
368
+ 141,5.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
369
+ 142,20.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
370
+ 143,78.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
371
+ 144,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
372
+ 145,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
373
+ 146,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
374
+ 147,76.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
375
+ 148,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
376
+ 149,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
377
+ 150,32.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
378
+ 151,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
379
+ 152,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
380
+ 153,19.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
381
+ 154,46.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
382
+ 155,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
383
+ 156,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
384
+ 157,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
385
+ 158,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
386
+ 159,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
387
+ 160,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
388
+ 161,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
389
+ 162,79.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
390
+ 163,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
391
+ 164,68.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
392
+ 165,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
393
+ 166,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
394
+ 167,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
395
+ 168,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
396
+ 169,16.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
397
+ 170,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
398
+ 171,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
399
+ 172,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
400
+ 173,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
401
+ 174,72.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
402
+ 175,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
403
+ 176,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
404
+ 177,15.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
405
+ 178,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
406
+ 179,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
407
+ 180,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
408
+ 181,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
409
+ 182,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
410
+ 183,79.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
411
+ 184,75.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
412
+ 185,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
413
+ 186,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
414
+ 187,73.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
415
+ 188,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
416
+ 189,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
417
+ 190,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
418
+ 191,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
419
+ 192,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
420
+ 193,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
421
+ 194,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
422
+ 195,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
423
+ 196,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
424
+ 197,83.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
425
+ 198,23.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
426
+ 199,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
427
+ 200,53.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
428
+ 201,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
429
+ 202,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
430
+ 203,82.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
431
+ 204,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
432
+ 205,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
433
+ 206,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
434
+ 207,54.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
435
+ 208,38.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
436
+ 209,8.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
437
+ 210,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
438
+ 211,9.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
439
+ 212,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
440
+ 213,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
441
+ 214,67.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
442
+ 215,81.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
443
+ 216,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
444
+ 217,22.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
445
+ 218,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
446
+ 219,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
447
+ 220,33.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
448
+ 221,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
449
+ 222,55.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
450
+ 223,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
451
+ 224,6.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
452
+ 225,77.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
453
+ 226,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
454
+ 227,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
455
+ 228,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
456
+ 229,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
457
+ 230,65.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
458
+ 231,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
459
+ 232,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
460
+ 233,84.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
461
+ 234,35.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
462
+ 235,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
463
+ 236,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
464
+ 237,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
465
+ 238,74.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
466
+ 239,18.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
467
+ 240,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
468
+ 241,40.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
469
+ 242,25.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
470
+ 243,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
471
+ 244,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
472
+ 245,31.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
473
+ 246,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
474
+ 247,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
475
+ 248,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
476
+ 249,87.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
477
+ 250,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
478
+ 251,80.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
479
+ 252,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
480
+ 253,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
481
+ 254,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
482
+ 255,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
483
+ 256,64.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
484
+ 257,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
485
+ 258,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
486
+ 259,34.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
487
+ 260,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
488
+ 261,75.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
489
+ 262,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
490
+ 263,58.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
491
+ 264,59.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
492
+ 265,66.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
493
+ 266,79.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
494
+ 267,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
495
+ 268,71.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
496
+ 269,47.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
497
+ 270,27.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
498
+ 271,80.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
499
+ 272,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
500
+ 273,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
501
+ 274,70.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
502
+ 275,39.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
503
+ 276,41.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
504
+ 277,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
505
+ 278,52.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
506
+ 279,60.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
507
+ 280,49.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
508
+ 281,17.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
509
+ 282,42.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
510
+ 283,82.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
511
+ 284,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
512
+ 285,57.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
513
+ 286,24.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
514
+ 287,69.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
515
+ 288,78.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
516
+ 289,75.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
517
+ 290,56.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
518
+ 291,43.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
519
+ 292,63.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
520
+ 293,61.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
521
+ 294,51.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
522
+ 295,21.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
523
+ 296,29.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
524
+ 297,37.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
525
+ 298,7.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
526
+ 299,50.0,10,nonhacking,step_kk-instruct-3ppl-qwen2.5-3b-instruct-og-rloo
20250922_054148/results_summary.txt ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Parallel AUC-Filtered SFT Experiments Results Summary
2
+ Generated at: Mon Sep 22 05:41:49 UTC 2025
3
+
4
+ Individual experiment results:
5
+ === filtered_top_10pct ===
6
+ Training completed
7
+
8
+ === filtered_top_20pct ===
9
+ Training completed
10
+
11
+ === filtered_top_50pct ===
12
+ Training completed
13
+
14
+ === full_data ===
15
+ Training completed
16
+
20250922_054148/run_analysis.sh ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # Results Analysis Job
3
+ # Generated at: 20250922_054148
4
+
5
+ echo "Starting results analysis..."
6
+ echo "Base output directory: checkpoints/auc_filtered_sft/20250922_054148"
7
+
8
+ # Create results summary
9
+ RESULTS_SUMMARY="checkpoints/auc_filtered_sft/20250922_054148/results_summary.txt"
10
+ echo "Parallel AUC-Filtered SFT Experiments Results Summary" > "$RESULTS_SUMMARY"
11
+ echo "Generated at: $(date)" >> "$RESULTS_SUMMARY"
12
+ echo "" >> "$RESULTS_SUMMARY"
13
+
14
+ echo "Individual experiment results:" >> "$RESULTS_SUMMARY"
15
+ for result_dir in checkpoints/auc_filtered_sft/20250922_054148/sft_results/*/; do
16
+ if [ -d "$result_dir" ]; then
17
+ filter_name=$(basename "$result_dir")
18
+ echo "=== $filter_name ===" >> "$RESULTS_SUMMARY"
19
+
20
+ if [ -f "$result_dir/training_log.txt" ]; then
21
+ echo "Training completed" >> "$RESULTS_SUMMARY"
22
+ # Extract key metrics if available
23
+ grep -i "loss\|accuracy\|perplexity" "$result_dir/training_log.txt" | tail -5 >> "$RESULTS_SUMMARY" 2>/dev/null || echo "No metrics found" >> "$RESULTS_SUMMARY"
24
+ else
25
+ echo "Training log not found" >> "$RESULTS_SUMMARY"
26
+ fi
27
+ echo "" >> "$RESULTS_SUMMARY"
28
+ fi
29
+ done
30
+
31
+ echo "Results analysis completed!"
32
+ echo "Summary saved to: $RESULTS_SUMMARY"
33
+
34
+ # Display filter information
35
+ echo ""
36
+ echo "Filter Information:"
37
+ if command -v jq &> /dev/null; then
38
+ jq . "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filter_info.json"
39
+ else
40
+ cat "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filter_info.json"
41
+ fi
42
+
43
+ echo ""
44
+ echo "Experiment completed! Results in: checkpoints/auc_filtered_sft/20250922_054148"
20250922_054148/sft_results/filtered_top_10pct/gpu_lock_wrapper.sh ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # GPU Lock Wrapper for job: sft_filtered_top_10pct
3
+ # GPUs: [1]
4
+
5
+ # Function to cleanup GPU locks on exit
6
+ cleanup_locks() {
7
+ echo "Cleaning up GPU locks for job: sft_filtered_top_10pct"
8
+ python3 -c "
9
+ import sys
10
+ sys.path.append('/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl')
11
+ from utils.utils import release_gpu_lock
12
+ gpus = [1]
13
+ for gpu_id in gpus:
14
+ release_gpu_lock(str(gpu_id))
15
+ print(f'Released lock for GPU {gpu_id}')
16
+ "
17
+ }
18
+
19
+ # Set up trap to cleanup locks on script exit (success, failure, or interrupt)
20
+ trap cleanup_locks EXIT INT TERM
21
+
22
+ # Run the actual job
23
+ echo "Starting job: sft_filtered_top_10pct on GPUs: [1]"
24
+ CUDA_VISIBLE_DEVICES=1 bash checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/run_sft_script.sh
25
+ JOB_EXIT_CODE=$?
26
+
27
+ # Explicit cleanup (trap will also run, but this ensures it happens)
28
+ cleanup_locks
29
+
30
+ # Exit with the same code as the job
31
+ exit $JOB_EXIT_CODE
20250922_054148/sft_results/filtered_top_10pct/run_sft_script.sh ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # SFT Job - filtered_top_10pct
3
+ # Generated at: 20250922_054148
4
+
5
+ export TRANSFORMERS_CACHE=/home/greenland-user/verl/checkpoints/huggingface
6
+ export PYTHONHASHSEED=42
7
+
8
+ echo "Starting SFT training for filtered_top_10pct"
9
+ echo "Data file: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet"
10
+ echo "Output: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct"
11
+ echo "Model: Qwen/Qwen2.5-1.5B"
12
+
13
+ # Log experiment start
14
+ echo "Filter: filtered_top_10pct - Started at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
15
+
16
+ # Run SFT training
17
+ timeout 7200 bash scripts_cot/run_simple_sft.sh \
18
+ --model_path "Qwen/Qwen2.5-1.5B" \
19
+ --train_data_path "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet" \
20
+ --test_data_path "data/big-math_tiny_instruct/test.parquet" \
21
+ --output_dir "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct" \
22
+ --sft_epochs "2" \
23
+ --sft_batch_size "4" \
24
+ --gpu "$CUDA_VISIBLE_DEVICES" \
25
+ 2>&1 | tee "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/training_log.txt"
26
+
27
+ SFT_EXIT_CODE=$?
28
+
29
+ if [ $SFT_EXIT_CODE -eq 0 ]; then
30
+ echo "SFT completed successfully for filtered_top_10pct"
31
+ echo "Filter: filtered_top_10pct - Completed successfully at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
32
+ elif [ $SFT_EXIT_CODE -eq 124 ]; then
33
+ echo "SFT timed out for filtered_top_10pct (2 hour limit)"
34
+ echo "Filter: filtered_top_10pct - Timed out at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
35
+ else
36
+ echo "SFT failed for filtered_top_10pct with exit code $SFT_EXIT_CODE"
37
+ echo "Filter: filtered_top_10pct - Failed with exit code $SFT_EXIT_CODE at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
38
+ fi
39
+
40
+ echo "Completed SFT training for filtered_top_10pct"
20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:451730a80246536b49d7d3d156a9b7726d517388ea4961ce638129d00b5b230f
3
+ size 205794
20250922_054148/sft_results/filtered_top_10pct/training_log.txt ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==================================================
2
+ Starting Simple SFT Training
3
+ ==================================================
4
+ Model: Qwen/Qwen2.5-1.5B
5
+ Train dataset: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet
6
+ Test dataset: data/big-math_tiny_instruct/test.parquet
7
+ Output directory: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct
8
+ SFT epochs: 2
9
+ SFT batch size: 4
10
+ GPU: 1
11
+ ==================================================
12
+ Resolved train dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet
13
+ Resolved test dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/data/big-math_tiny_instruct/test.parquet
14
+ Using 1 processes per node (GPUs: 1)
15
+
16
+ ==================================================
17
+ Step 1: Converting training data for SFT
18
+ ==================================================
19
+ Loaded 473 training examples from /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_10pct.parquet
20
+ Saved 473 SFT training examples to /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet
21
+ SFT training data saved to: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet
22
+
23
+ ==================================================
24
+ Step 2: Running SFT training
25
+ ==================================================
26
+ Executing: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/bin/torchrun --standalone --nnodes=1 --nproc_per_node=1 -m verl.trainer.fsdp_sft_trainer data.train_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet' data.val_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet' data.prompt_key=extra_info data.response_key=extra_info +data.prompt_dict_keys=['question'] +data.response_dict_keys=['answer'] data.micro_batch_size_per_gpu=4 data.train_batch_size=4 data.max_length=2048 data.truncation=right +data.apply_chat_template=false model.partial_pretrain='Qwen/Qwen2.5-1.5B' model.enable_gradient_checkpointing=True model.fsdp_config.cpu_offload=False model.fsdp_config.offload_params=False trainer.default_local_dir='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct' trainer.project_name=simple-sft trainer.experiment_name=qwen2.5-3b-simple-sft trainer.total_epochs=2 trainer.logger=['console','wandb'] trainer.default_hdfs_dir=null
27
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/transformers/utils/hub.py:111: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
28
+ warnings.warn(
29
+ INFO 09-22 05:48:36 [__init__.py:239] Automatically detected platform cuda.
30
+ Normalize batch size by dp 1
31
+ Using sequence parallel size: 1
32
+ Using remove padding: False
33
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
34
+ ls = ls[0]
35
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
36
+ ls = ls[0]
37
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
38
+ ls = ls[0]
39
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
40
+ ls = ls[0]
41
+ Using FSDP rank 0 and size 1 for data distribution
42
+ `torch_dtype` is deprecated! Use `dtype` instead!
43
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2ForCausalLM is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
44
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2Model is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
45
+ functools.partial(<function _or_policy at 0x7fabf6c392d0>, policies=[functools.partial(<function transformer_auto_wrap_policy at 0x7fabf6c391b0>, transformer_layer_cls={<class 'transformers.models.qwen2.modeling_qwen2.Qwen2DecoderLayer'>})])
46
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/torch/distributed/fsdp/_init_utils.py:444: UserWarning: FSDP is switching to use `NO_SHARD` instead of ShardingStrategy.FULL_SHARD since the world size is 1.
47
+ warnings.warn(
48
+ NCCL version 2.21.5+cuda12.4
49
+ Number of steps/epoch 118, number of epochs 2, total number of steps 236
50
+ {'data': {'train_batch_size': 4, 'micro_batch_size': None, 'micro_batch_size_per_gpu': 4, 'train_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet', 'val_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct/sft_training_data.parquet', 'prompt_key': 'extra_info', 'response_key': 'extra_info', 'max_length': 2048, 'truncation': 'right', 'balance_dp_token': False, 'num_workers': 8, 'chat_template': None, 'custom_cls': {'path': None}, 'prompt_dict_keys': ['question'], 'response_dict_keys': ['answer'], 'apply_chat_template': False}, 'model': {'partial_pretrain': 'Qwen/Qwen2.5-1.5B', 'fsdp_config': {'wrap_policy': {'min_num_params': 0}, 'cpu_offload': False, 'offload_params': False}, 'external_lib': None, 'enable_gradient_checkpointing': True, 'trust_remote_code': False, 'lora_rank': 0, 'lora_alpha': 16, 'target_modules': 'all-linear', 'use_liger': False}, 'optim': {'lr': 1e-05, 'betas': [0.9, 0.95], 'weight_decay': 0.01, 'warmup_steps_ratio': 0.1, 'clip_grad': 1.0}, 'ulysses_sequence_parallel_size': 1, 'use_remove_padding': False, 'trainer': {'default_local_dir': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_10pct', 'default_hdfs_dir': None, 'resume_path': None, 'project_name': 'simple-sft', 'experiment_name': 'qwen2.5-3b-simple-sft', 'total_epochs': 2, 'total_training_steps': None, 'logger': ['console', 'wandb'], 'seed': 1}}
20250922_054148/sft_results/filtered_top_20pct/gpu_lock_wrapper.sh ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # GPU Lock Wrapper for job: sft_filtered_top_20pct
3
+ # GPUs: [0]
4
+
5
+ # Function to cleanup GPU locks on exit
6
+ cleanup_locks() {
7
+ echo "Cleaning up GPU locks for job: sft_filtered_top_20pct"
8
+ python3 -c "
9
+ import sys
10
+ sys.path.append('/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl')
11
+ from utils.utils import release_gpu_lock
12
+ gpus = [0]
13
+ for gpu_id in gpus:
14
+ release_gpu_lock(str(gpu_id))
15
+ print(f'Released lock for GPU {gpu_id}')
16
+ "
17
+ }
18
+
19
+ # Set up trap to cleanup locks on script exit (success, failure, or interrupt)
20
+ trap cleanup_locks EXIT INT TERM
21
+
22
+ # Run the actual job
23
+ echo "Starting job: sft_filtered_top_20pct on GPUs: [0]"
24
+ CUDA_VISIBLE_DEVICES=0 bash checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/run_sft_script.sh
25
+ JOB_EXIT_CODE=$?
26
+
27
+ # Explicit cleanup (trap will also run, but this ensures it happens)
28
+ cleanup_locks
29
+
30
+ # Exit with the same code as the job
31
+ exit $JOB_EXIT_CODE
20250922_054148/sft_results/filtered_top_20pct/run_sft_script.sh ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # SFT Job - filtered_top_20pct
3
+ # Generated at: 20250922_054148
4
+
5
+ export TRANSFORMERS_CACHE=/home/greenland-user/verl/checkpoints/huggingface
6
+ export PYTHONHASHSEED=42
7
+
8
+ echo "Starting SFT training for filtered_top_20pct"
9
+ echo "Data file: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet"
10
+ echo "Output: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct"
11
+ echo "Model: Qwen/Qwen2.5-1.5B"
12
+
13
+ # Log experiment start
14
+ echo "Filter: filtered_top_20pct - Started at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
15
+
16
+ # Run SFT training
17
+ timeout 7200 bash scripts_cot/run_simple_sft.sh \
18
+ --model_path "Qwen/Qwen2.5-1.5B" \
19
+ --train_data_path "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet" \
20
+ --test_data_path "data/big-math_tiny_instruct/test.parquet" \
21
+ --output_dir "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct" \
22
+ --sft_epochs "2" \
23
+ --sft_batch_size "4" \
24
+ --gpu "$CUDA_VISIBLE_DEVICES" \
25
+ 2>&1 | tee "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/training_log.txt"
26
+
27
+ SFT_EXIT_CODE=$?
28
+
29
+ if [ $SFT_EXIT_CODE -eq 0 ]; then
30
+ echo "SFT completed successfully for filtered_top_20pct"
31
+ echo "Filter: filtered_top_20pct - Completed successfully at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
32
+ elif [ $SFT_EXIT_CODE -eq 124 ]; then
33
+ echo "SFT timed out for filtered_top_20pct (2 hour limit)"
34
+ echo "Filter: filtered_top_20pct - Timed out at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
35
+ else
36
+ echo "SFT failed for filtered_top_20pct with exit code $SFT_EXIT_CODE"
37
+ echo "Filter: filtered_top_20pct - Failed with exit code $SFT_EXIT_CODE at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
38
+ fi
39
+
40
+ echo "Completed SFT training for filtered_top_20pct"
20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:012da38b67d512890a27ac03e29e5029d344d89d3f47a9468dcee74ed3b3a573
3
+ size 185424
20250922_054148/sft_results/filtered_top_20pct/training_log.txt ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==================================================
2
+ Starting Simple SFT Training
3
+ ==================================================
4
+ Model: Qwen/Qwen2.5-1.5B
5
+ Train dataset: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet
6
+ Test dataset: data/big-math_tiny_instruct/test.parquet
7
+ Output directory: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct
8
+ SFT epochs: 2
9
+ SFT batch size: 4
10
+ GPU: 0
11
+ ==================================================
12
+ Resolved train dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet
13
+ Resolved test dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/data/big-math_tiny_instruct/test.parquet
14
+ Using 1 processes per node (GPUs: 0)
15
+
16
+ ==================================================
17
+ Step 1: Converting training data for SFT
18
+ ==================================================
19
+ Loaded 420 training examples from /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_20pct.parquet
20
+ Saved 420 SFT training examples to /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet
21
+ SFT training data saved to: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet
22
+
23
+ ==================================================
24
+ Step 2: Running SFT training
25
+ ==================================================
26
+ Executing: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/bin/torchrun --standalone --nnodes=1 --nproc_per_node=1 -m verl.trainer.fsdp_sft_trainer data.train_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet' data.val_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet' data.prompt_key=extra_info data.response_key=extra_info +data.prompt_dict_keys=['question'] +data.response_dict_keys=['answer'] data.micro_batch_size_per_gpu=4 data.train_batch_size=4 data.max_length=2048 data.truncation=right +data.apply_chat_template=false model.partial_pretrain='Qwen/Qwen2.5-1.5B' model.enable_gradient_checkpointing=True model.fsdp_config.cpu_offload=False model.fsdp_config.offload_params=False trainer.default_local_dir='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct' trainer.project_name=simple-sft trainer.experiment_name=qwen2.5-3b-simple-sft trainer.total_epochs=2 trainer.logger=['console','wandb'] trainer.default_hdfs_dir=null
27
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/transformers/utils/hub.py:111: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
28
+ warnings.warn(
29
+ INFO 09-22 05:48:36 [__init__.py:239] Automatically detected platform cuda.
30
+ Normalize batch size by dp 1
31
+ Using sequence parallel size: 1
32
+ Using remove padding: False
33
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
34
+ ls = ls[0]
35
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
36
+ ls = ls[0]
37
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
38
+ ls = ls[0]
39
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
40
+ ls = ls[0]
41
+ Using FSDP rank 0 and size 1 for data distribution
42
+ `torch_dtype` is deprecated! Use `dtype` instead!
43
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2ForCausalLM is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
44
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2Model is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
45
+ functools.partial(<function _or_policy at 0x7f34e63112d0>, policies=[functools.partial(<function transformer_auto_wrap_policy at 0x7f34e63111b0>, transformer_layer_cls={<class 'transformers.models.qwen2.modeling_qwen2.Qwen2DecoderLayer'>})])
46
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/torch/distributed/fsdp/_init_utils.py:444: UserWarning: FSDP is switching to use `NO_SHARD` instead of ShardingStrategy.FULL_SHARD since the world size is 1.
47
+ warnings.warn(
48
+ NCCL version 2.21.5+cuda12.4
49
+ Number of steps/epoch 105, number of epochs 2, total number of steps 210
50
+ {'data': {'train_batch_size': 4, 'micro_batch_size': None, 'micro_batch_size_per_gpu': 4, 'train_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet', 'val_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct/sft_training_data.parquet', 'prompt_key': 'extra_info', 'response_key': 'extra_info', 'max_length': 2048, 'truncation': 'right', 'balance_dp_token': False, 'num_workers': 8, 'chat_template': None, 'custom_cls': {'path': None}, 'prompt_dict_keys': ['question'], 'response_dict_keys': ['answer'], 'apply_chat_template': False}, 'model': {'partial_pretrain': 'Qwen/Qwen2.5-1.5B', 'fsdp_config': {'wrap_policy': {'min_num_params': 0}, 'cpu_offload': False, 'offload_params': False}, 'external_lib': None, 'enable_gradient_checkpointing': True, 'trust_remote_code': False, 'lora_rank': 0, 'lora_alpha': 16, 'target_modules': 'all-linear', 'use_liger': False}, 'optim': {'lr': 1e-05, 'betas': [0.9, 0.95], 'weight_decay': 0.01, 'warmup_steps_ratio': 0.1, 'clip_grad': 1.0}, 'ulysses_sequence_parallel_size': 1, 'use_remove_padding': False, 'trainer': {'default_local_dir': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_20pct', 'default_hdfs_dir': None, 'resume_path': None, 'project_name': 'simple-sft', 'experiment_name': 'qwen2.5-3b-simple-sft', 'total_epochs': 2, 'total_training_steps': None, 'logger': ['console', 'wandb'], 'seed': 1}}
20250922_054148/sft_results/filtered_top_50pct/gpu_lock_wrapper.sh ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # GPU Lock Wrapper for job: sft_filtered_top_50pct
3
+ # GPUs: [3]
4
+
5
+ # Function to cleanup GPU locks on exit
6
+ cleanup_locks() {
7
+ echo "Cleaning up GPU locks for job: sft_filtered_top_50pct"
8
+ python3 -c "
9
+ import sys
10
+ sys.path.append('/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl')
11
+ from utils.utils import release_gpu_lock
12
+ gpus = [3]
13
+ for gpu_id in gpus:
14
+ release_gpu_lock(str(gpu_id))
15
+ print(f'Released lock for GPU {gpu_id}')
16
+ "
17
+ }
18
+
19
+ # Set up trap to cleanup locks on script exit (success, failure, or interrupt)
20
+ trap cleanup_locks EXIT INT TERM
21
+
22
+ # Run the actual job
23
+ echo "Starting job: sft_filtered_top_50pct on GPUs: [3]"
24
+ CUDA_VISIBLE_DEVICES=3 bash checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/run_sft_script.sh
25
+ JOB_EXIT_CODE=$?
26
+
27
+ # Explicit cleanup (trap will also run, but this ensures it happens)
28
+ cleanup_locks
29
+
30
+ # Exit with the same code as the job
31
+ exit $JOB_EXIT_CODE
20250922_054148/sft_results/filtered_top_50pct/run_sft_script.sh ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # SFT Job - filtered_top_50pct
3
+ # Generated at: 20250922_054148
4
+
5
+ export TRANSFORMERS_CACHE=/home/greenland-user/verl/checkpoints/huggingface
6
+ export PYTHONHASHSEED=42
7
+
8
+ echo "Starting SFT training for filtered_top_50pct"
9
+ echo "Data file: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet"
10
+ echo "Output: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct"
11
+ echo "Model: Qwen/Qwen2.5-1.5B"
12
+
13
+ # Log experiment start
14
+ echo "Filter: filtered_top_50pct - Started at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
15
+
16
+ # Run SFT training
17
+ timeout 7200 bash scripts_cot/run_simple_sft.sh \
18
+ --model_path "Qwen/Qwen2.5-1.5B" \
19
+ --train_data_path "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet" \
20
+ --test_data_path "data/big-math_tiny_instruct/test.parquet" \
21
+ --output_dir "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct" \
22
+ --sft_epochs "2" \
23
+ --sft_batch_size "4" \
24
+ --gpu "$CUDA_VISIBLE_DEVICES" \
25
+ 2>&1 | tee "checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/training_log.txt"
26
+
27
+ SFT_EXIT_CODE=$?
28
+
29
+ if [ $SFT_EXIT_CODE -eq 0 ]; then
30
+ echo "SFT completed successfully for filtered_top_50pct"
31
+ echo "Filter: filtered_top_50pct - Completed successfully at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
32
+ elif [ $SFT_EXIT_CODE -eq 124 ]; then
33
+ echo "SFT timed out for filtered_top_50pct (2 hour limit)"
34
+ echo "Filter: filtered_top_50pct - Timed out at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
35
+ else
36
+ echo "SFT failed for filtered_top_50pct with exit code $SFT_EXIT_CODE"
37
+ echo "Filter: filtered_top_50pct - Failed with exit code $SFT_EXIT_CODE at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
38
+ fi
39
+
40
+ echo "Completed SFT training for filtered_top_50pct"
20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38c651033d6dd41e167c12f5e933b39eb0dad35ddc4741f277e17a3788a4bf78
3
+ size 122542
20250922_054148/sft_results/filtered_top_50pct/training_log.txt ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==================================================
2
+ Starting Simple SFT Training
3
+ ==================================================
4
+ Model: Qwen/Qwen2.5-1.5B
5
+ Train dataset: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet
6
+ Test dataset: data/big-math_tiny_instruct/test.parquet
7
+ Output directory: checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct
8
+ SFT epochs: 2
9
+ SFT batch size: 4
10
+ GPU: 3
11
+ ==================================================
12
+ Resolved train dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet
13
+ Resolved test dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/data/big-math_tiny_instruct/test.parquet
14
+ Using 1 processes per node (GPUs: 3)
15
+
16
+ ==================================================
17
+ Step 1: Converting training data for SFT
18
+ ==================================================
19
+ Loaded 263 training examples from /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/filtered_top_50pct.parquet
20
+ Saved 263 SFT training examples to /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet
21
+ SFT training data saved to: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet
22
+
23
+ ==================================================
24
+ Step 2: Running SFT training
25
+ ==================================================
26
+ Executing: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/bin/torchrun --standalone --nnodes=1 --nproc_per_node=1 -m verl.trainer.fsdp_sft_trainer data.train_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet' data.val_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet' data.prompt_key=extra_info data.response_key=extra_info +data.prompt_dict_keys=['question'] +data.response_dict_keys=['answer'] data.micro_batch_size_per_gpu=4 data.train_batch_size=4 data.max_length=2048 data.truncation=right +data.apply_chat_template=false model.partial_pretrain='Qwen/Qwen2.5-1.5B' model.enable_gradient_checkpointing=True model.fsdp_config.cpu_offload=False model.fsdp_config.offload_params=False trainer.default_local_dir='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct' trainer.project_name=simple-sft trainer.experiment_name=qwen2.5-3b-simple-sft trainer.total_epochs=2 trainer.logger=['console','wandb'] trainer.default_hdfs_dir=null
27
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/transformers/utils/hub.py:111: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
28
+ warnings.warn(
29
+ INFO 09-22 05:48:36 [__init__.py:239] Automatically detected platform cuda.
30
+ Normalize batch size by dp 1
31
+ Using sequence parallel size: 1
32
+ Using remove padding: False
33
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
34
+ ls = ls[0]
35
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
36
+ ls = ls[0]
37
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
38
+ ls = ls[0]
39
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
40
+ ls = ls[0]
41
+ Using FSDP rank 0 and size 1 for data distribution
42
+ `torch_dtype` is deprecated! Use `dtype` instead!
43
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2ForCausalLM is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
44
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2Model is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
45
+ functools.partial(<function _or_policy at 0x7fb0938292d0>, policies=[functools.partial(<function transformer_auto_wrap_policy at 0x7fb0938291b0>, transformer_layer_cls={<class 'transformers.models.qwen2.modeling_qwen2.Qwen2DecoderLayer'>})])
46
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/torch/distributed/fsdp/_init_utils.py:444: UserWarning: FSDP is switching to use `NO_SHARD` instead of ShardingStrategy.FULL_SHARD since the world size is 1.
47
+ warnings.warn(
48
+ NCCL version 2.21.5+cuda12.4
49
+ Number of steps/epoch 65, number of epochs 2, total number of steps 130
50
+ {'data': {'train_batch_size': 4, 'micro_batch_size': None, 'micro_batch_size_per_gpu': 4, 'train_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet', 'val_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct/sft_training_data.parquet', 'prompt_key': 'extra_info', 'response_key': 'extra_info', 'max_length': 2048, 'truncation': 'right', 'balance_dp_token': False, 'num_workers': 8, 'chat_template': None, 'custom_cls': {'path': None}, 'prompt_dict_keys': ['question'], 'response_dict_keys': ['answer'], 'apply_chat_template': False}, 'model': {'partial_pretrain': 'Qwen/Qwen2.5-1.5B', 'fsdp_config': {'wrap_policy': {'min_num_params': 0}, 'cpu_offload': False, 'offload_params': False}, 'external_lib': None, 'enable_gradient_checkpointing': True, 'trust_remote_code': False, 'lora_rank': 0, 'lora_alpha': 16, 'target_modules': 'all-linear', 'use_liger': False}, 'optim': {'lr': 1e-05, 'betas': [0.9, 0.95], 'weight_decay': 0.01, 'warmup_steps_ratio': 0.1, 'clip_grad': 1.0}, 'ulysses_sequence_parallel_size': 1, 'use_remove_padding': False, 'trainer': {'default_local_dir': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/filtered_top_50pct', 'default_hdfs_dir': None, 'resume_path': None, 'project_name': 'simple-sft', 'experiment_name': 'qwen2.5-3b-simple-sft', 'total_epochs': 2, 'total_training_steps': None, 'logger': ['console', 'wandb'], 'seed': 1}}
20250922_054148/sft_results/full_data/gpu_lock_wrapper.sh ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # GPU Lock Wrapper for job: sft_full_data
3
+ # GPUs: [2]
4
+
5
+ # Function to cleanup GPU locks on exit
6
+ cleanup_locks() {
7
+ echo "Cleaning up GPU locks for job: sft_full_data"
8
+ python3 -c "
9
+ import sys
10
+ sys.path.append('/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl')
11
+ from utils.utils import release_gpu_lock
12
+ gpus = [2]
13
+ for gpu_id in gpus:
14
+ release_gpu_lock(str(gpu_id))
15
+ print(f'Released lock for GPU {gpu_id}')
16
+ "
17
+ }
18
+
19
+ # Set up trap to cleanup locks on script exit (success, failure, or interrupt)
20
+ trap cleanup_locks EXIT INT TERM
21
+
22
+ # Run the actual job
23
+ echo "Starting job: sft_full_data on GPUs: [2]"
24
+ CUDA_VISIBLE_DEVICES=2 bash checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/run_sft_script.sh
25
+ JOB_EXIT_CODE=$?
26
+
27
+ # Explicit cleanup (trap will also run, but this ensures it happens)
28
+ cleanup_locks
29
+
30
+ # Exit with the same code as the job
31
+ exit $JOB_EXIT_CODE
20250922_054148/sft_results/full_data/run_sft_script.sh ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # SFT Job - full_data
3
+ # Generated at: 20250922_054148
4
+
5
+ export TRANSFORMERS_CACHE=/home/greenland-user/verl/checkpoints/huggingface
6
+ export PYTHONHASHSEED=42
7
+
8
+ echo "Starting SFT training for full_data"
9
+ echo "Data file: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet"
10
+ echo "Output: checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data"
11
+ echo "Model: Qwen/Qwen2.5-1.5B"
12
+
13
+ # Log experiment start
14
+ echo "Filter: full_data - Started at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
15
+
16
+ # Run SFT training
17
+ timeout 7200 bash scripts_cot/run_simple_sft.sh \
18
+ --model_path "Qwen/Qwen2.5-1.5B" \
19
+ --train_data_path "checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet" \
20
+ --test_data_path "data/big-math_tiny_instruct/test.parquet" \
21
+ --output_dir "checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data" \
22
+ --sft_epochs "2" \
23
+ --sft_batch_size "4" \
24
+ --gpu "$CUDA_VISIBLE_DEVICES" \
25
+ 2>&1 | tee "checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/training_log.txt"
26
+
27
+ SFT_EXIT_CODE=$?
28
+
29
+ if [ $SFT_EXIT_CODE -eq 0 ]; then
30
+ echo "SFT completed successfully for full_data"
31
+ echo "Filter: full_data - Completed successfully at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
32
+ elif [ $SFT_EXIT_CODE -eq 124 ]; then
33
+ echo "SFT timed out for full_data (2 hour limit)"
34
+ echo "Filter: full_data - Timed out at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
35
+ else
36
+ echo "SFT failed for full_data with exit code $SFT_EXIT_CODE"
37
+ echo "Filter: full_data - Failed with exit code $SFT_EXIT_CODE at $(date)" >> "checkpoints/auc_filtered_sft/20250922_054148/experiment_log.txt"
38
+ fi
39
+
40
+ echo "Completed SFT training for full_data"
20250922_054148/sft_results/full_data/sft_training_data.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eec728d52704a21a2fc0dae93e65185fc5c5ae244e991b8c56759422a86d559
3
+ size 224736
20250922_054148/sft_results/full_data/training_log.txt ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==================================================
2
+ Starting Simple SFT Training
3
+ ==================================================
4
+ Model: Qwen/Qwen2.5-1.5B
5
+ Train dataset: checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet
6
+ Test dataset: data/big-math_tiny_instruct/test.parquet
7
+ Output directory: checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data
8
+ SFT epochs: 2
9
+ SFT batch size: 4
10
+ GPU: 2
11
+ ==================================================
12
+ Resolved train dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet
13
+ Resolved test dataset path: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/data/big-math_tiny_instruct/test.parquet
14
+ Using 1 processes per node (GPUs: 2)
15
+
16
+ ==================================================
17
+ Step 1: Converting training data for SFT
18
+ ==================================================
19
+ Loaded 525 training examples from /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/filtered_data/full_data.parquet
20
+ Saved 525 SFT training examples to /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet
21
+ SFT training data saved to: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet
22
+
23
+ ==================================================
24
+ Step 2: Running SFT training
25
+ ==================================================
26
+ Executing: /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/bin/torchrun --standalone --nnodes=1 --nproc_per_node=1 -m verl.trainer.fsdp_sft_trainer data.train_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet' data.val_files='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet' data.prompt_key=extra_info data.response_key=extra_info +data.prompt_dict_keys=['question'] +data.response_dict_keys=['answer'] data.micro_batch_size_per_gpu=4 data.train_batch_size=4 data.max_length=2048 data.truncation=right +data.apply_chat_template=false model.partial_pretrain='Qwen/Qwen2.5-1.5B' model.enable_gradient_checkpointing=True model.fsdp_config.cpu_offload=False model.fsdp_config.offload_params=False trainer.default_local_dir='/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data' trainer.project_name=simple-sft trainer.experiment_name=qwen2.5-3b-simple-sft trainer.total_epochs=2 trainer.logger=['console','wandb'] trainer.default_hdfs_dir=null
27
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/transformers/utils/hub.py:111: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
28
+ warnings.warn(
29
+ INFO 09-22 05:48:36 [__init__.py:239] Automatically detected platform cuda.
30
+ Normalize batch size by dp 1
31
+ Using sequence parallel size: 1
32
+ Using remove padding: False
33
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
34
+ ls = ls[0]
35
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
36
+ ls = ls[0]
37
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
38
+ ls = ls[0]
39
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/verl/verl/utils/dataset/sft_dataset.py:83: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
40
+ ls = ls[0]
41
+ Using FSDP rank 0 and size 1 for data distribution
42
+ `torch_dtype` is deprecated! Use `dtype` instead!
43
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2ForCausalLM is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
44
+ Flash Attention 2 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in Qwen2Model is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", dtype=torch.float16)`
45
+ functools.partial(<function _or_policy at 0x7f760f2312d0>, policies=[functools.partial(<function transformer_auto_wrap_policy at 0x7f760f2311b0>, transformer_layer_cls={<class 'transformers.models.qwen2.modeling_qwen2.Qwen2DecoderLayer'>})])
46
+ /fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/miniconda3/envs/verl_persistent/lib/python3.10/site-packages/torch/distributed/fsdp/_init_utils.py:444: UserWarning: FSDP is switching to use `NO_SHARD` instead of ShardingStrategy.FULL_SHARD since the world size is 1.
47
+ warnings.warn(
48
+ NCCL version 2.21.5+cuda12.4
49
+ Number of steps/epoch 131, number of epochs 2, total number of steps 262
50
+ {'data': {'train_batch_size': 4, 'micro_batch_size': None, 'micro_batch_size_per_gpu': 4, 'train_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet', 'val_files': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data/sft_training_data.parquet', 'prompt_key': 'extra_info', 'response_key': 'extra_info', 'max_length': 2048, 'truncation': 'right', 'balance_dp_token': False, 'num_workers': 8, 'chat_template': None, 'custom_cls': {'path': None}, 'prompt_dict_keys': ['question'], 'response_dict_keys': ['answer'], 'apply_chat_template': False}, 'model': {'partial_pretrain': 'Qwen/Qwen2.5-1.5B', 'fsdp_config': {'wrap_policy': {'min_num_params': 0}, 'cpu_offload': False, 'offload_params': False}, 'external_lib': None, 'enable_gradient_checkpointing': True, 'trust_remote_code': False, 'lora_rank': 0, 'lora_alpha': 16, 'target_modules': 'all-linear', 'use_liger': False}, 'optim': {'lr': 1e-05, 'betas': [0.9, 0.95], 'weight_decay': 0.01, 'warmup_steps_ratio': 0.1, 'clip_grad': 1.0}, 'ulysses_sequence_parallel_size': 1, 'use_remove_padding': False, 'trainer': {'default_local_dir': '/fsx-sfai/dedicated-fsx-data-repo-pretraining-gl-ap-south-1/shared_experiments/wxpeng/checkpoints/auc_filtered_sft/20250922_054148/sft_results/full_data', 'default_hdfs_dir': None, 'resume_path': None, 'project_name': 'simple-sft', 'experiment_name': 'qwen2.5-3b-simple-sft', 'total_epochs': 2, 'total_training_steps': None, 'logger': ['console', 'wandb'], 'seed': 1}}