| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.866911427260211, | |
| "global_step": 21500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.885268471776045e-06, | |
| "loss": 1.4166, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.770536943552089e-06, | |
| "loss": 0.6545, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.655805415328132e-06, | |
| "loss": 0.5315, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.541073887104177e-06, | |
| "loss": 0.4715, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "exact_match": 67.5, | |
| "f1": 75.87103174603179, | |
| "step": 2179 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.4263423588802205e-06, | |
| "loss": 0.3414, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.311610830656265e-06, | |
| "loss": 0.2945, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.196879302432309e-06, | |
| "loss": 0.3019, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.0821477742083525e-06, | |
| "loss": 0.3033, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "exact_match": 68.33333333333333, | |
| "f1": 76.34986772486776, | |
| "step": 4358 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.967416245984397e-06, | |
| "loss": 0.2622, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.852684717760441e-06, | |
| "loss": 0.1801, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.7379531895364852e-06, | |
| "loss": 0.1697, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.623221661312529e-06, | |
| "loss": 0.2118, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.508490133088573e-06, | |
| "loss": 0.1559, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "exact_match": 67.91666666666667, | |
| "f1": 76.26147001147002, | |
| "step": 6537 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.393758604864617e-06, | |
| "loss": 0.1068, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.2790270766406612e-06, | |
| "loss": 0.0951, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.1642955484167054e-06, | |
| "loss": 0.0986, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.049564020192749e-06, | |
| "loss": 0.1122, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "exact_match": 67.5, | |
| "f1": 74.74419793169795, | |
| "step": 8716 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.934832491968793e-06, | |
| "loss": 0.0811, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 2.8201009637448373e-06, | |
| "loss": 0.0546, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 2.7053694355208814e-06, | |
| "loss": 0.056, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.5906379072969255e-06, | |
| "loss": 0.0731, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "exact_match": 70.41666666666667, | |
| "f1": 77.59175084175087, | |
| "step": 10895 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 2.4759063790729696e-06, | |
| "loss": 0.0687, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 2.3611748508490133e-06, | |
| "loss": 0.0471, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 2.2464433226250574e-06, | |
| "loss": 0.0426, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 2.1317117944011015e-06, | |
| "loss": 0.0437, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 2.0169802661771456e-06, | |
| "loss": 0.0299, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "exact_match": 66.66666666666667, | |
| "f1": 74.826330804272, | |
| "step": 13074 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 1.9022487379531897e-06, | |
| "loss": 0.0232, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 1.7875172097292336e-06, | |
| "loss": 0.0221, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 1.672785681505278e-06, | |
| "loss": 0.0242, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 1.5580541532813219e-06, | |
| "loss": 0.0327, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "exact_match": 67.08333333333333, | |
| "f1": 75.72919876963996, | |
| "step": 15253 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 1.443322625057366e-06, | |
| "loss": 0.0207, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 1.3285910968334099e-06, | |
| "loss": 0.0243, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 1.213859568609454e-06, | |
| "loss": 0.0127, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 1.099128040385498e-06, | |
| "loss": 0.0189, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "exact_match": 66.25, | |
| "f1": 73.67448496492618, | |
| "step": 17432 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 9.84396512161542e-07, | |
| "loss": 0.0234, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 8.696649839375861e-07, | |
| "loss": 0.0041, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 7.549334557136302e-07, | |
| "loss": 0.0115, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 6.402019274896742e-07, | |
| "loss": 0.0159, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 5.254703992657182e-07, | |
| "loss": 0.0116, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "exact_match": 69.16666666666667, | |
| "f1": 77.33799506593625, | |
| "step": 19611 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 4.107388710417623e-07, | |
| "loss": 0.0096, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 2.9600734281780635e-07, | |
| "loss": 0.0057, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 1.8127581459385043e-07, | |
| "loss": 0.005, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 6.654428636989445e-08, | |
| "loss": 0.0042, | |
| "step": 21500 | |
| } | |
| ], | |
| "max_steps": 21790, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.3296295768428288e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |