{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 2260, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.11, "learning_rate": 4.8893805309734514e-05, "loss": 5.1703, "step": 50 }, { "epoch": 0.22, "learning_rate": 4.778761061946903e-05, "loss": 4.528, "step": 100 }, { "epoch": 0.33, "learning_rate": 4.668141592920354e-05, "loss": 4.2274, "step": 150 }, { "epoch": 0.44, "learning_rate": 4.5575221238938055e-05, "loss": 4.183, "step": 200 }, { "epoch": 0.55, "learning_rate": 4.446902654867257e-05, "loss": 4.0013, "step": 250 }, { "epoch": 0.66, "learning_rate": 4.3362831858407084e-05, "loss": 4.0381, "step": 300 }, { "epoch": 0.77, "learning_rate": 4.2256637168141596e-05, "loss": 3.9149, "step": 350 }, { "epoch": 0.88, "learning_rate": 4.115044247787611e-05, "loss": 3.9236, "step": 400 }, { "epoch": 1.0, "learning_rate": 4.0044247787610625e-05, "loss": 3.9675, "step": 450 }, { "epoch": 1.11, "learning_rate": 3.893805309734514e-05, "loss": 3.8204, "step": 500 }, { "epoch": 1.22, "learning_rate": 3.783185840707965e-05, "loss": 3.833, "step": 550 }, { "epoch": 1.33, "learning_rate": 3.672566371681416e-05, "loss": 3.8205, "step": 600 }, { "epoch": 1.44, "learning_rate": 3.561946902654867e-05, "loss": 3.8247, "step": 650 }, { "epoch": 1.55, "learning_rate": 3.451327433628319e-05, "loss": 3.8613, "step": 700 }, { "epoch": 1.66, "learning_rate": 3.34070796460177e-05, "loss": 3.7906, "step": 750 }, { "epoch": 1.77, "learning_rate": 3.230088495575221e-05, "loss": 3.829, "step": 800 }, { "epoch": 1.88, "learning_rate": 3.119469026548672e-05, "loss": 3.7752, "step": 850 }, { "epoch": 1.99, "learning_rate": 3.008849557522124e-05, "loss": 3.7212, "step": 900 }, { "epoch": 2.1, "learning_rate": 2.8982300884955753e-05, "loss": 3.8136, "step": 950 }, { "epoch": 2.21, "learning_rate": 2.7876106194690264e-05, "loss": 3.7791, "step": 1000 }, { "epoch": 2.32, "learning_rate": 2.6769911504424782e-05, "loss": 3.8232, "step": 1050 }, { "epoch": 2.43, "learning_rate": 2.5663716814159294e-05, "loss": 3.6982, "step": 1100 }, { "epoch": 2.54, "learning_rate": 2.4557522123893805e-05, "loss": 3.7689, "step": 1150 }, { "epoch": 2.65, "learning_rate": 2.345132743362832e-05, "loss": 3.7717, "step": 1200 }, { "epoch": 2.77, "learning_rate": 2.234513274336283e-05, "loss": 3.6647, "step": 1250 }, { "epoch": 2.88, "learning_rate": 2.1238938053097346e-05, "loss": 3.6956, "step": 1300 }, { "epoch": 2.99, "learning_rate": 2.013274336283186e-05, "loss": 3.705, "step": 1350 }, { "epoch": 3.1, "learning_rate": 1.9026548672566372e-05, "loss": 3.5994, "step": 1400 }, { "epoch": 3.21, "learning_rate": 1.7920353982300887e-05, "loss": 3.8063, "step": 1450 }, { "epoch": 3.32, "learning_rate": 1.6814159292035402e-05, "loss": 3.6585, "step": 1500 }, { "epoch": 3.43, "learning_rate": 1.5707964601769913e-05, "loss": 3.705, "step": 1550 }, { "epoch": 3.54, "learning_rate": 1.4601769911504426e-05, "loss": 3.6719, "step": 1600 }, { "epoch": 3.65, "learning_rate": 1.3495575221238938e-05, "loss": 3.7247, "step": 1650 }, { "epoch": 3.76, "learning_rate": 1.2389380530973452e-05, "loss": 3.7643, "step": 1700 }, { "epoch": 3.87, "learning_rate": 1.1283185840707964e-05, "loss": 3.6094, "step": 1750 }, { "epoch": 3.98, "learning_rate": 1.0176991150442479e-05, "loss": 3.6419, "step": 1800 }, { "epoch": 4.09, "learning_rate": 9.070796460176992e-06, "loss": 3.6619, "step": 1850 }, { "epoch": 4.2, "learning_rate": 7.964601769911505e-06, "loss": 3.6914, "step": 1900 }, { "epoch": 4.31, "learning_rate": 6.858407079646018e-06, "loss": 3.6443, "step": 1950 }, { "epoch": 4.42, "learning_rate": 5.752212389380531e-06, "loss": 3.5983, "step": 2000 }, { "epoch": 4.54, "learning_rate": 4.646017699115045e-06, "loss": 3.7606, "step": 2050 }, { "epoch": 4.65, "learning_rate": 3.5398230088495575e-06, "loss": 3.6427, "step": 2100 }, { "epoch": 4.76, "learning_rate": 2.433628318584071e-06, "loss": 3.7245, "step": 2150 }, { "epoch": 4.87, "learning_rate": 1.3274336283185841e-06, "loss": 3.6501, "step": 2200 }, { "epoch": 4.98, "learning_rate": 2.2123893805309735e-07, "loss": 3.679, "step": 2250 }, { "epoch": 5.0, "step": 2260, "train_runtime": 938.0621, "train_samples_per_second": 2.409 } ], "max_steps": 2260, "num_train_epochs": 5, "total_flos": 2729754170757120, "trial_name": null, "trial_params": null }