| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 35586, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9297476535716295e-05, | |
| "loss": 2.948, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.859495307143259e-05, | |
| "loss": 2.7396, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.789242960714888e-05, | |
| "loss": 2.6771, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.7189906142865175e-05, | |
| "loss": 2.619, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.648738267858147e-05, | |
| "loss": 2.5869, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.578485921429776e-05, | |
| "loss": 2.5459, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.5082335750014054e-05, | |
| "loss": 2.5282, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.437981228573034e-05, | |
| "loss": 2.5131, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.367728882144664e-05, | |
| "loss": 2.5023, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.297476535716293e-05, | |
| "loss": 2.4828, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.227224189287922e-05, | |
| "loss": 2.4753, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.156971842859552e-05, | |
| "loss": 2.4594, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.086719496431181e-05, | |
| "loss": 2.4583, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.01646715000281e-05, | |
| "loss": 2.4519, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.94621480357444e-05, | |
| "loss": 2.4164, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.875962457146069e-05, | |
| "loss": 2.4297, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.805710110717698e-05, | |
| "loss": 2.4081, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.735457764289327e-05, | |
| "loss": 2.4227, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.665205417860957e-05, | |
| "loss": 2.4102, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.594953071432586e-05, | |
| "loss": 2.4211, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.524700725004215e-05, | |
| "loss": 2.3949, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.454448378575845e-05, | |
| "loss": 2.3758, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.384196032147474e-05, | |
| "loss": 2.3789, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.313943685719103e-05, | |
| "loss": 2.3722, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.243691339290733e-05, | |
| "loss": 2.366, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.173438992862362e-05, | |
| "loss": 2.3766, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.103186646433991e-05, | |
| "loss": 2.3588, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.0329343000056204e-05, | |
| "loss": 2.3629, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.9626819535772498e-05, | |
| "loss": 2.3556, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.8924296071488787e-05, | |
| "loss": 2.3642, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.8221772607205084e-05, | |
| "loss": 2.3566, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7519249142921377e-05, | |
| "loss": 2.3593, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.6816725678637666e-05, | |
| "loss": 2.3513, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.6114202214353963e-05, | |
| "loss": 2.3473, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.5411678750070256e-05, | |
| "loss": 2.3507, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.470915528578655e-05, | |
| "loss": 2.3396, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.400663182150284e-05, | |
| "loss": 2.3263, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 2.330410835721913e-05, | |
| "loss": 2.3359, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.2601584892935425e-05, | |
| "loss": 2.3396, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.1899061428651718e-05, | |
| "loss": 2.3202, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.119653796436801e-05, | |
| "loss": 2.3456, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.0494014500084304e-05, | |
| "loss": 2.316, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.9791491035800597e-05, | |
| "loss": 2.3161, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.908896757151689e-05, | |
| "loss": 2.3293, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.8386444107233183e-05, | |
| "loss": 2.3165, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.7683920642949476e-05, | |
| "loss": 2.3146, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.6981397178665766e-05, | |
| "loss": 2.298, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.6278873714382062e-05, | |
| "loss": 2.3104, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5576350250098355e-05, | |
| "loss": 2.3063, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.4873826785814647e-05, | |
| "loss": 2.3095, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4171303321530941e-05, | |
| "loss": 2.3047, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.3468779857247233e-05, | |
| "loss": 2.2978, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.2766256392963524e-05, | |
| "loss": 2.3043, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.2063732928679819e-05, | |
| "loss": 2.2825, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.136120946439611e-05, | |
| "loss": 2.3048, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.0658686000112405e-05, | |
| "loss": 2.3073, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.956162535828698e-06, | |
| "loss": 2.3025, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.25363907154499e-06, | |
| "loss": 2.2957, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.551115607261282e-06, | |
| "loss": 2.3087, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.848592142977575e-06, | |
| "loss": 2.3039, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7.146068678693869e-06, | |
| "loss": 2.2855, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.4435452144101614e-06, | |
| "loss": 2.2819, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 5.7410217501264545e-06, | |
| "loss": 2.3046, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.0384982858427475e-06, | |
| "loss": 2.2897, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.335974821559041e-06, | |
| "loss": 2.2965, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.6334513572753332e-06, | |
| "loss": 2.3033, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.9309278929916263e-06, | |
| "loss": 2.2887, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.228404428707919e-06, | |
| "loss": 2.2937, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.525880964424212e-06, | |
| "loss": 2.2957, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 8.233575001405047e-07, | |
| "loss": 2.2956, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.2083403585679763e-07, | |
| "loss": 2.2898, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 35586, | |
| "total_flos": 9.298088091765965e+16, | |
| "train_loss": 2.3804299571036704, | |
| "train_runtime": 13242.0355, | |
| "train_samples_per_second": 26.872, | |
| "train_steps_per_second": 2.687 | |
| } | |
| ], | |
| "max_steps": 35586, | |
| "num_train_epochs": 2, | |
| "total_flos": 9.298088091765965e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |