| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 11.580775911986104, | |
| "global_step": 40000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 6e-06, | |
| "loss": 4.3587, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.2e-05, | |
| "loss": 3.473, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4807435555099107e-05, | |
| "loss": 3.257, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.442230666529732e-05, | |
| "loss": 3.0935, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4037177775495533e-05, | |
| "loss": 3.0049, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.3652048885693745e-05, | |
| "loss": 2.932, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.326691999589196e-05, | |
| "loss": 2.8843, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.2881791106090172e-05, | |
| "loss": 2.7991, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.2496662216288385e-05, | |
| "loss": 2.7699, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.2111533326486598e-05, | |
| "loss": 2.7408, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.172640443668481e-05, | |
| "loss": 2.6829, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.1341275546883024e-05, | |
| "loss": 2.6527, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.0956146657081237e-05, | |
| "loss": 2.6339, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.057101776727945e-05, | |
| "loss": 2.6152, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 1.0185888877477662e-05, | |
| "loss": 2.5695, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 9.800759987675876e-06, | |
| "loss": 2.5439, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 9.41563109787409e-06, | |
| "loss": 2.5504, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 9.030502208072302e-06, | |
| "loss": 2.5106, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 8.645373318270514e-06, | |
| "loss": 2.4734, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 8.260244428468728e-06, | |
| "loss": 2.4931, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 7.875115538666941e-06, | |
| "loss": 2.4626, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 7.489986648865154e-06, | |
| "loss": 2.4341, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 7.104857759063367e-06, | |
| "loss": 2.4316, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 6.71972886926158e-06, | |
| "loss": 2.4158, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 6.334599979459793e-06, | |
| "loss": 2.4055, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 5.949471089658006e-06, | |
| "loss": 2.3773, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 5.564342199856218e-06, | |
| "loss": 2.3877, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 5.179213310054432e-06, | |
| "loss": 2.3826, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 4.794084420252645e-06, | |
| "loss": 2.3419, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 4.408955530450858e-06, | |
| "loss": 2.3447, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 4.023826640649071e-06, | |
| "loss": 2.3602, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 3.6386977508472838e-06, | |
| "loss": 2.3315, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 3.2535688610454968e-06, | |
| "loss": 2.3116, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 2.8684399712437098e-06, | |
| "loss": 2.3285, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 10.13, | |
| "learning_rate": 2.4833110814419224e-06, | |
| "loss": 2.3151, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 2.0981821916401354e-06, | |
| "loss": 2.2991, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 1.7130533018383486e-06, | |
| "loss": 2.2984, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 1.3279244120365616e-06, | |
| "loss": 2.31, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 11.29, | |
| "learning_rate": 9.427955222347745e-07, | |
| "loss": 2.2897, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 5.576666324329876e-07, | |
| "loss": 2.2826, | |
| "step": 40000 | |
| } | |
| ], | |
| "max_steps": 41448, | |
| "num_train_epochs": 12, | |
| "total_flos": 4.877556310867968e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |