| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 114970, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.697921196833957e-09, | |
| "loss": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.001304688179525e-06, | |
| "loss": 0.0024, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.000260937635905e-05, | |
| "loss": 0.0025, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.5003914064538576e-05, | |
| "loss": 0.0021, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.00052187527181e-05, | |
| "loss": 0.002, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.500652344089763e-05, | |
| "loss": 0.0024, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.0007828129077153e-05, | |
| "loss": 0.0021, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.500913281725668e-05, | |
| "loss": 0.0024, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.00104375054362e-05, | |
| "loss": 0.0021, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.501174219361573e-05, | |
| "loss": 0.0021, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.001304688179526e-05, | |
| "loss": 0.0021, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.501435156997478e-05, | |
| "loss": 0.0022, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.0015656258154306e-05, | |
| "loss": 0.002, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.501696094633383e-05, | |
| "loss": 0.0021, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.001826563451336e-05, | |
| "loss": 0.0018, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.501957032269288e-05, | |
| "loss": 0.0021, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.00208750108724e-05, | |
| "loss": 0.0019, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.502217969905193e-05, | |
| "loss": 0.0023, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.002348438723146e-05, | |
| "loss": 0.0021, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.502478907541099e-05, | |
| "loss": 0.0021, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.99971006929344e-05, | |
| "loss": 0.0022, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.944140017202556e-05, | |
| "loss": 0.0025, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.888569965111673e-05, | |
| "loss": 0.0022, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.832999913020789e-05, | |
| "loss": 0.0022, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.777429860929905e-05, | |
| "loss": 0.002, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.721859808839022e-05, | |
| "loss": 0.0021, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.666289756748138e-05, | |
| "loss": 0.0023, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.610719704657253e-05, | |
| "loss": 0.0023, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.55514965256637e-05, | |
| "loss": 0.0023, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.499579600475486e-05, | |
| "loss": 0.002, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.444009548384603e-05, | |
| "loss": 0.0025, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.388439496293719e-05, | |
| "loss": 0.0023, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.332869444202836e-05, | |
| "loss": 0.0026, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.277299392111952e-05, | |
| "loss": 0.0025, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.221729340021069e-05, | |
| "loss": 0.0027, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.166159287930185e-05, | |
| "loss": 0.0024, | |
| "step": 20125 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.110589235839302e-05, | |
| "loss": 0.0022, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.055019183748418e-05, | |
| "loss": 0.0026, | |
| "step": 21275 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.999449131657535e-05, | |
| "loss": 0.0025, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.94387907956665e-05, | |
| "loss": 0.0024, | |
| "step": 22425 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.02637363225221634, | |
| "eval_max_distance": 8, | |
| "eval_mean_distance": 0, | |
| "eval_runtime": 23.0018, | |
| "eval_samples_per_second": 11.217, | |
| "eval_steps_per_second": 0.783, | |
| "step": 22994 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.888309027475768e-05, | |
| "loss": 0.0026, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.832738975384883e-05, | |
| "loss": 0.0019, | |
| "step": 23575 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.777168923294e-05, | |
| "loss": 0.002, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.721598871203116e-05, | |
| "loss": 0.002, | |
| "step": 24725 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.666028819112233e-05, | |
| "loss": 0.0021, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.610458767021349e-05, | |
| "loss": 0.0022, | |
| "step": 25875 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.554888714930466e-05, | |
| "loss": 0.0022, | |
| "step": 26450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.499318662839582e-05, | |
| "loss": 0.0023, | |
| "step": 27025 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.443748610748699e-05, | |
| "loss": 0.0023, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.388178558657815e-05, | |
| "loss": 0.002, | |
| "step": 28175 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 8.33260850656693e-05, | |
| "loss": 0.0022, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.277038454476048e-05, | |
| "loss": 0.0022, | |
| "step": 29325 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.221468402385163e-05, | |
| "loss": 0.002, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 8.165898350294279e-05, | |
| "loss": 0.0023, | |
| "step": 30475 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 8.110328298203396e-05, | |
| "loss": 0.0026, | |
| "step": 31050 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.054758246112512e-05, | |
| "loss": 0.0022, | |
| "step": 31625 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.999188194021629e-05, | |
| "loss": 0.0023, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.943618141930745e-05, | |
| "loss": 0.0023, | |
| "step": 32775 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.888048089839862e-05, | |
| "loss": 0.0024, | |
| "step": 33350 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.832478037748978e-05, | |
| "loss": 0.0024, | |
| "step": 33925 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.776907985658095e-05, | |
| "loss": 0.0025, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.721337933567211e-05, | |
| "loss": 0.0023, | |
| "step": 35075 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.665767881476328e-05, | |
| "loss": 0.0021, | |
| "step": 35650 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 7.610197829385444e-05, | |
| "loss": 0.0022, | |
| "step": 36225 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.554627777294561e-05, | |
| "loss": 0.002, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 7.499057725203676e-05, | |
| "loss": 0.0021, | |
| "step": 37375 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.443487673112794e-05, | |
| "loss": 0.0024, | |
| "step": 37950 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 7.38791762102191e-05, | |
| "loss": 0.0024, | |
| "step": 38525 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 7.332347568931026e-05, | |
| "loss": 0.0022, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.276777516840142e-05, | |
| "loss": 0.0024, | |
| "step": 39675 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.221207464749259e-05, | |
| "loss": 0.0023, | |
| "step": 40250 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.165637412658375e-05, | |
| "loss": 0.0023, | |
| "step": 40825 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.110067360567492e-05, | |
| "loss": 0.0022, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.054497308476608e-05, | |
| "loss": 0.0023, | |
| "step": 41975 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.998927256385725e-05, | |
| "loss": 0.0023, | |
| "step": 42550 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.943357204294841e-05, | |
| "loss": 0.0023, | |
| "step": 43125 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.887787152203957e-05, | |
| "loss": 0.0024, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.832217100113074e-05, | |
| "loss": 0.0024, | |
| "step": 44275 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.77664704802219e-05, | |
| "loss": 0.0022, | |
| "step": 44850 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.721076995931305e-05, | |
| "loss": 0.0022, | |
| "step": 45425 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.025934860110282898, | |
| "eval_max_distance": 8, | |
| "eval_mean_distance": 0, | |
| "eval_runtime": 21.6745, | |
| "eval_samples_per_second": 11.903, | |
| "eval_steps_per_second": 0.83, | |
| "step": 45988 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.665506943840422e-05, | |
| "loss": 0.0024, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.609936891749538e-05, | |
| "loss": 0.0019, | |
| "step": 46575 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.554366839658655e-05, | |
| "loss": 0.0021, | |
| "step": 47150 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.498796787567771e-05, | |
| "loss": 0.0019, | |
| "step": 47725 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.443226735476888e-05, | |
| "loss": 0.0021, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.387656683386004e-05, | |
| "loss": 0.0018, | |
| "step": 48875 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 6.332086631295121e-05, | |
| "loss": 0.002, | |
| "step": 49450 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 6.276516579204237e-05, | |
| "loss": 0.0021, | |
| "step": 50025 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.220946527113354e-05, | |
| "loss": 0.002, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.16537647502247e-05, | |
| "loss": 0.0019, | |
| "step": 51175 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 6.109806422931587e-05, | |
| "loss": 0.002, | |
| "step": 51750 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 6.0542363708407024e-05, | |
| "loss": 0.0021, | |
| "step": 52325 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.9986663187498195e-05, | |
| "loss": 0.0021, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.943096266658935e-05, | |
| "loss": 0.0019, | |
| "step": 53475 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5.8875262145680524e-05, | |
| "loss": 0.0018, | |
| "step": 54050 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5.831956162477168e-05, | |
| "loss": 0.0021, | |
| "step": 54625 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.7763861103862846e-05, | |
| "loss": 0.0021, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.720816058295401e-05, | |
| "loss": 0.0019, | |
| "step": 55775 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5.6652460062045174e-05, | |
| "loss": 0.002, | |
| "step": 56350 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 5.609675954113633e-05, | |
| "loss": 0.0022, | |
| "step": 56925 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 5.55410590202275e-05, | |
| "loss": 0.0018, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 5.498535849931866e-05, | |
| "loss": 0.0023, | |
| "step": 58075 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 5.442965797840983e-05, | |
| "loss": 0.0021, | |
| "step": 58650 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.387395745750099e-05, | |
| "loss": 0.002, | |
| "step": 59225 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.331825693659216e-05, | |
| "loss": 0.0019, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.276255641568332e-05, | |
| "loss": 0.0021, | |
| "step": 60375 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.220685589477449e-05, | |
| "loss": 0.0022, | |
| "step": 60950 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.1651155373865647e-05, | |
| "loss": 0.0017, | |
| "step": 61525 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.109545485295682e-05, | |
| "loss": 0.0019, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 5.0539754332047975e-05, | |
| "loss": 0.002, | |
| "step": 62675 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.998405381113914e-05, | |
| "loss": 0.0019, | |
| "step": 63250 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.9428353290230304e-05, | |
| "loss": 0.0022, | |
| "step": 63825 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.887265276932147e-05, | |
| "loss": 0.0021, | |
| "step": 64400 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.831695224841263e-05, | |
| "loss": 0.0023, | |
| "step": 64975 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.77612517275038e-05, | |
| "loss": 0.002, | |
| "step": 65550 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.720555120659496e-05, | |
| "loss": 0.002, | |
| "step": 66125 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.6649850685686126e-05, | |
| "loss": 0.0021, | |
| "step": 66700 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.609415016477729e-05, | |
| "loss": 0.0021, | |
| "step": 67275 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.5538449643868454e-05, | |
| "loss": 0.002, | |
| "step": 67850 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.498274912295962e-05, | |
| "loss": 0.0019, | |
| "step": 68425 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.029171258211135864, | |
| "eval_max_distance": 8, | |
| "eval_mean_distance": 0, | |
| "eval_runtime": 20.5415, | |
| "eval_samples_per_second": 12.56, | |
| "eval_steps_per_second": 0.876, | |
| "step": 68982 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.442704860205078e-05, | |
| "loss": 0.0021, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.387134808114195e-05, | |
| "loss": 0.0017, | |
| "step": 69575 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.3315647560233105e-05, | |
| "loss": 0.002, | |
| "step": 70150 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.275994703932427e-05, | |
| "loss": 0.0016, | |
| "step": 70725 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.2204246518415434e-05, | |
| "loss": 0.0018, | |
| "step": 71300 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.16485459975066e-05, | |
| "loss": 0.0018, | |
| "step": 71875 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.109284547659776e-05, | |
| "loss": 0.0018, | |
| "step": 72450 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.0537144955688927e-05, | |
| "loss": 0.0018, | |
| "step": 73025 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.998144443478009e-05, | |
| "loss": 0.002, | |
| "step": 73600 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.9425743913871255e-05, | |
| "loss": 0.002, | |
| "step": 74175 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.887004339296242e-05, | |
| "loss": 0.0019, | |
| "step": 74750 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.8314342872053584e-05, | |
| "loss": 0.0018, | |
| "step": 75325 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.775864235114475e-05, | |
| "loss": 0.0018, | |
| "step": 75900 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.720294183023591e-05, | |
| "loss": 0.0017, | |
| "step": 76475 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.664724130932708e-05, | |
| "loss": 0.0017, | |
| "step": 77050 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.6091540788418234e-05, | |
| "loss": 0.0018, | |
| "step": 77625 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.55358402675094e-05, | |
| "loss": 0.0018, | |
| "step": 78200 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.498013974660056e-05, | |
| "loss": 0.0016, | |
| "step": 78775 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.442443922569173e-05, | |
| "loss": 0.0016, | |
| "step": 79350 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.386873870478289e-05, | |
| "loss": 0.0018, | |
| "step": 79925 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 3.3313038183874056e-05, | |
| "loss": 0.0017, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.275733766296522e-05, | |
| "loss": 0.0017, | |
| "step": 81075 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.2201637142056385e-05, | |
| "loss": 0.0016, | |
| "step": 81650 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 3.164593662114755e-05, | |
| "loss": 0.002, | |
| "step": 82225 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.1090236100238714e-05, | |
| "loss": 0.0018, | |
| "step": 82800 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 3.053453557932988e-05, | |
| "loss": 0.0016, | |
| "step": 83375 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.997883505842104e-05, | |
| "loss": 0.0017, | |
| "step": 83950 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.9423134537512203e-05, | |
| "loss": 0.0019, | |
| "step": 84525 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.8867434016603368e-05, | |
| "loss": 0.0018, | |
| "step": 85100 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.8311733495694532e-05, | |
| "loss": 0.0017, | |
| "step": 85675 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.7756032974785696e-05, | |
| "loss": 0.0017, | |
| "step": 86250 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.720033245387686e-05, | |
| "loss": 0.0019, | |
| "step": 86825 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.664463193296802e-05, | |
| "loss": 0.002, | |
| "step": 87400 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.6088931412059186e-05, | |
| "loss": 0.0019, | |
| "step": 87975 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 2.553323089115035e-05, | |
| "loss": 0.0017, | |
| "step": 88550 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.4977530370241514e-05, | |
| "loss": 0.0018, | |
| "step": 89125 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.442182984933268e-05, | |
| "loss": 0.0018, | |
| "step": 89700 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 2.3866129328423843e-05, | |
| "loss": 0.0018, | |
| "step": 90275 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 2.3310428807515004e-05, | |
| "loss": 0.0019, | |
| "step": 90850 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 2.275472828660617e-05, | |
| "loss": 0.0016, | |
| "step": 91425 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.02807791158556938, | |
| "eval_max_distance": 8, | |
| "eval_mean_distance": 0, | |
| "eval_runtime": 20.9058, | |
| "eval_samples_per_second": 12.341, | |
| "eval_steps_per_second": 0.861, | |
| "step": 91976 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.2199027765697333e-05, | |
| "loss": 0.0019, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.1643327244788497e-05, | |
| "loss": 0.0015, | |
| "step": 92575 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 2.108762672387966e-05, | |
| "loss": 0.0016, | |
| "step": 93150 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 2.0531926202970826e-05, | |
| "loss": 0.0015, | |
| "step": 93725 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.997622568206199e-05, | |
| "loss": 0.0017, | |
| "step": 94300 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 1.942052516115315e-05, | |
| "loss": 0.0016, | |
| "step": 94875 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.8864824640244315e-05, | |
| "loss": 0.0017, | |
| "step": 95450 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 1.830912411933548e-05, | |
| "loss": 0.0017, | |
| "step": 96025 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.7753423598426644e-05, | |
| "loss": 0.0015, | |
| "step": 96600 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.719772307751781e-05, | |
| "loss": 0.0018, | |
| "step": 97175 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 1.6642022556608973e-05, | |
| "loss": 0.0018, | |
| "step": 97750 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.6086322035700134e-05, | |
| "loss": 0.0014, | |
| "step": 98325 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 1.5530621514791298e-05, | |
| "loss": 0.0016, | |
| "step": 98900 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.4974920993882462e-05, | |
| "loss": 0.0018, | |
| "step": 99475 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.4419220472973627e-05, | |
| "loss": 0.0015, | |
| "step": 100050 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 1.3863519952064791e-05, | |
| "loss": 0.0019, | |
| "step": 100625 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 1.3307819431155954e-05, | |
| "loss": 0.0016, | |
| "step": 101200 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 1.2752118910247118e-05, | |
| "loss": 0.0015, | |
| "step": 101775 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 1.2196418389338282e-05, | |
| "loss": 0.0017, | |
| "step": 102350 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 1.1640717868429447e-05, | |
| "loss": 0.0016, | |
| "step": 102925 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 1.108501734752061e-05, | |
| "loss": 0.0015, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 1.0529316826611774e-05, | |
| "loss": 0.0017, | |
| "step": 104075 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 9.973616305702938e-06, | |
| "loss": 0.0016, | |
| "step": 104650 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 9.4179157847941e-06, | |
| "loss": 0.0016, | |
| "step": 105225 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 8.862215263885265e-06, | |
| "loss": 0.0017, | |
| "step": 105800 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 8.30651474297643e-06, | |
| "loss": 0.0015, | |
| "step": 106375 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 7.750814222067592e-06, | |
| "loss": 0.0014, | |
| "step": 106950 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 7.195113701158756e-06, | |
| "loss": 0.0017, | |
| "step": 107525 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 6.639413180249921e-06, | |
| "loss": 0.0017, | |
| "step": 108100 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 6.083712659341084e-06, | |
| "loss": 0.0016, | |
| "step": 108675 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 5.528012138432248e-06, | |
| "loss": 0.0017, | |
| "step": 109250 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 4.972311617523412e-06, | |
| "loss": 0.0018, | |
| "step": 109825 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 4.4166110966145756e-06, | |
| "loss": 0.0015, | |
| "step": 110400 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.86091057570574e-06, | |
| "loss": 0.0017, | |
| "step": 110975 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 3.3052100547969034e-06, | |
| "loss": 0.0014, | |
| "step": 111550 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.7495095338880677e-06, | |
| "loss": 0.0018, | |
| "step": 112125 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.1938090129792312e-06, | |
| "loss": 0.0017, | |
| "step": 112700 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 1.6381084920703951e-06, | |
| "loss": 0.0018, | |
| "step": 113275 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.082407971161559e-06, | |
| "loss": 0.0015, | |
| "step": 113850 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.26707450252723e-07, | |
| "loss": 0.0016, | |
| "step": 114425 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.027882983908057213, | |
| "eval_max_distance": 8, | |
| "eval_mean_distance": 0, | |
| "eval_runtime": 20.563, | |
| "eval_samples_per_second": 12.547, | |
| "eval_steps_per_second": 0.875, | |
| "step": 114970 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 114970, | |
| "total_flos": 2.9139726999711744e+16, | |
| "train_loss": 0.0019824859863435676, | |
| "train_runtime": 8494.2801, | |
| "train_samples_per_second": 203.021, | |
| "train_steps_per_second": 13.535 | |
| } | |
| ], | |
| "logging_steps": 575, | |
| "max_steps": 114970, | |
| "num_train_epochs": 5, | |
| "save_steps": 1150, | |
| "total_flos": 2.9139726999711744e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |