| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 114700, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.5566625155666254e-05, | |
| "loss": 8.9454, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.113325031133251e-05, | |
| "loss": 6.3983, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6699875466998756e-05, | |
| "loss": 5.7848, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.226650062266502e-05, | |
| "loss": 5.1787, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7.783312577833126e-05, | |
| "loss": 4.5006, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.339975093399751e-05, | |
| "loss": 4.016, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.990941915030131e-05, | |
| "loss": 3.6483, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.975216073068553e-05, | |
| "loss": 3.3951, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.959490231106974e-05, | |
| "loss": 3.2185, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.943764389145395e-05, | |
| "loss": 3.0927, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.928038547183816e-05, | |
| "loss": 2.9973, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.912312705222237e-05, | |
| "loss": 2.9138, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.89658686326066e-05, | |
| "loss": 2.8402, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.880861021299081e-05, | |
| "loss": 2.7818, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.865135179337502e-05, | |
| "loss": 2.7254, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.849409337375923e-05, | |
| "loss": 2.6782, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.833683495414345e-05, | |
| "loss": 2.6362, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.817957653452766e-05, | |
| "loss": 2.5982, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.802231811491189e-05, | |
| "loss": 2.5705, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.78650596952961e-05, | |
| "loss": 2.5337, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.77078012756803e-05, | |
| "loss": 2.5079, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.755054285606452e-05, | |
| "loss": 2.4798, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.739328443644873e-05, | |
| "loss": 2.4593, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.723602601683295e-05, | |
| "loss": 2.432, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.70790821140564e-05, | |
| "loss": 2.408, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.69218236944406e-05, | |
| "loss": 2.3853, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.676456527482482e-05, | |
| "loss": 2.3635, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.660730685520904e-05, | |
| "loss": 2.3485, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.645036295243249e-05, | |
| "loss": 2.3328, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.62931045328167e-05, | |
| "loss": 2.3185, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.61358461132009e-05, | |
| "loss": 2.3014, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.597858769358512e-05, | |
| "loss": 2.2902, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.582164379080856e-05, | |
| "loss": 2.2746, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.566438537119277e-05, | |
| "loss": 2.2592, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.5507126951577e-05, | |
| "loss": 2.248, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.53498685319612e-05, | |
| "loss": 2.24, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.519292462918465e-05, | |
| "loss": 2.2215, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.503566620956886e-05, | |
| "loss": 2.2128, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.487840778995309e-05, | |
| "loss": 2.2036, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.47211493703373e-05, | |
| "loss": 2.1933, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.456420546756074e-05, | |
| "loss": 2.1816, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.440694704794495e-05, | |
| "loss": 2.1657, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.424968862832916e-05, | |
| "loss": 2.1611, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.409243020871337e-05, | |
| "loss": 2.1546, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.393548630593683e-05, | |
| "loss": 2.1439, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6093312991188715, | |
| "eval_loss": 1.9991799592971802, | |
| "eval_runtime": 294.7692, | |
| "eval_samples_per_second": 603.699, | |
| "eval_steps_per_second": 37.731, | |
| "step": 22940 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.377822788632103e-05, | |
| "loss": 2.1305, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.362096946670525e-05, | |
| "loss": 2.1245, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.346371104708946e-05, | |
| "loss": 2.1178, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.330676714431291e-05, | |
| "loss": 2.1102, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.314950872469712e-05, | |
| "loss": 2.1007, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.299225030508134e-05, | |
| "loss": 2.0966, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.283499188546555e-05, | |
| "loss": 2.0878, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.2678047982689e-05, | |
| "loss": 2.0814, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 9.252078956307321e-05, | |
| "loss": 2.0756, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.236353114345743e-05, | |
| "loss": 2.0685, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.220627272384163e-05, | |
| "loss": 2.065, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 9.204932882106509e-05, | |
| "loss": 2.0592, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.18920704014493e-05, | |
| "loss": 2.0527, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.173481198183351e-05, | |
| "loss": 2.0459, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.157755356221772e-05, | |
| "loss": 2.0443, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.142060965944116e-05, | |
| "loss": 2.0367, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.126366575666462e-05, | |
| "loss": 2.0322, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.110640733704882e-05, | |
| "loss": 2.0228, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.094914891743304e-05, | |
| "loss": 2.0209, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.079189049781725e-05, | |
| "loss": 2.0156, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.063463207820148e-05, | |
| "loss": 2.0141, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 9.047737365858569e-05, | |
| "loss": 2.0093, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.032042975580913e-05, | |
| "loss": 2.0052, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.016317133619334e-05, | |
| "loss": 1.9973, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.000591291657756e-05, | |
| "loss": 1.995, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 8.984865449696177e-05, | |
| "loss": 1.9908, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.969139607734599e-05, | |
| "loss": 1.986, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.95341376577302e-05, | |
| "loss": 1.9825, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 8.937687923811441e-05, | |
| "loss": 1.9754, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.921993533533786e-05, | |
| "loss": 1.9733, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.906267691572208e-05, | |
| "loss": 1.9679, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.890541849610629e-05, | |
| "loss": 1.9611, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.87481600764905e-05, | |
| "loss": 1.961, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.859121617371395e-05, | |
| "loss": 1.9588, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.843395775409816e-05, | |
| "loss": 1.9578, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.827669933448237e-05, | |
| "loss": 1.9528, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.811944091486659e-05, | |
| "loss": 1.9486, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.796249701209002e-05, | |
| "loss": 1.9443, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 8.780523859247425e-05, | |
| "loss": 1.9406, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.764798017285846e-05, | |
| "loss": 1.9372, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 8.749072175324268e-05, | |
| "loss": 1.9351, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.733346333362689e-05, | |
| "loss": 1.9265, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.717651943085034e-05, | |
| "loss": 1.9272, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.701926101123455e-05, | |
| "loss": 1.9268, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.686200259161877e-05, | |
| "loss": 1.9242, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.670474417200297e-05, | |
| "loss": 1.9234, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6411638490813204, | |
| "eval_loss": 1.785447597503662, | |
| "eval_runtime": 294.391, | |
| "eval_samples_per_second": 604.475, | |
| "eval_steps_per_second": 37.78, | |
| "step": 45880 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.654780026922643e-05, | |
| "loss": 1.9157, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.639054184961062e-05, | |
| "loss": 1.9106, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.623328342999485e-05, | |
| "loss": 1.9071, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.607602501037906e-05, | |
| "loss": 1.9029, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.59190811076025e-05, | |
| "loss": 1.9037, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.576182268798671e-05, | |
| "loss": 1.9035, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.560456426837094e-05, | |
| "loss": 1.8995, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.544730584875515e-05, | |
| "loss": 1.8955, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.529036194597859e-05, | |
| "loss": 1.894, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.51331035263628e-05, | |
| "loss": 1.8871, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.497584510674703e-05, | |
| "loss": 1.8897, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.481858668713124e-05, | |
| "loss": 1.888, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.466164278435468e-05, | |
| "loss": 1.8866, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.450438436473889e-05, | |
| "loss": 1.8829, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.43471259451231e-05, | |
| "loss": 1.8787, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 8.418986752550731e-05, | |
| "loss": 1.8743, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 8.403260910589154e-05, | |
| "loss": 1.8758, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.387566520311498e-05, | |
| "loss": 1.8728, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 8.371840678349919e-05, | |
| "loss": 1.8713, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.35611483638834e-05, | |
| "loss": 1.8675, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.340388994426763e-05, | |
| "loss": 1.8653, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.324694604149107e-05, | |
| "loss": 1.8635, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.308968762187528e-05, | |
| "loss": 1.8589, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.293242920225949e-05, | |
| "loss": 1.8585, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 8.27751707826437e-05, | |
| "loss": 1.8592, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 8.261822687986716e-05, | |
| "loss": 1.8551, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 8.246096846025136e-05, | |
| "loss": 1.8564, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 8.230371004063558e-05, | |
| "loss": 1.851, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 8.214645162101979e-05, | |
| "loss": 1.8469, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 8.198950771824324e-05, | |
| "loss": 1.8461, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 8.183224929862745e-05, | |
| "loss": 1.8487, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 8.167499087901167e-05, | |
| "loss": 1.8435, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 8.151773245939588e-05, | |
| "loss": 1.8403, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 8.136078855661933e-05, | |
| "loss": 1.8435, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 8.120353013700354e-05, | |
| "loss": 1.8409, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 8.104627171738776e-05, | |
| "loss": 1.8359, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 8.088901329777196e-05, | |
| "loss": 1.833, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 8.073206939499542e-05, | |
| "loss": 1.8336, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 8.057481097537963e-05, | |
| "loss": 1.8348, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.041755255576384e-05, | |
| "loss": 1.8302, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.026029413614805e-05, | |
| "loss": 1.8253, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 8.01033502333715e-05, | |
| "loss": 1.8225, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.99460918137557e-05, | |
| "loss": 1.8245, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 7.978883339413993e-05, | |
| "loss": 1.8246, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.963157497452414e-05, | |
| "loss": 1.8203, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.947463107174758e-05, | |
| "loss": 1.8178, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6562844667040338, | |
| "eval_loss": 1.6895612478256226, | |
| "eval_runtime": 294.9578, | |
| "eval_samples_per_second": 603.314, | |
| "eval_steps_per_second": 37.707, | |
| "step": 68820 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.93173726521318e-05, | |
| "loss": 1.8202, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.916011423251602e-05, | |
| "loss": 1.8172, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.900285581290023e-05, | |
| "loss": 1.8126, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.884591191012367e-05, | |
| "loss": 1.8141, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.868865349050788e-05, | |
| "loss": 1.8052, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.85313950708921e-05, | |
| "loss": 1.8075, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 7.83741366512763e-05, | |
| "loss": 1.8115, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 7.821719274849975e-05, | |
| "loss": 1.8076, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 7.805993432888397e-05, | |
| "loss": 1.8078, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 7.790267590926819e-05, | |
| "loss": 1.8015, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 7.77454174896524e-05, | |
| "loss": 1.8043, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 7.758847358687584e-05, | |
| "loss": 1.8002, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 7.743121516726006e-05, | |
| "loss": 1.7954, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 7.727395674764428e-05, | |
| "loss": 1.7936, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 7.711669832802849e-05, | |
| "loss": 1.7903, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 7.695975442525193e-05, | |
| "loss": 1.7951, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 7.680249600563615e-05, | |
| "loss": 1.7946, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.664523758602035e-05, | |
| "loss": 1.7946, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 7.648797916640458e-05, | |
| "loss": 1.7902, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 7.633103526362802e-05, | |
| "loss": 1.7905, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 7.617377684401223e-05, | |
| "loss": 1.7898, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 7.601651842439644e-05, | |
| "loss": 1.7885, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 7.585926000478067e-05, | |
| "loss": 1.7852, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.57023161020041e-05, | |
| "loss": 1.7815, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.554505768238832e-05, | |
| "loss": 1.7833, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 7.538779926277253e-05, | |
| "loss": 1.7815, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 7.523054084315676e-05, | |
| "loss": 1.7778, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 7.507359694038019e-05, | |
| "loss": 1.7811, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.491633852076441e-05, | |
| "loss": 1.7786, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 7.475908010114862e-05, | |
| "loss": 1.7767, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 7.460182168153283e-05, | |
| "loss": 1.7804, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 7.444487777875628e-05, | |
| "loss": 1.7716, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 7.428761935914049e-05, | |
| "loss": 1.7777, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 7.41303609395247e-05, | |
| "loss": 1.7712, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 7.397310251990892e-05, | |
| "loss": 1.7678, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 7.381615861713237e-05, | |
| "loss": 1.7702, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 7.365890019751658e-05, | |
| "loss": 1.7701, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 7.350164177790079e-05, | |
| "loss": 1.7689, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 7.334438335828501e-05, | |
| "loss": 1.7665, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 7.318743945550846e-05, | |
| "loss": 1.7711, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 7.303018103589267e-05, | |
| "loss": 1.7626, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 7.287292261627688e-05, | |
| "loss": 1.7628, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 7.271566419666109e-05, | |
| "loss": 1.7589, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 7.255872029388455e-05, | |
| "loss": 1.7614, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 7.240146187426874e-05, | |
| "loss": 1.7646, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 7.224420345465297e-05, | |
| "loss": 1.7636, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6662888180155826, | |
| "eval_loss": 1.6273692846298218, | |
| "eval_runtime": 294.4726, | |
| "eval_samples_per_second": 604.307, | |
| "eval_steps_per_second": 37.769, | |
| "step": 91760 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 7.208694503503718e-05, | |
| "loss": 1.7576, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 7.193000113226062e-05, | |
| "loss": 1.7556, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 7.177274271264483e-05, | |
| "loss": 1.7484, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 7.161548429302906e-05, | |
| "loss": 1.7533, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 7.145822587341327e-05, | |
| "loss": 1.7508, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 7.130128197063671e-05, | |
| "loss": 1.7589, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 7.114402355102092e-05, | |
| "loss": 1.7485, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 7.098676513140515e-05, | |
| "loss": 1.7477, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 7.082950671178934e-05, | |
| "loss": 1.7523, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 7.06725628090128e-05, | |
| "loss": 1.7443, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 7.051530438939701e-05, | |
| "loss": 1.7481, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 7.035804596978122e-05, | |
| "loss": 1.7421, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 7.020078755016543e-05, | |
| "loss": 1.7449, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 7.004384364738888e-05, | |
| "loss": 1.7453, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 6.988658522777309e-05, | |
| "loss": 1.7443, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 6.972932680815731e-05, | |
| "loss": 1.7398, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.957206838854152e-05, | |
| "loss": 1.7387, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 6.941512448576497e-05, | |
| "loss": 1.7398, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 6.925786606614918e-05, | |
| "loss": 1.7383, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 6.91006076465334e-05, | |
| "loss": 1.7379, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 6.894334922691761e-05, | |
| "loss": 1.7365, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 6.878640532414106e-05, | |
| "loss": 1.7372, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 6.862914690452527e-05, | |
| "loss": 1.7336, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.847188848490948e-05, | |
| "loss": 1.7379, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 6.831463006529369e-05, | |
| "loss": 1.7325, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 6.815768616251714e-05, | |
| "loss": 1.7329, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 6.800042774290136e-05, | |
| "loss": 1.7335, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 6.784316932328557e-05, | |
| "loss": 1.7347, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 6.768591090366978e-05, | |
| "loss": 1.7293, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 6.752896700089323e-05, | |
| "loss": 1.7307, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 6.737170858127745e-05, | |
| "loss": 1.7264, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.721445016166166e-05, | |
| "loss": 1.7294, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 6.705719174204587e-05, | |
| "loss": 1.7324, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 6.690024783926932e-05, | |
| "loss": 1.7279, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 6.674298941965354e-05, | |
| "loss": 1.7228, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 6.658573100003774e-05, | |
| "loss": 1.7248, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 6.642847258042196e-05, | |
| "loss": 1.7278, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 6.62715286776454e-05, | |
| "loss": 1.7211, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 6.611427025802962e-05, | |
| "loss": 1.7227, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 6.595701183841383e-05, | |
| "loss": 1.7181, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 6.579975341879805e-05, | |
| "loss": 1.7188, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 6.564280951602148e-05, | |
| "loss": 1.7169, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 6.54855510964057e-05, | |
| "loss": 1.7186, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 6.532829267678992e-05, | |
| "loss": 1.7199, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 6.517103425717414e-05, | |
| "loss": 1.7216, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 6.501409035439757e-05, | |
| "loss": 1.7173, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.6715102448572298, | |
| "eval_loss": 1.5880473852157593, | |
| "eval_runtime": 294.7287, | |
| "eval_samples_per_second": 603.782, | |
| "eval_steps_per_second": 37.736, | |
| "step": 114700 | |
| } | |
| ], | |
| "max_steps": 321160, | |
| "num_train_epochs": 14, | |
| "total_flos": 8.721147965075726e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |