{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 500, "global_step": 522, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.29, "learning_rate": 1.8084291187739463e-05, "loss": 1.8183, "step": 50 }, { "epoch": 0.57, "learning_rate": 1.6168582375478928e-05, "loss": 1.7676, "step": 100 }, { "epoch": 0.86, "learning_rate": 1.4252873563218392e-05, "loss": 1.7023, "step": 150 }, { "epoch": 1.15, "learning_rate": 1.2337164750957855e-05, "loss": 1.5826, "step": 200 }, { "epoch": 1.44, "learning_rate": 1.0421455938697318e-05, "loss": 1.5113, "step": 250 }, { "epoch": 1.72, "learning_rate": 8.505747126436782e-06, "loss": 1.4513, "step": 300 }, { "epoch": 2.01, "learning_rate": 6.590038314176246e-06, "loss": 1.3732, "step": 350 }, { "epoch": 2.3, "learning_rate": 4.674329501915709e-06, "loss": 1.3101, "step": 400 }, { "epoch": 2.59, "learning_rate": 2.7586206896551725e-06, "loss": 1.3334, "step": 450 }, { "epoch": 2.87, "learning_rate": 8.429118773946361e-07, "loss": 1.2379, "step": 500 }, { "epoch": 2.87, "eval_accuracy": 0.5145772594752187, "eval_loss": 1.3184902667999268, "eval_runtime": 0.551, "eval_samples_per_second": 1244.933, "eval_steps_per_second": 78.035, "step": 500 }, { "epoch": 3.0, "step": 522, "total_flos": 485661087781800.0, "train_loss": 1.4998418946832532, "train_runtime": 84.0743, "train_samples_per_second": 197.968, "train_steps_per_second": 6.209 } ], "logging_steps": 50, "max_steps": 522, "num_train_epochs": 3, "save_steps": 500, "total_flos": 485661087781800.0, "trial_name": null, "trial_params": null }