| { | |
| "best_metric": 1.2345943450927734, | |
| "best_model_checkpoint": "outputs/checkpoint-200", | |
| "epoch": 0.064, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8184, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2e-05, | |
| "loss": 1.7858, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3e-05, | |
| "loss": 1.7218, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 1.7218, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 1.7009, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-05, | |
| "loss": 1.6714, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7e-05, | |
| "loss": 1.6582, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-05, | |
| "loss": 1.5659, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9e-05, | |
| "loss": 1.5343, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5252, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00011000000000000002, | |
| "loss": 1.414, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012, | |
| "loss": 1.3919, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013000000000000002, | |
| "loss": 1.352, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014, | |
| "loss": 1.3762, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 1.3238, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016, | |
| "loss": 1.3306, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00017, | |
| "loss": 1.3423, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00018, | |
| "loss": 1.3199, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019, | |
| "loss": 1.3506, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3315, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001998921832884097, | |
| "loss": 1.3142, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019978436657681943, | |
| "loss": 1.2994, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019967654986522912, | |
| "loss": 1.2972, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019956873315363883, | |
| "loss": 1.3135, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019946091644204851, | |
| "loss": 1.2903, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019935309973045823, | |
| "loss": 1.2783, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019924528301886794, | |
| "loss": 1.3044, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019913746630727762, | |
| "loss": 1.2947, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019902964959568736, | |
| "loss": 1.2759, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019892183288409705, | |
| "loss": 1.2746, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019881401617250676, | |
| "loss": 1.2869, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019870619946091644, | |
| "loss": 1.26, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019859838274932616, | |
| "loss": 1.258, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019849056603773587, | |
| "loss": 1.2649, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019838274932614555, | |
| "loss": 1.2549, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019827493261455526, | |
| "loss": 1.2859, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019816711590296498, | |
| "loss": 1.2934, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019805929919137466, | |
| "loss": 1.3011, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019795148247978437, | |
| "loss": 1.2304, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019784366576819408, | |
| "loss": 1.2849, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 1.2345943450927734, | |
| "eval_runtime": 710.2534, | |
| "eval_samples_per_second": 7.04, | |
| "eval_steps_per_second": 0.88, | |
| "step": 200 | |
| } | |
| ], | |
| "max_steps": 9375, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.19119709995008e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |