Commit
•
3a44d54
1
Parent(s):
22669ba
Update README.md (#1)
Browse files- Update README.md (f6ff0947da1d3ba31b99dae1b6e8015769fffad4)
README.md
CHANGED
|
@@ -8,4 +8,390 @@ tags:
|
|
| 8 |
|
| 9 |
Merge of top 7B models with PASS method
|
| 10 |
|
| 11 |
-
> mergekit is a toolkit for merging pre-trained language models. mergekit uses an out-of-core approach to perform unreasonably elaborate merges in resource-constrained situations. Merges can be run entirely on CPU or accelerated with as little as 8 GB of VRAM. Many merging algorithms are supported, with more coming as they catch my attention.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
Merge of top 7B models with PASS method
|
| 10 |
|
| 11 |
+
> mergekit is a toolkit for merging pre-trained language models. mergekit uses an out-of-core approach to perform unreasonably elaborate merges in resource-constrained situations. Merges can be run entirely on CPU or accelerated with as little as 8 GB of VRAM. Many merging algorithms are supported, with more coming as they catch my attention.
|
| 12 |
+
|
| 13 |
+
## Eval
|
| 14 |
+
|
| 15 |
+
```python
|
| 16 |
+
{
|
| 17 |
+
"all": {
|
| 18 |
+
"acc": 0.6152059168567449,
|
| 19 |
+
"acc_stderr": 0.031951119145286845,
|
| 20 |
+
"acc_norm": 0.6274010157580394,
|
| 21 |
+
"acc_norm_stderr": 0.032831804892806175,
|
| 22 |
+
"mc1": 0.25091799265605874,
|
| 23 |
+
"mc1_stderr": 0.015176985027707694,
|
| 24 |
+
"mc2": 0.5006656333594469,
|
| 25 |
+
"mc2_stderr": 0.01636490303268174
|
| 26 |
+
},
|
| 27 |
+
"harness|arc:challenge|25": {
|
| 28 |
+
"acc": 0.3447098976109215,
|
| 29 |
+
"acc_stderr": 0.013888816286782112,
|
| 30 |
+
"acc_norm": 0.3856655290102389,
|
| 31 |
+
"acc_norm_stderr": 0.01422425097325717
|
| 32 |
+
},
|
| 33 |
+
"harness|hellaswag|10": {
|
| 34 |
+
"acc": 0.34116709818761204,
|
| 35 |
+
"acc_stderr": 0.004731324409133264,
|
| 36 |
+
"acc_norm": 0.515435172276439,
|
| 37 |
+
"acc_norm_stderr": 0.004987403268345035
|
| 38 |
+
},
|
| 39 |
+
"harness|hendrycksTest-abstract_algebra|5": {
|
| 40 |
+
"acc": 0.27,
|
| 41 |
+
"acc_stderr": 0.04461960433384741,
|
| 42 |
+
"acc_norm": 0.27,
|
| 43 |
+
"acc_norm_stderr": 0.04461960433384741
|
| 44 |
+
},
|
| 45 |
+
"harness|hendrycksTest-anatomy|5": {
|
| 46 |
+
"acc": 0.5703703703703704,
|
| 47 |
+
"acc_stderr": 0.042763494943765995,
|
| 48 |
+
"acc_norm": 0.5703703703703704,
|
| 49 |
+
"acc_norm_stderr": 0.042763494943765995
|
| 50 |
+
},
|
| 51 |
+
"harness|hendrycksTest-astronomy|5": {
|
| 52 |
+
"acc": 0.6842105263157895,
|
| 53 |
+
"acc_stderr": 0.0378272898086547,
|
| 54 |
+
"acc_norm": 0.6842105263157895,
|
| 55 |
+
"acc_norm_stderr": 0.0378272898086547
|
| 56 |
+
},
|
| 57 |
+
"harness|hendrycksTest-business_ethics|5": {
|
| 58 |
+
"acc": 0.62,
|
| 59 |
+
"acc_stderr": 0.048783173121456316,
|
| 60 |
+
"acc_norm": 0.62,
|
| 61 |
+
"acc_norm_stderr": 0.048783173121456316
|
| 62 |
+
},
|
| 63 |
+
"harness|hendrycksTest-clinical_knowledge|5": {
|
| 64 |
+
"acc": 0.7169811320754716,
|
| 65 |
+
"acc_stderr": 0.027724236492700918,
|
| 66 |
+
"acc_norm": 0.7169811320754716,
|
| 67 |
+
"acc_norm_stderr": 0.027724236492700918
|
| 68 |
+
},
|
| 69 |
+
"harness|hendrycksTest-college_biology|5": {
|
| 70 |
+
"acc": 0.7638888888888888,
|
| 71 |
+
"acc_stderr": 0.03551446610810826,
|
| 72 |
+
"acc_norm": 0.7638888888888888,
|
| 73 |
+
"acc_norm_stderr": 0.03551446610810826
|
| 74 |
+
},
|
| 75 |
+
"harness|hendrycksTest-college_chemistry|5": {
|
| 76 |
+
"acc": 0.45,
|
| 77 |
+
"acc_stderr": 0.05,
|
| 78 |
+
"acc_norm": 0.45,
|
| 79 |
+
"acc_norm_stderr": 0.05
|
| 80 |
+
},
|
| 81 |
+
"harness|hendrycksTest-college_computer_science|5": {
|
| 82 |
+
"acc": 0.46,
|
| 83 |
+
"acc_stderr": 0.05009082659620333,
|
| 84 |
+
"acc_norm": 0.46,
|
| 85 |
+
"acc_norm_stderr": 0.05009082659620333
|
| 86 |
+
},
|
| 87 |
+
"harness|hendrycksTest-college_mathematics|5": {
|
| 88 |
+
"acc": 0.32,
|
| 89 |
+
"acc_stderr": 0.04688261722621504,
|
| 90 |
+
"acc_norm": 0.32,
|
| 91 |
+
"acc_norm_stderr": 0.04688261722621504
|
| 92 |
+
},
|
| 93 |
+
"harness|hendrycksTest-college_medicine|5": {
|
| 94 |
+
"acc": 0.6358381502890174,
|
| 95 |
+
"acc_stderr": 0.03669072477416907,
|
| 96 |
+
"acc_norm": 0.6358381502890174,
|
| 97 |
+
"acc_norm_stderr": 0.03669072477416907
|
| 98 |
+
},
|
| 99 |
+
"harness|hendrycksTest-college_physics|5": {
|
| 100 |
+
"acc": 0.4019607843137255,
|
| 101 |
+
"acc_stderr": 0.048786087144669955,
|
| 102 |
+
"acc_norm": 0.4019607843137255,
|
| 103 |
+
"acc_norm_stderr": 0.048786087144669955
|
| 104 |
+
},
|
| 105 |
+
"harness|hendrycksTest-computer_security|5": {
|
| 106 |
+
"acc": 0.76,
|
| 107 |
+
"acc_stderr": 0.04292346959909283,
|
| 108 |
+
"acc_norm": 0.76,
|
| 109 |
+
"acc_norm_stderr": 0.04292346959909283
|
| 110 |
+
},
|
| 111 |
+
"harness|hendrycksTest-conceptual_physics|5": {
|
| 112 |
+
"acc": 0.5446808510638298,
|
| 113 |
+
"acc_stderr": 0.03255525359340355,
|
| 114 |
+
"acc_norm": 0.5446808510638298,
|
| 115 |
+
"acc_norm_stderr": 0.03255525359340355
|
| 116 |
+
},
|
| 117 |
+
"harness|hendrycksTest-econometrics|5": {
|
| 118 |
+
"acc": 0.4824561403508772,
|
| 119 |
+
"acc_stderr": 0.04700708033551038,
|
| 120 |
+
"acc_norm": 0.4824561403508772,
|
| 121 |
+
"acc_norm_stderr": 0.04700708033551038
|
| 122 |
+
},
|
| 123 |
+
"harness|hendrycksTest-electrical_engineering|5": {
|
| 124 |
+
"acc": 0.5172413793103449,
|
| 125 |
+
"acc_stderr": 0.04164188720169375,
|
| 126 |
+
"acc_norm": 0.5172413793103449,
|
| 127 |
+
"acc_norm_stderr": 0.04164188720169375
|
| 128 |
+
},
|
| 129 |
+
"harness|hendrycksTest-elementary_mathematics|5": {
|
| 130 |
+
"acc": 0.42857142857142855,
|
| 131 |
+
"acc_stderr": 0.025487187147859372,
|
| 132 |
+
"acc_norm": 0.42857142857142855,
|
| 133 |
+
"acc_norm_stderr": 0.025487187147859372
|
| 134 |
+
},
|
| 135 |
+
"harness|hendrycksTest-formal_logic|5": {
|
| 136 |
+
"acc": 0.3968253968253968,
|
| 137 |
+
"acc_stderr": 0.043758884927270605,
|
| 138 |
+
"acc_norm": 0.3968253968253968,
|
| 139 |
+
"acc_norm_stderr": 0.043758884927270605
|
| 140 |
+
},
|
| 141 |
+
"harness|hendrycksTest-global_facts|5": {
|
| 142 |
+
"acc": 0.34,
|
| 143 |
+
"acc_stderr": 0.04760952285695236,
|
| 144 |
+
"acc_norm": 0.34,
|
| 145 |
+
"acc_norm_stderr": 0.04760952285695236
|
| 146 |
+
},
|
| 147 |
+
"harness|hendrycksTest-high_school_biology|5": {
|
| 148 |
+
"acc": 0.7741935483870968,
|
| 149 |
+
"acc_stderr": 0.023785577884181015,
|
| 150 |
+
"acc_norm": 0.7741935483870968,
|
| 151 |
+
"acc_norm_stderr": 0.023785577884181015
|
| 152 |
+
},
|
| 153 |
+
"harness|hendrycksTest-high_school_chemistry|5": {
|
| 154 |
+
"acc": 0.5123152709359606,
|
| 155 |
+
"acc_stderr": 0.035169204442208966,
|
| 156 |
+
"acc_norm": 0.5123152709359606,
|
| 157 |
+
"acc_norm_stderr": 0.035169204442208966
|
| 158 |
+
},
|
| 159 |
+
"harness|hendrycksTest-high_school_computer_science|5": {
|
| 160 |
+
"acc": 0.66,
|
| 161 |
+
"acc_stderr": 0.04760952285695237,
|
| 162 |
+
"acc_norm": 0.66,
|
| 163 |
+
"acc_norm_stderr": 0.04760952285695237
|
| 164 |
+
},
|
| 165 |
+
"harness|hendrycksTest-high_school_european_history|5": {
|
| 166 |
+
"acc": 0.7636363636363637,
|
| 167 |
+
"acc_stderr": 0.03317505930009181,
|
| 168 |
+
"acc_norm": 0.7636363636363637,
|
| 169 |
+
"acc_norm_stderr": 0.03317505930009181
|
| 170 |
+
},
|
| 171 |
+
"harness|hendrycksTest-high_school_geography|5": {
|
| 172 |
+
"acc": 0.7373737373737373,
|
| 173 |
+
"acc_stderr": 0.03135305009533085,
|
| 174 |
+
"acc_norm": 0.7373737373737373,
|
| 175 |
+
"acc_norm_stderr": 0.03135305009533085
|
| 176 |
+
},
|
| 177 |
+
"harness|hendrycksTest-high_school_government_and_politics|5": {
|
| 178 |
+
"acc": 0.8808290155440415,
|
| 179 |
+
"acc_stderr": 0.023381935348121437,
|
| 180 |
+
"acc_norm": 0.8808290155440415,
|
| 181 |
+
"acc_norm_stderr": 0.023381935348121437
|
| 182 |
+
},
|
| 183 |
+
"harness|hendrycksTest-high_school_macroeconomics|5": {
|
| 184 |
+
"acc": 0.617948717948718,
|
| 185 |
+
"acc_stderr": 0.024635549163908237,
|
| 186 |
+
"acc_norm": 0.617948717948718,
|
| 187 |
+
"acc_norm_stderr": 0.024635549163908237
|
| 188 |
+
},
|
| 189 |
+
"harness|hendrycksTest-high_school_mathematics|5": {
|
| 190 |
+
"acc": 0.2777777777777778,
|
| 191 |
+
"acc_stderr": 0.027309140588230203,
|
| 192 |
+
"acc_norm": 0.2777777777777778,
|
| 193 |
+
"acc_norm_stderr": 0.027309140588230203
|
| 194 |
+
},
|
| 195 |
+
"harness|hendrycksTest-high_school_microeconomics|5": {
|
| 196 |
+
"acc": 0.6512605042016807,
|
| 197 |
+
"acc_stderr": 0.030956636328566545,
|
| 198 |
+
"acc_norm": 0.6512605042016807,
|
| 199 |
+
"acc_norm_stderr": 0.030956636328566545
|
| 200 |
+
},
|
| 201 |
+
"harness|hendrycksTest-high_school_physics|5": {
|
| 202 |
+
"acc": 0.32450331125827814,
|
| 203 |
+
"acc_stderr": 0.038227469376587525,
|
| 204 |
+
"acc_norm": 0.32450331125827814,
|
| 205 |
+
"acc_norm_stderr": 0.038227469376587525
|
| 206 |
+
},
|
| 207 |
+
"harness|hendrycksTest-high_school_psychology|5": {
|
| 208 |
+
"acc": 0.8440366972477065,
|
| 209 |
+
"acc_stderr": 0.015555802713590158,
|
| 210 |
+
"acc_norm": 0.8440366972477065,
|
| 211 |
+
"acc_norm_stderr": 0.015555802713590158
|
| 212 |
+
},
|
| 213 |
+
"harness|hendrycksTest-high_school_statistics|5": {
|
| 214 |
+
"acc": 0.4722222222222222,
|
| 215 |
+
"acc_stderr": 0.0340470532865388,
|
| 216 |
+
"acc_norm": 0.4722222222222222,
|
| 217 |
+
"acc_norm_stderr": 0.0340470532865388
|
| 218 |
+
},
|
| 219 |
+
"harness|hendrycksTest-high_school_us_history|5": {
|
| 220 |
+
"acc": 0.8431372549019608,
|
| 221 |
+
"acc_stderr": 0.025524722324553346,
|
| 222 |
+
"acc_norm": 0.8431372549019608,
|
| 223 |
+
"acc_norm_stderr": 0.025524722324553346
|
| 224 |
+
},
|
| 225 |
+
"harness|hendrycksTest-high_school_world_history|5": {
|
| 226 |
+
"acc": 0.810126582278481,
|
| 227 |
+
"acc_stderr": 0.025530100460233497,
|
| 228 |
+
"acc_norm": 0.810126582278481,
|
| 229 |
+
"acc_norm_stderr": 0.025530100460233497
|
| 230 |
+
},
|
| 231 |
+
"harness|hendrycksTest-human_aging|5": {
|
| 232 |
+
"acc": 0.7174887892376681,
|
| 233 |
+
"acc_stderr": 0.03021683101150877,
|
| 234 |
+
"acc_norm": 0.7174887892376681,
|
| 235 |
+
"acc_norm_stderr": 0.03021683101150877
|
| 236 |
+
},
|
| 237 |
+
"harness|hendrycksTest-human_sexuality|5": {
|
| 238 |
+
"acc": 0.7786259541984732,
|
| 239 |
+
"acc_stderr": 0.0364129708131373,
|
| 240 |
+
"acc_norm": 0.7786259541984732,
|
| 241 |
+
"acc_norm_stderr": 0.0364129708131373
|
| 242 |
+
},
|
| 243 |
+
"harness|hendrycksTest-international_law|5": {
|
| 244 |
+
"acc": 0.7768595041322314,
|
| 245 |
+
"acc_stderr": 0.03800754475228733,
|
| 246 |
+
"acc_norm": 0.7768595041322314,
|
| 247 |
+
"acc_norm_stderr": 0.03800754475228733
|
| 248 |
+
},
|
| 249 |
+
"harness|hendrycksTest-jurisprudence|5": {
|
| 250 |
+
"acc": 0.8148148148148148,
|
| 251 |
+
"acc_stderr": 0.03755265865037181,
|
| 252 |
+
"acc_norm": 0.8148148148148148,
|
| 253 |
+
"acc_norm_stderr": 0.03755265865037181
|
| 254 |
+
},
|
| 255 |
+
"harness|hendrycksTest-logical_fallacies|5": {
|
| 256 |
+
"acc": 0.7914110429447853,
|
| 257 |
+
"acc_stderr": 0.031921934489347235,
|
| 258 |
+
"acc_norm": 0.7914110429447853,
|
| 259 |
+
"acc_norm_stderr": 0.031921934489347235
|
| 260 |
+
},
|
| 261 |
+
"harness|hendrycksTest-machine_learning|5": {
|
| 262 |
+
"acc": 0.5446428571428571,
|
| 263 |
+
"acc_stderr": 0.04726835553719097,
|
| 264 |
+
"acc_norm": 0.5446428571428571,
|
| 265 |
+
"acc_norm_stderr": 0.04726835553719097
|
| 266 |
+
},
|
| 267 |
+
"harness|hendrycksTest-management|5": {
|
| 268 |
+
"acc": 0.8349514563106796,
|
| 269 |
+
"acc_stderr": 0.036756688322331886,
|
| 270 |
+
"acc_norm": 0.8349514563106796,
|
| 271 |
+
"acc_norm_stderr": 0.036756688322331886
|
| 272 |
+
},
|
| 273 |
+
"harness|hendrycksTest-marketing|5": {
|
| 274 |
+
"acc": 0.8290598290598291,
|
| 275 |
+
"acc_stderr": 0.024662496845209804,
|
| 276 |
+
"acc_norm": 0.8290598290598291,
|
| 277 |
+
"acc_norm_stderr": 0.024662496845209804
|
| 278 |
+
},
|
| 279 |
+
"harness|hendrycksTest-medical_genetics|5": {
|
| 280 |
+
"acc": 0.69,
|
| 281 |
+
"acc_stderr": 0.04648231987117316,
|
| 282 |
+
"acc_norm": 0.69,
|
| 283 |
+
"acc_norm_stderr": 0.04648231987117316
|
| 284 |
+
},
|
| 285 |
+
"harness|hendrycksTest-miscellaneous|5": {
|
| 286 |
+
"acc": 0.8250319284802043,
|
| 287 |
+
"acc_stderr": 0.013586619219903324,
|
| 288 |
+
"acc_norm": 0.8250319284802043,
|
| 289 |
+
"acc_norm_stderr": 0.013586619219903324
|
| 290 |
+
},
|
| 291 |
+
"harness|hendrycksTest-moral_disputes|5": {
|
| 292 |
+
"acc": 0.7283236994219653,
|
| 293 |
+
"acc_stderr": 0.023948512905468348,
|
| 294 |
+
"acc_norm": 0.7283236994219653,
|
| 295 |
+
"acc_norm_stderr": 0.023948512905468348
|
| 296 |
+
},
|
| 297 |
+
"harness|hendrycksTest-moral_scenarios|5": {
|
| 298 |
+
"acc": 0.36312849162011174,
|
| 299 |
+
"acc_stderr": 0.016083749986853704,
|
| 300 |
+
"acc_norm": 0.36312849162011174,
|
| 301 |
+
"acc_norm_stderr": 0.016083749986853704
|
| 302 |
+
},
|
| 303 |
+
"harness|hendrycksTest-nutrition|5": {
|
| 304 |
+
"acc": 0.7450980392156863,
|
| 305 |
+
"acc_stderr": 0.02495418432487991,
|
| 306 |
+
"acc_norm": 0.7450980392156863,
|
| 307 |
+
"acc_norm_stderr": 0.02495418432487991
|
| 308 |
+
},
|
| 309 |
+
"harness|hendrycksTest-philosophy|5": {
|
| 310 |
+
"acc": 0.7202572347266881,
|
| 311 |
+
"acc_stderr": 0.02549425935069491,
|
| 312 |
+
"acc_norm": 0.7202572347266881,
|
| 313 |
+
"acc_norm_stderr": 0.02549425935069491
|
| 314 |
+
},
|
| 315 |
+
"harness|hendrycksTest-prehistory|5": {
|
| 316 |
+
"acc": 0.7530864197530864,
|
| 317 |
+
"acc_stderr": 0.023993501709042114,
|
| 318 |
+
"acc_norm": 0.7530864197530864,
|
| 319 |
+
"acc_norm_stderr": 0.023993501709042114
|
| 320 |
+
},
|
| 321 |
+
"harness|hendrycksTest-professional_accounting|5": {
|
| 322 |
+
"acc": 0.4787234042553192,
|
| 323 |
+
"acc_stderr": 0.029800481645628693,
|
| 324 |
+
"acc_norm": 0.4787234042553192,
|
| 325 |
+
"acc_norm_stderr": 0.029800481645628693
|
| 326 |
+
},
|
| 327 |
+
"harness|hendrycksTest-professional_law|5": {
|
| 328 |
+
"acc": 0.4367666232073012,
|
| 329 |
+
"acc_stderr": 0.01266770191960366,
|
| 330 |
+
"acc_norm": 0.4367666232073012,
|
| 331 |
+
"acc_norm_stderr": 0.01266770191960366
|
| 332 |
+
},
|
| 333 |
+
"harness|hendrycksTest-professional_medicine|5": {
|
| 334 |
+
"acc": 0.6176470588235294,
|
| 335 |
+
"acc_stderr": 0.029520095697687765,
|
| 336 |
+
"acc_norm": 0.6176470588235294,
|
| 337 |
+
"acc_norm_stderr": 0.029520095697687765
|
| 338 |
+
},
|
| 339 |
+
"harness|hendrycksTest-professional_psychology|5": {
|
| 340 |
+
"acc": 0.6699346405228758,
|
| 341 |
+
"acc_stderr": 0.019023726160724553,
|
| 342 |
+
"acc_norm": 0.6699346405228758,
|
| 343 |
+
"acc_norm_stderr": 0.019023726160724553
|
| 344 |
+
},
|
| 345 |
+
"harness|hendrycksTest-public_relations|5": {
|
| 346 |
+
"acc": 0.6545454545454545,
|
| 347 |
+
"acc_stderr": 0.04554619617541054,
|
| 348 |
+
"acc_norm": 0.6545454545454545,
|
| 349 |
+
"acc_norm_stderr": 0.04554619617541054
|
| 350 |
+
},
|
| 351 |
+
"harness|hendrycksTest-security_studies|5": {
|
| 352 |
+
"acc": 0.726530612244898,
|
| 353 |
+
"acc_stderr": 0.028535560337128445,
|
| 354 |
+
"acc_norm": 0.726530612244898,
|
| 355 |
+
"acc_norm_stderr": 0.028535560337128445
|
| 356 |
+
},
|
| 357 |
+
"harness|hendrycksTest-sociology|5": {
|
| 358 |
+
"acc": 0.845771144278607,
|
| 359 |
+
"acc_stderr": 0.025538433368578334,
|
| 360 |
+
"acc_norm": 0.845771144278607,
|
| 361 |
+
"acc_norm_stderr": 0.025538433368578334
|
| 362 |
+
},
|
| 363 |
+
"harness|hendrycksTest-us_foreign_policy|5": {
|
| 364 |
+
"acc": 0.85,
|
| 365 |
+
"acc_stderr": 0.03588702812826371,
|
| 366 |
+
"acc_norm": 0.85,
|
| 367 |
+
"acc_norm_stderr": 0.03588702812826371
|
| 368 |
+
},
|
| 369 |
+
"harness|hendrycksTest-virology|5": {
|
| 370 |
+
"acc": 0.5542168674698795,
|
| 371 |
+
"acc_stderr": 0.03869543323472101,
|
| 372 |
+
"acc_norm": 0.5542168674698795,
|
| 373 |
+
"acc_norm_stderr": 0.03869543323472101
|
| 374 |
+
},
|
| 375 |
+
"harness|hendrycksTest-world_religions|5": {
|
| 376 |
+
"acc": 0.8362573099415205,
|
| 377 |
+
"acc_stderr": 0.028380919596145866,
|
| 378 |
+
"acc_norm": 0.8362573099415205,
|
| 379 |
+
"acc_norm_stderr": 0.028380919596145866
|
| 380 |
+
},
|
| 381 |
+
"harness|truthfulqa:mc|0": {
|
| 382 |
+
"mc1": 0.25091799265605874,
|
| 383 |
+
"mc1_stderr": 0.015176985027707694,
|
| 384 |
+
"mc2": 0.5006656333594469,
|
| 385 |
+
"mc2_stderr": 0.01636490303268174
|
| 386 |
+
},
|
| 387 |
+
"harness|winogrande|5": {
|
| 388 |
+
"acc": 0.7261247040252565,
|
| 389 |
+
"acc_stderr": 0.012533292732620296
|
| 390 |
+
},
|
| 391 |
+
"harness|gsm8k|5": {
|
| 392 |
+
"acc": 0.0,
|
| 393 |
+
"acc_stderr": 0.0
|
| 394 |
+
}
|
| 395 |
+
}
|
| 396 |
+
|
| 397 |
+
```
|