| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.60555255092843, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9774652965566973e-05, | |
| "loss": 2.8362, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9549756625202815e-05, | |
| "loss": 2.8099, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.9324409590769786e-05, | |
| "loss": 2.8307, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.909951325040563e-05, | |
| "loss": 2.7755, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.88741662159726e-05, | |
| "loss": 2.7973, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.864881918153957e-05, | |
| "loss": 2.7982, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.842347214710655e-05, | |
| "loss": 2.7681, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.819812511267352e-05, | |
| "loss": 2.7785, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.79727780782405e-05, | |
| "loss": 2.7593, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.774743104380747e-05, | |
| "loss": 2.7396, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.752208400937444e-05, | |
| "loss": 2.7515, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.7296736974941416e-05, | |
| "loss": 2.5096, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.707138994050839e-05, | |
| "loss": 2.4782, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.684604290607536e-05, | |
| "loss": 2.5025, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.66211465657112e-05, | |
| "loss": 2.5009, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.639579953127817e-05, | |
| "loss": 2.4875, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.617045249684514e-05, | |
| "loss": 2.5066, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.594510546241212e-05, | |
| "loss": 2.5209, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.571975842797909e-05, | |
| "loss": 2.5225, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.549441139354607e-05, | |
| "loss": 2.5228, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.526906435911303e-05, | |
| "loss": 2.5173, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.5044168018748875e-05, | |
| "loss": 2.5319, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.481882098431585e-05, | |
| "loss": 2.2928, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.459347394988282e-05, | |
| "loss": 2.2681, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.4368126915449795e-05, | |
| "loss": 2.2761, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.4142779881016766e-05, | |
| "loss": 2.2751, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 4.391788354065261e-05, | |
| "loss": 2.2932, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.369253650621958e-05, | |
| "loss": 2.2924, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.346718947178655e-05, | |
| "loss": 2.2908, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.324184243735353e-05, | |
| "loss": 2.3225, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.30164954029205e-05, | |
| "loss": 2.3189, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.279114836848747e-05, | |
| "loss": 2.3191, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.256580133405445e-05, | |
| "loss": 2.3243, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.234090499369028e-05, | |
| "loss": 2.1313, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.211555795925726e-05, | |
| "loss": 2.0608, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.189021092482423e-05, | |
| "loss": 2.0668, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 4.16648638903912e-05, | |
| "loss": 2.0977, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 4.143951685595818e-05, | |
| "loss": 2.0911, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 4.121416982152515e-05, | |
| "loss": 2.1224, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.098882278709212e-05, | |
| "loss": 2.1214, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 110940, | |
| "num_train_epochs": 20, | |
| "total_flos": 3.714802778112e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |