{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.7161760674100723, "eval_steps": 500, "global_step": 8000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 5e-06, "loss": 6.8393, "step": 100 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 4.0541, "step": 200 }, { "epoch": 0.03, "learning_rate": 1.5e-05, "loss": 3.3832, "step": 300 }, { "epoch": 0.04, "learning_rate": 2e-05, "loss": 3.0121, "step": 400 }, { "epoch": 0.04, "learning_rate": 2.5e-05, "loss": 2.7978, "step": 500 }, { "epoch": 0.05, "learning_rate": 3e-05, "loss": 2.6647, "step": 600 }, { "epoch": 0.06, "learning_rate": 3.5e-05, "loss": 2.5648, "step": 700 }, { "epoch": 0.07, "learning_rate": 4e-05, "loss": 2.4824, "step": 800 }, { "epoch": 0.08, "learning_rate": 4.5e-05, "loss": 2.4302, "step": 900 }, { "epoch": 0.09, "learning_rate": 5e-05, "loss": 2.3927, "step": 1000 }, { "epoch": 0.1, "learning_rate": 5.500000000000001e-05, "loss": 2.3448, "step": 1100 }, { "epoch": 0.11, "learning_rate": 6e-05, "loss": 2.3187, "step": 1200 }, { "epoch": 0.12, "learning_rate": 6.500000000000001e-05, "loss": 2.2914, "step": 1300 }, { "epoch": 0.13, "learning_rate": 7e-05, "loss": 2.2567, "step": 1400 }, { "epoch": 0.13, "learning_rate": 7.500000000000001e-05, "loss": 2.2325, "step": 1500 }, { "epoch": 0.14, "learning_rate": 8e-05, "loss": 2.2113, "step": 1600 }, { "epoch": 0.15, "learning_rate": 8.5e-05, "loss": 2.1963, "step": 1700 }, { "epoch": 0.16, "learning_rate": 9e-05, "loss": 2.1679, "step": 1800 }, { "epoch": 0.17, "learning_rate": 9.5e-05, "loss": 2.1459, "step": 1900 }, { "epoch": 0.18, "learning_rate": 0.0001, "loss": 2.1606, "step": 2000 }, { "epoch": 0.19, "learning_rate": 9.890948745910579e-05, "loss": 2.1069, "step": 2100 }, { "epoch": 0.2, "learning_rate": 9.781897491821155e-05, "loss": 2.1075, "step": 2200 }, { "epoch": 0.21, "learning_rate": 9.672846237731734e-05, "loss": 2.0788, "step": 2300 }, { "epoch": 0.21, "learning_rate": 9.563794983642313e-05, "loss": 2.0525, "step": 2400 }, { "epoch": 0.22, "learning_rate": 9.45474372955289e-05, "loss": 2.043, "step": 2500 }, { "epoch": 0.23, "learning_rate": 9.345692475463468e-05, "loss": 2.0098, "step": 2600 }, { "epoch": 0.24, "learning_rate": 9.236641221374047e-05, "loss": 1.9921, "step": 2700 }, { "epoch": 0.25, "learning_rate": 9.127589967284624e-05, "loss": 1.9714, "step": 2800 }, { "epoch": 0.26, "learning_rate": 9.018538713195203e-05, "loss": 1.9696, "step": 2900 }, { "epoch": 0.27, "learning_rate": 8.90948745910578e-05, "loss": 1.9363, "step": 3000 }, { "epoch": 0.28, "learning_rate": 8.800436205016358e-05, "loss": 1.9166, "step": 3100 }, { "epoch": 0.29, "learning_rate": 8.691384950926937e-05, "loss": 1.9182, "step": 3200 }, { "epoch": 0.3, "learning_rate": 8.582333696837514e-05, "loss": 1.9062, "step": 3300 }, { "epoch": 0.3, "learning_rate": 8.473282442748092e-05, "loss": 1.8886, "step": 3400 }, { "epoch": 0.31, "learning_rate": 8.36423118865867e-05, "loss": 1.8874, "step": 3500 }, { "epoch": 0.32, "learning_rate": 8.255179934569248e-05, "loss": 1.8618, "step": 3600 }, { "epoch": 0.33, "learning_rate": 8.146128680479825e-05, "loss": 1.8587, "step": 3700 }, { "epoch": 0.34, "learning_rate": 8.037077426390404e-05, "loss": 1.8326, "step": 3800 }, { "epoch": 0.35, "learning_rate": 7.928026172300981e-05, "loss": 1.8549, "step": 3900 }, { "epoch": 0.36, "learning_rate": 7.81897491821156e-05, "loss": 1.8167, "step": 4000 }, { "epoch": 0.37, "learning_rate": 7.709923664122138e-05, "loss": 1.8316, "step": 4100 }, { "epoch": 0.38, "learning_rate": 7.600872410032715e-05, "loss": 1.8188, "step": 4200 }, { "epoch": 0.38, "learning_rate": 7.491821155943294e-05, "loss": 1.7976, "step": 4300 }, { "epoch": 0.39, "learning_rate": 7.382769901853873e-05, "loss": 1.7888, "step": 4400 }, { "epoch": 0.4, "learning_rate": 7.273718647764449e-05, "loss": 1.7899, "step": 4500 }, { "epoch": 0.41, "learning_rate": 7.164667393675028e-05, "loss": 1.79, "step": 4600 }, { "epoch": 0.42, "learning_rate": 7.055616139585605e-05, "loss": 1.7564, "step": 4700 }, { "epoch": 0.43, "learning_rate": 6.946564885496184e-05, "loss": 1.7666, "step": 4800 }, { "epoch": 0.44, "learning_rate": 6.837513631406761e-05, "loss": 1.7574, "step": 4900 }, { "epoch": 0.45, "learning_rate": 6.728462377317339e-05, "loss": 1.7588, "step": 5000 }, { "epoch": 0.46, "learning_rate": 6.619411123227918e-05, "loss": 1.7476, "step": 5100 }, { "epoch": 0.47, "learning_rate": 6.510359869138496e-05, "loss": 1.7251, "step": 5200 }, { "epoch": 0.47, "learning_rate": 6.401308615049072e-05, "loss": 1.7143, "step": 5300 }, { "epoch": 0.48, "learning_rate": 6.292257360959651e-05, "loss": 1.703, "step": 5400 }, { "epoch": 0.49, "learning_rate": 6.18320610687023e-05, "loss": 1.71, "step": 5500 }, { "epoch": 0.5, "learning_rate": 6.074154852780807e-05, "loss": 1.6915, "step": 5600 }, { "epoch": 0.51, "learning_rate": 5.965103598691385e-05, "loss": 1.6904, "step": 5700 }, { "epoch": 0.52, "learning_rate": 5.856052344601963e-05, "loss": 1.6986, "step": 5800 }, { "epoch": 0.53, "learning_rate": 5.747001090512541e-05, "loss": 1.686, "step": 5900 }, { "epoch": 0.54, "learning_rate": 5.6379498364231187e-05, "loss": 1.6657, "step": 6000 }, { "epoch": 0.55, "learning_rate": 5.528898582333697e-05, "loss": 1.658, "step": 6100 }, { "epoch": 0.56, "learning_rate": 5.419847328244275e-05, "loss": 1.6649, "step": 6200 }, { "epoch": 0.56, "learning_rate": 5.310796074154854e-05, "loss": 1.6459, "step": 6300 }, { "epoch": 0.57, "learning_rate": 5.2017448200654305e-05, "loss": 1.6521, "step": 6400 }, { "epoch": 0.58, "learning_rate": 5.0926935659760086e-05, "loss": 1.6366, "step": 6500 }, { "epoch": 0.59, "learning_rate": 4.9836423118865874e-05, "loss": 1.6311, "step": 6600 }, { "epoch": 0.6, "learning_rate": 4.874591057797165e-05, "loss": 1.6279, "step": 6700 }, { "epoch": 0.61, "learning_rate": 4.765539803707743e-05, "loss": 1.6249, "step": 6800 }, { "epoch": 0.62, "learning_rate": 4.656488549618321e-05, "loss": 1.6161, "step": 6900 }, { "epoch": 0.63, "learning_rate": 4.547437295528899e-05, "loss": 1.612, "step": 7000 }, { "epoch": 0.64, "learning_rate": 4.438386041439477e-05, "loss": 1.6116, "step": 7100 }, { "epoch": 0.64, "learning_rate": 4.329334787350055e-05, "loss": 1.589, "step": 7200 }, { "epoch": 0.65, "learning_rate": 4.220283533260633e-05, "loss": 1.6049, "step": 7300 }, { "epoch": 0.66, "learning_rate": 4.1112322791712104e-05, "loss": 1.5986, "step": 7400 }, { "epoch": 0.67, "learning_rate": 4.0021810250817885e-05, "loss": 1.5782, "step": 7500 }, { "epoch": 0.68, "learning_rate": 3.8931297709923666e-05, "loss": 1.5872, "step": 7600 }, { "epoch": 0.69, "learning_rate": 3.784078516902945e-05, "loss": 1.5697, "step": 7700 }, { "epoch": 0.7, "learning_rate": 3.675027262813522e-05, "loss": 1.5666, "step": 7800 }, { "epoch": 0.71, "learning_rate": 3.565976008724101e-05, "loss": 1.5668, "step": 7900 }, { "epoch": 0.72, "learning_rate": 3.4569247546346784e-05, "loss": 1.571, "step": 8000 } ], "logging_steps": 100, "max_steps": 11170, "num_train_epochs": 1, "save_steps": 1000, "total_flos": 1.90707419578368e+17, "trial_name": null, "trial_params": null }