{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "global_step": 22128, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 4.887020968908171e-05, "loss": 2.9996, "step": 500 }, { "epoch": 0.09, "learning_rate": 4.7740419378163416e-05, "loss": 2.3319, "step": 1000 }, { "epoch": 0.14, "learning_rate": 4.661062906724512e-05, "loss": 2.1701, "step": 1500 }, { "epoch": 0.18, "learning_rate": 4.548083875632683e-05, "loss": 2.0354, "step": 2000 }, { "epoch": 0.23, "learning_rate": 4.4351048445408536e-05, "loss": 1.9541, "step": 2500 }, { "epoch": 0.27, "learning_rate": 4.322125813449024e-05, "loss": 1.8943, "step": 3000 }, { "epoch": 0.32, "learning_rate": 4.209146782357194e-05, "loss": 1.8835, "step": 3500 }, { "epoch": 0.36, "learning_rate": 4.096167751265365e-05, "loss": 1.8138, "step": 4000 }, { "epoch": 0.41, "learning_rate": 3.9831887201735355e-05, "loss": 1.8035, "step": 4500 }, { "epoch": 0.45, "learning_rate": 3.870209689081707e-05, "loss": 1.732, "step": 5000 }, { "epoch": 0.5, "learning_rate": 3.7572306579898775e-05, "loss": 1.7372, "step": 5500 }, { "epoch": 0.54, "learning_rate": 3.644251626898048e-05, "loss": 1.7036, "step": 6000 }, { "epoch": 0.59, "learning_rate": 3.531272595806219e-05, "loss": 1.7085, "step": 6500 }, { "epoch": 0.63, "learning_rate": 3.4182935647143895e-05, "loss": 1.6925, "step": 7000 }, { "epoch": 0.68, "learning_rate": 3.30531453362256e-05, "loss": 1.6173, "step": 7500 }, { "epoch": 0.72, "learning_rate": 3.192335502530731e-05, "loss": 1.6409, "step": 8000 }, { "epoch": 0.77, "learning_rate": 3.079356471438901e-05, "loss": 1.6013, "step": 8500 }, { "epoch": 0.81, "learning_rate": 2.9663774403470718e-05, "loss": 1.6104, "step": 9000 }, { "epoch": 0.86, "learning_rate": 2.8533984092552425e-05, "loss": 1.59, "step": 9500 }, { "epoch": 0.9, "learning_rate": 2.740419378163413e-05, "loss": 1.5617, "step": 10000 }, { "epoch": 0.95, "learning_rate": 2.6274403470715835e-05, "loss": 1.5646, "step": 10500 }, { "epoch": 0.99, "learning_rate": 2.514461315979754e-05, "loss": 1.5452, "step": 11000 }, { "epoch": 1.04, "learning_rate": 2.4014822848879248e-05, "loss": 1.3452, "step": 11500 }, { "epoch": 1.08, "learning_rate": 2.2885032537960955e-05, "loss": 1.3119, "step": 12000 }, { "epoch": 1.13, "learning_rate": 2.1755242227042665e-05, "loss": 1.2932, "step": 12500 }, { "epoch": 1.17, "learning_rate": 2.0625451916124368e-05, "loss": 1.253, "step": 13000 }, { "epoch": 1.22, "learning_rate": 1.9495661605206075e-05, "loss": 1.2879, "step": 13500 }, { "epoch": 1.27, "learning_rate": 1.836587129428778e-05, "loss": 1.2861, "step": 14000 }, { "epoch": 1.31, "learning_rate": 1.7236080983369488e-05, "loss": 1.2526, "step": 14500 }, { "epoch": 1.36, "learning_rate": 1.6106290672451194e-05, "loss": 1.2673, "step": 15000 }, { "epoch": 1.4, "learning_rate": 1.49765003615329e-05, "loss": 1.263, "step": 15500 }, { "epoch": 1.45, "learning_rate": 1.3846710050614606e-05, "loss": 1.2778, "step": 16000 }, { "epoch": 1.49, "learning_rate": 1.2716919739696313e-05, "loss": 1.2665, "step": 16500 }, { "epoch": 1.54, "learning_rate": 1.158712942877802e-05, "loss": 1.2543, "step": 17000 }, { "epoch": 1.58, "learning_rate": 1.0457339117859726e-05, "loss": 1.244, "step": 17500 }, { "epoch": 1.63, "learning_rate": 9.32754880694143e-06, "loss": 1.2601, "step": 18000 }, { "epoch": 1.67, "learning_rate": 8.197758496023139e-06, "loss": 1.2237, "step": 18500 }, { "epoch": 1.72, "learning_rate": 7.067968185104845e-06, "loss": 1.2333, "step": 19000 }, { "epoch": 1.76, "learning_rate": 5.9381778741865515e-06, "loss": 1.2414, "step": 19500 }, { "epoch": 1.81, "learning_rate": 4.808387563268257e-06, "loss": 1.2307, "step": 20000 }, { "epoch": 1.85, "learning_rate": 3.678597252349964e-06, "loss": 1.2187, "step": 20500 }, { "epoch": 1.9, "learning_rate": 2.54880694143167e-06, "loss": 1.2226, "step": 21000 }, { "epoch": 1.94, "learning_rate": 1.4190166305133768e-06, "loss": 1.1874, "step": 21500 }, { "epoch": 1.99, "learning_rate": 2.8922631959508315e-07, "loss": 1.1936, "step": 22000 }, { "epoch": 2.0, "step": 22128, "total_flos": 5.203777169748787e+16, "train_loss": 1.539172684896363, "train_runtime": 4859.3462, "train_samples_per_second": 54.644, "train_steps_per_second": 4.554 } ], "max_steps": 22128, "num_train_epochs": 2, "total_flos": 5.203777169748787e+16, "trial_name": null, "trial_params": null }