lewtun HF Staff commited on
Commit
70f0381
·
1 Parent(s): 18af3b0

Clean up false PoT

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_hard_mmos/results_2024-05-24T09-30-06.964431.json +0 -91
  2. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_hard_pot/results_2024-05-27T14-37-04.090977.json +0 -193
  3. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_medium_mmos/results_2024-05-24T09-29-31.249987.json +0 -91
  4. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_medium_pot/results_2024-05-27T14-31-39.435993.json +0 -193
  5. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_mmos/results_2024-05-24T09-38-01.891596.json +0 -91
  6. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T12-10-10.985244.json +0 -91
  7. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T12-38-20.637098.json +0 -91
  8. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T13-19-13.846655.json +0 -193
  9. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_mmos/results_2024-05-24T09-36-20.897469.json +0 -91
  10. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T12-08-02.630027.json +0 -91
  11. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T12-36-15.235885.json +0 -91
  12. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T13-10-03.543903.json +0 -193
  13. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_hard_mmos/results_2024-05-24T09-36-06.066222.json +0 -91
  14. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_hard_pot/results_2024-05-27T14-35-43.671086.json +0 -193
  15. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_medium_mmos/results_2024-05-24T09-36-32.170044.json +0 -91
  16. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_medium_pot/results_2024-05-27T14-31-11.406681.json +0 -193
  17. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_hard_mmos/results_2024-05-24T09-48-10.979290.json +0 -91
  18. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_hard_pot/results_2024-05-27T14-35-53.533059.json +0 -193
  19. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_medium_mmos/results_2024-05-24T09-48-12.415661.json +0 -91
  20. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_medium_pot/results_2024-05-27T14-30-36.611690.json +0 -193
  21. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.13/aimo_kaggle_hard_mmos/results_2024-05-24T10-07-40.764237.json +0 -91
  22. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.13/aimo_kaggle_medium_mmos/results_2024-05-24T10-07-11.135359.json +0 -91
  23. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.14/aimo_kaggle_hard_mmos/results_2024-05-24T10-16-49.966984.json +0 -91
  24. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.14/aimo_kaggle_medium_mmos/results_2024-05-24T10-16-38.488727.json +0 -91
  25. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.15/aimo_kaggle_hard_mmos/results_2024-05-24T10-25-09.540279.json +0 -91
  26. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.15/aimo_kaggle_medium_mmos/results_2024-05-24T10-25-14.707607.json +0 -91
  27. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.01/aimo_kaggle_hard_pot/results_2024-05-27T22-52-32.480755.json +0 -193
  28. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.01/aimo_kaggle_medium_pot/results_2024-05-27T22-49-32.235858.json +0 -193
  29. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.02/aimo_kaggle_hard_pot/results_2024-05-27T23-01-26.617019.json +0 -193
  30. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.02/aimo_kaggle_medium_pot/results_2024-05-27T22-56-40.113739.json +0 -193
  31. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.03/aimo_kaggle_hard_pot/results_2024-05-27T22-50-33.482970.json +0 -193
  32. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.03/aimo_kaggle_medium_pot/results_2024-05-27T22-46-20.885308.json +0 -193
  33. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.04/aimo_kaggle_hard_pot/results_2024-05-27T23-02-16.395548.json +0 -193
  34. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.04/aimo_kaggle_medium_pot/results_2024-05-27T22-56-39.512900.json +0 -193
  35. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.05/aimo_kaggle_hard_pot/results_2024-05-27T22-47-26.728707.json +0 -193
  36. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.05/aimo_kaggle_medium_pot/results_2024-05-27T22-42-48.691321.json +0 -193
  37. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.06/aimo_kaggle_hard_pot/results_2024-05-27T23-22-45.715568.json +0 -193
  38. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.06/aimo_kaggle_medium_pot/results_2024-05-27T23-16-30.855912.json +0 -193
  39. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.07/aimo_kaggle_hard_pot/results_2024-05-27T22-43-29.392522.json +0 -193
  40. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.07/aimo_kaggle_medium_pot/results_2024-05-27T22-39-34.119815.json +0 -193
  41. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.08/aimo_kaggle_hard_pot/results_2024-05-27T22-45-02.958180.json +0 -193
  42. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.08/aimo_kaggle_medium_pot/results_2024-05-27T22-39-55.232091.json +0 -193
  43. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.09/aimo_kaggle_hard_pot/results_2024-05-27T23-01-23.916419.json +0 -193
  44. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.09/aimo_kaggle_medium_pot/results_2024-05-27T22-56-49.561125.json +0 -193
  45. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.10/aimo_kaggle_hard_pot/results_2024-05-27T22-56-26.205133.json +0 -193
  46. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.10/aimo_kaggle_medium_pot/results_2024-05-27T22-52-23.993586.json +0 -193
  47. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.11/aimo_kaggle_hard_pot/results_2024-05-27T23-28-10.769733.json +0 -193
  48. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.11/aimo_kaggle_medium_pot/results_2024-05-27T23-24-01.746924.json +0 -193
  49. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.12/aimo_kaggle_hard_pot/results_2024-05-27T22-58-24.076261.json +0 -193
  50. eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.12/aimo_kaggle_medium_pot/results_2024-05-27T22-54-41.166514.json +0 -193
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_hard_mmos/results_2024-05-24T09-30-06.964431.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 857446.223097524,
9
- "end_time": 857817.902641332,
10
- "total_evaluation_time_secondes": "371.6795438079862",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "e9f1f0034b3f35397651713c744779b27c7b0d2f",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.027994168488950606
21
- },
22
- "all": {
23
- "qem": 0.04,
24
- "qem_stderr": 0.027994168488950606
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "52b2b59de598fdd3"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "cc9bd188ea6976bc"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_hard_pot/results_2024-05-27T14-37-04.090977.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1143715.699540001,
9
- "end_time": 1144501.128154096,
10
- "total_evaluation_time_secondes": "785.4286140948534",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "e9f1f0034b3f35397651713c744779b27c7b0d2f",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.06,
20
- "qem_stderr": 0.0339266916772512
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.48,
24
- "qem_stderr": 0.07137140569598172
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.56,
28
- "qem_stderr": 0.07091242083423345
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3666666666666667,
32
- "qem_stderr": 0.058736839402488784
33
- },
34
- "all": {
35
- "qem": 0.3666666666666667,
36
- "qem_stderr": 0.058736839402488784
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "b73fa39a0f0c8082"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "1b2f354c5d65f92d"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "ddd3e2af9db6590b"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "a225423cb032047d"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_medium_mmos/results_2024-05-24T09-29-31.249987.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 727962.553320119,
9
- "end_time": 728298.349595936,
10
- "total_evaluation_time_secondes": "335.79627581697423",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "e9f1f0034b3f35397651713c744779b27c7b0d2f",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.075,
20
- "qem_stderr": 0.04217636961434869
21
- },
22
- "all": {
23
- "qem": 0.075,
24
- "qem_stderr": 0.04217636961434869
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "60bf7b05082d037a"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "572a44bb3cfcb49a"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.09/aimo_kaggle_medium_pot/results_2024-05-27T14-31-39.435993.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 438052.255329886,
9
- "end_time": 438511.635606925,
10
- "total_evaluation_time_secondes": "459.3802770390175",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "e9f1f0034b3f35397651713c744779b27c7b0d2f",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.125,
20
- "qem_stderr": 0.05295740910852021
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.15,
24
- "qem_stderr": 0.05717718748968655
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.2,
28
- "qem_stderr": 0.06405126152203487
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.15833333333333335,
32
- "qem_stderr": 0.05806195270674721
33
- },
34
- "all": {
35
- "qem": 0.15833333333333335,
36
- "qem_stderr": 0.05806195270674721
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "d3d0c2f5914eea85"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "70097c50fd00a4cd"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "a052161a23e608a2"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "480ba89e1c3b08f4"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_mmos/results_2024-05-24T09-38-01.891596.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2115203.534425676,
9
- "end_time": 2115616.753860881,
10
- "total_evaluation_time_secondes": "413.21943520521745",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.027994168488950612
21
- },
22
- "all": {
23
- "qem": 0.04,
24
- "qem_stderr": 0.027994168488950612
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "70ab15813457a0e7"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "932433668b9d10d1"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T12-10-10.985244.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 4049391.660600598,
9
- "end_time": 4049852.921236317,
10
- "total_evaluation_time_secondes": "461.26063571916893",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.08,
20
- "qem_stderr": 0.038756171332144415
21
- },
22
- "all": {
23
- "qem": 0.08,
24
- "qem_stderr": 0.038756171332144415
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_pot:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_pot:v0": {
32
- "name": "aimo_kaggle_hard_pot:v0",
33
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code_and_math"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_pot:v0|0": {
64
- "hashes": {
65
- "hash_examples": "303213a38d9f7512",
66
- "hash_full_prompts": "3c670fa0e80bc301",
67
- "hash_input_tokens": "69dabd3aaebed332",
68
- "hash_cont_tokens": "38540998e511c7fa"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "cae48890a7c47904",
81
- "hash_full_prompts": "6f120f8d8bb938e6",
82
- "hash_input_tokens": "ff7094ff98f5f808",
83
- "hash_cont_tokens": "d081df183e2788f4"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T12-38-20.637098.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 9223358.910966022,
9
- "end_time": 9223818.340137284,
10
- "total_evaluation_time_secondes": "459.42917126230896",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.08,
20
- "qem_stderr": 0.038756171332144415
21
- },
22
- "all": {
23
- "qem": 0.08,
24
- "qem_stderr": 0.038756171332144415
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_pot:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_pot:v0": {
32
- "name": "aimo_kaggle_hard_pot:v0",
33
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code_and_math"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_pot:v0|0": {
64
- "hashes": {
65
- "hash_examples": "303213a38d9f7512",
66
- "hash_full_prompts": "3c670fa0e80bc301",
67
- "hash_input_tokens": "69dabd3aaebed332",
68
- "hash_cont_tokens": "38540998e511c7fa"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "cae48890a7c47904",
81
- "hash_full_prompts": "6f120f8d8bb938e6",
82
- "hash_input_tokens": "ff7094ff98f5f808",
83
- "hash_cont_tokens": "d081df183e2788f4"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_hard_pot/results_2024-05-27T13-19-13.846655.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2072215.893674265,
9
- "end_time": 2073130.940811355,
10
- "total_evaluation_time_secondes": "915.047137090005",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.48,
24
- "qem_stderr": 0.07137140569598172
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.58,
28
- "qem_stderr": 0.07050835816716033
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3666666666666667,
32
- "qem_stderr": 0.056624644117364224
33
- },
34
- "all": {
35
- "qem": 0.3666666666666667,
36
- "qem_stderr": 0.056624644117364224
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "1847a73ee97b84cf"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "07a33d50dbb61bc0"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "81be89d98a712a48"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "be0f2fadf1d23c89"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_mmos/results_2024-05-24T09-36-20.897469.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2117005.70480802,
9
- "end_time": 2117317.858500111,
10
- "total_evaluation_time_secondes": "312.15369209088385",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.175,
20
- "qem_stderr": 0.060843430844447585
21
- },
22
- "all": {
23
- "qem": 0.175,
24
- "qem_stderr": 0.060843430844447585
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "dcf1a76b60e91789"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "4e52fc07aed5d27d"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T12-08-02.630027.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2108115.015177626,
9
- "end_time": 2108455.320385662,
10
- "total_evaluation_time_secondes": "340.3052080357447",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.175,
20
- "qem_stderr": 0.06084343084444758
21
- },
22
- "all": {
23
- "qem": 0.175,
24
- "qem_stderr": 0.06084343084444758
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_pot:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_pot:v0": {
32
- "name": "aimo_kaggle_medium_pot:v0",
33
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code_and_math"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_pot:v0|0": {
64
- "hashes": {
65
- "hash_examples": "2799c24461029dc3",
66
- "hash_full_prompts": "de6572b914d5bba6",
67
- "hash_input_tokens": "bb5427e5227cc6b4",
68
- "hash_cont_tokens": "d8fff7d9380f6289"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "c72202b0a18ef5f9",
81
- "hash_full_prompts": "a0429b0dea3e33db",
82
- "hash_input_tokens": "8e6ef9684e80b71f",
83
- "hash_cont_tokens": "4b3f922cf74d16d1"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T12-36-15.235885.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2106210.420319901,
9
- "end_time": 2106551.181892837,
10
- "total_evaluation_time_secondes": "340.7615729360841",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.175,
20
- "qem_stderr": 0.06084343084444758
21
- },
22
- "all": {
23
- "qem": 0.175,
24
- "qem_stderr": 0.06084343084444758
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_pot:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_pot:v0": {
32
- "name": "aimo_kaggle_medium_pot:v0",
33
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code_and_math"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_pot:v0|0": {
64
- "hashes": {
65
- "hash_examples": "2799c24461029dc3",
66
- "hash_full_prompts": "de6572b914d5bba6",
67
- "hash_input_tokens": "bb5427e5227cc6b4",
68
- "hash_cont_tokens": "d8fff7d9380f6289"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "c72202b0a18ef5f9",
81
- "hash_full_prompts": "a0429b0dea3e33db",
82
- "hash_input_tokens": "8e6ef9684e80b71f",
83
- "hash_cont_tokens": "4b3f922cf74d16d1"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.10/aimo_kaggle_medium_pot/results_2024-05-27T13-10-03.543903.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 9225308.278480008,
9
- "end_time": 9225721.246559735,
10
- "total_evaluation_time_secondes": "412.96807972714305",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "61a9aad6e4c0fc64852cc86f9381183b734be0e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.175,
20
- "qem_stderr": 0.060843430844447585
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.15,
24
- "qem_stderr": 0.05717718748968655
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.225,
28
- "qem_stderr": 0.06686668711812965
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.18333333333333332,
32
- "qem_stderr": 0.06162910181742126
33
- },
34
- "all": {
35
- "qem": 0.18333333333333332,
36
- "qem_stderr": 0.06162910181742126
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "7bb95d2c217c1b5c"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "6c0f57c20d743daf"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "8adfed93bcb46450"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "bd6abfe65da8351f"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_hard_mmos/results_2024-05-24T09-36-06.066222.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1870349.343120157,
9
- "end_time": 1870680.547112474,
10
- "total_evaluation_time_secondes": "331.203992316965",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "0b525f62d709e3bb0fed36d4055acadaeffb2fea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "747646d721b689b3"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "6df83d700244dcf8"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_hard_pot/results_2024-05-27T14-35-43.671086.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 342864.500085301,
9
- "end_time": 343553.631159885,
10
- "total_evaluation_time_secondes": "689.1310745839728",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "0b525f62d709e3bb0fed36d4055acadaeffb2fea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.22,
24
- "qem_stderr": 0.05917804336345136
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.5,
28
- "qem_stderr": 0.07142857142857142
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.25333333333333335,
32
- "qem_stderr": 0.05286692776032447
33
- },
34
- "all": {
35
- "qem": 0.25333333333333335,
36
- "qem_stderr": 0.05286692776032447
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "954b24cc2bdc4e6c"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "3a75a70b0409cc6d"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "277be9de6dc4234e"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "5297800052500dc2"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_medium_mmos/results_2024-05-24T09-36-32.170044.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2420257.809069017,
9
- "end_time": 2420609.398094652,
10
- "total_evaluation_time_secondes": "351.58902563527226",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "0b525f62d709e3bb0fed36d4055acadaeffb2fea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "e0e2d443b31f87e8"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "67c15a3d33fd85be"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.11/aimo_kaggle_medium_pot/results_2024-05-27T14-31-11.406681.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 277940.525956996,
9
- "end_time": 278355.570149596,
10
- "total_evaluation_time_secondes": "415.0441926000058",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "0b525f62d709e3bb0fed36d4055acadaeffb2fea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.1,
20
- "qem_stderr": 0.04803844614152612
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.1,
24
- "qem_stderr": 0.04803844614152612
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.125,
28
- "qem_stderr": 0.05295740910852021
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.10833333333333334,
32
- "qem_stderr": 0.04967810046385748
33
- },
34
- "all": {
35
- "qem": 0.10833333333333334,
36
- "qem_stderr": 0.04967810046385748
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "8aca826f529dccef"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "2beb929a9d1f02a0"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "547712b40f6f642f"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "7494ce6ef42cbe4e"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_hard_mmos/results_2024-05-24T09-48-10.979290.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1871072.599385947,
9
- "end_time": 1871405.4602378,
10
- "total_evaluation_time_secondes": "332.86085185292177",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "feac1ee8323888504974e4c56f4dd29dec9f5eea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "6431505c900c5db1"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "6b37585ba81ea148"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_hard_pot/results_2024-05-27T14-35-53.533059.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 9230225.89268124,
9
- "end_time": 9230871.207996167,
10
- "total_evaluation_time_secondes": "645.3153149280697",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "feac1ee8323888504974e4c56f4dd29dec9f5eea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.02,
20
- "qem_stderr": 0.02
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.06,
24
- "qem_stderr": 0.033926691677251195
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.42,
28
- "qem_stderr": 0.07050835816716035
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.16666666666666666,
32
- "qem_stderr": 0.041478349948137185
33
- },
34
- "all": {
35
- "qem": 0.16666666666666666,
36
- "qem_stderr": 0.041478349948137185
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "dc39a3565f517db9"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "318061967e30b047"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "a2d575236416e3ca"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "9a548ef00ebb9273"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_medium_mmos/results_2024-05-24T09-48-12.415661.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2115893.229903704,
9
- "end_time": 2116227.277975012,
10
- "total_evaluation_time_secondes": "334.0480713080615",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "feac1ee8323888504974e4c56f4dd29dec9f5eea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "22010b252d6c7aee"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "97d4e941c9cdda6b"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.12/aimo_kaggle_medium_pot/results_2024-05-27T14-30-36.611690.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 4057927.003502535,
9
- "end_time": 4058278.547280695,
10
- "total_evaluation_time_secondes": "351.54377815965563",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "feac1ee8323888504974e4c56f4dd29dec9f5eea",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.025,
20
- "qem_stderr": 0.024999999999999998
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.15,
28
- "qem_stderr": 0.05717718748968655
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.05833333333333333,
32
- "qem_stderr": 0.027392395829895517
33
- },
34
- "all": {
35
- "qem": 0.05833333333333333,
36
- "qem_stderr": 0.027392395829895517
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "dea1ea4cd88ac25c"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "c30f828e6479e03f"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "4631fc712773916c"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "f61db1e864829049"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.13/aimo_kaggle_hard_mmos/results_2024-05-24T10-07-40.764237.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1060.262766582,
9
- "end_time": 1408.920575274,
10
- "total_evaluation_time_secondes": "348.6578086919999",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "f2387bf0ced9308beb238e535b4d8334e7487ab2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "4f6b08399af03ffd"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "575c9bad78c37155"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.13/aimo_kaggle_medium_mmos/results_2024-05-24T10-07-11.135359.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1251998.579556463,
9
- "end_time": 1252318.380801528,
10
- "total_evaluation_time_secondes": "319.8012450649403",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "f2387bf0ced9308beb238e535b4d8334e7487ab2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "cb917267759cd0c9"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "b39be9d73f59030c"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.14/aimo_kaggle_hard_mmos/results_2024-05-24T10-16-49.966984.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1579.416964497,
9
- "end_time": 1958.123144323,
10
- "total_evaluation_time_secondes": "378.7061798259999",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "868636f68df95b6db35b4113753003b88836a86a",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.02,
20
- "qem_stderr": 0.02
21
- },
22
- "all": {
23
- "qem": 0.02,
24
- "qem_stderr": 0.02
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "0592c8e0d76ec685"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "6cc5093a5dfe5f31"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.14/aimo_kaggle_medium_mmos/results_2024-05-24T10-16-38.488727.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 1252518.683488359,
9
- "end_time": 1252885.734003984,
10
- "total_evaluation_time_secondes": "367.05051562492736",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "868636f68df95b6db35b4113753003b88836a86a",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.075,
20
- "qem_stderr": 0.042176369614348695
21
- },
22
- "all": {
23
- "qem": 0.075,
24
- "qem_stderr": 0.042176369614348695
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "5bd7b16ad1cce5f7"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "32fd13771d23980f"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.15/aimo_kaggle_hard_mmos/results_2024-05-24T10-25-09.540279.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 466.001152748,
9
- "end_time": 788.580502107,
10
- "total_evaluation_time_secondes": "322.57934935900005",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "c599676fdc00a821d271b74201c28a69b8dae2e7",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_hard_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_hard_mmos:v0": {
32
- "name": "aimo_kaggle_hard_mmos:v0",
33
- "prompt_function": "mmos_hard_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 50,
56
- "effective_num_docs": 50,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_hard_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "b40b6a493a95bf77",
66
- "hash_full_prompts": "bf4ae1359925ab3b",
67
- "hash_input_tokens": "5f11181c1839f9c3",
68
- "hash_cont_tokens": "d220e4e743de8a9a"
69
- },
70
- "truncated": 50,
71
- "non_truncated": 0,
72
- "padded": 35,
73
- "non_padded": 15,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "79dbebcff6acad9e",
81
- "hash_full_prompts": "86a15c8410a494ef",
82
- "hash_input_tokens": "28eff78f8cb1ac9f",
83
- "hash_cont_tokens": "426a138c8b85ec49"
84
- },
85
- "truncated": 50,
86
- "non_truncated": 0,
87
- "padded": 35,
88
- "non_padded": 15,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v01.15/aimo_kaggle_medium_mmos/results_2024-05-24T10-25-14.707607.json DELETED
@@ -1,91 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2423204.564062136,
9
- "end_time": 2423531.935635801,
10
- "total_evaluation_time_secondes": "327.37157366471365",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "c599676fdc00a821d271b74201c28a69b8dae2e7",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_mmos:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "all": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- }
26
- },
27
- "versions": {
28
- "custom|aimo_kaggle_medium_mmos:v0|0": 0
29
- },
30
- "config_tasks": {
31
- "custom|aimo_kaggle_medium_mmos:v0": {
32
- "name": "aimo_kaggle_medium_mmos:v0",
33
- "prompt_function": "mmos_medium_prompt_fn",
34
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
35
- "hf_subset": "v0",
36
- "metric": [
37
- "quasi_exact_match_code"
38
- ],
39
- "hf_avail_splits": [
40
- "train"
41
- ],
42
- "evaluation_splits": [
43
- "train"
44
- ],
45
- "few_shots_split": null,
46
- "few_shots_select": null,
47
- "generation_size": 2048,
48
- "stop_sequence": null,
49
- "output_regex": null,
50
- "num_samples": null,
51
- "frozen": false,
52
- "suite": [
53
- "custom"
54
- ],
55
- "original_num_docs": 40,
56
- "effective_num_docs": 40,
57
- "trust_dataset": null,
58
- "must_remove_duplicate_docs": null,
59
- "version": 0
60
- }
61
- },
62
- "summary_tasks": {
63
- "custom|aimo_kaggle_medium_mmos:v0|0": {
64
- "hashes": {
65
- "hash_examples": "3401efda8b0cbcb5",
66
- "hash_full_prompts": "9553e2c8c7c87406",
67
- "hash_input_tokens": "4f9c9b13fc9c2736",
68
- "hash_cont_tokens": "b3bfc1c6165ee81c"
69
- },
70
- "truncated": 40,
71
- "non_truncated": 0,
72
- "padded": 31,
73
- "non_padded": 9,
74
- "effective_few_shots": 0.0,
75
- "num_truncated_few_shots": 0
76
- }
77
- },
78
- "summary_general": {
79
- "hashes": {
80
- "hash_examples": "4c81d27cfdb9d737",
81
- "hash_full_prompts": "466bf00fd3e3d325",
82
- "hash_input_tokens": "f3be8baebf6e5c0d",
83
- "hash_cont_tokens": "6e3f84e0c25bbd6c"
84
- },
85
- "truncated": 40,
86
- "non_truncated": 0,
87
- "padded": 31,
88
- "non_padded": 9,
89
- "num_truncated_few_shots": 0
90
- }
91
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.01/aimo_kaggle_hard_pot/results_2024-05-27T22-52-32.480755.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 10719.748736187,
9
- "end_time": 11433.49436362,
10
- "total_evaluation_time_secondes": "713.7456274329998",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "ab83077ac83ffbbb8abe282b3dd8f4228cbbdcba",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.06,
20
- "qem_stderr": 0.0339266916772512
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.38,
24
- "qem_stderr": 0.06934092056863767
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.5,
28
- "qem_stderr": 0.07142857142857142
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3133333333333333,
32
- "qem_stderr": 0.0582320612248201
33
- },
34
- "all": {
35
- "qem": 0.3133333333333333,
36
- "qem_stderr": 0.0582320612248201
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "c2831c3f8adfd2a5"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "e37a489b906d968e"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "43a13ab9e5e2cd6c"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "9ca954caaded26b5"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.01/aimo_kaggle_medium_pot/results_2024-05-27T22-49-32.235858.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 41425.043779027,
9
- "end_time": 41783.252532695,
10
- "total_evaluation_time_secondes": "358.20875366799737",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "ab83077ac83ffbbb8abe282b3dd8f4228cbbdcba",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.15,
20
- "qem_stderr": 0.05717718748968655
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.225,
24
- "qem_stderr": 0.06686668711812965
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.2,
28
- "qem_stderr": 0.06405126152203487
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.19166666666666665,
32
- "qem_stderr": 0.06269837870995036
33
- },
34
- "all": {
35
- "qem": 0.19166666666666665,
36
- "qem_stderr": 0.06269837870995036
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "129f73f798992c2d"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "c213703222b7b9d6"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "710c61a10724eb58"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "46841f1d968bb396"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.02/aimo_kaggle_hard_pot/results_2024-05-27T23-01-26.617019.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 9260481.786180781,
9
- "end_time": 9261204.31955451,
10
- "total_evaluation_time_secondes": "722.5333737283945",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "c579155dded92356530d24203be65dd06b745a34",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.06,
20
- "qem_stderr": 0.0339266916772512
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.36,
24
- "qem_stderr": 0.06857142857142856
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.5,
28
- "qem_stderr": 0.07142857142857142
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.30666666666666664,
32
- "qem_stderr": 0.057975563892417065
33
- },
34
- "all": {
35
- "qem": 0.30666666666666664,
36
- "qem_stderr": 0.057975563892417065
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "9aabf9d4ff47b254"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "9bf8175c55528051"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "3f123f6b38b6ffd5"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "afa8c82042b153d3"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.02/aimo_kaggle_medium_pot/results_2024-05-27T22-56-40.113739.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2107342.128751614,
9
- "end_time": 2107777.207887237,
10
- "total_evaluation_time_secondes": "435.0791356228292",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "c579155dded92356530d24203be65dd06b745a34",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.15,
20
- "qem_stderr": 0.05717718748968655
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.15,
24
- "qem_stderr": 0.05717718748968655
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.15,
28
- "qem_stderr": 0.05717718748968655
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.15,
32
- "qem_stderr": 0.05717718748968655
33
- },
34
- "all": {
35
- "qem": 0.15,
36
- "qem_stderr": 0.05717718748968655
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "9ed765ebd58d3013"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "2f085c109bc10f7d"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "1aedc4a3ff7a75a1"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "70d3921740f7ab02"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.03/aimo_kaggle_hard_pot/results_2024-05-27T22-50-33.482970.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 14259.254878757,
9
- "end_time": 14914.427603002,
10
- "total_evaluation_time_secondes": "655.1727242449997",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "8ccdf6776522c8613835892a71912fe2d8366885",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.12,
24
- "qem_stderr": 0.04642307659791979
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.54,
28
- "qem_stderr": 0.07119963311072637
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.23333333333333336,
32
- "qem_stderr": 0.048538959399198923
33
- },
34
- "all": {
35
- "qem": 0.23333333333333336,
36
- "qem_stderr": 0.048538959399198923
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "8fa9e18efe49021a"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "c0e55e5137310b5c"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "a1da9b9deee659aa"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "75cb29ad60e745ed"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.03/aimo_kaggle_medium_pot/results_2024-05-27T22-46-20.885308.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 106996.846626916,
9
- "end_time": 107359.816964207,
10
- "total_evaluation_time_secondes": "362.97033729100076",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "8ccdf6776522c8613835892a71912fe2d8366885",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.025,
20
- "qem_stderr": 0.024999999999999994
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.05,
24
- "qem_stderr": 0.03489912202260563
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.125,
28
- "qem_stderr": 0.05295740910852021
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.06666666666666667,
32
- "qem_stderr": 0.03761884371037528
33
- },
34
- "all": {
35
- "qem": 0.06666666666666667,
36
- "qem_stderr": 0.03761884371037528
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "398c838fac2a32cb"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "ad46c87b88dbba2f"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "246adbd385dd4579"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "2f97b0cecd2c9ec6"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.04/aimo_kaggle_hard_pot/results_2024-05-27T23-02-16.395548.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 304641.79705412,
9
- "end_time": 305415.434942523,
10
- "total_evaluation_time_secondes": "773.6378884030273",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "9d352e2a08647c995abd486f03c103c328f94b5e",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.08,
20
- "qem_stderr": 0.03875617133214439
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.44,
24
- "qem_stderr": 0.07091242083423345
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.38,
28
- "qem_stderr": 0.06934092056863767
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3,
32
- "qem_stderr": 0.0596698375783385
33
- },
34
- "all": {
35
- "qem": 0.3,
36
- "qem_stderr": 0.0596698375783385
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "566a96752b9c4a94"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "7ab56adc56fc6b33"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "5c601ac7277feacc"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "3f013a410775284d"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.04/aimo_kaggle_medium_pot/results_2024-05-27T22-56-39.512900.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 18377.382677895,
9
- "end_time": 18811.76337702,
10
- "total_evaluation_time_secondes": "434.38069912500214",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "9d352e2a08647c995abd486f03c103c328f94b5e",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.15,
20
- "qem_stderr": 0.05717718748968655
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.15,
24
- "qem_stderr": 0.05717718748968655
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.175,
28
- "qem_stderr": 0.060843430844447585
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.15833333333333333,
32
- "qem_stderr": 0.05839926860794023
33
- },
34
- "all": {
35
- "qem": 0.15833333333333333,
36
- "qem_stderr": 0.05839926860794023
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "c0bcec869e50357e"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "175d7969de36ab1b"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "f3c3ab633ac76bd0"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "ed6a5a9ce37315ba"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.05/aimo_kaggle_hard_pot/results_2024-05-27T22-47-26.728707.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 211019.661163806,
9
- "end_time": 211758.414953224,
10
- "total_evaluation_time_secondes": "738.7537894179986",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "1b1c05ec5ff47d465eaaa801c306a11637d140a4",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.36,
24
- "qem_stderr": 0.06857142857142856
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.6,
28
- "qem_stderr": 0.06998542122237653
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3333333333333333,
32
- "qem_stderr": 0.0555170060942519
33
- },
34
- "all": {
35
- "qem": 0.3333333333333333,
36
- "qem_stderr": 0.0555170060942519
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "6756ac049a892978"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "dfdc87594f75c999"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "63fc4eabc1edb20f"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "5cc4dd7edcddd58a"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.05/aimo_kaggle_medium_pot/results_2024-05-27T22-42-48.691321.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 40984.915370342,
9
- "end_time": 41379.707945993,
10
- "total_evaluation_time_secondes": "394.79257565100124",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "1b1c05ec5ff47d465eaaa801c306a11637d140a4",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.225,
20
- "qem_stderr": 0.06686668711812965
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.2,
24
- "qem_stderr": 0.06405126152203486
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.175,
28
- "qem_stderr": 0.060843430844447585
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.20000000000000004,
32
- "qem_stderr": 0.06392045982820403
33
- },
34
- "all": {
35
- "qem": 0.20000000000000004,
36
- "qem_stderr": 0.06392045982820403
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "a2da112f28b7b3dd"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "86ee58a3785c67b7"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "f4b14a0f9c0626a6"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "5b986fd2338ac693"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.06/aimo_kaggle_hard_pot/results_2024-05-27T23-22-45.715568.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 2382128.191168791,
9
- "end_time": 2382914.608837258,
10
- "total_evaluation_time_secondes": "786.4176684669219",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "86c0089055d54cd0922ae0d3b6a0455e83fe19a1",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.36,
24
- "qem_stderr": 0.06857142857142856
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.5,
28
- "qem_stderr": 0.07142857142857142
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3,
32
- "qem_stderr": 0.055998056162983534
33
- },
34
- "all": {
35
- "qem": 0.3,
36
- "qem_stderr": 0.055998056162983534
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "96626341015b155a"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "944cb6439171b0d6"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "e20041a1a89e4258"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "81eae61419effeda"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.06/aimo_kaggle_medium_pot/results_2024-05-27T23-16-30.855912.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 44794.384758256,
9
- "end_time": 45210.722487146,
10
- "total_evaluation_time_secondes": "416.337728890001",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "86c0089055d54cd0922ae0d3b6a0455e83fe19a1",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.175,
20
- "qem_stderr": 0.060843430844447585
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.125,
24
- "qem_stderr": 0.05295740910852021
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.175,
28
- "qem_stderr": 0.060843430844447564
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.15833333333333333,
32
- "qem_stderr": 0.058214756932471794
33
- },
34
- "all": {
35
- "qem": 0.15833333333333333,
36
- "qem_stderr": 0.058214756932471794
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "f64111597c7c5f38"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "384432275bdfee30"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "5cd657852f93ead5"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "1d9394689bab0335"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.07/aimo_kaggle_hard_pot/results_2024-05-27T22-43-29.392522.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 15640.261235424,
9
- "end_time": 16281.29244475,
10
- "total_evaluation_time_secondes": "641.0312093260018",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "d572ef7608311c43ad723a3ac133e4503e7303a8",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.02,
20
- "qem_stderr": 0.02
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.22,
24
- "qem_stderr": 0.05917804336345138
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.56,
28
- "qem_stderr": 0.07091242083423345
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.26666666666666666,
32
- "qem_stderr": 0.050030154732561616
33
- },
34
- "all": {
35
- "qem": 0.26666666666666666,
36
- "qem_stderr": 0.050030154732561616
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "e8d46b691518bf31"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "96ef4d972ae45fd3"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "de4121a908cc86cf"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "974e7c9b71028589"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.07/aimo_kaggle_medium_pot/results_2024-05-27T22-39-34.119815.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 106589.039780829,
9
- "end_time": 106953.051437736,
10
- "total_evaluation_time_secondes": "364.0116569069942",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "d572ef7608311c43ad723a3ac133e4503e7303a8",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.1,
24
- "qem_stderr": 0.04803844614152612
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.075,
28
- "qem_stderr": 0.04217636961434867
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.05833333333333333,
32
- "qem_stderr": 0.03007160525195826
33
- },
34
- "all": {
35
- "qem": 0.05833333333333333,
36
- "qem_stderr": 0.03007160525195826
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "af7a2a7c3bf735ce"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "3492201abfa537d3"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "b23881bd7634dfdf"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "37a711876b56979b"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.08/aimo_kaggle_hard_pot/results_2024-05-27T22-45-02.958180.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 3181770.247112947,
9
- "end_time": 3182453.504957875,
10
- "total_evaluation_time_secondes": "683.2578449277207",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "dc1193beda83f0cca3b7e60a158c36f6804af333",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.027994168488950606
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.1,
24
- "qem_stderr": 0.04285714285714285
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.34,
28
- "qem_stderr": 0.0676726816132972
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.16,
32
- "qem_stderr": 0.04617466431979689
33
- },
34
- "all": {
35
- "qem": 0.16,
36
- "qem_stderr": 0.04617466431979689
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "aec6eb9abe870d9c"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "f5a8786338ef07ef"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "19cc51ae4f8a174c"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "0bced5d39b0648e2"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.08/aimo_kaggle_medium_pot/results_2024-05-27T22-39-55.232091.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 10322.250336088,
9
- "end_time": 10676.245850647,
10
- "total_evaluation_time_secondes": "353.9955145589993",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "dc1193beda83f0cca3b7e60a158c36f6804af333",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.0,
20
- "qem_stderr": 0.0
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.0,
24
- "qem_stderr": 0.0
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.1,
28
- "qem_stderr": 0.048038446141526144
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.03333333333333333,
32
- "qem_stderr": 0.016012815380508715
33
- },
34
- "all": {
35
- "qem": 0.03333333333333333,
36
- "qem_stderr": 0.016012815380508715
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "f6ef25c13e3b9beb"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "d2e29a088eb3ef40"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "05fe807135396e54"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "42dc377e96ff07f4"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.09/aimo_kaggle_hard_pot/results_2024-05-27T23-01-23.916419.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 45437.752892193,
9
- "end_time": 46158.771446592,
10
- "total_evaluation_time_secondes": "721.018554399001",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "4c5a6b0364eeff00e757545ad918edf9938087d7",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.38,
24
- "qem_stderr": 0.06934092056863767
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.5,
28
- "qem_stderr": 0.07142857142857142
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.30666666666666664,
32
- "qem_stderr": 0.05625455349538657
33
- },
34
- "all": {
35
- "qem": 0.30666666666666664,
36
- "qem_stderr": 0.05625455349538657
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "ee8d8deb951e1d20"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "4873c8bc9c937aea"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "0f9b2a78a3b11cef"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "28102952776d50f9"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.09/aimo_kaggle_medium_pot/results_2024-05-27T22-56-49.561125.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 43584.407888815,
9
- "end_time": 44029.427719081,
10
- "total_evaluation_time_secondes": "445.0198302659992",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "4c5a6b0364eeff00e757545ad918edf9938087d7",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.1,
20
- "qem_stderr": 0.04803844614152612
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.1,
24
- "qem_stderr": 0.04803844614152612
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.125,
28
- "qem_stderr": 0.05295740910852021
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.10833333333333334,
32
- "qem_stderr": 0.04967810046385748
33
- },
34
- "all": {
35
- "qem": 0.10833333333333334,
36
- "qem_stderr": 0.04967810046385748
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "fe53ab7e7343a382"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "afec954f4fad91f4"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "47fe73a6dd818fe2"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "8efd6f1fe2208bcb"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.10/aimo_kaggle_hard_pot/results_2024-05-27T22-56-26.205133.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 16333.701216751,
9
- "end_time": 17058.105147357,
10
- "total_evaluation_time_secondes": "724.4039306060004",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "18af21a11f19e82ae0b5c4807f6061cb2eb1458b",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.06,
20
- "qem_stderr": 0.0339266916772512
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.44,
24
- "qem_stderr": 0.07091242083423345
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.6,
28
- "qem_stderr": 0.06998542122237653
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.3666666666666667,
32
- "qem_stderr": 0.058274844577953726
33
- },
34
- "all": {
35
- "qem": 0.3666666666666667,
36
- "qem_stderr": 0.058274844577953726
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "28667310ce047385"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "20523b0a9736ca9a"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "fa8e50486f1f0438"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "f27ba9fc232dd906"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.10/aimo_kaggle_medium_pot/results_2024-05-27T22-52-23.993586.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 3182508.308928725,
9
- "end_time": 3182894.568337251,
10
- "total_evaluation_time_secondes": "386.25940852612257",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "18af21a11f19e82ae0b5c4807f6061cb2eb1458b",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.25,
20
- "qem_stderr": 0.06933752452815363
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.15,
24
- "qem_stderr": 0.05717718748968655
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.175,
28
- "qem_stderr": 0.060843430844447585
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.19166666666666665,
32
- "qem_stderr": 0.062452714287429247
33
- },
34
- "all": {
35
- "qem": 0.19166666666666665,
36
- "qem_stderr": 0.062452714287429247
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "33e1217843971331"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "0317f6719f9f3cd1"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "847c8e8d9dcacdd4"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "4d3a9d65f57acd85"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.11/aimo_kaggle_hard_pot/results_2024-05-27T23-28-10.769733.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 109144.315993371,
9
- "end_time": 109869.701287677,
10
- "total_evaluation_time_secondes": "725.3852943060047",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "5fbd6d2b74f2755ce0b32890432d6af62410c3e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.12,
20
- "qem_stderr": 0.046423076597919784
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.34,
24
- "qem_stderr": 0.0676726816132972
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.52,
28
- "qem_stderr": 0.07137140569598172
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.32666666666666666,
32
- "qem_stderr": 0.06182238796906623
33
- },
34
- "all": {
35
- "qem": 0.32666666666666666,
36
- "qem_stderr": 0.06182238796906623
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "e730032197dfa045"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "7b27c9de11d4cbd5"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "dc5b74898fd59f58"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "b759b18b73f3f7fe"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.11/aimo_kaggle_medium_pot/results_2024-05-27T23-24-01.746924.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 45275.122268848,
9
- "end_time": 45661.613526662,
10
- "total_evaluation_time_secondes": "386.49125781399925",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "5fbd6d2b74f2755ce0b32890432d6af62410c3e2",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.125,
20
- "qem_stderr": 0.05295740910852021
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.125,
24
- "qem_stderr": 0.05295740910852021
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.225,
28
- "qem_stderr": 0.06686668711812965
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.15833333333333333,
32
- "qem_stderr": 0.057593835111723356
33
- },
34
- "all": {
35
- "qem": 0.15833333333333333,
36
- "qem_stderr": 0.057593835111723356
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "fa3b910dfca55474"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "b1ae92916dde27d0"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "2febceafe1d903ce"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "4e6bfc37d6e77d6a"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.12/aimo_kaggle_hard_pot/results_2024-05-27T22-58-24.076261.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 107404.931245741,
9
- "end_time": 108083.007317716,
10
- "total_evaluation_time_secondes": "678.076071974996",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "651246d76fe6c8aedf46454ea474eca76136ab73",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_hard_pot:v0|0": {
19
- "qem": 0.04,
20
- "qem_stderr": 0.02799416848895062
21
- },
22
- "custom|aimo_kaggle_hard_pot:v1|0": {
23
- "qem": 0.3,
24
- "qem_stderr": 0.06546536707079771
25
- },
26
- "custom|aimo_kaggle_hard_pot:v2|0": {
27
- "qem": 0.52,
28
- "qem_stderr": 0.07137140569598172
29
- },
30
- "custom|aimo_kaggle_hard_pot:_average|0": {
31
- "qem": 0.2866666666666667,
32
- "qem_stderr": 0.05494364708524335
33
- },
34
- "all": {
35
- "qem": 0.2866666666666667,
36
- "qem_stderr": 0.05494364708524335
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_hard_pot:v0|0": 0,
41
- "custom|aimo_kaggle_hard_pot:v1|0": 0,
42
- "custom|aimo_kaggle_hard_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_hard_pot:v0": {
46
- "name": "aimo_kaggle_hard_pot:v0",
47
- "prompt_function": "kaggle_hard_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 50,
70
- "effective_num_docs": 50,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_hard_pot:v1": {
76
- "name": "aimo_kaggle_hard_pot:v1",
77
- "prompt_function": "kaggle_hard_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 50,
100
- "effective_num_docs": 50,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_hard_pot:v2": {
106
- "name": "aimo_kaggle_hard_pot:v2",
107
- "prompt_function": "kaggle_hard_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-hard",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 50,
130
- "effective_num_docs": 50,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_hard_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "303213a38d9f7512",
140
- "hash_full_prompts": "3c670fa0e80bc301",
141
- "hash_input_tokens": "1bd4187ad6963415",
142
- "hash_cont_tokens": "713e9ea3fd7a21cd"
143
- },
144
- "truncated": 50,
145
- "non_truncated": 0,
146
- "padded": 37,
147
- "non_padded": 13,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_hard_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "e4234b97ad92862f",
154
- "hash_full_prompts": "09e7759a96f64e59",
155
- "hash_input_tokens": "dc7f68cf14be3d61",
156
- "hash_cont_tokens": "a24d886da74b19a1"
157
- },
158
- "truncated": 50,
159
- "non_truncated": 0,
160
- "padded": 32,
161
- "non_padded": 18,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_hard_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "6396eb8833e13ba0",
168
- "hash_full_prompts": "65f8b95ea99c7087",
169
- "hash_input_tokens": "bddb67459e7601c5",
170
- "hash_cont_tokens": "81abaeff7a78b548"
171
- },
172
- "truncated": 50,
173
- "non_truncated": 0,
174
- "padded": 30,
175
- "non_padded": 20,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "648c9a107d279e1e",
183
- "hash_full_prompts": "7f4c38eb08b3bb41",
184
- "hash_input_tokens": "16f0fd76e0a2a165",
185
- "hash_cont_tokens": "417fec7e4d72917f"
186
- },
187
- "truncated": 150,
188
- "non_truncated": 0,
189
- "padded": 99,
190
- "non_padded": 51,
191
- "num_truncated_few_shots": 0
192
- }
193
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_results/AI-MO/deepseek-math-7b-rl-sft/aimo_v02.12/aimo_kaggle_medium_pot/results_2024-05-27T22-54-41.166514.json DELETED
@@ -1,193 +0,0 @@
1
- {
2
- "config_general": {
3
- "lighteval_sha": "?",
4
- "num_fewshot_seeds": 1,
5
- "override_batch_size": 4,
6
- "max_samples": null,
7
- "job_id": "",
8
- "start_time": 211814.762291854,
9
- "end_time": 212192.853241965,
10
- "total_evaluation_time_secondes": "378.09095011101454",
11
- "model_name": "AI-MO/deepseek-math-7b-rl-sft",
12
- "model_sha": "651246d76fe6c8aedf46454ea474eca76136ab73",
13
- "model_dtype": "torch.bfloat16",
14
- "model_size": "12.93 GB",
15
- "config": null
16
- },
17
- "results": {
18
- "custom|aimo_kaggle_medium_pot:v0|0": {
19
- "qem": 0.1,
20
- "qem_stderr": 0.04803844614152612
21
- },
22
- "custom|aimo_kaggle_medium_pot:v1|0": {
23
- "qem": 0.125,
24
- "qem_stderr": 0.05295740910852021
25
- },
26
- "custom|aimo_kaggle_medium_pot:v2|0": {
27
- "qem": 0.1,
28
- "qem_stderr": 0.04803844614152612
29
- },
30
- "custom|aimo_kaggle_medium_pot:_average|0": {
31
- "qem": 0.10833333333333334,
32
- "qem_stderr": 0.04967810046385749
33
- },
34
- "all": {
35
- "qem": 0.10833333333333334,
36
- "qem_stderr": 0.04967810046385749
37
- }
38
- },
39
- "versions": {
40
- "custom|aimo_kaggle_medium_pot:v0|0": 0,
41
- "custom|aimo_kaggle_medium_pot:v1|0": 0,
42
- "custom|aimo_kaggle_medium_pot:v2|0": 0
43
- },
44
- "config_tasks": {
45
- "custom|aimo_kaggle_medium_pot:v0": {
46
- "name": "aimo_kaggle_medium_pot:v0",
47
- "prompt_function": "kaggle_medium_pot_prompt_fn_v0",
48
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
49
- "hf_subset": "v0",
50
- "metric": [
51
- "quasi_exact_match_code_and_math"
52
- ],
53
- "hf_avail_splits": [
54
- "train"
55
- ],
56
- "evaluation_splits": [
57
- "train"
58
- ],
59
- "few_shots_split": null,
60
- "few_shots_select": null,
61
- "generation_size": 2048,
62
- "stop_sequence": null,
63
- "output_regex": null,
64
- "num_samples": null,
65
- "frozen": false,
66
- "suite": [
67
- "custom"
68
- ],
69
- "original_num_docs": 40,
70
- "effective_num_docs": 40,
71
- "trust_dataset": null,
72
- "must_remove_duplicate_docs": null,
73
- "version": 0
74
- },
75
- "custom|aimo_kaggle_medium_pot:v1": {
76
- "name": "aimo_kaggle_medium_pot:v1",
77
- "prompt_function": "kaggle_medium_pot_prompt_fn_v1",
78
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
79
- "hf_subset": "v0",
80
- "metric": [
81
- "quasi_exact_match_code_and_math"
82
- ],
83
- "hf_avail_splits": [
84
- "train"
85
- ],
86
- "evaluation_splits": [
87
- "train"
88
- ],
89
- "few_shots_split": null,
90
- "few_shots_select": null,
91
- "generation_size": 2048,
92
- "stop_sequence": null,
93
- "output_regex": null,
94
- "num_samples": null,
95
- "frozen": false,
96
- "suite": [
97
- "custom"
98
- ],
99
- "original_num_docs": 40,
100
- "effective_num_docs": 40,
101
- "trust_dataset": null,
102
- "must_remove_duplicate_docs": null,
103
- "version": 0
104
- },
105
- "custom|aimo_kaggle_medium_pot:v2": {
106
- "name": "aimo_kaggle_medium_pot:v2",
107
- "prompt_function": "kaggle_medium_pot_prompt_fn_v2",
108
- "hf_repo": "AI-MO/kaggle-validation-set-medium",
109
- "hf_subset": "v0",
110
- "metric": [
111
- "quasi_exact_match_code_and_math"
112
- ],
113
- "hf_avail_splits": [
114
- "train"
115
- ],
116
- "evaluation_splits": [
117
- "train"
118
- ],
119
- "few_shots_split": null,
120
- "few_shots_select": null,
121
- "generation_size": 2048,
122
- "stop_sequence": null,
123
- "output_regex": null,
124
- "num_samples": null,
125
- "frozen": false,
126
- "suite": [
127
- "custom"
128
- ],
129
- "original_num_docs": 40,
130
- "effective_num_docs": 40,
131
- "trust_dataset": null,
132
- "must_remove_duplicate_docs": null,
133
- "version": 0
134
- }
135
- },
136
- "summary_tasks": {
137
- "custom|aimo_kaggle_medium_pot:v0|0": {
138
- "hashes": {
139
- "hash_examples": "2799c24461029dc3",
140
- "hash_full_prompts": "de6572b914d5bba6",
141
- "hash_input_tokens": "5af82ee284ccce29",
142
- "hash_cont_tokens": "dfc8b121255332df"
143
- },
144
- "truncated": 40,
145
- "non_truncated": 0,
146
- "padded": 29,
147
- "non_padded": 11,
148
- "effective_few_shots": 0.0,
149
- "num_truncated_few_shots": 0
150
- },
151
- "custom|aimo_kaggle_medium_pot:v1|0": {
152
- "hashes": {
153
- "hash_examples": "806b2e2056b41f84",
154
- "hash_full_prompts": "427d5de6e4d90df2",
155
- "hash_input_tokens": "be3dcf9f8a2350d0",
156
- "hash_cont_tokens": "5530d0bf94ec3456"
157
- },
158
- "truncated": 40,
159
- "non_truncated": 0,
160
- "padded": 26,
161
- "non_padded": 14,
162
- "effective_few_shots": 0.0,
163
- "num_truncated_few_shots": 0
164
- },
165
- "custom|aimo_kaggle_medium_pot:v2|0": {
166
- "hashes": {
167
- "hash_examples": "d8534375acc5d427",
168
- "hash_full_prompts": "6bb90c129d6d6123",
169
- "hash_input_tokens": "22eeaedbeb3f56f0",
170
- "hash_cont_tokens": "e03d999936114bfc"
171
- },
172
- "truncated": 40,
173
- "non_truncated": 0,
174
- "padded": 31,
175
- "non_padded": 9,
176
- "effective_few_shots": 0.0,
177
- "num_truncated_few_shots": 0
178
- }
179
- },
180
- "summary_general": {
181
- "hashes": {
182
- "hash_examples": "623505a45a4910c2",
183
- "hash_full_prompts": "3aca0fad61a42921",
184
- "hash_input_tokens": "c7d2c8bb76f8ecb1",
185
- "hash_cont_tokens": "45672d8b1cdafda4"
186
- },
187
- "truncated": 120,
188
- "non_truncated": 0,
189
- "padded": 86,
190
- "non_padded": 34,
191
- "num_truncated_few_shots": 0
192
- }
193
- }