| { | |
| "best_metric": 0.7430530679085001, | |
| "best_model_checkpoint": "bertbase-uncased-2-actual/checkpoint-20000", | |
| "epoch": 1.0, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.751e-06, | |
| "loss": 0.6643, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.501000000000001e-06, | |
| "loss": 0.6215, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.251000000000002e-06, | |
| "loss": 0.6165, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.0015e-06, | |
| "loss": 0.6052, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.7515e-06, | |
| "loss": 0.5804, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.5015e-06, | |
| "loss": 0.5753, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.2515e-06, | |
| "loss": 0.601, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.001500000000001e-06, | |
| "loss": 0.5831, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.7515e-06, | |
| "loss": 0.5854, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.502e-06, | |
| "loss": 0.5882, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.252e-06, | |
| "loss": 0.5659, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.002000000000001e-06, | |
| "loss": 0.564, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.752000000000001e-06, | |
| "loss": 0.5692, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.502000000000001e-06, | |
| "loss": 0.5755, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.2520000000000004e-06, | |
| "loss": 0.5547, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.002e-06, | |
| "loss": 0.5676, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.752500000000001e-06, | |
| "loss": 0.5788, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.5025e-06, | |
| "loss": 0.5486, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.2525e-06, | |
| "loss": 0.5601, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.0025e-06, | |
| "loss": 0.5331, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.752500000000001e-06, | |
| "loss": 0.5659, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.5025000000000005e-06, | |
| "loss": 0.5652, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.2525e-06, | |
| "loss": 0.5529, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0025e-06, | |
| "loss": 0.5531, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.7530000000000003e-06, | |
| "loss": 0.5556, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.503e-06, | |
| "loss": 0.5323, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.253e-06, | |
| "loss": 0.5809, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.0030000000000003e-06, | |
| "loss": 0.5367, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.753e-06, | |
| "loss": 0.5706, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.5040000000000005e-06, | |
| "loss": 0.5386, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.254e-06, | |
| "loss": 0.552, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.004e-06, | |
| "loss": 0.5468, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.754e-06, | |
| "loss": 0.5546, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5045000000000002e-06, | |
| "loss": 0.5495, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2545000000000002e-06, | |
| "loss": 0.5439, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0045e-06, | |
| "loss": 0.562, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.545000000000001e-07, | |
| "loss": 0.5589, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.045000000000001e-07, | |
| "loss": 0.5585, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.545e-07, | |
| "loss": 0.5288, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.500000000000001e-09, | |
| "loss": 0.5205, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.74895, | |
| "eval_f1": 0.7430530679085001, | |
| "eval_loss": 0.5389962196350098, | |
| "eval_runtime": 136.8617, | |
| "eval_samples_per_second": 146.133, | |
| "eval_steps_per_second": 36.533, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 20000, | |
| "num_train_epochs": 1, | |
| "total_flos": 6597569324250480.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |