{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 22413, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.07, "learning_rate": 4.8884575915763174e-05, "loss": 2.9446, "step": 500 }, { "epoch": 0.13, "learning_rate": 4.776915183152635e-05, "loss": 2.844, "step": 1000 }, { "epoch": 0.2, "learning_rate": 4.665372774728952e-05, "loss": 2.8468, "step": 1500 }, { "epoch": 0.27, "learning_rate": 4.5538303663052694e-05, "loss": 2.7941, "step": 2000 }, { "epoch": 0.33, "learning_rate": 4.4422879578815865e-05, "loss": 2.7849, "step": 2500 }, { "epoch": 0.4, "learning_rate": 4.330745549457904e-05, "loss": 2.7557, "step": 3000 }, { "epoch": 0.47, "learning_rate": 4.2192031410342214e-05, "loss": 2.7485, "step": 3500 }, { "epoch": 0.54, "learning_rate": 4.1076607326105386e-05, "loss": 2.7153, "step": 4000 }, { "epoch": 0.6, "learning_rate": 3.9961183241868564e-05, "loss": 2.7089, "step": 4500 }, { "epoch": 0.67, "learning_rate": 3.8845759157631735e-05, "loss": 2.6941, "step": 5000 }, { "epoch": 0.74, "learning_rate": 3.7730335073394906e-05, "loss": 2.6581, "step": 5500 }, { "epoch": 0.8, "learning_rate": 3.661491098915808e-05, "loss": 2.6449, "step": 6000 }, { "epoch": 0.87, "learning_rate": 3.5499486904921255e-05, "loss": 2.6534, "step": 6500 }, { "epoch": 0.94, "learning_rate": 3.4384062820684427e-05, "loss": 2.6297, "step": 7000 }, { "epoch": 1.0, "learning_rate": 3.32686387364476e-05, "loss": 2.6267, "step": 7500 }, { "epoch": 1.07, "learning_rate": 3.2153214652210776e-05, "loss": 2.4631, "step": 8000 }, { "epoch": 1.14, "learning_rate": 3.103779056797395e-05, "loss": 2.4614, "step": 8500 }, { "epoch": 1.2, "learning_rate": 2.9922366483737118e-05, "loss": 2.4762, "step": 9000 }, { "epoch": 1.27, "learning_rate": 2.8806942399500293e-05, "loss": 2.4324, "step": 9500 }, { "epoch": 1.34, "learning_rate": 2.7691518315263464e-05, "loss": 2.4394, "step": 10000 }, { "epoch": 1.41, "learning_rate": 2.657609423102664e-05, "loss": 2.4331, "step": 10500 }, { "epoch": 1.47, "learning_rate": 2.546067014678981e-05, "loss": 2.4087, "step": 11000 }, { "epoch": 1.54, "learning_rate": 2.4345246062552984e-05, "loss": 2.4048, "step": 11500 }, { "epoch": 1.61, "learning_rate": 2.322982197831616e-05, "loss": 2.4241, "step": 12000 }, { "epoch": 1.67, "learning_rate": 2.211439789407933e-05, "loss": 2.4255, "step": 12500 }, { "epoch": 1.74, "learning_rate": 2.0998973809842505e-05, "loss": 2.3951, "step": 13000 }, { "epoch": 1.81, "learning_rate": 1.9883549725605676e-05, "loss": 2.4398, "step": 13500 }, { "epoch": 1.87, "learning_rate": 1.876812564136885e-05, "loss": 2.395, "step": 14000 }, { "epoch": 1.94, "learning_rate": 1.7652701557132022e-05, "loss": 2.3734, "step": 14500 }, { "epoch": 2.01, "learning_rate": 1.6537277472895197e-05, "loss": 2.37, "step": 15000 }, { "epoch": 2.07, "learning_rate": 1.5421853388658368e-05, "loss": 2.2448, "step": 15500 }, { "epoch": 2.14, "learning_rate": 1.4306429304421542e-05, "loss": 2.2514, "step": 16000 }, { "epoch": 2.21, "learning_rate": 1.3191005220184715e-05, "loss": 2.2704, "step": 16500 }, { "epoch": 2.28, "learning_rate": 1.2075581135947887e-05, "loss": 2.2553, "step": 17000 }, { "epoch": 2.34, "learning_rate": 1.0960157051711061e-05, "loss": 2.2649, "step": 17500 }, { "epoch": 2.41, "learning_rate": 9.844732967474234e-06, "loss": 2.2569, "step": 18000 }, { "epoch": 2.48, "learning_rate": 8.729308883237407e-06, "loss": 2.217, "step": 18500 }, { "epoch": 2.54, "learning_rate": 7.61388479900058e-06, "loss": 2.2623, "step": 19000 }, { "epoch": 2.61, "learning_rate": 6.498460714763753e-06, "loss": 2.2607, "step": 19500 }, { "epoch": 2.68, "learning_rate": 5.3830366305269266e-06, "loss": 2.244, "step": 20000 }, { "epoch": 2.74, "learning_rate": 4.2676125462901e-06, "loss": 2.2591, "step": 20500 }, { "epoch": 2.81, "learning_rate": 3.1521884620532732e-06, "loss": 2.2415, "step": 21000 }, { "epoch": 2.88, "learning_rate": 2.0367643778164457e-06, "loss": 2.2263, "step": 21500 }, { "epoch": 2.94, "learning_rate": 9.21340293579619e-07, "loss": 2.2415, "step": 22000 }, { "epoch": 3.0, "step": 22413, "total_flos": 1.1708869345542144e+16, "train_loss": 2.470616279322582, "train_runtime": 9527.188, "train_samples_per_second": 2.353, "train_steps_per_second": 2.353 } ], "max_steps": 22413, "num_train_epochs": 3, "total_flos": 1.1708869345542144e+16, "trial_name": null, "trial_params": null }