{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 21050, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.12, "learning_rate": 1.9527790973871736e-05, "loss": 0.3444, "step": 500 }, { "epoch": 0.24, "learning_rate": 1.905273159144893e-05, "loss": 0.2812, "step": 1000 }, { "epoch": 0.36, "learning_rate": 1.857767220902613e-05, "loss": 0.265, "step": 1500 }, { "epoch": 0.48, "learning_rate": 1.8102612826603327e-05, "loss": 0.2375, "step": 2000 }, { "epoch": 0.59, "learning_rate": 1.7628503562945368e-05, "loss": 0.2282, "step": 2500 }, { "epoch": 0.71, "learning_rate": 1.7153444180522567e-05, "loss": 0.2192, "step": 3000 }, { "epoch": 0.83, "learning_rate": 1.6678384798099765e-05, "loss": 0.2187, "step": 3500 }, { "epoch": 0.95, "learning_rate": 1.620332541567696e-05, "loss": 0.2216, "step": 4000 }, { "epoch": 1.07, "learning_rate": 1.5729216152019004e-05, "loss": 0.1802, "step": 4500 }, { "epoch": 1.19, "learning_rate": 1.5254156769596201e-05, "loss": 0.1474, "step": 5000 }, { "epoch": 1.31, "learning_rate": 1.4779097387173397e-05, "loss": 0.144, "step": 5500 }, { "epoch": 1.43, "learning_rate": 1.4304038004750596e-05, "loss": 0.1607, "step": 6000 }, { "epoch": 1.54, "learning_rate": 1.382897862232779e-05, "loss": 0.1622, "step": 6500 }, { "epoch": 1.66, "learning_rate": 1.3354869358669835e-05, "loss": 0.1581, "step": 7000 }, { "epoch": 1.78, "learning_rate": 1.2879809976247033e-05, "loss": 0.1535, "step": 7500 }, { "epoch": 1.9, "learning_rate": 1.2404750593824228e-05, "loss": 0.1528, "step": 8000 }, { "epoch": 2.02, "learning_rate": 1.1929691211401427e-05, "loss": 0.1475, "step": 8500 }, { "epoch": 2.14, "learning_rate": 1.1455581947743468e-05, "loss": 0.1033, "step": 9000 }, { "epoch": 2.26, "learning_rate": 1.0981472684085512e-05, "loss": 0.0884, "step": 9500 }, { "epoch": 2.38, "learning_rate": 1.0506413301662707e-05, "loss": 0.102, "step": 10000 }, { "epoch": 2.49, "learning_rate": 1.0031353919239906e-05, "loss": 0.104, "step": 10500 }, { "epoch": 2.61, "learning_rate": 9.556294536817104e-06, "loss": 0.1026, "step": 11000 }, { "epoch": 2.73, "learning_rate": 9.0812351543943e-06, "loss": 0.1116, "step": 11500 }, { "epoch": 2.85, "learning_rate": 8.606175771971497e-06, "loss": 0.1002, "step": 12000 }, { "epoch": 2.97, "learning_rate": 8.131116389548695e-06, "loss": 0.0996, "step": 12500 }, { "epoch": 3.09, "learning_rate": 7.656057007125892e-06, "loss": 0.0686, "step": 13000 }, { "epoch": 3.21, "learning_rate": 7.180997624703089e-06, "loss": 0.0594, "step": 13500 }, { "epoch": 3.33, "learning_rate": 6.7068883610451305e-06, "loss": 0.0562, "step": 14000 }, { "epoch": 3.44, "learning_rate": 6.231828978622329e-06, "loss": 0.0523, "step": 14500 }, { "epoch": 3.56, "learning_rate": 5.757719714964372e-06, "loss": 0.0576, "step": 15000 }, { "epoch": 3.68, "learning_rate": 5.282660332541568e-06, "loss": 0.067, "step": 15500 }, { "epoch": 3.8, "learning_rate": 4.807600950118766e-06, "loss": 0.0664, "step": 16000 }, { "epoch": 3.92, "learning_rate": 4.332541567695962e-06, "loss": 0.064, "step": 16500 }, { "epoch": 4.04, "learning_rate": 3.857482185273159e-06, "loss": 0.0441, "step": 17000 }, { "epoch": 4.16, "learning_rate": 3.382422802850356e-06, "loss": 0.0306, "step": 17500 }, { "epoch": 4.28, "learning_rate": 2.9073634204275536e-06, "loss": 0.0287, "step": 18000 }, { "epoch": 4.39, "learning_rate": 2.4323040380047506e-06, "loss": 0.0314, "step": 18500 }, { "epoch": 4.51, "learning_rate": 1.957244655581948e-06, "loss": 0.0304, "step": 19000 }, { "epoch": 4.63, "learning_rate": 1.4821852731591448e-06, "loss": 0.0281, "step": 19500 }, { "epoch": 4.75, "learning_rate": 1.007125890736342e-06, "loss": 0.0257, "step": 20000 }, { "epoch": 4.87, "learning_rate": 5.320665083135392e-07, "loss": 0.0385, "step": 20500 }, { "epoch": 4.99, "learning_rate": 5.700712589073635e-08, "loss": 0.0246, "step": 21000 }, { "epoch": 5.0, "step": 21050, "total_flos": 7.845474691055616e+16, "train_loss": 0.11896094931842595, "train_runtime": 1931.7504, "train_samples_per_second": 174.321, "train_steps_per_second": 10.897 } ], "max_steps": 21050, "num_train_epochs": 5, "total_flos": 7.845474691055616e+16, "trial_name": null, "trial_params": null }