Update README.md
Browse files
README.md
CHANGED
|
@@ -8,4 +8,392 @@ tags:
|
|
| 8 |
|
| 9 |
Merge of top 7B models with TIES 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 TIES 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 |
+
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
```python
|
| 19 |
+
{
|
| 20 |
+
"all": {
|
| 21 |
+
"acc": 0.6487801278765712,
|
| 22 |
+
"acc_stderr": 0.03219011246717665,
|
| 23 |
+
"acc_norm": 0.6479445077777353,
|
| 24 |
+
"acc_norm_stderr": 0.032868022907407396,
|
| 25 |
+
"mc1": 0.5862913096695227,
|
| 26 |
+
"mc1_stderr": 0.0172408618120998,
|
| 27 |
+
"mc2": 0.7078078883926877,
|
| 28 |
+
"mc2_stderr": 0.015097515102384168
|
| 29 |
+
},
|
| 30 |
+
"harness|arc:challenge|25": {
|
| 31 |
+
"acc": 0.7167235494880546,
|
| 32 |
+
"acc_stderr": 0.013167478735134575,
|
| 33 |
+
"acc_norm": 0.7363481228668942,
|
| 34 |
+
"acc_norm_stderr": 0.012875929151297044
|
| 35 |
+
},
|
| 36 |
+
"harness|hellaswag|10": {
|
| 37 |
+
"acc": 0.7321250746863175,
|
| 38 |
+
"acc_stderr": 0.004419469983939178,
|
| 39 |
+
"acc_norm": 0.8884684325831508,
|
| 40 |
+
"acc_norm_stderr": 0.0031414591751392717
|
| 41 |
+
},
|
| 42 |
+
"harness|hendrycksTest-abstract_algebra|5": {
|
| 43 |
+
"acc": 0.31,
|
| 44 |
+
"acc_stderr": 0.04648231987117316,
|
| 45 |
+
"acc_norm": 0.31,
|
| 46 |
+
"acc_norm_stderr": 0.04648231987117316
|
| 47 |
+
},
|
| 48 |
+
"harness|hendrycksTest-anatomy|5": {
|
| 49 |
+
"acc": 0.6518518518518519,
|
| 50 |
+
"acc_stderr": 0.041153246103369526,
|
| 51 |
+
"acc_norm": 0.6518518518518519,
|
| 52 |
+
"acc_norm_stderr": 0.041153246103369526
|
| 53 |
+
},
|
| 54 |
+
"harness|hendrycksTest-astronomy|5": {
|
| 55 |
+
"acc": 0.7039473684210527,
|
| 56 |
+
"acc_stderr": 0.03715062154998904,
|
| 57 |
+
"acc_norm": 0.7039473684210527,
|
| 58 |
+
"acc_norm_stderr": 0.03715062154998904
|
| 59 |
+
},
|
| 60 |
+
"harness|hendrycksTest-business_ethics|5": {
|
| 61 |
+
"acc": 0.61,
|
| 62 |
+
"acc_stderr": 0.04902071300001975,
|
| 63 |
+
"acc_norm": 0.61,
|
| 64 |
+
"acc_norm_stderr": 0.04902071300001975
|
| 65 |
+
},
|
| 66 |
+
"harness|hendrycksTest-clinical_knowledge|5": {
|
| 67 |
+
"acc": 0.7132075471698113,
|
| 68 |
+
"acc_stderr": 0.02783491252754407,
|
| 69 |
+
"acc_norm": 0.7132075471698113,
|
| 70 |
+
"acc_norm_stderr": 0.02783491252754407
|
| 71 |
+
},
|
| 72 |
+
"harness|hendrycksTest-college_biology|5": {
|
| 73 |
+
"acc": 0.75,
|
| 74 |
+
"acc_stderr": 0.03621034121889507,
|
| 75 |
+
"acc_norm": 0.75,
|
| 76 |
+
"acc_norm_stderr": 0.03621034121889507
|
| 77 |
+
},
|
| 78 |
+
"harness|hendrycksTest-college_chemistry|5": {
|
| 79 |
+
"acc": 0.46,
|
| 80 |
+
"acc_stderr": 0.05009082659620333,
|
| 81 |
+
"acc_norm": 0.46,
|
| 82 |
+
"acc_norm_stderr": 0.05009082659620333
|
| 83 |
+
},
|
| 84 |
+
"harness|hendrycksTest-college_computer_science|5": {
|
| 85 |
+
"acc": 0.55,
|
| 86 |
+
"acc_stderr": 0.05,
|
| 87 |
+
"acc_norm": 0.55,
|
| 88 |
+
"acc_norm_stderr": 0.05
|
| 89 |
+
},
|
| 90 |
+
"harness|hendrycksTest-college_mathematics|5": {
|
| 91 |
+
"acc": 0.3,
|
| 92 |
+
"acc_stderr": 0.046056618647183814,
|
| 93 |
+
"acc_norm": 0.3,
|
| 94 |
+
"acc_norm_stderr": 0.046056618647183814
|
| 95 |
+
},
|
| 96 |
+
"harness|hendrycksTest-college_medicine|5": {
|
| 97 |
+
"acc": 0.6589595375722543,
|
| 98 |
+
"acc_stderr": 0.036146654241808254,
|
| 99 |
+
"acc_norm": 0.6589595375722543,
|
| 100 |
+
"acc_norm_stderr": 0.036146654241808254
|
| 101 |
+
},
|
| 102 |
+
"harness|hendrycksTest-college_physics|5": {
|
| 103 |
+
"acc": 0.43137254901960786,
|
| 104 |
+
"acc_stderr": 0.04928099597287534,
|
| 105 |
+
"acc_norm": 0.43137254901960786,
|
| 106 |
+
"acc_norm_stderr": 0.04928099597287534
|
| 107 |
+
},
|
| 108 |
+
"harness|hendrycksTest-computer_security|5": {
|
| 109 |
+
"acc": 0.77,
|
| 110 |
+
"acc_stderr": 0.04229525846816506,
|
| 111 |
+
"acc_norm": 0.77,
|
| 112 |
+
"acc_norm_stderr": 0.04229525846816506
|
| 113 |
+
},
|
| 114 |
+
"harness|hendrycksTest-conceptual_physics|5": {
|
| 115 |
+
"acc": 0.548936170212766,
|
| 116 |
+
"acc_stderr": 0.032529096196131965,
|
| 117 |
+
"acc_norm": 0.548936170212766,
|
| 118 |
+
"acc_norm_stderr": 0.032529096196131965
|
| 119 |
+
},
|
| 120 |
+
"harness|hendrycksTest-econometrics|5": {
|
| 121 |
+
"acc": 0.49122807017543857,
|
| 122 |
+
"acc_stderr": 0.04702880432049615,
|
| 123 |
+
"acc_norm": 0.49122807017543857,
|
| 124 |
+
"acc_norm_stderr": 0.04702880432049615
|
| 125 |
+
},
|
| 126 |
+
"harness|hendrycksTest-electrical_engineering|5": {
|
| 127 |
+
"acc": 0.5517241379310345,
|
| 128 |
+
"acc_stderr": 0.04144311810878152,
|
| 129 |
+
"acc_norm": 0.5517241379310345,
|
| 130 |
+
"acc_norm_stderr": 0.04144311810878152
|
| 131 |
+
},
|
| 132 |
+
"harness|hendrycksTest-elementary_mathematics|5": {
|
| 133 |
+
"acc": 0.4126984126984127,
|
| 134 |
+
"acc_stderr": 0.025355741263055277,
|
| 135 |
+
"acc_norm": 0.4126984126984127,
|
| 136 |
+
"acc_norm_stderr": 0.025355741263055277
|
| 137 |
+
},
|
| 138 |
+
"harness|hendrycksTest-formal_logic|5": {
|
| 139 |
+
"acc": 0.49206349206349204,
|
| 140 |
+
"acc_stderr": 0.044715725362943486,
|
| 141 |
+
"acc_norm": 0.49206349206349204,
|
| 142 |
+
"acc_norm_stderr": 0.044715725362943486
|
| 143 |
+
},
|
| 144 |
+
"harness|hendrycksTest-global_facts|5": {
|
| 145 |
+
"acc": 0.35,
|
| 146 |
+
"acc_stderr": 0.047937248544110196,
|
| 147 |
+
"acc_norm": 0.35,
|
| 148 |
+
"acc_norm_stderr": 0.047937248544110196
|
| 149 |
+
},
|
| 150 |
+
"harness|hendrycksTest-high_school_biology|5": {
|
| 151 |
+
"acc": 0.7967741935483871,
|
| 152 |
+
"acc_stderr": 0.02289168798455496,
|
| 153 |
+
"acc_norm": 0.7967741935483871,
|
| 154 |
+
"acc_norm_stderr": 0.02289168798455496
|
| 155 |
+
},
|
| 156 |
+
"harness|hendrycksTest-high_school_chemistry|5": {
|
| 157 |
+
"acc": 0.5024630541871922,
|
| 158 |
+
"acc_stderr": 0.035179450386910616,
|
| 159 |
+
"acc_norm": 0.5024630541871922,
|
| 160 |
+
"acc_norm_stderr": 0.035179450386910616
|
| 161 |
+
},
|
| 162 |
+
"harness|hendrycksTest-high_school_computer_science|5": {
|
| 163 |
+
"acc": 0.7,
|
| 164 |
+
"acc_stderr": 0.046056618647183814,
|
| 165 |
+
"acc_norm": 0.7,
|
| 166 |
+
"acc_norm_stderr": 0.046056618647183814
|
| 167 |
+
},
|
| 168 |
+
"harness|hendrycksTest-high_school_european_history|5": {
|
| 169 |
+
"acc": 0.7575757575757576,
|
| 170 |
+
"acc_stderr": 0.03346409881055953,
|
| 171 |
+
"acc_norm": 0.7575757575757576,
|
| 172 |
+
"acc_norm_stderr": 0.03346409881055953
|
| 173 |
+
},
|
| 174 |
+
"harness|hendrycksTest-high_school_geography|5": {
|
| 175 |
+
"acc": 0.803030303030303,
|
| 176 |
+
"acc_stderr": 0.028335609732463362,
|
| 177 |
+
"acc_norm": 0.803030303030303,
|
| 178 |
+
"acc_norm_stderr": 0.028335609732463362
|
| 179 |
+
},
|
| 180 |
+
"harness|hendrycksTest-high_school_government_and_politics|5": {
|
| 181 |
+
"acc": 0.9067357512953368,
|
| 182 |
+
"acc_stderr": 0.020986854593289733,
|
| 183 |
+
"acc_norm": 0.9067357512953368,
|
| 184 |
+
"acc_norm_stderr": 0.020986854593289733
|
| 185 |
+
},
|
| 186 |
+
"harness|hendrycksTest-high_school_macroeconomics|5": {
|
| 187 |
+
"acc": 0.6487179487179487,
|
| 188 |
+
"acc_stderr": 0.024203665177902803,
|
| 189 |
+
"acc_norm": 0.6487179487179487,
|
| 190 |
+
"acc_norm_stderr": 0.024203665177902803
|
| 191 |
+
},
|
| 192 |
+
"harness|hendrycksTest-high_school_mathematics|5": {
|
| 193 |
+
"acc": 0.3333333333333333,
|
| 194 |
+
"acc_stderr": 0.02874204090394848,
|
| 195 |
+
"acc_norm": 0.3333333333333333,
|
| 196 |
+
"acc_norm_stderr": 0.02874204090394848
|
| 197 |
+
},
|
| 198 |
+
"harness|hendrycksTest-high_school_microeconomics|5": {
|
| 199 |
+
"acc": 0.6554621848739496,
|
| 200 |
+
"acc_stderr": 0.03086868260412162,
|
| 201 |
+
"acc_norm": 0.6554621848739496,
|
| 202 |
+
"acc_norm_stderr": 0.03086868260412162
|
| 203 |
+
},
|
| 204 |
+
"harness|hendrycksTest-high_school_physics|5": {
|
| 205 |
+
"acc": 0.32450331125827814,
|
| 206 |
+
"acc_stderr": 0.038227469376587525,
|
| 207 |
+
"acc_norm": 0.32450331125827814,
|
| 208 |
+
"acc_norm_stderr": 0.038227469376587525
|
| 209 |
+
},
|
| 210 |
+
"harness|hendrycksTest-high_school_psychology|5": {
|
| 211 |
+
"acc": 0.8403669724770643,
|
| 212 |
+
"acc_stderr": 0.015703498348461763,
|
| 213 |
+
"acc_norm": 0.8403669724770643,
|
| 214 |
+
"acc_norm_stderr": 0.015703498348461763
|
| 215 |
+
},
|
| 216 |
+
"harness|hendrycksTest-high_school_statistics|5": {
|
| 217 |
+
"acc": 0.5046296296296297,
|
| 218 |
+
"acc_stderr": 0.03409825519163572,
|
| 219 |
+
"acc_norm": 0.5046296296296297,
|
| 220 |
+
"acc_norm_stderr": 0.03409825519163572
|
| 221 |
+
},
|
| 222 |
+
"harness|hendrycksTest-high_school_us_history|5": {
|
| 223 |
+
"acc": 0.8235294117647058,
|
| 224 |
+
"acc_stderr": 0.026756401538078962,
|
| 225 |
+
"acc_norm": 0.8235294117647058,
|
| 226 |
+
"acc_norm_stderr": 0.026756401538078962
|
| 227 |
+
},
|
| 228 |
+
"harness|hendrycksTest-high_school_world_history|5": {
|
| 229 |
+
"acc": 0.7721518987341772,
|
| 230 |
+
"acc_stderr": 0.02730348459906944,
|
| 231 |
+
"acc_norm": 0.7721518987341772,
|
| 232 |
+
"acc_norm_stderr": 0.02730348459906944
|
| 233 |
+
},
|
| 234 |
+
"harness|hendrycksTest-human_aging|5": {
|
| 235 |
+
"acc": 0.6816143497757847,
|
| 236 |
+
"acc_stderr": 0.03126580522513713,
|
| 237 |
+
"acc_norm": 0.6816143497757847,
|
| 238 |
+
"acc_norm_stderr": 0.03126580522513713
|
| 239 |
+
},
|
| 240 |
+
"harness|hendrycksTest-human_sexuality|5": {
|
| 241 |
+
"acc": 0.7862595419847328,
|
| 242 |
+
"acc_stderr": 0.0359546161177469,
|
| 243 |
+
"acc_norm": 0.7862595419847328,
|
| 244 |
+
"acc_norm_stderr": 0.0359546161177469
|
| 245 |
+
},
|
| 246 |
+
"harness|hendrycksTest-international_law|5": {
|
| 247 |
+
"acc": 0.7851239669421488,
|
| 248 |
+
"acc_stderr": 0.037494924487096966,
|
| 249 |
+
"acc_norm": 0.7851239669421488,
|
| 250 |
+
"acc_norm_stderr": 0.037494924487096966
|
| 251 |
+
},
|
| 252 |
+
"harness|hendrycksTest-jurisprudence|5": {
|
| 253 |
+
"acc": 0.7777777777777778,
|
| 254 |
+
"acc_stderr": 0.0401910747255735,
|
| 255 |
+
"acc_norm": 0.7777777777777778,
|
| 256 |
+
"acc_norm_stderr": 0.0401910747255735
|
| 257 |
+
},
|
| 258 |
+
"harness|hendrycksTest-logical_fallacies|5": {
|
| 259 |
+
"acc": 0.7423312883435583,
|
| 260 |
+
"acc_stderr": 0.03436150827846917,
|
| 261 |
+
"acc_norm": 0.7423312883435583,
|
| 262 |
+
"acc_norm_stderr": 0.03436150827846917
|
| 263 |
+
},
|
| 264 |
+
"harness|hendrycksTest-machine_learning|5": {
|
| 265 |
+
"acc": 0.42857142857142855,
|
| 266 |
+
"acc_stderr": 0.04697113923010212,
|
| 267 |
+
"acc_norm": 0.42857142857142855,
|
| 268 |
+
"acc_norm_stderr": 0.04697113923010212
|
| 269 |
+
},
|
| 270 |
+
"harness|hendrycksTest-management|5": {
|
| 271 |
+
"acc": 0.7475728155339806,
|
| 272 |
+
"acc_stderr": 0.04301250399690878,
|
| 273 |
+
"acc_norm": 0.7475728155339806,
|
| 274 |
+
"acc_norm_stderr": 0.04301250399690878
|
| 275 |
+
},
|
| 276 |
+
"harness|hendrycksTest-marketing|5": {
|
| 277 |
+
"acc": 0.8846153846153846,
|
| 278 |
+
"acc_stderr": 0.02093019318517933,
|
| 279 |
+
"acc_norm": 0.8846153846153846,
|
| 280 |
+
"acc_norm_stderr": 0.02093019318517933
|
| 281 |
+
},
|
| 282 |
+
"harness|hendrycksTest-medical_genetics|5": {
|
| 283 |
+
"acc": 0.7,
|
| 284 |
+
"acc_stderr": 0.046056618647183814,
|
| 285 |
+
"acc_norm": 0.7,
|
| 286 |
+
"acc_norm_stderr": 0.046056618647183814
|
| 287 |
+
},
|
| 288 |
+
"harness|hendrycksTest-miscellaneous|5": {
|
| 289 |
+
"acc": 0.80970625798212,
|
| 290 |
+
"acc_stderr": 0.014036945850381396,
|
| 291 |
+
"acc_norm": 0.80970625798212,
|
| 292 |
+
"acc_norm_stderr": 0.014036945850381396
|
| 293 |
+
},
|
| 294 |
+
"harness|hendrycksTest-moral_disputes|5": {
|
| 295 |
+
"acc": 0.7369942196531792,
|
| 296 |
+
"acc_stderr": 0.023703099525258172,
|
| 297 |
+
"acc_norm": 0.7369942196531792,
|
| 298 |
+
"acc_norm_stderr": 0.023703099525258172
|
| 299 |
+
},
|
| 300 |
+
"harness|hendrycksTest-moral_scenarios|5": {
|
| 301 |
+
"acc": 0.47150837988826816,
|
| 302 |
+
"acc_stderr": 0.016695329746015796,
|
| 303 |
+
"acc_norm": 0.47150837988826816,
|
| 304 |
+
"acc_norm_stderr": 0.016695329746015796
|
| 305 |
+
},
|
| 306 |
+
"harness|hendrycksTest-nutrition|5": {
|
| 307 |
+
"acc": 0.7189542483660131,
|
| 308 |
+
"acc_stderr": 0.025738854797818733,
|
| 309 |
+
"acc_norm": 0.7189542483660131,
|
| 310 |
+
"acc_norm_stderr": 0.025738854797818733
|
| 311 |
+
},
|
| 312 |
+
"harness|hendrycksTest-philosophy|5": {
|
| 313 |
+
"acc": 0.7170418006430869,
|
| 314 |
+
"acc_stderr": 0.025583062489984813,
|
| 315 |
+
"acc_norm": 0.7170418006430869,
|
| 316 |
+
"acc_norm_stderr": 0.025583062489984813
|
| 317 |
+
},
|
| 318 |
+
"harness|hendrycksTest-prehistory|5": {
|
| 319 |
+
"acc": 0.7407407407407407,
|
| 320 |
+
"acc_stderr": 0.024383665531035457,
|
| 321 |
+
"acc_norm": 0.7407407407407407,
|
| 322 |
+
"acc_norm_stderr": 0.024383665531035457
|
| 323 |
+
},
|
| 324 |
+
"harness|hendrycksTest-professional_accounting|5": {
|
| 325 |
+
"acc": 0.475177304964539,
|
| 326 |
+
"acc_stderr": 0.029790719243829727,
|
| 327 |
+
"acc_norm": 0.475177304964539,
|
| 328 |
+
"acc_norm_stderr": 0.029790719243829727
|
| 329 |
+
},
|
| 330 |
+
"harness|hendrycksTest-professional_law|5": {
|
| 331 |
+
"acc": 0.470013037809648,
|
| 332 |
+
"acc_stderr": 0.01274724896707906,
|
| 333 |
+
"acc_norm": 0.470013037809648,
|
| 334 |
+
"acc_norm_stderr": 0.01274724896707906
|
| 335 |
+
},
|
| 336 |
+
"harness|hendrycksTest-professional_medicine|5": {
|
| 337 |
+
"acc": 0.6691176470588235,
|
| 338 |
+
"acc_stderr": 0.028582709753898445,
|
| 339 |
+
"acc_norm": 0.6691176470588235,
|
| 340 |
+
"acc_norm_stderr": 0.028582709753898445
|
| 341 |
+
},
|
| 342 |
+
"harness|hendrycksTest-professional_psychology|5": {
|
| 343 |
+
"acc": 0.6584967320261438,
|
| 344 |
+
"acc_stderr": 0.019184639328092487,
|
| 345 |
+
"acc_norm": 0.6584967320261438,
|
| 346 |
+
"acc_norm_stderr": 0.019184639328092487
|
| 347 |
+
},
|
| 348 |
+
"harness|hendrycksTest-public_relations|5": {
|
| 349 |
+
"acc": 0.6818181818181818,
|
| 350 |
+
"acc_stderr": 0.044612721759105085,
|
| 351 |
+
"acc_norm": 0.6818181818181818,
|
| 352 |
+
"acc_norm_stderr": 0.044612721759105085
|
| 353 |
+
},
|
| 354 |
+
"harness|hendrycksTest-security_studies|5": {
|
| 355 |
+
"acc": 0.7306122448979592,
|
| 356 |
+
"acc_stderr": 0.02840125202902294,
|
| 357 |
+
"acc_norm": 0.7306122448979592,
|
| 358 |
+
"acc_norm_stderr": 0.02840125202902294
|
| 359 |
+
},
|
| 360 |
+
"harness|hendrycksTest-sociology|5": {
|
| 361 |
+
"acc": 0.835820895522388,
|
| 362 |
+
"acc_stderr": 0.026193923544454125,
|
| 363 |
+
"acc_norm": 0.835820895522388,
|
| 364 |
+
"acc_norm_stderr": 0.026193923544454125
|
| 365 |
+
},
|
| 366 |
+
"harness|hendrycksTest-us_foreign_policy|5": {
|
| 367 |
+
"acc": 0.85,
|
| 368 |
+
"acc_stderr": 0.03588702812826371,
|
| 369 |
+
"acc_norm": 0.85,
|
| 370 |
+
"acc_norm_stderr": 0.03588702812826371
|
| 371 |
+
},
|
| 372 |
+
"harness|hendrycksTest-virology|5": {
|
| 373 |
+
"acc": 0.5542168674698795,
|
| 374 |
+
"acc_stderr": 0.03869543323472101,
|
| 375 |
+
"acc_norm": 0.5542168674698795,
|
| 376 |
+
"acc_norm_stderr": 0.03869543323472101
|
| 377 |
+
},
|
| 378 |
+
"harness|hendrycksTest-world_religions|5": {
|
| 379 |
+
"acc": 0.8245614035087719,
|
| 380 |
+
"acc_stderr": 0.029170885500727665,
|
| 381 |
+
"acc_norm": 0.8245614035087719,
|
| 382 |
+
"acc_norm_stderr": 0.029170885500727665
|
| 383 |
+
},
|
| 384 |
+
"harness|truthfulqa:mc|0": {
|
| 385 |
+
"mc1": 0.5862913096695227,
|
| 386 |
+
"mc1_stderr": 0.0172408618120998,
|
| 387 |
+
"mc2": 0.7078078883926877,
|
| 388 |
+
"mc2_stderr": 0.015097515102384168
|
| 389 |
+
},
|
| 390 |
+
"harness|winogrande|5": {
|
| 391 |
+
"acc": 0.8579321231254933,
|
| 392 |
+
"acc_stderr": 0.009812000391679367
|
| 393 |
+
},
|
| 394 |
+
"harness|gsm8k|5": {
|
| 395 |
+
"acc": 0.6648976497346475,
|
| 396 |
+
"acc_stderr": 0.013001948176422954
|
| 397 |
+
}
|
| 398 |
+
}
|
| 399 |
+
```
|