| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.73913043478261, | |
| "global_step": 140, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.285714285714286e-05, | |
| "loss": 3.3723, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 2.261305332183838, | |
| "eval_runtime": 1.7879, | |
| "eval_samples_per_second": 189.047, | |
| "eval_steps_per_second": 24.05, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.642857142857143e-05, | |
| "loss": 2.4888, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "eval_loss": 2.087858200073242, | |
| "eval_runtime": 1.8469, | |
| "eval_samples_per_second": 183.011, | |
| "eval_steps_per_second": 23.283, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.928571428571429e-05, | |
| "loss": 2.3065, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.214285714285714e-05, | |
| "loss": 2.1694, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "eval_loss": 2.0183253288269043, | |
| "eval_runtime": 1.7709, | |
| "eval_samples_per_second": 190.86, | |
| "eval_steps_per_second": 24.281, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 2.0846, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "eval_loss": 1.983121395111084, | |
| "eval_runtime": 1.724, | |
| "eval_samples_per_second": 196.056, | |
| "eval_steps_per_second": 24.942, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 5.785714285714287e-05, | |
| "loss": 2.026, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 5.0714285714285716e-05, | |
| "loss": 1.9812, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "eval_loss": 1.9576869010925293, | |
| "eval_runtime": 2.058, | |
| "eval_samples_per_second": 164.239, | |
| "eval_steps_per_second": 20.894, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 4.3571428571428576e-05, | |
| "loss": 1.9352, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "eval_loss": 1.9370218515396118, | |
| "eval_runtime": 1.7855, | |
| "eval_samples_per_second": 189.305, | |
| "eval_steps_per_second": 24.083, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 3.642857142857143e-05, | |
| "loss": 1.9164, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 2.9285714285714288e-05, | |
| "loss": 1.8822, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "eval_loss": 1.9341849088668823, | |
| "eval_runtime": 1.8263, | |
| "eval_samples_per_second": 185.074, | |
| "eval_steps_per_second": 23.545, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 2.214285714285714e-05, | |
| "loss": 1.853, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 1.9238194227218628, | |
| "eval_runtime": 1.8358, | |
| "eval_samples_per_second": 184.114, | |
| "eval_steps_per_second": 23.423, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.8446, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "eval_loss": 1.9229856729507446, | |
| "eval_runtime": 1.8246, | |
| "eval_samples_per_second": 185.25, | |
| "eval_steps_per_second": 23.567, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 7.857142857142858e-06, | |
| "loss": 1.8343, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 7.142857142857143e-07, | |
| "loss": 1.8211, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "eval_loss": 1.9205037355422974, | |
| "eval_runtime": 1.9363, | |
| "eval_samples_per_second": 174.556, | |
| "eval_steps_per_second": 22.207, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "step": 140, | |
| "total_flos": 2.181454126645248e+16, | |
| "train_loss": 2.1082653863089424, | |
| "train_runtime": 679.206, | |
| "train_samples_per_second": 54.078, | |
| "train_steps_per_second": 0.206 | |
| } | |
| ], | |
| "max_steps": 140, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.181454126645248e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |