| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 22160, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.9323104693140794e-05, | |
| "loss": 1.365, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.864620938628159e-05, | |
| "loss": 0.9493, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.7969314079422385e-05, | |
| "loss": 0.8795, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.729241877256318e-05, | |
| "loss": 0.8473, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.6615523465703972e-05, | |
| "loss": 0.8401, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.5938628158844765e-05, | |
| "loss": 0.8402, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.526173285198556e-05, | |
| "loss": 0.8296, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.4584837545126353e-05, | |
| "loss": 0.7466, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.390794223826715e-05, | |
| "loss": 0.7833, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3231046931407943e-05, | |
| "loss": 0.8053, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.2554151624548737e-05, | |
| "loss": 0.7537, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.1877256317689534e-05, | |
| "loss": 0.7497, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1200361010830327e-05, | |
| "loss": 0.7674, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.0523465703971117e-05, | |
| "loss": 0.7417, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.9846570397111914e-05, | |
| "loss": 0.7584, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9169675090252708e-05, | |
| "loss": 0.7545, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.84927797833935e-05, | |
| "loss": 0.766, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7815884476534298e-05, | |
| "loss": 0.7339, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.7138989169675092e-05, | |
| "loss": 0.714, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6462093862815885e-05, | |
| "loss": 0.7173, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.578519855595668e-05, | |
| "loss": 0.7207, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.5108303249097474e-05, | |
| "loss": 0.6918, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 88.30652790917692, | |
| "eval_f1": 94.12994072622395, | |
| "step": 11080 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.4431407942238268e-05, | |
| "loss": 0.5424, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.3754512635379063e-05, | |
| "loss": 0.4775, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.3077617328519855e-05, | |
| "loss": 0.483, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.240072202166065e-05, | |
| "loss": 0.4584, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.1723826714801445e-05, | |
| "loss": 0.4604, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.1046931407942239e-05, | |
| "loss": 0.4531, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.0370036101083034e-05, | |
| "loss": 0.4533, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.693140794223826e-06, | |
| "loss": 0.4583, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.016245487364621e-06, | |
| "loss": 0.4623, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 8.339350180505416e-06, | |
| "loss": 0.4756, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.66245487364621e-06, | |
| "loss": 0.4339, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.985559566787004e-06, | |
| "loss": 0.4502, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.308664259927798e-06, | |
| "loss": 0.44, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 5.631768953068592e-06, | |
| "loss": 0.4057, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.954873646209386e-06, | |
| "loss": 0.4528, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.277978339350181e-06, | |
| "loss": 0.4447, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.6010830324909746e-06, | |
| "loss": 0.4338, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.924187725631769e-06, | |
| "loss": 0.4449, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.247292418772563e-06, | |
| "loss": 0.4263, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.5703971119133576e-06, | |
| "loss": 0.4382, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.935018050541516e-07, | |
| "loss": 0.397, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.1660649819494586e-07, | |
| "loss": 0.435, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 88.31598864711448, | |
| "eval_f1": 94.44474801701165, | |
| "step": 22160 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 22160, | |
| "total_flos": 8.247163627136614e+16, | |
| "train_loss": 0.6275850592107118, | |
| "train_runtime": 29795.4441, | |
| "train_samples_per_second": 5.95, | |
| "train_steps_per_second": 0.744 | |
| } | |
| ], | |
| "max_steps": 22160, | |
| "num_train_epochs": 2, | |
| "total_flos": 8.247163627136614e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |