{ "best_metric": null, "best_model_checkpoint": null, "epoch": 6.0, "global_step": 35574, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.08, "learning_rate": 4.9297239556979816e-05, "loss": 3.4444, "step": 500 }, { "epoch": 0.17, "learning_rate": 4.8594479113959636e-05, "loss": 3.339, "step": 1000 }, { "epoch": 0.25, "learning_rate": 4.789171867093945e-05, "loss": 3.3261, "step": 1500 }, { "epoch": 0.34, "learning_rate": 4.718895822791927e-05, "loss": 3.2812, "step": 2000 }, { "epoch": 0.42, "learning_rate": 4.648619778489909e-05, "loss": 3.25, "step": 2500 }, { "epoch": 0.51, "learning_rate": 4.5783437341878904e-05, "loss": 3.2331, "step": 3000 }, { "epoch": 0.59, "learning_rate": 4.508067689885872e-05, "loss": 3.2195, "step": 3500 }, { "epoch": 0.67, "learning_rate": 4.437791645583854e-05, "loss": 3.2063, "step": 4000 }, { "epoch": 0.76, "learning_rate": 4.367515601281835e-05, "loss": 3.2028, "step": 4500 }, { "epoch": 0.84, "learning_rate": 4.297239556979817e-05, "loss": 3.1897, "step": 5000 }, { "epoch": 0.93, "learning_rate": 4.2269635126777984e-05, "loss": 3.1792, "step": 5500 }, { "epoch": 1.01, "learning_rate": 4.15668746837578e-05, "loss": 3.1564, "step": 6000 }, { "epoch": 1.1, "learning_rate": 4.0864114240737625e-05, "loss": 3.0599, "step": 6500 }, { "epoch": 1.18, "learning_rate": 4.016135379771744e-05, "loss": 3.0641, "step": 7000 }, { "epoch": 1.26, "learning_rate": 3.945859335469725e-05, "loss": 3.0661, "step": 7500 }, { "epoch": 1.35, "learning_rate": 3.875583291167707e-05, "loss": 3.0731, "step": 8000 }, { "epoch": 1.43, "learning_rate": 3.8053072468656886e-05, "loss": 3.0555, "step": 8500 }, { "epoch": 1.52, "learning_rate": 3.73503120256367e-05, "loss": 3.0592, "step": 9000 }, { "epoch": 1.6, "learning_rate": 3.664755158261652e-05, "loss": 3.0618, "step": 9500 }, { "epoch": 1.69, "learning_rate": 3.594479113959633e-05, "loss": 3.0592, "step": 10000 }, { "epoch": 1.77, "learning_rate": 3.524203069657615e-05, "loss": 3.0496, "step": 10500 }, { "epoch": 1.86, "learning_rate": 3.453927025355597e-05, "loss": 3.0396, "step": 11000 }, { "epoch": 1.94, "learning_rate": 3.383650981053579e-05, "loss": 3.0479, "step": 11500 }, { "epoch": 2.02, "learning_rate": 3.313374936751561e-05, "loss": 3.0178, "step": 12000 }, { "epoch": 2.11, "learning_rate": 3.243098892449542e-05, "loss": 2.9502, "step": 12500 }, { "epoch": 2.19, "learning_rate": 3.1728228481475234e-05, "loss": 2.9585, "step": 13000 }, { "epoch": 2.28, "learning_rate": 3.1025468038455054e-05, "loss": 2.968, "step": 13500 }, { "epoch": 2.36, "learning_rate": 3.0322707595434867e-05, "loss": 2.9591, "step": 14000 }, { "epoch": 2.45, "learning_rate": 2.9619947152414684e-05, "loss": 2.9585, "step": 14500 }, { "epoch": 2.53, "learning_rate": 2.89171867093945e-05, "loss": 2.9628, "step": 15000 }, { "epoch": 2.61, "learning_rate": 2.821442626637432e-05, "loss": 2.9712, "step": 15500 }, { "epoch": 2.7, "learning_rate": 2.7511665823354138e-05, "loss": 2.9583, "step": 16000 }, { "epoch": 2.78, "learning_rate": 2.6808905380333955e-05, "loss": 2.9491, "step": 16500 }, { "epoch": 2.87, "learning_rate": 2.610614493731377e-05, "loss": 2.9594, "step": 17000 }, { "epoch": 2.95, "learning_rate": 2.5403384494293585e-05, "loss": 2.9673, "step": 17500 }, { "epoch": 3.04, "learning_rate": 2.4700624051273402e-05, "loss": 2.9344, "step": 18000 }, { "epoch": 3.12, "learning_rate": 2.3997863608253223e-05, "loss": 2.8771, "step": 18500 }, { "epoch": 3.2, "learning_rate": 2.3295103165233036e-05, "loss": 2.8863, "step": 19000 }, { "epoch": 3.29, "learning_rate": 2.2592342722212853e-05, "loss": 2.8845, "step": 19500 }, { "epoch": 3.37, "learning_rate": 2.188958227919267e-05, "loss": 2.9015, "step": 20000 }, { "epoch": 3.46, "learning_rate": 2.1186821836172487e-05, "loss": 2.8927, "step": 20500 }, { "epoch": 3.54, "learning_rate": 2.0484061393152303e-05, "loss": 2.889, "step": 21000 }, { "epoch": 3.63, "learning_rate": 1.978130095013212e-05, "loss": 2.9037, "step": 21500 }, { "epoch": 3.71, "learning_rate": 1.9078540507111937e-05, "loss": 2.8995, "step": 22000 }, { "epoch": 3.79, "learning_rate": 1.837578006409175e-05, "loss": 2.8922, "step": 22500 }, { "epoch": 3.88, "learning_rate": 1.767301962107157e-05, "loss": 2.8954, "step": 23000 }, { "epoch": 3.96, "learning_rate": 1.6970259178051388e-05, "loss": 2.9017, "step": 23500 }, { "epoch": 4.05, "learning_rate": 1.6267498735031205e-05, "loss": 2.8701, "step": 24000 }, { "epoch": 4.13, "learning_rate": 1.5564738292011018e-05, "loss": 2.8549, "step": 24500 }, { "epoch": 4.22, "learning_rate": 1.4861977848990838e-05, "loss": 2.8502, "step": 25000 }, { "epoch": 4.3, "learning_rate": 1.4159217405970653e-05, "loss": 2.846, "step": 25500 }, { "epoch": 4.39, "learning_rate": 1.345645696295047e-05, "loss": 2.8497, "step": 26000 }, { "epoch": 4.47, "learning_rate": 1.2753696519930285e-05, "loss": 2.8284, "step": 26500 }, { "epoch": 4.55, "learning_rate": 1.2050936076910104e-05, "loss": 2.8432, "step": 27000 }, { "epoch": 4.64, "learning_rate": 1.1348175633889919e-05, "loss": 2.8468, "step": 27500 }, { "epoch": 4.72, "learning_rate": 1.0645415190869738e-05, "loss": 2.8425, "step": 28000 }, { "epoch": 4.81, "learning_rate": 9.942654747849553e-06, "loss": 2.8403, "step": 28500 }, { "epoch": 4.89, "learning_rate": 9.23989430482937e-06, "loss": 2.8276, "step": 29000 }, { "epoch": 4.98, "learning_rate": 8.537133861809187e-06, "loss": 2.8549, "step": 29500 }, { "epoch": 5.06, "learning_rate": 7.834373418789003e-06, "loss": 2.8061, "step": 30000 }, { "epoch": 5.14, "learning_rate": 7.13161297576882e-06, "loss": 2.8111, "step": 30500 }, { "epoch": 5.23, "learning_rate": 6.428852532748637e-06, "loss": 2.8153, "step": 31000 }, { "epoch": 5.31, "learning_rate": 5.726092089728453e-06, "loss": 2.8129, "step": 31500 }, { "epoch": 5.4, "learning_rate": 5.02333164670827e-06, "loss": 2.812, "step": 32000 }, { "epoch": 5.48, "learning_rate": 4.320571203688087e-06, "loss": 2.8134, "step": 32500 }, { "epoch": 5.57, "learning_rate": 3.6178107606679037e-06, "loss": 2.8172, "step": 33000 }, { "epoch": 5.65, "learning_rate": 2.9150503176477205e-06, "loss": 2.809, "step": 33500 }, { "epoch": 5.73, "learning_rate": 2.212289874627537e-06, "loss": 2.8115, "step": 34000 }, { "epoch": 5.82, "learning_rate": 1.5095294316073538e-06, "loss": 2.81, "step": 34500 }, { "epoch": 5.9, "learning_rate": 8.067689885871704e-07, "loss": 2.8087, "step": 35000 }, { "epoch": 5.99, "learning_rate": 1.0400854556698713e-07, "loss": 2.799, "step": 35500 }, { "epoch": 6.0, "step": 35574, "total_flos": 1.8590405492736e+16, "train_loss": 2.969814732685026, "train_runtime": 9127.501, "train_samples_per_second": 3.897, "train_steps_per_second": 3.897 } ], "max_steps": 35574, "num_train_epochs": 6, "total_flos": 1.8590405492736e+16, "trial_name": null, "trial_params": null }