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": 20.0001028,
|
| 4 |
+
"effective_l0": 10,
|
| 5 |
+
"cross_entropy": 78.67905505371094,
|
| 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": 5004.448766946793,
|
| 14 |
+
"script_runtime_minutes": 83.40747944911321,
|
| 15 |
+
"history": {
|
| 16 |
+
"train_loss": [
|
| 17 |
+
134.85628967285157,
|
| 18 |
+
127.95163928440638,
|
| 19 |
+
122.03448693411691,
|
| 20 |
+
117.12716505868094,
|
| 21 |
+
113.01315977913993,
|
| 22 |
+
109.62399384634836,
|
| 23 |
+
106.74624633789062,
|
| 24 |
+
104.3243599482945,
|
| 25 |
+
102.1401078905378,
|
| 26 |
+
100.1238767351423,
|
| 27 |
+
100.04065633501325,
|
| 28 |
+
98.25842034476143,
|
| 29 |
+
96.74119698660714,
|
| 30 |
+
95.30732519967215,
|
| 31 |
+
93.89328417096819,
|
| 32 |
+
92.80071939740863,
|
| 33 |
+
91.7336601802281,
|
| 34 |
+
90.81961157662528,
|
| 35 |
+
89.7965257372175,
|
| 36 |
+
89.08478028433663,
|
| 37 |
+
88.26737207685198,
|
| 38 |
+
88.21992280142648,
|
| 39 |
+
87.36145090375628,
|
| 40 |
+
86.75979183741978,
|
| 41 |
+
86.07088748386929,
|
| 42 |
+
85.54736295427595,
|
| 43 |
+
85.01100474766322,
|
| 44 |
+
84.55322516305107,
|
| 45 |
+
84.0957924434117,
|
| 46 |
+
83.66296664646694,
|
| 47 |
+
83.28380519321986,
|
| 48 |
+
82.91472789219448,
|
| 49 |
+
82.90812541416713,
|
| 50 |
+
82.38021087646484,
|
| 51 |
+
82.08990516662598,
|
| 52 |
+
81.78877470833915,
|
| 53 |
+
81.52134170532227,
|
| 54 |
+
81.20402723039899,
|
| 55 |
+
80.98520082746234,
|
| 56 |
+
80.74978223528181,
|
| 57 |
+
80.55749991280692,
|
| 58 |
+
80.32966079711915,
|
| 59 |
+
80.16898558480399,
|
| 60 |
+
80.1669516972133,
|
| 61 |
+
79.85717146737235,
|
| 62 |
+
79.61502233232771,
|
| 63 |
+
79.43960898263114,
|
| 64 |
+
79.34254237583706,
|
| 65 |
+
79.18994750976563,
|
| 66 |
+
78.99718780517578,
|
| 67 |
+
78.91671006338937,
|
| 68 |
+
78.74960626874652,
|
| 69 |
+
78.69427800859724,
|
| 70 |
+
78.55073492867606,
|
| 71 |
+
78.5346915108817
|
| 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 |
+
131.15583425292968,
|
| 132 |
+
124.80445671386718,
|
| 133 |
+
119.39876856689453,
|
| 134 |
+
114.91041881103516,
|
| 135 |
+
111.19660385742188,
|
| 136 |
+
108.0870958251953,
|
| 137 |
+
105.4311197265625,
|
| 138 |
+
103.11729665527344,
|
| 139 |
+
101.07299282226562,
|
| 140 |
+
99.24316971435547,
|
| 141 |
+
99.15732747802734,
|
| 142 |
+
97.52390229492187,
|
| 143 |
+
96.04291837158203,
|
| 144 |
+
94.69359848632813,
|
| 145 |
+
93.46102884521484,
|
| 146 |
+
92.33263518066406,
|
| 147 |
+
91.29743673095703,
|
| 148 |
+
90.34595754394532,
|
| 149 |
+
89.47052893066406,
|
| 150 |
+
88.66277144775391,
|
| 151 |
+
87.9165919921875,
|
| 152 |
+
87.88123934326173,
|
| 153 |
+
87.1975422241211,
|
| 154 |
+
86.56350412597656,
|
| 155 |
+
85.97446087646485,
|
| 156 |
+
85.4267031982422,
|
| 157 |
+
84.91648664550782,
|
| 158 |
+
84.43963762207031,
|
| 159 |
+
83.99465645751953,
|
| 160 |
+
83.57803491210937,
|
| 161 |
+
83.18774576416016,
|
| 162 |
+
82.82235047607422,
|
| 163 |
+
82.80524810791016,
|
| 164 |
+
82.46637570800782,
|
| 165 |
+
82.14948515625,
|
| 166 |
+
81.85052250976563,
|
| 167 |
+
81.5690076538086,
|
| 168 |
+
81.3041095336914,
|
| 169 |
+
81.05420622558594,
|
| 170 |
+
80.81840328369141,
|
| 171 |
+
80.59531881103516,
|
| 172 |
+
80.38440787353515,
|
| 173 |
+
80.18435812988281,
|
| 174 |
+
80.17528757324219,
|
| 175 |
+
79.98983715820313,
|
| 176 |
+
79.81431489257812,
|
| 177 |
+
79.64884621582031,
|
| 178 |
+
79.49099832763672,
|
| 179 |
+
79.34037728271484,
|
| 180 |
+
79.19644326171876,
|
| 181 |
+
79.05946433105468,
|
| 182 |
+
78.92887561035157,
|
| 183 |
+
78.80397731933594,
|
| 184 |
+
78.68422974853516,
|
| 185 |
+
78.67905505371094
|
| 186 |
+
],
|
| 187 |
+
"val_cross_entropy": [
|
| 188 |
+
131.15583425292968,
|
| 189 |
+
124.80445671386718,
|
| 190 |
+
119.39876856689453,
|
| 191 |
+
114.91041881103516,
|
| 192 |
+
111.19660385742188,
|
| 193 |
+
108.0870958251953,
|
| 194 |
+
105.4311197265625,
|
| 195 |
+
103.11729665527344,
|
| 196 |
+
101.07299282226562,
|
| 197 |
+
99.24316971435547,
|
| 198 |
+
99.15732747802734,
|
| 199 |
+
97.52390229492187,
|
| 200 |
+
96.04291837158203,
|
| 201 |
+
94.69359848632813,
|
| 202 |
+
93.46102884521484,
|
| 203 |
+
92.33263518066406,
|
| 204 |
+
91.29743673095703,
|
| 205 |
+
90.34595754394532,
|
| 206 |
+
89.47052893066406,
|
| 207 |
+
88.66277144775391,
|
| 208 |
+
87.9165919921875,
|
| 209 |
+
87.88123934326173,
|
| 210 |
+
87.1975422241211,
|
| 211 |
+
86.56350412597656,
|
| 212 |
+
85.97446087646485,
|
| 213 |
+
85.4267031982422,
|
| 214 |
+
84.91648664550782,
|
| 215 |
+
84.43963762207031,
|
| 216 |
+
83.99465645751953,
|
| 217 |
+
83.57803491210937,
|
| 218 |
+
83.18774576416016,
|
| 219 |
+
82.82235047607422,
|
| 220 |
+
82.80524810791016,
|
| 221 |
+
82.46637570800782,
|
| 222 |
+
82.14948515625,
|
| 223 |
+
81.85052250976563,
|
| 224 |
+
81.5690076538086,
|
| 225 |
+
81.3041095336914,
|
| 226 |
+
81.05420622558594,
|
| 227 |
+
80.81840328369141,
|
| 228 |
+
80.59531881103516,
|
| 229 |
+
80.38440787353515,
|
| 230 |
+
80.18435812988281,
|
| 231 |
+
80.17528757324219,
|
| 232 |
+
79.98983715820313,
|
| 233 |
+
79.81431489257812,
|
| 234 |
+
79.64884621582031,
|
| 235 |
+
79.49099832763672,
|
| 236 |
+
79.34037728271484,
|
| 237 |
+
79.19644326171876,
|
| 238 |
+
79.05946433105468,
|
| 239 |
+
78.92887561035157,
|
| 240 |
+
78.80397731933594,
|
| 241 |
+
78.68422974853516,
|
| 242 |
+
78.67905505371094
|
| 243 |
+
],
|
| 244 |
+
"effective_l0": [
|
| 245 |
+
17,
|
| 246 |
+
17,
|
| 247 |
+
16,
|
| 248 |
+
15,
|
| 249 |
+
15,
|
| 250 |
+
14,
|
| 251 |
+
14,
|
| 252 |
+
13,
|
| 253 |
+
13,
|
| 254 |
+
13,
|
| 255 |
+
13,
|
| 256 |
+
13,
|
| 257 |
+
12,
|
| 258 |
+
12,
|
| 259 |
+
12,
|
| 260 |
+
12,
|
| 261 |
+
12,
|
| 262 |
+
12,
|
| 263 |
+
12,
|
| 264 |
+
11,
|
| 265 |
+
11,
|
| 266 |
+
11,
|
| 267 |
+
11,
|
| 268 |
+
11,
|
| 269 |
+
11,
|
| 270 |
+
11,
|
| 271 |
+
11,
|
| 272 |
+
11,
|
| 273 |
+
11,
|
| 274 |
+
11,
|
| 275 |
+
11,
|
| 276 |
+
11,
|
| 277 |
+
11,
|
| 278 |
+
11,
|
| 279 |
+
11,
|
| 280 |
+
10,
|
| 281 |
+
10,
|
| 282 |
+
10,
|
| 283 |
+
10,
|
| 284 |
+
10,
|
| 285 |
+
10,
|
| 286 |
+
10,
|
| 287 |
+
10,
|
| 288 |
+
10,
|
| 289 |
+
10,
|
| 290 |
+
10,
|
| 291 |
+
10,
|
| 292 |
+
10,
|
| 293 |
+
10,
|
| 294 |
+
10,
|
| 295 |
+
10,
|
| 296 |
+
10,
|
| 297 |
+
10,
|
| 298 |
+
10,
|
| 299 |
+
10
|
| 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_0"
|
| 360 |
+
}
|