{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.998099762470309, "global_step": 2104, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.95, "learning_rate": 3.811787072243346e-05, "loss": 0.2524, "step": 500 }, { "epoch": 1.0, "eval_accuracy": 0.908256880733945, "eval_loss": 0.21832765638828278, "eval_runtime": 5.5594, "eval_samples_per_second": 156.85, "eval_steps_per_second": 5.036, "step": 526 }, { "epoch": 1.9, "learning_rate": 2.6235741444866924e-05, "loss": 0.1152, "step": 1000 }, { "epoch": 2.0, "eval_accuracy": 0.8944954128440367, "eval_loss": 0.27505654096603394, "eval_runtime": 5.5334, "eval_samples_per_second": 157.59, "eval_steps_per_second": 5.06, "step": 1052 }, { "epoch": 2.85, "learning_rate": 1.4353612167300381e-05, "loss": 0.0647, "step": 1500 }, { "epoch": 3.0, "eval_accuracy": 0.9059633027522935, "eval_loss": 0.3214015066623688, "eval_runtime": 5.5423, "eval_samples_per_second": 157.334, "eval_steps_per_second": 5.052, "step": 1578 }, { "epoch": 3.8, "learning_rate": 2.4714828897338406e-06, "loss": 0.0322, "step": 2000 }, { "epoch": 4.0, "eval_accuracy": 0.9208715596330275, "eval_loss": 0.3048288822174072, "eval_runtime": 5.5583, "eval_samples_per_second": 156.883, "eval_steps_per_second": 5.038, "step": 2104 }, { "epoch": 4.0, "step": 2104, "total_flos": 2.509461556171899e+17, "train_loss": 0.11159764983808133, "train_runtime": 5170.0912, "train_samples_per_second": 52.107, "train_steps_per_second": 0.407 } ], "max_steps": 2104, "num_train_epochs": 4, "total_flos": 2.509461556171899e+17, "trial_name": null, "trial_params": null }