{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 21650, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.23, "learning_rate": 4.884526558891455e-05, "loss": 4.2127, "step": 500 }, { "epoch": 0.46, "learning_rate": 4.7690531177829104e-05, "loss": 3.9681, "step": 1000 }, { "epoch": 0.69, "learning_rate": 4.653579676674365e-05, "loss": 3.8763, "step": 1500 }, { "epoch": 0.92, "learning_rate": 4.53810623556582e-05, "loss": 3.8201, "step": 2000 }, { "epoch": 1.15, "learning_rate": 4.422632794457275e-05, "loss": 3.7144, "step": 2500 }, { "epoch": 1.39, "learning_rate": 4.30715935334873e-05, "loss": 3.6592, "step": 3000 }, { "epoch": 1.62, "learning_rate": 4.1916859122401844e-05, "loss": 3.6401, "step": 3500 }, { "epoch": 1.85, "learning_rate": 4.07621247113164e-05, "loss": 3.6187, "step": 4000 }, { "epoch": 2.08, "learning_rate": 3.960739030023095e-05, "loss": 3.5668, "step": 4500 }, { "epoch": 2.31, "learning_rate": 3.84526558891455e-05, "loss": 3.5025, "step": 5000 }, { "epoch": 2.54, "learning_rate": 3.729792147806005e-05, "loss": 3.4936, "step": 5500 }, { "epoch": 2.77, "learning_rate": 3.61431870669746e-05, "loss": 3.484, "step": 6000 }, { "epoch": 3.0, "learning_rate": 3.498845265588915e-05, "loss": 3.4812, "step": 6500 }, { "epoch": 3.23, "learning_rate": 3.38337182448037e-05, "loss": 3.3856, "step": 7000 }, { "epoch": 3.46, "learning_rate": 3.2678983833718243e-05, "loss": 3.3908, "step": 7500 }, { "epoch": 3.7, "learning_rate": 3.1524249422632794e-05, "loss": 3.387, "step": 8000 }, { "epoch": 3.93, "learning_rate": 3.0369515011547345e-05, "loss": 3.3845, "step": 8500 }, { "epoch": 4.16, "learning_rate": 2.9214780600461896e-05, "loss": 3.3198, "step": 9000 }, { "epoch": 4.39, "learning_rate": 2.8060046189376443e-05, "loss": 3.3053, "step": 9500 }, { "epoch": 4.62, "learning_rate": 2.6905311778290994e-05, "loss": 3.3024, "step": 10000 }, { "epoch": 4.85, "learning_rate": 2.575057736720554e-05, "loss": 3.3116, "step": 10500 }, { "epoch": 5.08, "learning_rate": 2.4595842956120095e-05, "loss": 3.2744, "step": 11000 }, { "epoch": 5.31, "learning_rate": 2.3441108545034643e-05, "loss": 3.2322, "step": 11500 }, { "epoch": 5.54, "learning_rate": 2.2286374133949193e-05, "loss": 3.2407, "step": 12000 }, { "epoch": 5.77, "learning_rate": 2.113163972286374e-05, "loss": 3.242, "step": 12500 }, { "epoch": 6.0, "learning_rate": 1.997690531177829e-05, "loss": 3.2379, "step": 13000 }, { "epoch": 6.24, "learning_rate": 1.8822170900692842e-05, "loss": 3.1766, "step": 13500 }, { "epoch": 6.47, "learning_rate": 1.7667436489607393e-05, "loss": 3.1793, "step": 14000 }, { "epoch": 6.7, "learning_rate": 1.651270207852194e-05, "loss": 3.1928, "step": 14500 }, { "epoch": 6.93, "learning_rate": 1.535796766743649e-05, "loss": 3.1859, "step": 15000 }, { "epoch": 7.16, "learning_rate": 1.420323325635104e-05, "loss": 3.156, "step": 15500 }, { "epoch": 7.39, "learning_rate": 1.304849884526559e-05, "loss": 3.1379, "step": 16000 }, { "epoch": 7.62, "learning_rate": 1.189376443418014e-05, "loss": 3.1447, "step": 16500 }, { "epoch": 7.85, "learning_rate": 1.0739030023094689e-05, "loss": 3.1469, "step": 17000 }, { "epoch": 8.08, "learning_rate": 9.584295612009238e-06, "loss": 3.1291, "step": 17500 }, { "epoch": 8.31, "learning_rate": 8.429561200923789e-06, "loss": 3.1071, "step": 18000 }, { "epoch": 8.55, "learning_rate": 7.274826789838338e-06, "loss": 3.1128, "step": 18500 }, { "epoch": 8.78, "learning_rate": 6.120092378752887e-06, "loss": 3.1132, "step": 19000 }, { "epoch": 9.01, "learning_rate": 4.965357967667437e-06, "loss": 3.1141, "step": 19500 }, { "epoch": 9.24, "learning_rate": 3.810623556581986e-06, "loss": 3.0843, "step": 20000 }, { "epoch": 9.47, "learning_rate": 2.655889145496536e-06, "loss": 3.0897, "step": 20500 }, { "epoch": 9.7, "learning_rate": 1.5011547344110855e-06, "loss": 3.0902, "step": 21000 }, { "epoch": 9.93, "learning_rate": 3.4642032332563515e-07, "loss": 3.0911, "step": 21500 }, { "epoch": 10.0, "step": 21650, "total_flos": 2.262266413056e+16, "train_loss": 3.354035396443779, "train_runtime": 7135.8784, "train_samples_per_second": 6.067, "train_steps_per_second": 3.034 } ], "max_steps": 21650, "num_train_epochs": 10, "total_flos": 2.262266413056e+16, "trial_name": null, "trial_params": null }