{ "best_metric": 2.523940324783325, "best_model_checkpoint": "traffic_sign_detection/checkpoint-2895", "epoch": 5.0, "eval_steps": 500, "global_step": 2895, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.8635578583765112, "grad_norm": 1.785918951034546, "learning_rate": 8.418277680140598e-07, "loss": 2.6578, "step": 500 }, { "epoch": 1.0, "eval_accuracy": 0.849978383052313, "eval_loss": 2.7859294414520264, "eval_runtime": 47.349, "eval_samples_per_second": 97.7, "eval_steps_per_second": 12.228, "step": 579 }, { "epoch": 1.7271157167530224, "grad_norm": 1.761311411857605, "learning_rate": 6.660808435852372e-07, "loss": 2.5366, "step": 1000 }, { "epoch": 2.0, "eval_accuracy": 0.8769995676610463, "eval_loss": 2.669792413711548, "eval_runtime": 47.4363, "eval_samples_per_second": 97.52, "eval_steps_per_second": 12.206, "step": 1158 }, { "epoch": 2.5906735751295336, "grad_norm": 1.8694429397583008, "learning_rate": 4.903339191564148e-07, "loss": 2.4436, "step": 1500 }, { "epoch": 3.0, "eval_accuracy": 0.9005620406398617, "eval_loss": 2.5879104137420654, "eval_runtime": 47.3175, "eval_samples_per_second": 97.765, "eval_steps_per_second": 12.236, "step": 1737 }, { "epoch": 3.454231433506045, "grad_norm": 1.893329381942749, "learning_rate": 3.1458699472759226e-07, "loss": 2.3753, "step": 2000 }, { "epoch": 4.0, "eval_accuracy": 0.9051015996541288, "eval_loss": 2.5401721000671387, "eval_runtime": 48.1616, "eval_samples_per_second": 96.052, "eval_steps_per_second": 12.022, "step": 2316 }, { "epoch": 4.317789291882556, "grad_norm": 1.803151249885559, "learning_rate": 1.3884007029876976e-07, "loss": 2.3365, "step": 2500 }, { "epoch": 5.0, "eval_accuracy": 0.9055339386078686, "eval_loss": 2.523940324783325, "eval_runtime": 47.5865, "eval_samples_per_second": 97.212, "eval_steps_per_second": 12.167, "step": 2895 } ], "logging_steps": 500, "max_steps": 2895, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "total_flos": 7.172321994146857e+18, "train_batch_size": 32, "trial_name": null, "trial_params": null }