{ "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": 1.9528741092636582e-05, "loss": 0.3246, "step": 500 }, { "epoch": 0.24, "learning_rate": 1.9053681710213777e-05, "loss": 0.2672, "step": 1000 }, { "epoch": 0.36, "learning_rate": 1.8578622327790975e-05, "loss": 0.2608, "step": 1500 }, { "epoch": 0.48, "learning_rate": 1.810356294536817e-05, "loss": 0.227, "step": 2000 }, { "epoch": 0.59, "learning_rate": 1.7628503562945368e-05, "loss": 0.209, "step": 2500 }, { "epoch": 0.71, "learning_rate": 1.7154394299287413e-05, "loss": 0.2195, "step": 3000 }, { "epoch": 0.83, "learning_rate": 1.6679334916864608e-05, "loss": 0.2098, "step": 3500 }, { "epoch": 0.95, "learning_rate": 1.6204275534441806e-05, "loss": 0.2085, "step": 4000 }, { "epoch": 1.07, "learning_rate": 1.5730166270783847e-05, "loss": 0.1726, "step": 4500 }, { "epoch": 1.19, "learning_rate": 1.5255106888361047e-05, "loss": 0.1403, "step": 5000 }, { "epoch": 1.31, "learning_rate": 1.4780047505938244e-05, "loss": 0.1555, "step": 5500 }, { "epoch": 1.43, "learning_rate": 1.430498812351544e-05, "loss": 0.1444, "step": 6000 }, { "epoch": 1.54, "learning_rate": 1.3829928741092637e-05, "loss": 0.1581, "step": 6500 }, { "epoch": 1.66, "learning_rate": 1.335581947743468e-05, "loss": 0.1368, "step": 7000 }, { "epoch": 1.78, "learning_rate": 1.2880760095011876e-05, "loss": 0.1543, "step": 7500 }, { "epoch": 1.9, "learning_rate": 1.2406650831353921e-05, "loss": 0.1345, "step": 8000 }, { "epoch": 2.02, "learning_rate": 1.1931591448931118e-05, "loss": 0.1271, "step": 8500 }, { "epoch": 2.14, "learning_rate": 1.1456532066508314e-05, "loss": 0.091, "step": 9000 }, { "epoch": 2.26, "learning_rate": 1.0981472684085512e-05, "loss": 0.0895, "step": 9500 }, { "epoch": 2.38, "learning_rate": 1.0506413301662707e-05, "loss": 0.095, "step": 10000 }, { "epoch": 2.49, "learning_rate": 1.0031353919239906e-05, "loss": 0.1101, "step": 10500 }, { "epoch": 2.61, "learning_rate": 9.556294536817104e-06, "loss": 0.0967, "step": 11000 }, { "epoch": 2.73, "learning_rate": 9.082185273159145e-06, "loss": 0.1037, "step": 11500 }, { "epoch": 2.85, "learning_rate": 8.607125890736343e-06, "loss": 0.1038, "step": 12000 }, { "epoch": 2.97, "learning_rate": 8.13206650831354e-06, "loss": 0.1006, "step": 12500 }, { "epoch": 3.09, "learning_rate": 7.657007125890738e-06, "loss": 0.0771, "step": 13000 }, { "epoch": 3.21, "learning_rate": 7.181947743467934e-06, "loss": 0.0608, "step": 13500 }, { "epoch": 3.33, "learning_rate": 6.7068883610451305e-06, "loss": 0.0604, "step": 14000 }, { "epoch": 3.44, "learning_rate": 6.231828978622329e-06, "loss": 0.0593, "step": 14500 }, { "epoch": 3.56, "learning_rate": 5.756769596199525e-06, "loss": 0.0635, "step": 15000 }, { "epoch": 3.68, "learning_rate": 5.281710213776723e-06, "loss": 0.0715, "step": 15500 }, { "epoch": 3.8, "learning_rate": 4.807600950118766e-06, "loss": 0.0572, "step": 16000 }, { "epoch": 3.92, "learning_rate": 4.332541567695962e-06, "loss": 0.0603, "step": 16500 }, { "epoch": 4.04, "learning_rate": 3.857482185273159e-06, "loss": 0.0446, "step": 17000 }, { "epoch": 4.16, "learning_rate": 3.382422802850356e-06, "loss": 0.0347, "step": 17500 }, { "epoch": 4.28, "learning_rate": 2.9073634204275536e-06, "loss": 0.0305, "step": 18000 }, { "epoch": 4.39, "learning_rate": 2.4332541567695966e-06, "loss": 0.0331, "step": 18500 }, { "epoch": 4.51, "learning_rate": 1.9581947743467935e-06, "loss": 0.0299, "step": 19000 }, { "epoch": 4.63, "learning_rate": 1.4840855106888363e-06, "loss": 0.0343, "step": 19500 }, { "epoch": 4.75, "learning_rate": 1.0090261282660335e-06, "loss": 0.0347, "step": 20000 }, { "epoch": 4.87, "learning_rate": 5.339667458432305e-07, "loss": 0.0368, "step": 20500 }, { "epoch": 4.99, "learning_rate": 5.8907363420427555e-08, "loss": 0.031, "step": 21000 }, { "epoch": 5.0, "step": 21050, "total_flos": 2.199759356952576e+16, "train_loss": 0.11555842199121688, "train_runtime": 974.5285, "train_samples_per_second": 345.547, "train_steps_per_second": 21.6 } ], "max_steps": 21050, "num_train_epochs": 5, "total_flos": 2.199759356952576e+16, "trial_name": null, "trial_params": null }