| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 25000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 1.558, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 1.4467, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.88e-05, | |
| "loss": 1.4187, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8400000000000003e-05, | |
| "loss": 1.4271, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.4207, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.76e-05, | |
| "loss": 1.4212, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.72e-05, | |
| "loss": 1.4041, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 1.4028, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.64e-05, | |
| "loss": 1.3807, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.3853, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.2580437660217285, | |
| "eval_runtime": 314.0588, | |
| "eval_samples_per_second": 31.841, | |
| "eval_steps_per_second": 3.98, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.5600000000000003e-05, | |
| "loss": 1.3783, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.5200000000000002e-05, | |
| "loss": 1.3674, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.48e-05, | |
| "loss": 1.369, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.4400000000000001e-05, | |
| "loss": 1.3729, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.4e-05, | |
| "loss": 1.3774, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 1.3641, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.3200000000000002e-05, | |
| "loss": 1.3798, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 1.3602, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2400000000000002e-05, | |
| "loss": 1.3641, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.3537, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.234650731086731, | |
| "eval_runtime": 312.8974, | |
| "eval_samples_per_second": 31.959, | |
| "eval_steps_per_second": 3.995, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.16e-05, | |
| "loss": 1.3352, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 1.3482, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.0800000000000002e-05, | |
| "loss": 1.3437, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.04e-05, | |
| "loss": 1.3391, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1e-05, | |
| "loss": 1.3408, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.335, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 1.3334, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 8.8e-06, | |
| "loss": 1.3466, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.3279, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.3286, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.2182681560516357, | |
| "eval_runtime": 313.6417, | |
| "eval_samples_per_second": 31.884, | |
| "eval_steps_per_second": 3.985, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 1.3196, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 1.3267, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 1.3224, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.3191, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6e-06, | |
| "loss": 1.3106, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 1.3268, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.2e-06, | |
| "loss": 1.3112, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 1.3199, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.3137, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.3216, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.1937603950500488, | |
| "eval_runtime": 291.0578, | |
| "eval_samples_per_second": 34.357, | |
| "eval_steps_per_second": 4.295, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 1.3143, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 1.327, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 1.3079, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.3193, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.3018, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 1.3092, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 1.3076, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 1.3124, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.3079, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.3194, | |
| "step": 25000 | |
| } | |
| ], | |
| "max_steps": 25000, | |
| "num_train_epochs": 5, | |
| "total_flos": 2.65243613184e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |