{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 15340, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.16, "learning_rate": 9.674054758800522e-06, "loss": 0.7901, "step": 500 }, { "epoch": 0.33, "learning_rate": 9.348109517601044e-06, "loss": 0.6318, "step": 1000 }, { "epoch": 0.49, "learning_rate": 9.022164276401565e-06, "loss": 0.5709, "step": 1500 }, { "epoch": 0.65, "learning_rate": 8.696219035202086e-06, "loss": 0.5524, "step": 2000 }, { "epoch": 0.81, "learning_rate": 8.370273794002608e-06, "loss": 0.5292, "step": 2500 }, { "epoch": 0.98, "learning_rate": 8.04432855280313e-06, "loss": 0.5161, "step": 3000 }, { "epoch": 1.14, "learning_rate": 7.71838331160365e-06, "loss": 0.4633, "step": 3500 }, { "epoch": 1.3, "learning_rate": 7.392438070404173e-06, "loss": 0.4524, "step": 4000 }, { "epoch": 1.47, "learning_rate": 7.066492829204694e-06, "loss": 0.4503, "step": 4500 }, { "epoch": 1.63, "learning_rate": 6.7405475880052155e-06, "loss": 0.4442, "step": 5000 }, { "epoch": 1.79, "learning_rate": 6.414602346805737e-06, "loss": 0.4399, "step": 5500 }, { "epoch": 1.96, "learning_rate": 6.088657105606258e-06, "loss": 0.4328, "step": 6000 }, { "epoch": 2.12, "learning_rate": 5.7627118644067805e-06, "loss": 0.3998, "step": 6500 }, { "epoch": 2.28, "learning_rate": 5.436766623207301e-06, "loss": 0.3769, "step": 7000 }, { "epoch": 2.44, "learning_rate": 5.110821382007823e-06, "loss": 0.3771, "step": 7500 }, { "epoch": 2.61, "learning_rate": 4.784876140808345e-06, "loss": 0.3796, "step": 8000 }, { "epoch": 2.77, "learning_rate": 4.458930899608866e-06, "loss": 0.3845, "step": 8500 }, { "epoch": 2.93, "learning_rate": 4.1329856584093875e-06, "loss": 0.3815, "step": 9000 }, { "epoch": 3.1, "learning_rate": 3.807040417209909e-06, "loss": 0.3546, "step": 9500 }, { "epoch": 3.26, "learning_rate": 3.4810951760104307e-06, "loss": 0.3405, "step": 10000 }, { "epoch": 3.42, "learning_rate": 3.155149934810952e-06, "loss": 0.3405, "step": 10500 }, { "epoch": 3.59, "learning_rate": 2.829204693611474e-06, "loss": 0.3432, "step": 11000 }, { "epoch": 3.75, "learning_rate": 2.5032594524119952e-06, "loss": 0.3351, "step": 11500 }, { "epoch": 3.91, "learning_rate": 2.1773142112125166e-06, "loss": 0.3377, "step": 12000 }, { "epoch": 4.07, "learning_rate": 1.851368970013038e-06, "loss": 0.3255, "step": 12500 }, { "epoch": 4.24, "learning_rate": 1.5254237288135596e-06, "loss": 0.3138, "step": 13000 }, { "epoch": 4.4, "learning_rate": 1.199478487614081e-06, "loss": 0.3123, "step": 13500 }, { "epoch": 4.56, "learning_rate": 8.735332464146024e-07, "loss": 0.314, "step": 14000 }, { "epoch": 4.73, "learning_rate": 5.475880052151239e-07, "loss": 0.3078, "step": 14500 }, { "epoch": 4.89, "learning_rate": 2.216427640156454e-07, "loss": 0.316, "step": 15000 }, { "epoch": 5.0, "step": 15340, "total_flos": 3.2666546994015744e+17, "train_runtime": 10826.9111, "train_samples_per_second": 1.417 } ], "max_steps": 15340, "num_train_epochs": 5, "total_flos": 3.2666546994015744e+17, "trial_name": null, "trial_params": null }