| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0387498870924037, | |
| "global_step": 11500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.954836961430765e-05, | |
| "loss": 1.1847, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9096739228615304e-05, | |
| "loss": 0.4241, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.8645108842922955e-05, | |
| "loss": 0.3364, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8193478457230606e-05, | |
| "loss": 0.2869, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.774184807153825e-05, | |
| "loss": 0.2546, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.729021768584591e-05, | |
| "loss": 0.2338, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.683858730015356e-05, | |
| "loss": 0.2185, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.63869569144612e-05, | |
| "loss": 0.2037, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.593532652876886e-05, | |
| "loss": 0.1931, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.548369614307651e-05, | |
| "loss": 0.1868, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.5032065757384155e-05, | |
| "loss": 0.1779, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.4580435371691806e-05, | |
| "loss": 0.1688, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.4128804985999464e-05, | |
| "loss": 0.1628, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.367717460030711e-05, | |
| "loss": 0.1602, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.322554421461476e-05, | |
| "loss": 0.1533, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.277391382892241e-05, | |
| "loss": 0.1493, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.232228344323007e-05, | |
| "loss": 0.1457, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.187065305753771e-05, | |
| "loss": 0.1421, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.141902267184536e-05, | |
| "loss": 0.1393, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.096739228615302e-05, | |
| "loss": 0.134, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.0515761900460664e-05, | |
| "loss": 0.1312, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.0064131514768315e-05, | |
| "loss": 0.128, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.9612501129075966e-05, | |
| "loss": 0.121, | |
| "step": 11500 | |
| } | |
| ], | |
| "max_steps": 55355, | |
| "num_train_epochs": 5, | |
| "total_flos": 2.80476865363968e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |