Upload results/lr0.0001-fixed4x1_history.json with huggingface_hub
Browse files
results/lr0.0001-fixed4x1_history.json
ADDED
|
@@ -0,0 +1,302 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
{
|
| 4 |
+
"step": 0,
|
| 5 |
+
"loss": 4.3324,
|
| 6 |
+
"lr": 1.0000000000000001e-07,
|
| 7 |
+
"iters": 1,
|
| 8 |
+
"steps_per_s": 0.0,
|
| 9 |
+
"flops_cum": 87407198208.0
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"step": 100,
|
| 13 |
+
"loss": 4.2427,
|
| 14 |
+
"lr": 1.0100000000000002e-05,
|
| 15 |
+
"iters": 1,
|
| 16 |
+
"steps_per_s": 22.5,
|
| 17 |
+
"flops_cum": 8828127019008.0
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"step": 200,
|
| 21 |
+
"loss": 4.192,
|
| 22 |
+
"lr": 2.01e-05,
|
| 23 |
+
"iters": 1,
|
| 24 |
+
"steps_per_s": 22.26,
|
| 25 |
+
"flops_cum": 17568846839808.0
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"step": 300,
|
| 29 |
+
"loss": 4.1813,
|
| 30 |
+
"lr": 3.0100000000000003e-05,
|
| 31 |
+
"iters": 1,
|
| 32 |
+
"steps_per_s": 22.15,
|
| 33 |
+
"flops_cum": 26309566660608.0
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"step": 400,
|
| 37 |
+
"loss": 4.1795,
|
| 38 |
+
"lr": 4.0100000000000006e-05,
|
| 39 |
+
"iters": 1,
|
| 40 |
+
"steps_per_s": 22.13,
|
| 41 |
+
"flops_cum": 35050286481408.0
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"step": 500,
|
| 45 |
+
"loss": 4.1726,
|
| 46 |
+
"lr": 5.0100000000000005e-05,
|
| 47 |
+
"iters": 1,
|
| 48 |
+
"steps_per_s": 22.15,
|
| 49 |
+
"flops_cum": 43791006302208.0
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"step": 600,
|
| 53 |
+
"loss": 4.1603,
|
| 54 |
+
"lr": 6.0100000000000004e-05,
|
| 55 |
+
"iters": 1,
|
| 56 |
+
"steps_per_s": 22.05,
|
| 57 |
+
"flops_cum": 52531726123008.0
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"step": 700,
|
| 61 |
+
"loss": 4.1656,
|
| 62 |
+
"lr": 7.010000000000001e-05,
|
| 63 |
+
"iters": 1,
|
| 64 |
+
"steps_per_s": 22.05,
|
| 65 |
+
"flops_cum": 61272445943808.0
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"step": 800,
|
| 69 |
+
"loss": 4.1566,
|
| 70 |
+
"lr": 8.010000000000001e-05,
|
| 71 |
+
"iters": 1,
|
| 72 |
+
"steps_per_s": 22.0,
|
| 73 |
+
"flops_cum": 70013165764608.0
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"step": 900,
|
| 77 |
+
"loss": 4.156,
|
| 78 |
+
"lr": 9.01e-05,
|
| 79 |
+
"iters": 1,
|
| 80 |
+
"steps_per_s": 21.93,
|
| 81 |
+
"flops_cum": 78753885585408.0
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"step": 1000,
|
| 85 |
+
"loss": 4.1562,
|
| 86 |
+
"lr": 0.0001,
|
| 87 |
+
"iters": 1,
|
| 88 |
+
"steps_per_s": 21.9,
|
| 89 |
+
"flops_cum": 87494605406208.0
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"step": 1100,
|
| 93 |
+
"loss": 4.1626,
|
| 94 |
+
"lr": 9.938441702975689e-05,
|
| 95 |
+
"iters": 1,
|
| 96 |
+
"steps_per_s": 21.91,
|
| 97 |
+
"flops_cum": 96235325227008.0
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"step": 1200,
|
| 101 |
+
"loss": 4.144,
|
| 102 |
+
"lr": 9.755282581475769e-05,
|
| 103 |
+
"iters": 1,
|
| 104 |
+
"steps_per_s": 21.84,
|
| 105 |
+
"flops_cum": 104976045047808.0
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"step": 1300,
|
| 109 |
+
"loss": 4.1266,
|
| 110 |
+
"lr": 9.45503262094184e-05,
|
| 111 |
+
"iters": 1,
|
| 112 |
+
"steps_per_s": 21.84,
|
| 113 |
+
"flops_cum": 113716764868608.0
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"step": 1400,
|
| 117 |
+
"loss": 4.1128,
|
| 118 |
+
"lr": 9.045084971874738e-05,
|
| 119 |
+
"iters": 1,
|
| 120 |
+
"steps_per_s": 21.82,
|
| 121 |
+
"flops_cum": 122457484689408.0
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"step": 1500,
|
| 125 |
+
"loss": 4.0892,
|
| 126 |
+
"lr": 8.535533905932738e-05,
|
| 127 |
+
"iters": 1,
|
| 128 |
+
"steps_per_s": 21.78,
|
| 129 |
+
"flops_cum": 131198204510208.0
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"step": 1600,
|
| 133 |
+
"loss": 4.0926,
|
| 134 |
+
"lr": 7.938926261462366e-05,
|
| 135 |
+
"iters": 1,
|
| 136 |
+
"steps_per_s": 21.75,
|
| 137 |
+
"flops_cum": 139938924331008.0
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"step": 1700,
|
| 141 |
+
"loss": 4.1051,
|
| 142 |
+
"lr": 7.269952498697734e-05,
|
| 143 |
+
"iters": 1,
|
| 144 |
+
"steps_per_s": 21.76,
|
| 145 |
+
"flops_cum": 148679644151808.0
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"step": 1800,
|
| 149 |
+
"loss": 4.0713,
|
| 150 |
+
"lr": 6.545084971874738e-05,
|
| 151 |
+
"iters": 1,
|
| 152 |
+
"steps_per_s": 21.78,
|
| 153 |
+
"flops_cum": 157420363972608.0
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"step": 1900,
|
| 157 |
+
"loss": 4.0529,
|
| 158 |
+
"lr": 5.782172325201155e-05,
|
| 159 |
+
"iters": 1,
|
| 160 |
+
"steps_per_s": 21.7,
|
| 161 |
+
"flops_cum": 166161083793408.0
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"step": 2000,
|
| 165 |
+
"loss": 4.0843,
|
| 166 |
+
"lr": 5e-05,
|
| 167 |
+
"iters": 1,
|
| 168 |
+
"steps_per_s": 21.72,
|
| 169 |
+
"flops_cum": 174901803614208.0
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"step": 2100,
|
| 173 |
+
"loss": 4.0869,
|
| 174 |
+
"lr": 4.2178276747988446e-05,
|
| 175 |
+
"iters": 1,
|
| 176 |
+
"steps_per_s": 21.75,
|
| 177 |
+
"flops_cum": 183642523435008.0
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"step": 2200,
|
| 181 |
+
"loss": 4.059,
|
| 182 |
+
"lr": 3.4549150281252636e-05,
|
| 183 |
+
"iters": 1,
|
| 184 |
+
"steps_per_s": 21.69,
|
| 185 |
+
"flops_cum": 192383243255808.0
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"step": 2300,
|
| 189 |
+
"loss": 4.036,
|
| 190 |
+
"lr": 2.7300475013022663e-05,
|
| 191 |
+
"iters": 1,
|
| 192 |
+
"steps_per_s": 21.7,
|
| 193 |
+
"flops_cum": 201123963076608.0
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"step": 2400,
|
| 197 |
+
"loss": 4.0677,
|
| 198 |
+
"lr": 2.061073738537635e-05,
|
| 199 |
+
"iters": 1,
|
| 200 |
+
"steps_per_s": 21.63,
|
| 201 |
+
"flops_cum": 209864682897408.0
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"step": 2500,
|
| 205 |
+
"loss": 4.0587,
|
| 206 |
+
"lr": 1.4644660940672627e-05,
|
| 207 |
+
"iters": 1,
|
| 208 |
+
"steps_per_s": 20.25,
|
| 209 |
+
"flops_cum": 218605402718208.0
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"step": 2600,
|
| 213 |
+
"loss": 4.0276,
|
| 214 |
+
"lr": 9.549150281252633e-06,
|
| 215 |
+
"iters": 1,
|
| 216 |
+
"steps_per_s": 21.64,
|
| 217 |
+
"flops_cum": 227346122539008.0
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"step": 2700,
|
| 221 |
+
"loss": 4.0375,
|
| 222 |
+
"lr": 5.449673790581611e-06,
|
| 223 |
+
"iters": 1,
|
| 224 |
+
"steps_per_s": 21.61,
|
| 225 |
+
"flops_cum": 236086842359808.0
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"step": 2800,
|
| 229 |
+
"loss": 4.0483,
|
| 230 |
+
"lr": 2.4471741852423237e-06,
|
| 231 |
+
"iters": 1,
|
| 232 |
+
"steps_per_s": 21.63,
|
| 233 |
+
"flops_cum": 244827562180608.0
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"step": 2900,
|
| 237 |
+
"loss": 4.0707,
|
| 238 |
+
"lr": 6.15582970243117e-07,
|
| 239 |
+
"iters": 1,
|
| 240 |
+
"steps_per_s": 21.62,
|
| 241 |
+
"flops_cum": 253568282001408.0
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"step": 2999,
|
| 245 |
+
"loss": 4.0661,
|
| 246 |
+
"lr": 6.168501482384237e-11,
|
| 247 |
+
"iters": 1,
|
| 248 |
+
"steps_per_s": 21.81,
|
| 249 |
+
"flops_cum": 262221594624000.0
|
| 250 |
+
}
|
| 251 |
+
],
|
| 252 |
+
"id_acc": [
|
| 253 |
+
{
|
| 254 |
+
"step": 2500,
|
| 255 |
+
"mean": 0.04249999881722033,
|
| 256 |
+
"per_k": {
|
| 257 |
+
"1": 0.03999999910593033,
|
| 258 |
+
"2": 0.022499999031424522,
|
| 259 |
+
"3": 0.03500000014901161,
|
| 260 |
+
"4": 0.05249999836087227,
|
| 261 |
+
"5": 0.03749999776482582,
|
| 262 |
+
"6": 0.05249999836087227,
|
| 263 |
+
"7": 0.054999999701976776,
|
| 264 |
+
"8": 0.044999998062849045
|
| 265 |
+
}
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"step": 3000,
|
| 269 |
+
"mean": 0.04562499839812517,
|
| 270 |
+
"per_k": {
|
| 271 |
+
"1": 0.03749999776482582,
|
| 272 |
+
"2": 0.042500000447034836,
|
| 273 |
+
"3": 0.032499998807907104,
|
| 274 |
+
"4": 0.032499998807907104,
|
| 275 |
+
"5": 0.044999998062849045,
|
| 276 |
+
"6": 0.07499999552965164,
|
| 277 |
+
"7": 0.042500000447034836,
|
| 278 |
+
"8": 0.057499997317790985
|
| 279 |
+
}
|
| 280 |
+
}
|
| 281 |
+
],
|
| 282 |
+
"ood": [
|
| 283 |
+
{
|
| 284 |
+
"step": 2500,
|
| 285 |
+
"acc": {
|
| 286 |
+
"16": 0.02499999850988388,
|
| 287 |
+
"32": 0.042500000447034836,
|
| 288 |
+
"64": 0.027499999850988388
|
| 289 |
+
}
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"step": 3000,
|
| 293 |
+
"acc": {
|
| 294 |
+
"16": 0.04749999940395355,
|
| 295 |
+
"32": 0.027499999850988388,
|
| 296 |
+
"64": 0.014999999664723873
|
| 297 |
+
}
|
| 298 |
+
}
|
| 299 |
+
],
|
| 300 |
+
"iters": [],
|
| 301 |
+
"flops_cum": 262221594624000.0
|
| 302 |
+
}
|