| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 21960, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.9316939890710385e-05, | |
| "loss": 2.4099, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.8633879781420765e-05, | |
| "loss": 1.6744, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.795081967213115e-05, | |
| "loss": 1.5233, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.7267759562841533e-05, | |
| "loss": 1.402, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.6584699453551913e-05, | |
| "loss": 1.3482, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.5901639344262294e-05, | |
| "loss": 1.2935, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.5218579234972678e-05, | |
| "loss": 1.2617, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.453551912568306e-05, | |
| "loss": 1.2702, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.3852459016393442e-05, | |
| "loss": 1.1879, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.3169398907103826e-05, | |
| "loss": 1.1748, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.248633879781421e-05, | |
| "loss": 1.1447, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.180327868852459e-05, | |
| "loss": 1.1318, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1120218579234974e-05, | |
| "loss": 1.1303, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.0437158469945358e-05, | |
| "loss": 1.0937, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.975409836065574e-05, | |
| "loss": 1.0495, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.907103825136612e-05, | |
| "loss": 1.0518, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8387978142076503e-05, | |
| "loss": 1.0632, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7704918032786887e-05, | |
| "loss": 1.0331, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.7021857923497267e-05, | |
| "loss": 1.0486, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.633879781420765e-05, | |
| "loss": 1.0162, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.5655737704918035e-05, | |
| "loss": 1.0093, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.4972677595628415e-05, | |
| "loss": 0.9826, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.4289617486338798e-05, | |
| "loss": 0.7505, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.3606557377049181e-05, | |
| "loss": 0.716, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.2923497267759564e-05, | |
| "loss": 0.7136, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.2240437158469946e-05, | |
| "loss": 0.7486, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1557377049180328e-05, | |
| "loss": 0.7063, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.087431693989071e-05, | |
| "loss": 0.6988, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.0191256830601094e-05, | |
| "loss": 0.7327, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.508196721311476e-06, | |
| "loss": 0.7099, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 8.825136612021857e-06, | |
| "loss": 0.6963, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.14207650273224e-06, | |
| "loss": 0.7096, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.459016393442623e-06, | |
| "loss": 0.6755, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.775956284153006e-06, | |
| "loss": 0.6778, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.092896174863388e-06, | |
| "loss": 0.7141, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5.409836065573771e-06, | |
| "loss": 0.6712, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.726775956284153e-06, | |
| "loss": 0.6862, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.043715846994535e-06, | |
| "loss": 0.66, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.3606557377049183e-06, | |
| "loss": 0.652, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.6775956284153005e-06, | |
| "loss": 0.694, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.994535519125683e-06, | |
| "loss": 0.6673, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.3114754098360657e-06, | |
| "loss": 0.6625, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.284153005464482e-07, | |
| "loss": 0.6728, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 21960, | |
| "total_flos": 5.164033933049242e+16, | |
| "train_loss": 0.9679226359383004, | |
| "train_runtime": 5136.3695, | |
| "train_samples_per_second": 51.302, | |
| "train_steps_per_second": 4.275 | |
| } | |
| ], | |
| "max_steps": 21960, | |
| "num_train_epochs": 2, | |
| "total_flos": 5.164033933049242e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |