Upload evals/e003_e001s1337_test.json with huggingface_hub
Browse files- evals/e003_e001s1337_test.json +245 -0
evals/e003_e001s1337_test.json
ADDED
|
@@ -0,0 +1,245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "ckpt:runs/e001_gpt2_124m_2p5B_s1337/ckpt_004768.pt",
|
| 3 |
+
"split": "test",
|
| 4 |
+
"tasks": {
|
| 5 |
+
"popqa_oracle_surr": {
|
| 6 |
+
"n": 6533,
|
| 7 |
+
"acc": 0.2472064901270473,
|
| 8 |
+
"acc_ci95": [
|
| 9 |
+
0.23664472677177406,
|
| 10 |
+
0.2574659421399051
|
| 11 |
+
],
|
| 12 |
+
"acc_norm": 0.2547068727996326,
|
| 13 |
+
"acc_norm_ci95": [
|
| 14 |
+
0.24475355885504363,
|
| 15 |
+
0.2660378080514312
|
| 16 |
+
],
|
| 17 |
+
"gold_logprob_per_tok": -4.197760838047277,
|
| 18 |
+
"gold_logprob_per_tok_ci95": [
|
| 19 |
+
-4.2696923724006135,
|
| 20 |
+
-4.125128099136931
|
| 21 |
+
],
|
| 22 |
+
"majority_baseline": 0.27812643502219503,
|
| 23 |
+
"buckets": {
|
| 24 |
+
"tail": {
|
| 25 |
+
"n": 2206,
|
| 26 |
+
"acc": 0.24161378059836808,
|
| 27 |
+
"acc_ci95": [
|
| 28 |
+
0.22484134179510426,
|
| 29 |
+
0.2606527651858567
|
| 30 |
+
],
|
| 31 |
+
"acc_norm": 0.25657298277425206,
|
| 32 |
+
"acc_norm_ci95": [
|
| 33 |
+
0.2384406165004533,
|
| 34 |
+
0.2756119673617407
|
| 35 |
+
],
|
| 36 |
+
"gold_logprob_per_tok": -4.983078423697757,
|
| 37 |
+
"gold_logprob_per_tok_ci95": [
|
| 38 |
+
-5.1388833758214325,
|
| 39 |
+
-4.834178411346546
|
| 40 |
+
],
|
| 41 |
+
"majority_baseline": 0.29782411604714415
|
| 42 |
+
},
|
| 43 |
+
"torso": {
|
| 44 |
+
"n": 2189,
|
| 45 |
+
"acc": 0.25719506624029237,
|
| 46 |
+
"acc_ci95": [
|
| 47 |
+
0.2384650525354043,
|
| 48 |
+
0.2745545911375057
|
| 49 |
+
],
|
| 50 |
+
"acc_norm": 0.2814070351758794,
|
| 51 |
+
"acc_norm_ci95": [
|
| 52 |
+
0.2635906806761078,
|
| 53 |
+
0.2996802192782092
|
| 54 |
+
],
|
| 55 |
+
"gold_logprob_per_tok": -3.719702163895861,
|
| 56 |
+
"gold_logprob_per_tok_ci95": [
|
| 57 |
+
-3.8340724114053755,
|
| 58 |
+
-3.6117194435916566
|
| 59 |
+
],
|
| 60 |
+
"majority_baseline": 0.2745545911375057
|
| 61 |
+
},
|
| 62 |
+
"head": {
|
| 63 |
+
"n": 2138,
|
| 64 |
+
"acc": 0.24275023386342376,
|
| 65 |
+
"acc_ci95": [
|
| 66 |
+
0.22404115996258187,
|
| 67 |
+
0.2605355472404116
|
| 68 |
+
],
|
| 69 |
+
"acc_norm": 0.225444340505145,
|
| 70 |
+
"acc_norm_ci95": [
|
| 71 |
+
0.20860617399438727,
|
| 72 |
+
0.24276192703461177
|
| 73 |
+
],
|
| 74 |
+
"gold_logprob_per_tok": -3.8769282111868884,
|
| 75 |
+
"gold_logprob_per_tok_ci95": [
|
| 76 |
+
-3.991595347775711,
|
| 77 |
+
-3.7685890932431336
|
| 78 |
+
],
|
| 79 |
+
"majority_baseline": 0.2614593077642657
|
| 80 |
+
}
|
| 81 |
+
},
|
| 82 |
+
"seconds": 18.9
|
| 83 |
+
},
|
| 84 |
+
"popqa_oracle_tag": {
|
| 85 |
+
"n": 6533,
|
| 86 |
+
"acc": 0.3375172202663401,
|
| 87 |
+
"acc_ci95": [
|
| 88 |
+
0.32603704270626055,
|
| 89 |
+
0.34930736262054185
|
| 90 |
+
],
|
| 91 |
+
"acc_norm": 0.3321598040716363,
|
| 92 |
+
"acc_norm_ci95": [
|
| 93 |
+
0.32067962651155674,
|
| 94 |
+
0.3442599112199602
|
| 95 |
+
],
|
| 96 |
+
"gold_logprob_per_tok": -3.101713042738754,
|
| 97 |
+
"gold_logprob_per_tok_ci95": [
|
| 98 |
+
-3.158646165370071,
|
| 99 |
+
-3.0423221162350074
|
| 100 |
+
],
|
| 101 |
+
"majority_baseline": 0.27812643502219503,
|
| 102 |
+
"buckets": {
|
| 103 |
+
"tail": {
|
| 104 |
+
"n": 2206,
|
| 105 |
+
"acc": 0.2751586582048957,
|
| 106 |
+
"acc_ci95": [
|
| 107 |
+
0.25611967361740706,
|
| 108 |
+
0.2932910244786945
|
| 109 |
+
],
|
| 110 |
+
"acc_norm": 0.2633726201269266,
|
| 111 |
+
"acc_norm_ci95": [
|
| 112 |
+
0.2443336355394379,
|
| 113 |
+
0.28059836808703537
|
| 114 |
+
],
|
| 115 |
+
"gold_logprob_per_tok": -3.5070108537588927,
|
| 116 |
+
"gold_logprob_per_tok_ci95": [
|
| 117 |
+
-3.6294212114900106,
|
| 118 |
+
-3.391425857636802
|
| 119 |
+
],
|
| 120 |
+
"majority_baseline": 0.29782411604714415
|
| 121 |
+
},
|
| 122 |
+
"torso": {
|
| 123 |
+
"n": 2189,
|
| 124 |
+
"acc": 0.32754682503426225,
|
| 125 |
+
"acc_ci95": [
|
| 126 |
+
0.3065326633165829,
|
| 127 |
+
0.34810415714938325
|
| 128 |
+
],
|
| 129 |
+
"acc_norm": 0.32891731384193695,
|
| 130 |
+
"acc_norm_ci95": [
|
| 131 |
+
0.3074463225216994,
|
| 132 |
+
0.3485609867519415
|
| 133 |
+
],
|
| 134 |
+
"gold_logprob_per_tok": -2.9954604698022567,
|
| 135 |
+
"gold_logprob_per_tok_ci95": [
|
| 136 |
+
-3.084774050218739,
|
| 137 |
+
-2.9129059081018944
|
| 138 |
+
],
|
| 139 |
+
"majority_baseline": 0.2745545911375057
|
| 140 |
+
},
|
| 141 |
+
"head": {
|
| 142 |
+
"n": 2138,
|
| 143 |
+
"acc": 0.41206735266604305,
|
| 144 |
+
"acc_ci95": [
|
| 145 |
+
0.3914873713751169,
|
| 146 |
+
0.4331150608044902
|
| 147 |
+
],
|
| 148 |
+
"acc_norm": 0.40645463049579045,
|
| 149 |
+
"acc_norm_ci95": [
|
| 150 |
+
0.38681010289990647,
|
| 151 |
+
0.42844948550046774
|
| 152 |
+
],
|
| 153 |
+
"gold_logprob_per_tok": -2.7923116914981394,
|
| 154 |
+
"gold_logprob_per_tok_ci95": [
|
| 155 |
+
-2.8772600195085034,
|
| 156 |
+
-2.7115851148481607
|
| 157 |
+
],
|
| 158 |
+
"majority_baseline": 0.2614593077642657
|
| 159 |
+
}
|
| 160 |
+
},
|
| 161 |
+
"seconds": 20.0
|
| 162 |
+
},
|
| 163 |
+
"popqa_closed_surr": {
|
| 164 |
+
"n": 6533,
|
| 165 |
+
"acc": 0.249808663707332,
|
| 166 |
+
"acc_ci95": [
|
| 167 |
+
0.23970610745446197,
|
| 168 |
+
0.2606880453084341
|
| 169 |
+
],
|
| 170 |
+
"acc_norm": 0.24965559467319762,
|
| 171 |
+
"acc_norm_ci95": [
|
| 172 |
+
0.23924307362620542,
|
| 173 |
+
0.2603704270626052
|
| 174 |
+
],
|
| 175 |
+
"gold_logprob_per_tok": -9.787470660189092,
|
| 176 |
+
"gold_logprob_per_tok_ci95": [
|
| 177 |
+
-9.841889293214633,
|
| 178 |
+
-9.730295559747786
|
| 179 |
+
],
|
| 180 |
+
"majority_baseline": 0.27812643502219503,
|
| 181 |
+
"buckets": {
|
| 182 |
+
"tail": {
|
| 183 |
+
"n": 2206,
|
| 184 |
+
"acc": 0.23526745240253852,
|
| 185 |
+
"acc_ci95": [
|
| 186 |
+
0.21848368087035358,
|
| 187 |
+
0.25339981867633726
|
| 188 |
+
],
|
| 189 |
+
"acc_norm": 0.24342701722574797,
|
| 190 |
+
"acc_norm_ci95": [
|
| 191 |
+
0.2257479601087942,
|
| 192 |
+
0.2615593834995467
|
| 193 |
+
],
|
| 194 |
+
"gold_logprob_per_tok": -10.05581066532182,
|
| 195 |
+
"gold_logprob_per_tok_ci95": [
|
| 196 |
+
-10.161450151947953,
|
| 197 |
+
-9.954193737742122
|
| 198 |
+
],
|
| 199 |
+
"majority_baseline": 0.29782411604714415
|
| 200 |
+
},
|
| 201 |
+
"torso": {
|
| 202 |
+
"n": 2189,
|
| 203 |
+
"acc": 0.22795797167656465,
|
| 204 |
+
"acc_ci95": [
|
| 205 |
+
0.2115121059844678,
|
| 206 |
+
0.245317496573778
|
| 207 |
+
],
|
| 208 |
+
"acc_norm": 0.26907263590680675,
|
| 209 |
+
"acc_norm_ci95": [
|
| 210 |
+
0.24941754225673823,
|
| 211 |
+
0.2882594792142531
|
| 212 |
+
],
|
| 213 |
+
"gold_logprob_per_tok": -9.588150490942883,
|
| 214 |
+
"gold_logprob_per_tok_ci95": [
|
| 215 |
+
-9.68061342681789,
|
| 216 |
+
-9.497315112281589
|
| 217 |
+
],
|
| 218 |
+
"majority_baseline": 0.2745545911375057
|
| 219 |
+
},
|
| 220 |
+
"head": {
|
| 221 |
+
"n": 2138,
|
| 222 |
+
"acc": 0.2871842843779233,
|
| 223 |
+
"acc_ci95": [
|
| 224 |
+
0.27034611786716556,
|
| 225 |
+
0.3068288119738073
|
| 226 |
+
],
|
| 227 |
+
"acc_norm": 0.2362020579981291,
|
| 228 |
+
"acc_norm_ci95": [
|
| 229 |
+
0.2184284377923293,
|
| 230 |
+
0.254922825070159
|
| 231 |
+
],
|
| 232 |
+
"gold_logprob_per_tok": -9.71467075334023,
|
| 233 |
+
"gold_logprob_per_tok_ci95": [
|
| 234 |
+
-9.81772005508157,
|
| 235 |
+
-9.617779475225195
|
| 236 |
+
],
|
| 237 |
+
"majority_baseline": 0.2614593077642657
|
| 238 |
+
}
|
| 239 |
+
},
|
| 240 |
+
"seconds": 9.1
|
| 241 |
+
}
|
| 242 |
+
},
|
| 243 |
+
"suite_version": "evals-v0.3-dev",
|
| 244 |
+
"limit": null
|
| 245 |
+
}
|