acroma/output_dir
06
1{2 "best_metric": 99.44547134935306,3 "best_model_checkpoint": "./output_dir/checkpoint-1000",4 "epoch": 5.0761421319796955,5 "eval_steps": 1000,6 "global_step": 5000,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.10152284263959391,13 "grad_norm": 45.86384582519531,14 "learning_rate": 1.9000000000000002e-06,15 "loss": 2.9237,16 "step": 10017 },18 {19 "epoch": 0.20304568527918782,20 "grad_norm": 45.55335998535156,21 "learning_rate": 3.900000000000001e-06,22 "loss": 2.3206,23 "step": 20024 },25 {26 "epoch": 0.30456852791878175,27 "grad_norm": 48.86751174926758,28 "learning_rate": 5.9e-06,29 "loss": 2.3718,30 "step": 30031 },32 {33 "epoch": 0.40609137055837563,34 "grad_norm": 57.0309944152832,35 "learning_rate": 7.9e-06,36 "loss": 2.2004,37 "step": 40038 },39 {40 "epoch": 0.5076142131979695,41 "grad_norm": 50.58991241455078,42 "learning_rate": 9.9e-06,43 "loss": 2.1753,44 "step": 50045 },46 {47 "epoch": 0.6091370558375635,48 "grad_norm": 55.411651611328125,49 "learning_rate": 9.78888888888889e-06,50 "loss": 2.1806,51 "step": 60052 },53 {54 "epoch": 0.7106598984771574,55 "grad_norm": 52.69567108154297,56 "learning_rate": 9.566666666666668e-06,57 "loss": 2.1058,58 "step": 70059 },60 {61 "epoch": 0.8121827411167513,62 "grad_norm": 54.5142822265625,63 "learning_rate": 9.344444444444446e-06,64 "loss": 1.9591,65 "step": 80066 },67 {68 "epoch": 0.9137055837563451,69 "grad_norm": 34.65861511230469,70 "learning_rate": 9.124444444444444e-06,71 "loss": 2.0829,72 "step": 90073 },74 {75 "epoch": 1.015228426395939,76 "grad_norm": 47.46334457397461,77 "learning_rate": 8.902222222222224e-06,78 "loss": 1.9297,79 "step": 100080 },81 {82 "epoch": 1.015228426395939,83 "eval_loss": 1.9972329139709473,84 "eval_runtime": 143.8629,85 "eval_samples_per_second": 3.476,86 "eval_steps_per_second": 1.738,87 "eval_wer": 99.44547134935306,88 "step": 100089 },90 {91 "epoch": 1.116751269035533,92 "grad_norm": 35.19301986694336,93 "learning_rate": 8.68e-06,94 "loss": 1.3307,95 "step": 110096 },97 {98 "epoch": 1.218274111675127,99 "grad_norm": 32.941097259521484,100 "learning_rate": 8.457777777777778e-06,101 "loss": 1.3622,102 "step": 1200103 },104 {105 "epoch": 1.3197969543147208,106 "grad_norm": 60.84815216064453,107 "learning_rate": 8.235555555555557e-06,108 "loss": 1.3763,109 "step": 1300110 },111 {112 "epoch": 1.4213197969543148,113 "grad_norm": 41.705345153808594,114 "learning_rate": 8.013333333333333e-06,115 "loss": 1.3411,116 "step": 1400117 },118 {119 "epoch": 1.5228426395939088,120 "grad_norm": 46.46513366699219,121 "learning_rate": 7.791111111111111e-06,122 "loss": 1.3745,123 "step": 1500124 },125 {126 "epoch": 1.6243654822335025,127 "grad_norm": 52.48996353149414,128 "learning_rate": 7.56888888888889e-06,129 "loss": 1.4233,130 "step": 1600131 },132 {133 "epoch": 1.7258883248730963,134 "grad_norm": 71.2730712890625,135 "learning_rate": 7.346666666666668e-06,136 "loss": 1.2765,137 "step": 1700138 },139 {140 "epoch": 1.8274111675126905,141 "grad_norm": 41.1866455078125,142 "learning_rate": 7.124444444444445e-06,143 "loss": 1.4133,144 "step": 1800145 },146 {147 "epoch": 1.9289340101522843,148 "grad_norm": 40.041481018066406,149 "learning_rate": 6.902222222222223e-06,150 "loss": 1.2927,151 "step": 1900152 },153 {154 "epoch": 2.030456852791878,155 "grad_norm": 29.876537322998047,156 "learning_rate": 6.680000000000001e-06,157 "loss": 1.1678,158 "step": 2000159 },160 {161 "epoch": 2.030456852791878,162 "eval_loss": 1.9988778829574585,163 "eval_runtime": 150.1664,164 "eval_samples_per_second": 3.33,165 "eval_steps_per_second": 1.665,166 "eval_wer": 99.44547134935306,167 "step": 2000168 },169 {170 "epoch": 2.1319796954314723,171 "grad_norm": 31.42411231994629,172 "learning_rate": 6.457777777777778e-06,173 "loss": 0.6302,174 "step": 2100175 },176 {177 "epoch": 2.233502538071066,178 "grad_norm": 29.325563430786133,179 "learning_rate": 6.235555555555556e-06,180 "loss": 0.7149,181 "step": 2200182 },183 {184 "epoch": 2.33502538071066,185 "grad_norm": 28.083486557006836,186 "learning_rate": 6.013333333333335e-06,187 "loss": 0.7032,188 "step": 2300189 },190 {191 "epoch": 2.436548223350254,192 "grad_norm": 59.06522750854492,193 "learning_rate": 5.791111111111112e-06,194 "loss": 0.7279,195 "step": 2400196 },197 {198 "epoch": 2.5380710659898478,199 "grad_norm": 25.313156127929688,200 "learning_rate": 5.56888888888889e-06,201 "loss": 0.7762,202 "step": 2500203 },204 {205 "epoch": 2.6395939086294415,206 "grad_norm": 42.13996887207031,207 "learning_rate": 5.346666666666667e-06,208 "loss": 0.6803,209 "step": 2600210 },211 {212 "epoch": 2.7411167512690353,213 "grad_norm": 40.32999038696289,214 "learning_rate": 5.124444444444445e-06,215 "loss": 0.7569,216 "step": 2700217 },218 {219 "epoch": 2.8426395939086295,220 "grad_norm": 26.812973022460938,221 "learning_rate": 4.902222222222222e-06,222 "loss": 0.7796,223 "step": 2800224 },225 {226 "epoch": 2.9441624365482233,227 "grad_norm": 35.564395904541016,228 "learning_rate": 4.680000000000001e-06,229 "loss": 0.7115,230 "step": 2900231 },232 {233 "epoch": 3.045685279187817,234 "grad_norm": 26.95085906982422,235 "learning_rate": 4.457777777777778e-06,236 "loss": 0.5123,237 "step": 3000238 },239 {240 "epoch": 3.045685279187817,241 "eval_loss": 2.0867390632629395,242 "eval_runtime": 147.2269,243 "eval_samples_per_second": 3.396,244 "eval_steps_per_second": 1.698,245 "eval_wer": 102.40295748613678,246 "step": 3000247 },248 {249 "epoch": 3.1472081218274113,250 "grad_norm": 10.427111625671387,251 "learning_rate": 4.235555555555556e-06,252 "loss": 0.3293,253 "step": 3100254 },255 {256 "epoch": 3.248730964467005,257 "grad_norm": 25.181922912597656,258 "learning_rate": 4.013333333333334e-06,259 "loss": 0.3875,260 "step": 3200261 },262 {263 "epoch": 3.350253807106599,264 "grad_norm": 25.681047439575195,265 "learning_rate": 3.7911111111111114e-06,266 "loss": 0.336,267 "step": 3300268 },269 {270 "epoch": 3.451776649746193,271 "grad_norm": 22.395158767700195,272 "learning_rate": 3.568888888888889e-06,273 "loss": 0.3491,274 "step": 3400275 },276 {277 "epoch": 3.553299492385787,278 "grad_norm": 24.35469627380371,279 "learning_rate": 3.346666666666667e-06,280 "loss": 0.3499,281 "step": 3500282 },283 {284 "epoch": 3.6548223350253806,285 "grad_norm": 43.59064865112305,286 "learning_rate": 3.124444444444445e-06,287 "loss": 0.3475,288 "step": 3600289 },290 {291 "epoch": 3.7563451776649748,292 "grad_norm": 3.5348989963531494,293 "learning_rate": 2.9022222222222223e-06,294 "loss": 0.3032,295 "step": 3700296 },297 {298 "epoch": 3.8578680203045685,299 "grad_norm": 41.54439926147461,300 "learning_rate": 2.68e-06,301 "loss": 0.3408,302 "step": 3800303 },304 {305 "epoch": 3.9593908629441623,306 "grad_norm": 20.793079376220703,307 "learning_rate": 2.457777777777778e-06,308 "loss": 0.2949,309 "step": 3900310 },311 {312 "epoch": 4.060913705583756,313 "grad_norm": 17.149005889892578,314 "learning_rate": 2.235555555555556e-06,315 "loss": 0.2025,316 "step": 4000317 },318 {319 "epoch": 4.060913705583756,320 "eval_loss": 2.169760227203369,321 "eval_runtime": 148.2971,322 "eval_samples_per_second": 3.372,323 "eval_steps_per_second": 1.686,324 "eval_wer": 103.32717190388169,325 "step": 4000326 },327 {328 "epoch": 4.16243654822335,329 "grad_norm": 1.9773640632629395,330 "learning_rate": 2.0133333333333337e-06,331 "loss": 0.1643,332 "step": 4100333 },334 {335 "epoch": 4.2639593908629445,336 "grad_norm": 10.312366485595703,337 "learning_rate": 1.7911111111111113e-06,338 "loss": 0.1461,339 "step": 4200340 },341 {342 "epoch": 4.365482233502538,343 "grad_norm": 1.339500069618225,344 "learning_rate": 1.568888888888889e-06,345 "loss": 0.1546,346 "step": 4300347 },348 {349 "epoch": 4.467005076142132,350 "grad_norm": 23.366968154907227,351 "learning_rate": 1.3466666666666668e-06,352 "loss": 0.1184,353 "step": 4400354 },355 {356 "epoch": 4.568527918781726,357 "grad_norm": 3.0241127014160156,358 "learning_rate": 1.1244444444444446e-06,359 "loss": 0.18,360 "step": 4500361 },362 {363 "epoch": 4.67005076142132,364 "grad_norm": 21.15237045288086,365 "learning_rate": 9.022222222222222e-07,366 "loss": 0.2008,367 "step": 4600368 },369 {370 "epoch": 4.771573604060913,371 "grad_norm": 4.779166221618652,372 "learning_rate": 6.800000000000001e-07,373 "loss": 0.1546,374 "step": 4700375 },376 {377 "epoch": 4.873096446700508,378 "grad_norm": 1.9334161281585693,379 "learning_rate": 4.5777777777777784e-07,380 "loss": 0.1429,381 "step": 4800382 },383 {384 "epoch": 4.974619289340102,385 "grad_norm": 18.270540237426758,386 "learning_rate": 2.3555555555555556e-07,387 "loss": 0.1373,388 "step": 4900389 },390 {391 "epoch": 5.0761421319796955,392 "grad_norm": 1.0400763750076294,393 "learning_rate": 1.3333333333333334e-08,394 "loss": 0.0843,395 "step": 5000396 },397 {398 "epoch": 5.0761421319796955,399 "eval_loss": 2.2213687896728516,400 "eval_runtime": 142.0959,401 "eval_samples_per_second": 3.519,402 "eval_steps_per_second": 1.759,403 "eval_wer": 104.25138632162663,404 "step": 5000405 },406 {407 "epoch": 5.0761421319796955,408 "step": 5000,409 "total_flos": 2.8844110761984e+18,410 "train_loss": 0.9465061870574951,411 "train_runtime": 2908.7621,412 "train_samples_per_second": 3.438,413 "train_steps_per_second": 1.719414 }415 ],416 "logging_steps": 100,417 "max_steps": 5000,418 "num_input_tokens_seen": 0,419 "num_train_epochs": 6,420 "save_steps": 1000,421 "stateful_callbacks": {422 "TrainerControl": {423 "args": {424 "should_epoch_stop": false,425 "should_evaluate": false,426 "should_log": false,427 "should_save": true,428 "should_training_stop": true429 },430 "attributes": {}431 }432 },433 "total_flos": 2.8844110761984e+18,434 "train_batch_size": 2,435 "trial_name": null,436 "trial_params": null437}438 