| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.887269193391642, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9912109375000002e-05, | |
| "loss": 1.9656, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.98193359375e-05, | |
| "loss": 1.8445, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.97216796875e-05, | |
| "loss": 1.59, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9624023437500002e-05, | |
| "loss": 1.6275, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.9526367187500002e-05, | |
| "loss": 1.473, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.9428710937500003e-05, | |
| "loss": 1.3701, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.93310546875e-05, | |
| "loss": 1.3224, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.92333984375e-05, | |
| "loss": 1.1423, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.91357421875e-05, | |
| "loss": 1.1652, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.9038085937500002e-05, | |
| "loss": 1.1422, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.8940429687500002e-05, | |
| "loss": 1.002, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8842773437500003e-05, | |
| "loss": 1.0779, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.87451171875e-05, | |
| "loss": 0.9887, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.86474609375e-05, | |
| "loss": 0.9543, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.85498046875e-05, | |
| "loss": 0.9371, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.8452148437500002e-05, | |
| "loss": 0.8701, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.8354492187500003e-05, | |
| "loss": 0.875, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.82568359375e-05, | |
| "loss": 0.7843, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.81591796875e-05, | |
| "loss": 0.7945, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.80615234375e-05, | |
| "loss": 0.799, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.79638671875e-05, | |
| "loss": 0.7623, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.7866210937500002e-05, | |
| "loss": 0.7263, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 1.7768554687500003e-05, | |
| "loss": 0.7779, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 1.76708984375e-05, | |
| "loss": 0.695, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.75732421875e-05, | |
| "loss": 0.7344, | |
| "step": 500 | |
| } | |
| ], | |
| "max_steps": 4096, | |
| "num_train_epochs": 32, | |
| "total_flos": 6.497819467776e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |