Upload results.json with huggingface_hub
Browse files- results.json +360 -0
results.json
ADDED
|
@@ -0,0 +1,360 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"d_sae": 16384,
|
| 3 |
+
"actual_l0": 329.3830754,
|
| 4 |
+
"effective_l0": 162,
|
| 5 |
+
"cross_entropy": 905.6413854492188,
|
| 6 |
+
"sae_type": "saebench",
|
| 7 |
+
"sae_release": "canrager/saebench_gemma-2-2b_width-2pow14_date-0107",
|
| 8 |
+
"num_tokens": 10000000,
|
| 9 |
+
"num_epochs": 5,
|
| 10 |
+
"lr": 0.0005,
|
| 11 |
+
"loaded_dag": null,
|
| 12 |
+
"allow_negative_influence": false,
|
| 13 |
+
"script_runtime_seconds": 6022.779719829559,
|
| 14 |
+
"script_runtime_minutes": 100.37966199715932,
|
| 15 |
+
"history": {
|
| 16 |
+
"train_loss": [
|
| 17 |
+
1420.3783656529017,
|
| 18 |
+
1094.5040370396205,
|
| 19 |
+
1026.2416063581195,
|
| 20 |
+
992.3883841378348,
|
| 21 |
+
972.653471156529,
|
| 22 |
+
960.1246682303292,
|
| 23 |
+
950.369138445173,
|
| 24 |
+
943.7700535365514,
|
| 25 |
+
937.6258479527065,
|
| 26 |
+
934.2963500976563,
|
| 27 |
+
933.885109601702,
|
| 28 |
+
929.750561959403,
|
| 29 |
+
927.3069841657366,
|
| 30 |
+
923.8627319335938,
|
| 31 |
+
922.0174486432757,
|
| 32 |
+
920.4228502546038,
|
| 33 |
+
918.5081556047712,
|
| 34 |
+
918.1200487409319,
|
| 35 |
+
916.3387438092913,
|
| 36 |
+
914.9147931780134,
|
| 37 |
+
914.2666098458426,
|
| 38 |
+
914.2695595877511,
|
| 39 |
+
912.6019465855189,
|
| 40 |
+
912.2011243547712,
|
| 41 |
+
911.7710750034877,
|
| 42 |
+
911.1132032121931,
|
| 43 |
+
909.8142521449498,
|
| 44 |
+
909.6818006243025,
|
| 45 |
+
910.0303174700056,
|
| 46 |
+
909.4441000802176,
|
| 47 |
+
908.6869995117188,
|
| 48 |
+
908.6951054164341,
|
| 49 |
+
908.3833173479353,
|
| 50 |
+
907.2083766392299,
|
| 51 |
+
906.7662227085658,
|
| 52 |
+
907.3851693289621,
|
| 53 |
+
906.7015110560826,
|
| 54 |
+
906.4677764892579,
|
| 55 |
+
906.2122715541294,
|
| 56 |
+
906.6031829833985,
|
| 57 |
+
904.8416756766183,
|
| 58 |
+
904.7858734130859,
|
| 59 |
+
905.7477992466518,
|
| 60 |
+
905.9446166992187,
|
| 61 |
+
903.9516919817244,
|
| 62 |
+
905.1117710658482,
|
| 63 |
+
904.4221426827567,
|
| 64 |
+
904.0226558140346,
|
| 65 |
+
903.8429635184152,
|
| 66 |
+
904.1890664236886,
|
| 67 |
+
904.3355904715402,
|
| 68 |
+
903.3897670200893,
|
| 69 |
+
903.8843309674944,
|
| 70 |
+
902.9499965122768,
|
| 71 |
+
902.9998753138951
|
| 72 |
+
],
|
| 73 |
+
"train_steps": [
|
| 74 |
+
140,
|
| 75 |
+
280,
|
| 76 |
+
420,
|
| 77 |
+
560,
|
| 78 |
+
700,
|
| 79 |
+
840,
|
| 80 |
+
980,
|
| 81 |
+
1120,
|
| 82 |
+
1260,
|
| 83 |
+
1400,
|
| 84 |
+
1407,
|
| 85 |
+
1547,
|
| 86 |
+
1687,
|
| 87 |
+
1827,
|
| 88 |
+
1967,
|
| 89 |
+
2107,
|
| 90 |
+
2247,
|
| 91 |
+
2387,
|
| 92 |
+
2527,
|
| 93 |
+
2667,
|
| 94 |
+
2807,
|
| 95 |
+
2814,
|
| 96 |
+
2954,
|
| 97 |
+
3094,
|
| 98 |
+
3234,
|
| 99 |
+
3374,
|
| 100 |
+
3514,
|
| 101 |
+
3654,
|
| 102 |
+
3794,
|
| 103 |
+
3934,
|
| 104 |
+
4074,
|
| 105 |
+
4214,
|
| 106 |
+
4221,
|
| 107 |
+
4361,
|
| 108 |
+
4501,
|
| 109 |
+
4641,
|
| 110 |
+
4781,
|
| 111 |
+
4921,
|
| 112 |
+
5061,
|
| 113 |
+
5201,
|
| 114 |
+
5341,
|
| 115 |
+
5481,
|
| 116 |
+
5621,
|
| 117 |
+
5628,
|
| 118 |
+
5768,
|
| 119 |
+
5908,
|
| 120 |
+
6048,
|
| 121 |
+
6188,
|
| 122 |
+
6328,
|
| 123 |
+
6468,
|
| 124 |
+
6608,
|
| 125 |
+
6748,
|
| 126 |
+
6888,
|
| 127 |
+
7028,
|
| 128 |
+
7035
|
| 129 |
+
],
|
| 130 |
+
"val_loss": [
|
| 131 |
+
1157.242766015625,
|
| 132 |
+
1051.6337751953124,
|
| 133 |
+
1006.9013120117188,
|
| 134 |
+
982.134125390625,
|
| 135 |
+
966.36363984375,
|
| 136 |
+
955.4471161132813,
|
| 137 |
+
947.4677446289063,
|
| 138 |
+
941.3503963867188,
|
| 139 |
+
936.4663686523437,
|
| 140 |
+
932.56058984375,
|
| 141 |
+
933.4227443359375,
|
| 142 |
+
929.2720069335937,
|
| 143 |
+
926.6685235351563,
|
| 144 |
+
924.4171421875,
|
| 145 |
+
922.4184139648437,
|
| 146 |
+
920.751034765625,
|
| 147 |
+
919.2843603515626,
|
| 148 |
+
917.9932103515625,
|
| 149 |
+
916.7930734375,
|
| 150 |
+
915.7856177734375,
|
| 151 |
+
914.8475911132813,
|
| 152 |
+
915.826787890625,
|
| 153 |
+
914.0802901367188,
|
| 154 |
+
913.3703591796875,
|
| 155 |
+
912.714684375,
|
| 156 |
+
912.0590510742187,
|
| 157 |
+
911.4835946289063,
|
| 158 |
+
910.949715625,
|
| 159 |
+
910.43665703125,
|
| 160 |
+
910.0339327148438,
|
| 161 |
+
909.6020287109375,
|
| 162 |
+
909.1736103515625,
|
| 163 |
+
910.112005859375,
|
| 164 |
+
908.8126391601562,
|
| 165 |
+
908.5509233398437,
|
| 166 |
+
908.2231232421875,
|
| 167 |
+
907.8818041992188,
|
| 168 |
+
907.58483828125,
|
| 169 |
+
907.31560234375,
|
| 170 |
+
907.0377946289062,
|
| 171 |
+
906.7497403320313,
|
| 172 |
+
906.5649120117188,
|
| 173 |
+
906.3602444335937,
|
| 174 |
+
907.458184375,
|
| 175 |
+
906.1869296875,
|
| 176 |
+
905.9936012695313,
|
| 177 |
+
905.83822421875,
|
| 178 |
+
905.6570301757813,
|
| 179 |
+
905.4729654296875,
|
| 180 |
+
905.3166546875,
|
| 181 |
+
905.1094328125,
|
| 182 |
+
904.9633212890625,
|
| 183 |
+
904.7684698242188,
|
| 184 |
+
904.632570703125,
|
| 185 |
+
905.6413854492188
|
| 186 |
+
],
|
| 187 |
+
"val_cross_entropy": [
|
| 188 |
+
1157.242766015625,
|
| 189 |
+
1051.6337751953124,
|
| 190 |
+
1006.9013120117188,
|
| 191 |
+
982.134125390625,
|
| 192 |
+
966.36363984375,
|
| 193 |
+
955.4471161132813,
|
| 194 |
+
947.4677446289063,
|
| 195 |
+
941.3503963867188,
|
| 196 |
+
936.4663686523437,
|
| 197 |
+
932.56058984375,
|
| 198 |
+
933.4227443359375,
|
| 199 |
+
929.2720069335937,
|
| 200 |
+
926.6685235351563,
|
| 201 |
+
924.4171421875,
|
| 202 |
+
922.4184139648437,
|
| 203 |
+
920.751034765625,
|
| 204 |
+
919.2843603515626,
|
| 205 |
+
917.9932103515625,
|
| 206 |
+
916.7930734375,
|
| 207 |
+
915.7856177734375,
|
| 208 |
+
914.8475911132813,
|
| 209 |
+
915.826787890625,
|
| 210 |
+
914.0802901367188,
|
| 211 |
+
913.3703591796875,
|
| 212 |
+
912.714684375,
|
| 213 |
+
912.0590510742187,
|
| 214 |
+
911.4835946289063,
|
| 215 |
+
910.949715625,
|
| 216 |
+
910.43665703125,
|
| 217 |
+
910.0339327148438,
|
| 218 |
+
909.6020287109375,
|
| 219 |
+
909.1736103515625,
|
| 220 |
+
910.112005859375,
|
| 221 |
+
908.8126391601562,
|
| 222 |
+
908.5509233398437,
|
| 223 |
+
908.2231232421875,
|
| 224 |
+
907.8818041992188,
|
| 225 |
+
907.58483828125,
|
| 226 |
+
907.31560234375,
|
| 227 |
+
907.0377946289062,
|
| 228 |
+
906.7497403320313,
|
| 229 |
+
906.5649120117188,
|
| 230 |
+
906.3602444335937,
|
| 231 |
+
907.458184375,
|
| 232 |
+
906.1869296875,
|
| 233 |
+
905.9936012695313,
|
| 234 |
+
905.83822421875,
|
| 235 |
+
905.6570301757813,
|
| 236 |
+
905.4729654296875,
|
| 237 |
+
905.3166546875,
|
| 238 |
+
905.1094328125,
|
| 239 |
+
904.9633212890625,
|
| 240 |
+
904.7684698242188,
|
| 241 |
+
904.632570703125,
|
| 242 |
+
905.6413854492188
|
| 243 |
+
],
|
| 244 |
+
"effective_l0": [
|
| 245 |
+
219,
|
| 246 |
+
195,
|
| 247 |
+
185,
|
| 248 |
+
179,
|
| 249 |
+
176,
|
| 250 |
+
173,
|
| 251 |
+
171,
|
| 252 |
+
170,
|
| 253 |
+
169,
|
| 254 |
+
168,
|
| 255 |
+
168,
|
| 256 |
+
167,
|
| 257 |
+
167,
|
| 258 |
+
166,
|
| 259 |
+
166,
|
| 260 |
+
166,
|
| 261 |
+
165,
|
| 262 |
+
165,
|
| 263 |
+
165,
|
| 264 |
+
165,
|
| 265 |
+
164,
|
| 266 |
+
165,
|
| 267 |
+
164,
|
| 268 |
+
164,
|
| 269 |
+
164,
|
| 270 |
+
164,
|
| 271 |
+
164,
|
| 272 |
+
163,
|
| 273 |
+
163,
|
| 274 |
+
163,
|
| 275 |
+
163,
|
| 276 |
+
163,
|
| 277 |
+
163,
|
| 278 |
+
163,
|
| 279 |
+
163,
|
| 280 |
+
163,
|
| 281 |
+
163,
|
| 282 |
+
163,
|
| 283 |
+
163,
|
| 284 |
+
163,
|
| 285 |
+
163,
|
| 286 |
+
163,
|
| 287 |
+
162,
|
| 288 |
+
163,
|
| 289 |
+
162,
|
| 290 |
+
162,
|
| 291 |
+
162,
|
| 292 |
+
162,
|
| 293 |
+
162,
|
| 294 |
+
162,
|
| 295 |
+
162,
|
| 296 |
+
162,
|
| 297 |
+
162,
|
| 298 |
+
162,
|
| 299 |
+
162
|
| 300 |
+
],
|
| 301 |
+
"val_steps": [
|
| 302 |
+
140,
|
| 303 |
+
280,
|
| 304 |
+
420,
|
| 305 |
+
560,
|
| 306 |
+
700,
|
| 307 |
+
840,
|
| 308 |
+
980,
|
| 309 |
+
1120,
|
| 310 |
+
1260,
|
| 311 |
+
1400,
|
| 312 |
+
1407,
|
| 313 |
+
1547,
|
| 314 |
+
1687,
|
| 315 |
+
1827,
|
| 316 |
+
1967,
|
| 317 |
+
2107,
|
| 318 |
+
2247,
|
| 319 |
+
2387,
|
| 320 |
+
2527,
|
| 321 |
+
2667,
|
| 322 |
+
2807,
|
| 323 |
+
2814,
|
| 324 |
+
2954,
|
| 325 |
+
3094,
|
| 326 |
+
3234,
|
| 327 |
+
3374,
|
| 328 |
+
3514,
|
| 329 |
+
3654,
|
| 330 |
+
3794,
|
| 331 |
+
3934,
|
| 332 |
+
4074,
|
| 333 |
+
4214,
|
| 334 |
+
4221,
|
| 335 |
+
4361,
|
| 336 |
+
4501,
|
| 337 |
+
4641,
|
| 338 |
+
4781,
|
| 339 |
+
4921,
|
| 340 |
+
5061,
|
| 341 |
+
5201,
|
| 342 |
+
5341,
|
| 343 |
+
5481,
|
| 344 |
+
5621,
|
| 345 |
+
5628,
|
| 346 |
+
5768,
|
| 347 |
+
5908,
|
| 348 |
+
6048,
|
| 349 |
+
6188,
|
| 350 |
+
6328,
|
| 351 |
+
6468,
|
| 352 |
+
6608,
|
| 353 |
+
6748,
|
| 354 |
+
6888,
|
| 355 |
+
7028,
|
| 356 |
+
7035
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"sae_id": "gemma-2-2b_jump_relu_width-2pow14_date-0107/resid_post_layer_12/trainer_4"
|
| 360 |
+
}
|