| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 10325, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.7094430992736077e-05, | |
| "loss": 1.8248, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_HasAns_exact": 52.27732793522267, | |
| "eval_HasAns_f1": 58.04863749892537, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 71.35407905803196, | |
| "eval_NoAns_f1": 71.35407905803196, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 61.8798955613577, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 64.74808302537699, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 61.82936073443948, | |
| "eval_f1": 64.71088377778418, | |
| "eval_total": 11873, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.4188861985472156e-05, | |
| "loss": 1.312, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_HasAns_exact": 61.40350877192982, | |
| "eval_HasAns_f1": 67.48064450139083, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 70.52985702270816, | |
| "eval_NoAns_f1": 70.52985702270816, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 65.98163901288638, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 69.01585619508543, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 65.97321654173335, | |
| "eval_f1": 69.00743372393222, | |
| "eval_total": 11873, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.1283292978208232e-05, | |
| "loss": 1.0311, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_HasAns_exact": 65.18218623481782, | |
| "eval_HasAns_f1": 72.17951939458223, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 66.57695542472666, | |
| "eval_NoAns_f1": 66.57695542472666, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 65.88056935904994, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 69.37422647781422, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 65.88056935904994, | |
| "eval_f1": 69.37422647781396, | |
| "eval_total": 11873, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.8377723970944308e-05, | |
| "loss": 0.972, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_HasAns_exact": 59.14304993252362, | |
| "eval_HasAns_f1": 64.53303485957208, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 81.05971404541631, | |
| "eval_NoAns_f1": 81.05971404541631, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 70.11707234902721, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.8082060681837, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 70.11707234902721, | |
| "eval_f1": 72.80820606818367, | |
| "eval_total": 11873, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.547215496368039e-05, | |
| "loss": 0.7884, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_HasAns_exact": 65.50269905533064, | |
| "eval_HasAns_f1": 72.13556041501943, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 73.38940285954584, | |
| "eval_NoAns_f1": 73.38940285954584, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.45169712793734, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.76337927568756, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.45169712793734, | |
| "eval_f1": 72.76337927568737, | |
| "eval_total": 11873, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.2566585956416464e-05, | |
| "loss": 0.7493, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "eval_HasAns_exact": 65.36774628879893, | |
| "eval_HasAns_f1": 72.29399398264833, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 74.3313708999159, | |
| "eval_NoAns_f1": 74.3313708999159, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.85597574328308, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 73.31414101988895, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.85597574328308, | |
| "eval_f1": 73.3141410198889, | |
| "eval_total": 11873, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 9.661016949152542e-06, | |
| "loss": 0.6217, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "eval_HasAns_exact": 67.57759784075573, | |
| "eval_HasAns_f1": 74.89750206922953, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 69.80656013456687, | |
| "eval_NoAns_f1": 69.80656013456687, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 68.69367472416407, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.34838644541345, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 68.69367472416407, | |
| "eval_f1": 72.34838644541342, | |
| "eval_total": 11873, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.75544794188862e-06, | |
| "loss": 0.598, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "eval_HasAns_exact": 66.44736842105263, | |
| "eval_HasAns_f1": 73.58893809834937, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 70.69806560134568, | |
| "eval_NoAns_f1": 70.69806560134568, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 68.57576012802156, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.1414322451796, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 68.57576012802156, | |
| "eval_f1": 72.14143224517956, | |
| "eval_total": 11873, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 3.849878934624698e-06, | |
| "loss": 0.522, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "eval_HasAns_exact": 66.26180836707152, | |
| "eval_HasAns_f1": 73.56604079226567, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 71.9932716568545, | |
| "eval_NoAns_f1": 71.9932716568545, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.13164322412196, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.778530263333, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.13164322412196, | |
| "eval_f1": 72.77853026333297, | |
| "eval_total": 11873, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 9.44309927360775e-07, | |
| "loss": 0.4903, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "eval_HasAns_exact": 66.14372469635627, | |
| "eval_HasAns_f1": 73.19227567032453, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 72.1951219512195, | |
| "eval_NoAns_f1": 72.1951219512195, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.17375557988714, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.69298493840527, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.17375557988714, | |
| "eval_f1": 72.69298493840525, | |
| "eval_total": 11873, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 10325, | |
| "total_flos": 1.294545788106624e+17, | |
| "train_loss": 0.8785118254506847, | |
| "train_runtime": 11833.1715, | |
| "train_samples_per_second": 55.824, | |
| "train_steps_per_second": 0.873 | |
| } | |
| ], | |
| "max_steps": 10325, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.294545788106624e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |