Multi-Domain-Expert-Learning/expert-github
129
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 0.015009324542872259,5 "global_step": 1000,6 "is_hyper_param_search": false,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [10 {11 "epoch": 0.0,12 "learning_rate": 9.8e-05,13 "loss": 1.7437,14 "step": 2015 },16 {17 "epoch": 0.0,18 "learning_rate": 9.6e-05,19 "loss": 1.7193,20 "step": 4021 },22 {23 "epoch": 0.0,24 "learning_rate": 9.4e-05,25 "loss": 1.7594,26 "step": 6027 },28 {29 "epoch": 0.0,30 "learning_rate": 9.200000000000001e-05,31 "loss": 1.6967,32 "step": 8033 },34 {35 "epoch": 0.0,36 "learning_rate": 9e-05,37 "loss": 1.6901,38 "step": 10039 },40 {41 "epoch": 0.0,42 "learning_rate": 8.800000000000001e-05,43 "loss": 1.6982,44 "step": 12045 },46 {47 "epoch": 0.0,48 "learning_rate": 8.6e-05,49 "loss": 1.7172,50 "step": 14051 },52 {53 "epoch": 0.0,54 "learning_rate": 8.4e-05,55 "loss": 1.7211,56 "step": 16057 },58 {59 "epoch": 0.0,60 "learning_rate": 8.2e-05,61 "loss": 1.7745,62 "step": 18063 },64 {65 "epoch": 0.0,66 "learning_rate": 8e-05,67 "loss": 1.7274,68 "step": 20069 },70 {71 "epoch": 0.0,72 "eval_accuracy": 0.621273737842038,73 "eval_loss": 1.8277987241744995,74 "eval_runtime": 1953.3722,75 "eval_samples_per_second": 83.488,76 "eval_steps_per_second": 1.305,77 "step": 20078 },79 {80 "epoch": 0.0,81 "learning_rate": 7.800000000000001e-05,82 "loss": 1.7197,83 "step": 22084 },85 {86 "epoch": 0.0,87 "learning_rate": 7.6e-05,88 "loss": 1.7018,89 "step": 24090 },91 {92 "epoch": 0.0,93 "learning_rate": 7.4e-05,94 "loss": 1.7197,95 "step": 26096 },97 {98 "epoch": 0.0,99 "learning_rate": 7.2e-05,100 "loss": 1.7273,101 "step": 280102 },103 {104 "epoch": 0.0,105 "learning_rate": 7e-05,106 "loss": 1.7183,107 "step": 300108 },109 {110 "epoch": 0.0,111 "learning_rate": 6.800000000000001e-05,112 "loss": 1.6984,113 "step": 320114 },115 {116 "epoch": 0.01,117 "learning_rate": 6.6e-05,118 "loss": 1.7386,119 "step": 340120 },121 {122 "epoch": 0.01,123 "learning_rate": 6.400000000000001e-05,124 "loss": 1.7249,125 "step": 360126 },127 {128 "epoch": 0.01,129 "learning_rate": 6.2e-05,130 "loss": 1.7773,131 "step": 380132 },133 {134 "epoch": 0.01,135 "learning_rate": 6e-05,136 "loss": 1.6666,137 "step": 400138 },139 {140 "epoch": 0.01,141 "eval_accuracy": 0.6223302329977166,142 "eval_loss": 1.8232501745224,143 "eval_runtime": 1951.452,144 "eval_samples_per_second": 83.571,145 "eval_steps_per_second": 1.306,146 "step": 400147 },148 {149 "epoch": 0.01,150 "learning_rate": 5.8e-05,151 "loss": 1.7198,152 "step": 420153 },154 {155 "epoch": 0.01,156 "learning_rate": 5.6000000000000006e-05,157 "loss": 1.7502,158 "step": 440159 },160 {161 "epoch": 0.01,162 "learning_rate": 5.4000000000000005e-05,163 "loss": 1.7609,164 "step": 460165 },166 {167 "epoch": 0.01,168 "learning_rate": 5.2000000000000004e-05,169 "loss": 1.6984,170 "step": 480171 },172 {173 "epoch": 0.01,174 "learning_rate": 5e-05,175 "loss": 1.7254,176 "step": 500177 },178 {179 "epoch": 0.01,180 "learning_rate": 4.8e-05,181 "loss": 1.7266,182 "step": 520183 },184 {185 "epoch": 0.01,186 "learning_rate": 4.600000000000001e-05,187 "loss": 1.6992,188 "step": 540189 },190 {191 "epoch": 0.01,192 "learning_rate": 4.4000000000000006e-05,193 "loss": 1.6693,194 "step": 560195 },196 {197 "epoch": 0.01,198 "learning_rate": 4.2e-05,199 "loss": 1.7121,200 "step": 580201 },202 {203 "epoch": 0.01,204 "learning_rate": 4e-05,205 "loss": 1.7053,206 "step": 600207 },208 {209 "epoch": 0.01,210 "eval_accuracy": 0.6236126691293421,211 "eval_loss": 1.8155388832092285,212 "eval_runtime": 1969.8681,213 "eval_samples_per_second": 82.789,214 "eval_steps_per_second": 1.294,215 "step": 600216 },217 {218 "epoch": 0.01,219 "learning_rate": 3.8e-05,220 "loss": 1.711,221 "step": 620222 },223 {224 "epoch": 0.01,225 "learning_rate": 3.6e-05,226 "loss": 1.6866,227 "step": 640228 },229 {230 "epoch": 0.01,231 "learning_rate": 3.4000000000000007e-05,232 "loss": 1.7273,233 "step": 660234 },235 {236 "epoch": 0.01,237 "learning_rate": 3.2000000000000005e-05,238 "loss": 1.64,239 "step": 680240 },241 {242 "epoch": 0.01,243 "learning_rate": 3e-05,244 "loss": 1.67,245 "step": 700246 },247 {248 "epoch": 0.01,249 "learning_rate": 2.8000000000000003e-05,250 "loss": 1.6873,251 "step": 720252 },253 {254 "epoch": 0.01,255 "learning_rate": 2.6000000000000002e-05,256 "loss": 1.7368,257 "step": 740258 },259 {260 "epoch": 0.01,261 "learning_rate": 2.4e-05,262 "loss": 1.7325,263 "step": 760264 },265 {266 "epoch": 0.01,267 "learning_rate": 2.2000000000000003e-05,268 "loss": 1.6948,269 "step": 780270 },271 {272 "epoch": 0.01,273 "learning_rate": 2e-05,274 "loss": 1.7117,275 "step": 800276 },277 {278 "epoch": 0.01,279 "eval_accuracy": 0.6249809385925811,280 "eval_loss": 1.808415412902832,281 "eval_runtime": 1947.0907,282 "eval_samples_per_second": 83.758,283 "eval_steps_per_second": 1.309,284 "step": 800285 },286 {287 "epoch": 0.01,288 "learning_rate": 1.8e-05,289 "loss": 1.7079,290 "step": 820291 },292 {293 "epoch": 0.01,294 "learning_rate": 1.6000000000000003e-05,295 "loss": 1.6933,296 "step": 840297 },298 {299 "epoch": 0.01,300 "learning_rate": 1.4000000000000001e-05,301 "loss": 1.6583,302 "step": 860303 },304 {305 "epoch": 0.01,306 "learning_rate": 1.2e-05,307 "loss": 1.6468,308 "step": 880309 },310 {311 "epoch": 0.01,312 "learning_rate": 1e-05,313 "loss": 1.6779,314 "step": 900315 },316 {317 "epoch": 0.01,318 "learning_rate": 8.000000000000001e-06,319 "loss": 1.717,320 "step": 920321 },322 {323 "epoch": 0.01,324 "learning_rate": 6e-06,325 "loss": 1.6988,326 "step": 940327 },328 {329 "epoch": 0.01,330 "learning_rate": 4.000000000000001e-06,331 "loss": 1.7096,332 "step": 960333 },334 {335 "epoch": 0.01,336 "learning_rate": 2.0000000000000003e-06,337 "loss": 1.7079,338 "step": 980339 },340 {341 "epoch": 0.02,342 "learning_rate": 0.0,343 "loss": 1.678,344 "step": 1000345 },346 {347 "epoch": 0.02,348 "eval_accuracy": 0.6257492027087862,349 "eval_loss": 1.8040720224380493,350 "eval_runtime": 1943.062,351 "eval_samples_per_second": 83.931,352 "eval_steps_per_second": 1.312,353 "step": 1000354 },355 {356 "epoch": 0.02,357 "step": 1000,358 "total_flos": 1.7866929733632e+17,359 "train_loss": 1.710025089263916,360 "train_runtime": 11151.3123,361 "train_samples_per_second": 5.739,362 "train_steps_per_second": 0.09363 }364 ],365 "max_steps": 1000,366 "num_train_epochs": 1,367 "total_flos": 1.7866929733632e+17,368 "trial_name": null,369 "trial_params": null370}371 