| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.028762080073630927, | |
| "eval_steps": 500, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019920000000000002, | |
| "loss": 1.0145, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001984, | |
| "loss": 0.9763, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001976, | |
| "loss": 0.9834, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001968, | |
| "loss": 1.0162, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000196, | |
| "loss": 0.8783, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001952, | |
| "loss": 1.0353, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001944, | |
| "loss": 1.0571, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019360000000000002, | |
| "loss": 0.9313, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001928, | |
| "loss": 0.9828, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000192, | |
| "loss": 1.0549, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001912, | |
| "loss": 1.0297, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001904, | |
| "loss": 0.8975, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001896, | |
| "loss": 0.9403, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001888, | |
| "loss": 1.0143, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000188, | |
| "loss": 0.9905, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018720000000000002, | |
| "loss": 0.931, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018640000000000003, | |
| "loss": 1.0687, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001856, | |
| "loss": 1.1008, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018480000000000002, | |
| "loss": 1.0267, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 0.9282, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001832, | |
| "loss": 0.9845, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018240000000000002, | |
| "loss": 1.0082, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018160000000000002, | |
| "loss": 0.9368, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001808, | |
| "loss": 1.0412, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018, | |
| "loss": 0.9716, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017920000000000002, | |
| "loss": 0.978, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001784, | |
| "loss": 1.0301, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001776, | |
| "loss": 1.0605, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017680000000000001, | |
| "loss": 0.9383, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 1.0683, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001752, | |
| "loss": 1.0155, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001744, | |
| "loss": 0.9944, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017360000000000002, | |
| "loss": 0.9694, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001728, | |
| "loss": 1.0904, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000172, | |
| "loss": 1.0099, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017120000000000001, | |
| "loss": 0.9639, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001704, | |
| "loss": 0.9827, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001696, | |
| "loss": 0.9929, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001688, | |
| "loss": 1.0025, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000168, | |
| "loss": 1.0533, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001672, | |
| "loss": 0.9881, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001664, | |
| "loss": 1.0109, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001656, | |
| "loss": 0.9747, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001648, | |
| "loss": 1.0304, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000164, | |
| "loss": 1.0282, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001632, | |
| "loss": 0.9555, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016240000000000002, | |
| "loss": 1.0386, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016160000000000002, | |
| "loss": 0.931, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001608, | |
| "loss": 0.9508, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00016, | |
| "loss": 0.9486, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2500, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 3.27221445132288e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |