{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 48090, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.1, "learning_rate": 4.948014140153878e-05, "loss": 0.7397, "step": 500 }, { "epoch": 0.21, "learning_rate": 4.8960282803077565e-05, "loss": 0.2767, "step": 1000 }, { "epoch": 0.31, "learning_rate": 4.844042420461635e-05, "loss": 0.4767, "step": 1500 }, { "epoch": 0.42, "learning_rate": 4.792056560615513e-05, "loss": 0.4352, "step": 2000 }, { "epoch": 0.52, "learning_rate": 4.740070700769391e-05, "loss": 0.6092, "step": 2500 }, { "epoch": 0.62, "learning_rate": 4.688084840923269e-05, "loss": 0.1961, "step": 3000 }, { "epoch": 0.73, "learning_rate": 4.636098981077147e-05, "loss": 0.2384, "step": 3500 }, { "epoch": 0.83, "learning_rate": 4.5841131212310254e-05, "loss": 0.2925, "step": 4000 }, { "epoch": 0.94, "learning_rate": 4.532127261384904e-05, "loss": 0.1461, "step": 4500 }, { "epoch": 1.04, "learning_rate": 4.4801414015387816e-05, "loss": 0.1366, "step": 5000 }, { "epoch": 1.14, "learning_rate": 4.4281555416926594e-05, "loss": 0.0943, "step": 5500 }, { "epoch": 1.25, "learning_rate": 4.376169681846538e-05, "loss": 0.098, "step": 6000 }, { "epoch": 1.35, "learning_rate": 4.324183822000416e-05, "loss": 0.2768, "step": 6500 }, { "epoch": 1.46, "learning_rate": 4.272197962154294e-05, "loss": 0.1589, "step": 7000 }, { "epoch": 1.56, "learning_rate": 4.220212102308173e-05, "loss": 0.2075, "step": 7500 }, { "epoch": 1.66, "learning_rate": 4.1682262424620505e-05, "loss": 0.1032, "step": 8000 }, { "epoch": 1.77, "learning_rate": 4.116240382615928e-05, "loss": 0.092, "step": 8500 }, { "epoch": 1.87, "learning_rate": 4.064254522769807e-05, "loss": 0.0877, "step": 9000 }, { "epoch": 1.98, "learning_rate": 4.012268662923685e-05, "loss": 0.1081, "step": 9500 }, { "epoch": 2.08, "learning_rate": 3.960282803077563e-05, "loss": 0.0561, "step": 10000 }, { "epoch": 2.18, "learning_rate": 3.9082969432314415e-05, "loss": 0.0463, "step": 10500 }, { "epoch": 2.29, "learning_rate": 3.856311083385319e-05, "loss": 0.0453, "step": 11000 }, { "epoch": 2.39, "learning_rate": 3.804325223539197e-05, "loss": 0.0466, "step": 11500 }, { "epoch": 2.5, "learning_rate": 3.7523393636930756e-05, "loss": 0.0484, "step": 12000 }, { "epoch": 2.6, "learning_rate": 3.700353503846954e-05, "loss": 0.0442, "step": 12500 }, { "epoch": 2.7, "learning_rate": 3.648367644000832e-05, "loss": 0.1853, "step": 13000 }, { "epoch": 2.81, "learning_rate": 3.59638178415471e-05, "loss": 0.0908, "step": 13500 }, { "epoch": 2.91, "learning_rate": 3.544395924308588e-05, "loss": 0.0353, "step": 14000 }, { "epoch": 3.02, "learning_rate": 3.4924100644624667e-05, "loss": 0.0333, "step": 14500 }, { "epoch": 3.12, "learning_rate": 3.4404242046163445e-05, "loss": 0.0379, "step": 15000 }, { "epoch": 3.22, "learning_rate": 3.388438344770223e-05, "loss": 0.0439, "step": 15500 }, { "epoch": 3.33, "learning_rate": 3.336452484924101e-05, "loss": 0.0419, "step": 16000 }, { "epoch": 3.43, "learning_rate": 3.2844666250779785e-05, "loss": 0.0355, "step": 16500 }, { "epoch": 3.54, "learning_rate": 3.232480765231857e-05, "loss": 0.0368, "step": 17000 }, { "epoch": 3.64, "learning_rate": 3.1804949053857355e-05, "loss": 0.0336, "step": 17500 }, { "epoch": 3.74, "learning_rate": 3.128509045539613e-05, "loss": 0.176, "step": 18000 }, { "epoch": 3.85, "learning_rate": 3.076523185693492e-05, "loss": 0.2009, "step": 18500 }, { "epoch": 3.95, "learning_rate": 3.0245373258473696e-05, "loss": 0.1369, "step": 19000 }, { "epoch": 4.05, "learning_rate": 2.9725514660012477e-05, "loss": 0.1509, "step": 19500 }, { "epoch": 4.16, "learning_rate": 2.9205656061551262e-05, "loss": 0.1126, "step": 20000 }, { "epoch": 4.26, "learning_rate": 2.868579746309004e-05, "loss": 0.0878, "step": 20500 }, { "epoch": 4.37, "learning_rate": 2.816593886462882e-05, "loss": 0.0961, "step": 21000 }, { "epoch": 4.47, "learning_rate": 2.76460802661676e-05, "loss": 0.1387, "step": 21500 }, { "epoch": 4.57, "learning_rate": 2.7126221667706388e-05, "loss": 0.0979, "step": 22000 }, { "epoch": 4.68, "learning_rate": 2.6606363069245166e-05, "loss": 0.0647, "step": 22500 }, { "epoch": 4.78, "learning_rate": 2.6086504470783947e-05, "loss": 0.05, "step": 23000 }, { "epoch": 4.89, "learning_rate": 2.5566645872322732e-05, "loss": 0.0221, "step": 23500 }, { "epoch": 4.99, "learning_rate": 2.504678727386151e-05, "loss": 0.0252, "step": 24000 }, { "epoch": 5.09, "learning_rate": 2.452692867540029e-05, "loss": 0.0245, "step": 24500 }, { "epoch": 5.2, "learning_rate": 2.4007070076939073e-05, "loss": 0.0195, "step": 25000 }, { "epoch": 5.3, "learning_rate": 2.3487211478477854e-05, "loss": 0.023, "step": 25500 }, { "epoch": 5.41, "learning_rate": 2.296735288001664e-05, "loss": 0.0204, "step": 26000 }, { "epoch": 5.51, "learning_rate": 2.2447494281555417e-05, "loss": 0.0789, "step": 26500 }, { "epoch": 5.61, "learning_rate": 2.19276356830942e-05, "loss": 0.056, "step": 27000 }, { "epoch": 5.72, "learning_rate": 2.140777708463298e-05, "loss": 0.0301, "step": 27500 }, { "epoch": 5.82, "learning_rate": 2.088791848617176e-05, "loss": 0.0439, "step": 28000 }, { "epoch": 5.93, "learning_rate": 2.0368059887710546e-05, "loss": 0.0254, "step": 28500 }, { "epoch": 6.03, "learning_rate": 1.9848201289249324e-05, "loss": 0.0132, "step": 29000 }, { "epoch": 6.13, "learning_rate": 1.9328342690788105e-05, "loss": 0.019, "step": 29500 }, { "epoch": 6.24, "learning_rate": 1.880848409232689e-05, "loss": 0.0187, "step": 30000 }, { "epoch": 6.34, "learning_rate": 1.8288625493865668e-05, "loss": 0.0196, "step": 30500 }, { "epoch": 6.45, "learning_rate": 1.776876689540445e-05, "loss": 0.062, "step": 31000 }, { "epoch": 6.55, "learning_rate": 1.7248908296943234e-05, "loss": 0.0454, "step": 31500 }, { "epoch": 6.65, "learning_rate": 1.6729049698482012e-05, "loss": 0.0153, "step": 32000 }, { "epoch": 6.76, "learning_rate": 1.6209191100020797e-05, "loss": 0.0133, "step": 32500 }, { "epoch": 6.86, "learning_rate": 1.5689332501559575e-05, "loss": 0.0158, "step": 33000 }, { "epoch": 6.97, "learning_rate": 1.5169473903098358e-05, "loss": 0.0143, "step": 33500 }, { "epoch": 7.07, "learning_rate": 1.464961530463714e-05, "loss": 0.0196, "step": 34000 }, { "epoch": 7.17, "learning_rate": 1.412975670617592e-05, "loss": 0.01, "step": 34500 }, { "epoch": 7.28, "learning_rate": 1.3609898107714703e-05, "loss": 0.0089, "step": 35000 }, { "epoch": 7.38, "learning_rate": 1.3090039509253486e-05, "loss": 0.0106, "step": 35500 }, { "epoch": 7.49, "learning_rate": 1.2570180910792265e-05, "loss": 0.0093, "step": 36000 }, { "epoch": 7.59, "learning_rate": 1.2050322312331047e-05, "loss": 0.0082, "step": 36500 }, { "epoch": 7.69, "learning_rate": 1.1530463713869828e-05, "loss": 0.0112, "step": 37000 }, { "epoch": 7.8, "learning_rate": 1.101060511540861e-05, "loss": 0.0132, "step": 37500 }, { "epoch": 7.9, "learning_rate": 1.0490746516947391e-05, "loss": 0.0092, "step": 38000 }, { "epoch": 8.01, "learning_rate": 9.970887918486172e-06, "loss": 0.0109, "step": 38500 }, { "epoch": 8.11, "learning_rate": 9.451029320024954e-06, "loss": 0.0083, "step": 39000 }, { "epoch": 8.21, "learning_rate": 8.931170721563735e-06, "loss": 0.0049, "step": 39500 }, { "epoch": 8.32, "learning_rate": 8.411312123102517e-06, "loss": 0.0076, "step": 40000 }, { "epoch": 8.42, "learning_rate": 7.891453524641298e-06, "loss": 0.0054, "step": 40500 }, { "epoch": 8.53, "learning_rate": 7.3715949261800795e-06, "loss": 0.0033, "step": 41000 }, { "epoch": 8.63, "learning_rate": 6.851736327718861e-06, "loss": 0.0051, "step": 41500 }, { "epoch": 8.73, "learning_rate": 6.3318777292576415e-06, "loss": 0.0057, "step": 42000 }, { "epoch": 8.84, "learning_rate": 5.812019130796424e-06, "loss": 0.008, "step": 42500 }, { "epoch": 8.94, "learning_rate": 5.292160532335205e-06, "loss": 0.005, "step": 43000 }, { "epoch": 9.05, "learning_rate": 4.7723019338739865e-06, "loss": 0.0058, "step": 43500 }, { "epoch": 9.15, "learning_rate": 4.252443335412768e-06, "loss": 0.0042, "step": 44000 }, { "epoch": 9.25, "learning_rate": 3.7325847369515494e-06, "loss": 0.0034, "step": 44500 }, { "epoch": 9.36, "learning_rate": 3.2127261384903308e-06, "loss": 0.0029, "step": 45000 }, { "epoch": 9.46, "learning_rate": 2.692867540029112e-06, "loss": 0.0036, "step": 45500 }, { "epoch": 9.57, "learning_rate": 2.1730089415678936e-06, "loss": 0.0034, "step": 46000 }, { "epoch": 9.67, "learning_rate": 1.6531503431066752e-06, "loss": 0.0021, "step": 46500 }, { "epoch": 9.77, "learning_rate": 1.1332917446454564e-06, "loss": 0.0039, "step": 47000 }, { "epoch": 9.88, "learning_rate": 6.134331461842379e-07, "loss": 0.0033, "step": 47500 }, { "epoch": 9.98, "learning_rate": 9.357454772301934e-08, "loss": 0.0068, "step": 48000 }, { "epoch": 10.0, "step": 48090, "total_flos": 1.579934745091154e+17, "train_runtime": 28299.6455, "train_samples_per_second": 1.699 } ], "max_steps": 48090, "num_train_epochs": 10, "total_flos": 1.579934745091154e+17, "trial_name": null, "trial_params": null }