| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.453662604578592, | |
| "global_step": 11400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "eval_exact_match": 14.738231412230533, | |
| "eval_f1": 19.992446610541194, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005, | |
| "loss": 3.6292, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "eval_exact_match": 21.381434227892655, | |
| "eval_f1": 31.43956554541397, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_exact_match": 24.175098988121427, | |
| "eval_f1": 34.101042049094296, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005, | |
| "loss": 2.7193, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "eval_exact_match": 24.950592885375492, | |
| "eval_f1": 38.217244544211226, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005, | |
| "loss": 2.4559, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "eval_exact_match": 28.40909090909091, | |
| "eval_f1": 44.64064089265683, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0005, | |
| "loss": 2.2801, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_exact_match": 26.630434782608695, | |
| "eval_f1": 45.504003307885625, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0005, | |
| "loss": 2.1285, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_exact_match": 30.8300395256917, | |
| "eval_f1": 50.359736600917515, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0005, | |
| "loss": 2.026, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_exact_match": 31.175889328063242, | |
| "eval_f1": 51.623685774579855, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0005, | |
| "loss": 1.9281, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_exact_match": 35.869565217391305, | |
| "eval_f1": 56.21242269358832, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0005, | |
| "loss": 1.8836, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "eval_exact_match": 35.0296442687747, | |
| "eval_f1": 57.21764925319378, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0005, | |
| "loss": 1.8377, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_exact_match": 36.61067193675889, | |
| "eval_f1": 58.4380904170952, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005, | |
| "loss": 1.7926, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_exact_match": 36.85770750988142, | |
| "eval_f1": 59.677808943848156, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005, | |
| "loss": 1.7356, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_exact_match": 35.62252964426877, | |
| "eval_f1": 59.02520911162799, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005, | |
| "loss": 1.7001, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "eval_exact_match": 36.31422924901186, | |
| "eval_f1": 60.89754377178488, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005, | |
| "loss": 1.675, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_exact_match": 36.26482213438735, | |
| "eval_f1": 59.783069358423525, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005, | |
| "loss": 1.6594, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_exact_match": 36.56126482213439, | |
| "eval_f1": 61.449547007432926, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0005, | |
| "loss": 1.6435, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_exact_match": 39.92094861660079, | |
| "eval_f1": 63.03510761568375, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0005, | |
| "loss": 1.6116, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "eval_exact_match": 37.30237154150198, | |
| "eval_f1": 62.953985227470696, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0005, | |
| "loss": 1.5701, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_exact_match": 36.6600790513834, | |
| "eval_f1": 61.820603199741704, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0005, | |
| "loss": 1.5627, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_exact_match": 38.04347826086956, | |
| "eval_f1": 63.56696764038496, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0005, | |
| "loss": 1.5525, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "eval_exact_match": 39.03162055335969, | |
| "eval_f1": 64.72028168259759, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4518, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4245, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4164, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4627, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_exact_match": 42.5, | |
| "eval_f1": 67.3159921257088, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4369, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4205, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4189, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4127, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_exact_match": 44.95, | |
| "eval_f1": 68.47203537029696, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4329, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4233, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4426, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3969, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "eval_exact_match": 46.9, | |
| "eval_f1": 69.7906665651929, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4589, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4199, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3795, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3794, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_exact_match": 50.95, | |
| "eval_f1": 71.26934663680022, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3577, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3633, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3672, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3568, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_exact_match": 49.5, | |
| "eval_f1": 71.44716194869687, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0005, | |
| "loss": 1.334, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3173, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3332, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3858, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "eval_exact_match": 50.6, | |
| "eval_f1": 71.4031953884389, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3724, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3356, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0005, | |
| "loss": 1.4173, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3499, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "eval_exact_match": 50.45, | |
| "eval_f1": 70.96365965036652, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3853, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.0005, | |
| "loss": 1.337, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3366, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0005, | |
| "loss": 1.342, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_exact_match": 51.25, | |
| "eval_f1": 72.29202655492661, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3289, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3894, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3393, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2983, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "eval_exact_match": 52.15, | |
| "eval_f1": 72.57033769389315, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3181, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3672, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3064, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3316, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_exact_match": 52.6, | |
| "eval_f1": 72.91182293082619, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3099, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0005, | |
| "loss": 1.293, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3386, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3475, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_exact_match": 50.7, | |
| "eval_f1": 72.13671184582012, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2548, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3243, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0005, | |
| "loss": 1.322, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3049, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "eval_exact_match": 51.25, | |
| "eval_f1": 72.62424474108035, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0005, | |
| "loss": 1.274, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3015, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0005, | |
| "loss": 1.303, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2727, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "eval_exact_match": 52.35, | |
| "eval_f1": 73.1319111267331, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2866, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3302, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2281, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3037, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_exact_match": 53.2, | |
| "eval_f1": 73.01971722088678, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2822, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2871, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3176, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2828, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "eval_exact_match": 52.85, | |
| "eval_f1": 73.01360270382577, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3448, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3102, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3273, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0005, | |
| "loss": 1.3016, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "eval_exact_match": 54.0, | |
| "eval_f1": 73.98223052649728, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2767, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0005, | |
| "loss": 1.262, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2846, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2642, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "eval_exact_match": 52.45, | |
| "eval_f1": 73.66687625952305, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2875, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0005, | |
| "loss": 1.247, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2501, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0005, | |
| "loss": 1.293, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_exact_match": 52.3, | |
| "eval_f1": 73.46398897242841, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2373, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2477, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2647, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2947, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 53.05, | |
| "eval_f1": 73.40471749318357, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2237, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2996, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2833, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2663, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "eval_exact_match": 53.95, | |
| "eval_f1": 74.59089034170034, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2519, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2365, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2152, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2309, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_exact_match": 53.35, | |
| "eval_f1": 74.02883287579421, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2532, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2607, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1885, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2169, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "eval_exact_match": 54.35, | |
| "eval_f1": 74.273655582778, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2543, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2352, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2142, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0005, | |
| "loss": 1.223, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 52.7, | |
| "eval_f1": 73.99393883664169, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2145, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1973, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2277, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2552, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "eval_exact_match": 53.8, | |
| "eval_f1": 74.61861738151018, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.0005, | |
| "loss": 1.228, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2117, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2429, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2279, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "eval_exact_match": 54.4, | |
| "eval_f1": 74.93345914261613, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2134, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.0005, | |
| "loss": 1.245, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2179, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2006, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_exact_match": 55.4, | |
| "eval_f1": 75.75427604493635, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2037, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1985, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2012, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2194, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "eval_exact_match": 54.0, | |
| "eval_f1": 74.89545612203896, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2173, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0005, | |
| "loss": 1.2616, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1785, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1664, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_exact_match": 55.2, | |
| "eval_f1": 75.50544659346157, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1271, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1806, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1865, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0005, | |
| "loss": 1.198, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "eval_exact_match": 54.65, | |
| "eval_f1": 75.02151439633785, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1726, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1799, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1107, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1585, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_exact_match": 53.9, | |
| "eval_f1": 74.15646190187827, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1436, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1806, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1444, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1738, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "eval_exact_match": 54.85, | |
| "eval_f1": 75.11255802008874, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1095, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1511, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1297, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0005, | |
| "loss": 1.1505, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_exact_match": 54.85, | |
| "eval_f1": 75.52882358671673, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1081, | |
| "step": 10210 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1572, | |
| "step": 10220 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0839, | |
| "step": 10230 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0971, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0737, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1049, | |
| "step": 10260 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.202, | |
| "step": 10270 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1848, | |
| "step": 10280 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1095, | |
| "step": 10290 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2268, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1422, | |
| "step": 10310 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1222, | |
| "step": 10320 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1251, | |
| "step": 10330 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0694, | |
| "step": 10340 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1571, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1546, | |
| "step": 10360 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2171, | |
| "step": 10370 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1283, | |
| "step": 10380 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1215, | |
| "step": 10390 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.127, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "eval_exact_match": 55.35, | |
| "eval_f1": 75.70498337751204, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0492, | |
| "step": 10410 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1658, | |
| "step": 10420 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1677, | |
| "step": 10430 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1758, | |
| "step": 10440 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1968, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1315, | |
| "step": 10460 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1138, | |
| "step": 10470 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1375, | |
| "step": 10480 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1892, | |
| "step": 10490 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.131, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1195, | |
| "step": 10510 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1044, | |
| "step": 10520 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1371, | |
| "step": 10530 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1382, | |
| "step": 10540 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1273, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1566, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.209, | |
| "step": 10570 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1336, | |
| "step": 10580 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0522, | |
| "step": 10590 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2019, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "eval_exact_match": 55.55, | |
| "eval_f1": 76.02151178378499, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1227, | |
| "step": 10610 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2202, | |
| "step": 10620 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0762, | |
| "step": 10630 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2074, | |
| "step": 10640 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1283, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1631, | |
| "step": 10660 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0954, | |
| "step": 10670 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2253, | |
| "step": 10680 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1141, | |
| "step": 10690 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2454, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0997, | |
| "step": 10710 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1408, | |
| "step": 10720 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.098, | |
| "step": 10730 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2331, | |
| "step": 10740 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1006, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1057, | |
| "step": 10760 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2075, | |
| "step": 10770 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0302, | |
| "step": 10780 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1484, | |
| "step": 10790 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1272, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "eval_exact_match": 55.6, | |
| "eval_f1": 75.90896740958824, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1306, | |
| "step": 10810 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1356, | |
| "step": 10820 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1776, | |
| "step": 10830 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1216, | |
| "step": 10840 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1211, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2693, | |
| "step": 10860 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1473, | |
| "step": 10870 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0641, | |
| "step": 10880 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2424, | |
| "step": 10890 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1371, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1217, | |
| "step": 10910 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2007, | |
| "step": 10920 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1501, | |
| "step": 10930 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1135, | |
| "step": 10940 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.137, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1422, | |
| "step": 10960 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1248, | |
| "step": 10970 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1899, | |
| "step": 10980 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.102, | |
| "step": 10990 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1877, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "eval_exact_match": 55.95, | |
| "eval_f1": 76.45373153349801, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1474, | |
| "step": 11010 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1536, | |
| "step": 11020 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2218, | |
| "step": 11030 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0967, | |
| "step": 11040 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1305, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.087, | |
| "step": 11060 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0908, | |
| "step": 11070 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0134, | |
| "step": 11080 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0997, | |
| "step": 11090 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.066, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.2601, | |
| "step": 11110 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1191, | |
| "step": 11120 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1025, | |
| "step": 11130 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.112, | |
| "step": 11140 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0794, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.112, | |
| "step": 11160 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1411, | |
| "step": 11170 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1118, | |
| "step": 11180 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1464, | |
| "step": 11190 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1496, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "eval_exact_match": 56.15, | |
| "eval_f1": 76.4356306603872, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1588, | |
| "step": 11210 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1369, | |
| "step": 11220 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1026, | |
| "step": 11230 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1764, | |
| "step": 11240 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1449, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1712, | |
| "step": 11260 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0465, | |
| "step": 11270 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1362, | |
| "step": 11280 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1022, | |
| "step": 11290 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1401, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0913, | |
| "step": 11310 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1922, | |
| "step": 11320 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.101, | |
| "step": 11330 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0261, | |
| "step": 11340 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1109, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0958, | |
| "step": 11360 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.027, | |
| "step": 11370 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.117, | |
| "step": 11380 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1295, | |
| "step": 11390 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 5e-05, | |
| "loss": 1.1344, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "eval_exact_match": 56.45, | |
| "eval_f1": 76.81870981288014, | |
| "step": 11400 | |
| } | |
| ], | |
| "max_steps": 92920, | |
| "num_train_epochs": 20, | |
| "total_flos": 3.858021871727411e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |