| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.999886413115499, | |
| "global_step": 26410, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999242711094284e-05, | |
| "loss": 1.3163, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9963650132525562e-05, | |
| "loss": 1.0093, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.99265429761454e-05, | |
| "loss": 0.8939, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9888678530859524e-05, | |
| "loss": 0.8368, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.985081408557365e-05, | |
| "loss": 0.8794, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9812949640287772e-05, | |
| "loss": 0.843, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9775085195001894e-05, | |
| "loss": 0.8488, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9737220749716017e-05, | |
| "loss": 0.8465, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9699356304430143e-05, | |
| "loss": 0.8591, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9661491859144265e-05, | |
| "loss": 0.8978, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9623627413858387e-05, | |
| "loss": 0.8035, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9585762968572513e-05, | |
| "loss": 0.8153, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9547898523286635e-05, | |
| "loss": 0.8079, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9510034078000758e-05, | |
| "loss": 0.8092, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.947216963271488e-05, | |
| "loss": 0.8294, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9434305187429006e-05, | |
| "loss": 0.8099, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.939644074214313e-05, | |
| "loss": 0.85, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.935857629685725e-05, | |
| "loss": 0.8509, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9320711851571376e-05, | |
| "loss": 0.8229, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.92828474062855e-05, | |
| "loss": 0.8251, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9244982960999625e-05, | |
| "loss": 0.8116, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9207118515713747e-05, | |
| "loss": 0.8277, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.916925407042787e-05, | |
| "loss": 0.8004, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9131389625141995e-05, | |
| "loss": 0.8347, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9093525179856118e-05, | |
| "loss": 0.8103, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.905566073457024e-05, | |
| "loss": 0.8566, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9017796289284362e-05, | |
| "loss": 0.7921, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.8979931843998488e-05, | |
| "loss": 0.7959, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.894206739871261e-05, | |
| "loss": 0.7827, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.8904202953426733e-05, | |
| "loss": 0.785, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.886633850814086e-05, | |
| "loss": 0.8061, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.882847406285498e-05, | |
| "loss": 0.7717, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.8790609617569103e-05, | |
| "loss": 0.8362, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.8752745172283226e-05, | |
| "loss": 0.7975, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.871488072699735e-05, | |
| "loss": 0.8123, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.8677016281711474e-05, | |
| "loss": 0.8526, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.8639151836425596e-05, | |
| "loss": 0.808, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.8601287391139722e-05, | |
| "loss": 0.7989, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.8563422945853844e-05, | |
| "loss": 0.7755, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8525558500567967e-05, | |
| "loss": 0.8467, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8487694055282093e-05, | |
| "loss": 0.8221, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8449829609996215e-05, | |
| "loss": 0.7851, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.841196516471034e-05, | |
| "loss": 0.8159, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.837410071942446e-05, | |
| "loss": 0.8027, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8336236274138586e-05, | |
| "loss": 0.7585, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8298371828852708e-05, | |
| "loss": 0.8287, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8260507383566834e-05, | |
| "loss": 0.8127, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8222642938280956e-05, | |
| "loss": 0.8015, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.818477849299508e-05, | |
| "loss": 0.8228, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8146914047709204e-05, | |
| "loss": 0.7865, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8109049602423327e-05, | |
| "loss": 0.8093, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.807118515713745e-05, | |
| "loss": 0.7812, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.803332071185157e-05, | |
| "loss": 0.805, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.7995456266565697e-05, | |
| "loss": 0.7844, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.795759182127982e-05, | |
| "loss": 0.7976, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7919727375993942e-05, | |
| "loss": 0.7683, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7881862930708068e-05, | |
| "loss": 0.8157, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.784399848542219e-05, | |
| "loss": 0.7801, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7806134040136312e-05, | |
| "loss": 0.7817, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7768269594850435e-05, | |
| "loss": 0.8286, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.773040514956456e-05, | |
| "loss": 0.7994, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7692540704278683e-05, | |
| "loss": 0.758, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7654676258992805e-05, | |
| "loss": 0.8116, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.761681181370693e-05, | |
| "loss": 0.7762, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7578947368421054e-05, | |
| "loss": 0.7603, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7541082923135176e-05, | |
| "loss": 0.825, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7503218477849302e-05, | |
| "loss": 0.7649, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7465354032563424e-05, | |
| "loss": 0.7886, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.742748958727755e-05, | |
| "loss": 0.796, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7389625141991672e-05, | |
| "loss": 0.7712, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7351760696705795e-05, | |
| "loss": 0.8061, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7313896251419917e-05, | |
| "loss": 0.7691, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7276031806134043e-05, | |
| "loss": 0.8038, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7238167360848165e-05, | |
| "loss": 0.7679, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7200302915562288e-05, | |
| "loss": 0.7909, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.7162438470276413e-05, | |
| "loss": 0.8001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7124574024990536e-05, | |
| "loss": 0.7369, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7086709579704658e-05, | |
| "loss": 0.7716, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.704884513441878e-05, | |
| "loss": 0.8003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7010980689132906e-05, | |
| "loss": 0.7973, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.697311624384703e-05, | |
| "loss": 0.7745, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.693525179856115e-05, | |
| "loss": 0.7266, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6897387353275277e-05, | |
| "loss": 0.7917, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.68595229079894e-05, | |
| "loss": 0.7393, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.682165846270352e-05, | |
| "loss": 0.7822, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6783794017417647e-05, | |
| "loss": 0.7919, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.674592957213177e-05, | |
| "loss": 0.7943, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.6708065126845892e-05, | |
| "loss": 0.7731, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.6670200681560014e-05, | |
| "loss": 0.7623, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.663233623627414e-05, | |
| "loss": 0.7891, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.6594471790988263e-05, | |
| "loss": 0.7769, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.655660734570239e-05, | |
| "loss": 0.7958, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.651874290041651e-05, | |
| "loss": 0.7555, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.6480878455130633e-05, | |
| "loss": 0.7993, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.644301400984476e-05, | |
| "loss": 0.7916, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.640514956455888e-05, | |
| "loss": 0.8101, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6367285119273004e-05, | |
| "loss": 0.8066, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6329420673987126e-05, | |
| "loss": 0.761, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6291556228701252e-05, | |
| "loss": 0.7919, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6253691783415374e-05, | |
| "loss": 0.7484, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.6215827338129497e-05, | |
| "loss": 0.809, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.6177962892843622e-05, | |
| "loss": 0.7683, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.6140098447557745e-05, | |
| "loss": 0.7622, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.6102234002271867e-05, | |
| "loss": 0.7618, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.6064369556985993e-05, | |
| "loss": 0.7687, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.6026505111700115e-05, | |
| "loss": 0.7598, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.5988640666414238e-05, | |
| "loss": 0.7245, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.595077622112836e-05, | |
| "loss": 0.7751, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.5912911775842486e-05, | |
| "loss": 0.7903, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.5875047330556608e-05, | |
| "loss": 0.7399, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.583718288527073e-05, | |
| "loss": 0.7529, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.5799318439984856e-05, | |
| "loss": 0.7777, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.576145399469898e-05, | |
| "loss": 0.7411, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.5723589549413105e-05, | |
| "loss": 0.7805, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.5685725104127224e-05, | |
| "loss": 0.7585, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.564786065884135e-05, | |
| "loss": 0.7908, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.560999621355547e-05, | |
| "loss": 0.7302, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.5572131768269597e-05, | |
| "loss": 0.7492, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.553426732298372e-05, | |
| "loss": 0.8124, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.5496402877697842e-05, | |
| "loss": 0.7609, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.5458538432411968e-05, | |
| "loss": 0.7383, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.542067398712609e-05, | |
| "loss": 0.7585, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.538356683074593e-05, | |
| "loss": 0.7852, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.5345702385460056e-05, | |
| "loss": 0.7737, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.5307837940174178e-05, | |
| "loss": 0.7881, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.52699734948883e-05, | |
| "loss": 0.7716, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5232109049602425e-05, | |
| "loss": 0.7478, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5194244604316549e-05, | |
| "loss": 0.7681, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5156380159030673e-05, | |
| "loss": 0.7703, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.5118515713744795e-05, | |
| "loss": 0.8031, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.508065126845892e-05, | |
| "loss": 0.7812, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.5042786823173042e-05, | |
| "loss": 0.8151, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.5004922377887166e-05, | |
| "loss": 0.8384, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.4967057932601288e-05, | |
| "loss": 0.7861, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4929193487315412e-05, | |
| "loss": 0.7473, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4891329042029536e-05, | |
| "loss": 0.7741, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4853464596743659e-05, | |
| "loss": 0.799, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.4815600151457783e-05, | |
| "loss": 0.7929, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.4777735706171905e-05, | |
| "loss": 0.7594, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4739871260886029e-05, | |
| "loss": 0.736, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4702006815600151e-05, | |
| "loss": 0.7608, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4664142370314276e-05, | |
| "loss": 0.7927, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.46262779250284e-05, | |
| "loss": 0.7753, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.4588413479742522e-05, | |
| "loss": 0.7476, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4550549034456646e-05, | |
| "loss": 0.7907, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4512684589170768e-05, | |
| "loss": 0.7874, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4474820143884894e-05, | |
| "loss": 0.7626, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4436955698599018e-05, | |
| "loss": 0.7527, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.439909125331314e-05, | |
| "loss": 0.7451, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4361226808027265e-05, | |
| "loss": 0.7615, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4323362362741387e-05, | |
| "loss": 0.7365, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4285497917455511e-05, | |
| "loss": 0.7627, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4247633472169634e-05, | |
| "loss": 0.7769, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4209769026883758e-05, | |
| "loss": 0.7253, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4171904581597882e-05, | |
| "loss": 0.7422, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4134040136312004e-05, | |
| "loss": 0.7459, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4096175691026128e-05, | |
| "loss": 0.7355, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.405831124574025e-05, | |
| "loss": 0.7658, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4020446800454375e-05, | |
| "loss": 0.8133, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.3982582355168497e-05, | |
| "loss": 0.7574, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3944717909882621e-05, | |
| "loss": 0.7755, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3906853464596745e-05, | |
| "loss": 0.7434, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3868989019310868e-05, | |
| "loss": 0.7565, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.3831124574024992e-05, | |
| "loss": 0.7705, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.3793260128739114e-05, | |
| "loss": 0.7699, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.3755395683453238e-05, | |
| "loss": 0.7834, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.3718288527073081e-05, | |
| "loss": 0.7788, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.3680424081787202e-05, | |
| "loss": 0.7329, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.3642559636501328e-05, | |
| "loss": 0.7774, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.3604695191215448e-05, | |
| "loss": 0.7427, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.3566830745929574e-05, | |
| "loss": 0.7482, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.3528966300643698e-05, | |
| "loss": 0.7888, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.349110185535782e-05, | |
| "loss": 0.7535, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.3453237410071945e-05, | |
| "loss": 0.7456, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.3415372964786067e-05, | |
| "loss": 0.7964, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.3377508519500191e-05, | |
| "loss": 0.7913, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3339644074214313e-05, | |
| "loss": 0.7745, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3301779628928438e-05, | |
| "loss": 0.7595, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3263915183642562e-05, | |
| "loss": 0.7117, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3226050738356684e-05, | |
| "loss": 0.8142, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3188186293070808e-05, | |
| "loss": 0.7539, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.315032184778493e-05, | |
| "loss": 0.7135, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3112457402499055e-05, | |
| "loss": 0.7883, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3074592957213177e-05, | |
| "loss": 0.7335, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.3036728511927301e-05, | |
| "loss": 0.771, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2998864066641425e-05, | |
| "loss": 0.7341, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2960999621355547e-05, | |
| "loss": 0.7177, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.2923135176069672e-05, | |
| "loss": 0.7784, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.2885270730783794e-05, | |
| "loss": 0.7906, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.2847406285497918e-05, | |
| "loss": 0.7762, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.280954184021204e-05, | |
| "loss": 0.7482, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.2771677394926164e-05, | |
| "loss": 0.7069, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.2734570238546007e-05, | |
| "loss": 0.761, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.269670579326013e-05, | |
| "loss": 0.7375, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.2658841347974254e-05, | |
| "loss": 0.7937, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2620976902688378e-05, | |
| "loss": 0.7983, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.25831124574025e-05, | |
| "loss": 0.7339, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2545248012116624e-05, | |
| "loss": 0.7671, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2507383566830747e-05, | |
| "loss": 0.7311, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2469519121544871e-05, | |
| "loss": 0.7652, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2431654676258993e-05, | |
| "loss": 0.7549, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2393790230973117e-05, | |
| "loss": 0.7802, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2355925785687241e-05, | |
| "loss": 0.7573, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2318061340401364e-05, | |
| "loss": 0.695, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2280196895115488e-05, | |
| "loss": 0.7214, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.224233244982961e-05, | |
| "loss": 0.772, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2204468004543734e-05, | |
| "loss": 0.7855, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2166603559257857e-05, | |
| "loss": 0.7345, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.212873911397198e-05, | |
| "loss": 0.7416, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.2090874668686105e-05, | |
| "loss": 0.7723, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2053010223400227e-05, | |
| "loss": 0.7446, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2015145778114351e-05, | |
| "loss": 0.7657, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.1977281332828474e-05, | |
| "loss": 0.7944, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1939416887542598e-05, | |
| "loss": 0.7598, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1901552442256724e-05, | |
| "loss": 0.7212, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1863687996970846e-05, | |
| "loss": 0.7564, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.182582355168497e-05, | |
| "loss": 0.7552, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.1787959106399092e-05, | |
| "loss": 0.7582, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1750094661113217e-05, | |
| "loss": 0.7506, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1712230215827339e-05, | |
| "loss": 0.7738, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1674365770541463e-05, | |
| "loss": 0.7295, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1636501325255587e-05, | |
| "loss": 0.7627, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.159863687996971e-05, | |
| "loss": 0.7918, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1560772434683833e-05, | |
| "loss": 0.7235, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1522907989397956e-05, | |
| "loss": 0.7988, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.148504354411208e-05, | |
| "loss": 0.7646, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1447179098826202e-05, | |
| "loss": 0.7794, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1409314653540326e-05, | |
| "loss": 0.79, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.137145020825445e-05, | |
| "loss": 0.7291, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.1333585762968573e-05, | |
| "loss": 0.7682, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.1295721317682697e-05, | |
| "loss": 0.8078, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.125785687239682e-05, | |
| "loss": 0.8196, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.1219992427110943e-05, | |
| "loss": 0.7173, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.1182127981825066e-05, | |
| "loss": 0.7908, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.114426353653919e-05, | |
| "loss": 0.7784, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1106399091253316e-05, | |
| "loss": 0.7317, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1068534645967436e-05, | |
| "loss": 0.7497, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1030670200681562e-05, | |
| "loss": 0.7504, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0992805755395683e-05, | |
| "loss": 0.7962, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0954941310109809e-05, | |
| "loss": 0.7615, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0917076864823933e-05, | |
| "loss": 0.7485, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0879969708443772e-05, | |
| "loss": 0.7088, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0842105263157896e-05, | |
| "loss": 0.7479, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0804240817872019e-05, | |
| "loss": 0.7571, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0766376372586143e-05, | |
| "loss": 0.7576, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.0728511927300267e-05, | |
| "loss": 0.7267, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.069064748201439e-05, | |
| "loss": 0.7521, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0652783036728513e-05, | |
| "loss": 0.7616, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0614918591442636e-05, | |
| "loss": 0.7232, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.057705414615676e-05, | |
| "loss": 0.797, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0539189700870882e-05, | |
| "loss": 0.7575, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0501325255585006e-05, | |
| "loss": 0.7384, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.046346081029913e-05, | |
| "loss": 0.7523, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0425596365013253e-05, | |
| "loss": 0.7271, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0387731919727377e-05, | |
| "loss": 0.7746, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.03498674744415e-05, | |
| "loss": 0.7463, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0312003029155623e-05, | |
| "loss": 0.7234, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0274138583869746e-05, | |
| "loss": 0.7462, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.023627413858387e-05, | |
| "loss": 0.7391, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0198409693297995e-05, | |
| "loss": 0.7761, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0160545248012116e-05, | |
| "loss": 0.7849, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0122680802726242e-05, | |
| "loss": 0.7391, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0084816357440364e-05, | |
| "loss": 0.7203, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0046951912154488e-05, | |
| "loss": 0.7439, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0009087466868612e-05, | |
| "loss": 0.7551, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.971223021582735e-06, | |
| "loss": 0.6935, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.933358576296857e-06, | |
| "loss": 0.7541, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.895494131010981e-06, | |
| "loss": 0.7828, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.857629685725105e-06, | |
| "loss": 0.7204, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.819765240439228e-06, | |
| "loss": 0.7374, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.781900795153352e-06, | |
| "loss": 0.7477, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.744036349867474e-06, | |
| "loss": 0.7516, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.706171904581598e-06, | |
| "loss": 0.727, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.668307459295722e-06, | |
| "loss": 0.7343, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.630443014009846e-06, | |
| "loss": 0.7739, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.592578568723969e-06, | |
| "loss": 0.7447, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.554714123438093e-06, | |
| "loss": 0.7607, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.516849678152215e-06, | |
| "loss": 0.79, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.47898523286634e-06, | |
| "loss": 0.7209, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.441120787580462e-06, | |
| "loss": 0.7745, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.403256342294586e-06, | |
| "loss": 0.7179, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.36539189700871e-06, | |
| "loss": 0.774, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.327527451722832e-06, | |
| "loss": 0.7592, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.289663006436956e-06, | |
| "loss": 0.7598, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.25179856115108e-06, | |
| "loss": 0.7665, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.213934115865203e-06, | |
| "loss": 0.7529, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.176069670579327e-06, | |
| "loss": 0.6873, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.138205225293451e-06, | |
| "loss": 0.7574, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.100340780007573e-06, | |
| "loss": 0.7215, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.062476334721697e-06, | |
| "loss": 0.7411, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.02461188943582e-06, | |
| "loss": 0.7627, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.986747444149944e-06, | |
| "loss": 0.7313, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.948882998864066e-06, | |
| "loss": 0.765, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.911018553578192e-06, | |
| "loss": 0.7696, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.873154108292314e-06, | |
| "loss": 0.7305, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.835289663006439e-06, | |
| "loss": 0.7653, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.797425217720561e-06, | |
| "loss": 0.7444, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.759560772434685e-06, | |
| "loss": 0.7858, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.721696327148807e-06, | |
| "loss": 0.7634, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.683831881862931e-06, | |
| "loss": 0.7471, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.645967436577056e-06, | |
| "loss": 0.7065, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.608102991291178e-06, | |
| "loss": 0.7659, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.570238546005302e-06, | |
| "loss": 0.7085, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.532374100719424e-06, | |
| "loss": 0.7312, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.494509655433548e-06, | |
| "loss": 0.814, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.45664521014767e-06, | |
| "loss": 0.7194, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.418780764861797e-06, | |
| "loss": 0.7266, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.380916319575919e-06, | |
| "loss": 0.7345, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.343051874290043e-06, | |
| "loss": 0.7121, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.305187429004165e-06, | |
| "loss": 0.7716, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.26732298371829e-06, | |
| "loss": 0.7601, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.229458538432412e-06, | |
| "loss": 0.736, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.191594093146536e-06, | |
| "loss": 0.7659, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.15372964786066e-06, | |
| "loss": 0.7504, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.115865202574782e-06, | |
| "loss": 0.7367, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.078000757288907e-06, | |
| "loss": 0.739, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.040136312003029e-06, | |
| "loss": 0.7421, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.002271866717153e-06, | |
| "loss": 0.7181, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.964407421431277e-06, | |
| "loss": 0.7699, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.926542976145401e-06, | |
| "loss": 0.7725, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.888678530859524e-06, | |
| "loss": 0.7658, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.850814085573648e-06, | |
| "loss": 0.6961, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.813706929193487e-06, | |
| "loss": 0.7598, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.775842483907611e-06, | |
| "loss": 0.7684, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.737978038621735e-06, | |
| "loss": 0.7258, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.700113593335858e-06, | |
| "loss": 0.7602, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.662249148049982e-06, | |
| "loss": 0.775, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.624384702764105e-06, | |
| "loss": 0.7311, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.586520257478228e-06, | |
| "loss": 0.7252, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.548655812192352e-06, | |
| "loss": 0.7078, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.5107913669064756e-06, | |
| "loss": 0.7607, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.472926921620599e-06, | |
| "loss": 0.7462, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.435062476334722e-06, | |
| "loss": 0.7574, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.397198031048845e-06, | |
| "loss": 0.7223, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 7.3593335857629685e-06, | |
| "loss": 0.7306, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.321469140477092e-06, | |
| "loss": 0.7638, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.283604695191217e-06, | |
| "loss": 0.7236, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.24574024990534e-06, | |
| "loss": 0.7381, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.207875804619463e-06, | |
| "loss": 0.7257, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.170011359333586e-06, | |
| "loss": 0.7516, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.132904202953427e-06, | |
| "loss": 0.7189, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.095039757667551e-06, | |
| "loss": 0.7672, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.057175312381674e-06, | |
| "loss": 0.7289, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.019310867095797e-06, | |
| "loss": 0.7194, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.9814464218099205e-06, | |
| "loss": 0.6933, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.943581976524045e-06, | |
| "loss": 0.8152, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.905717531238168e-06, | |
| "loss": 0.76, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.867853085952292e-06, | |
| "loss": 0.7271, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.829988640666415e-06, | |
| "loss": 0.7494, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.792124195380538e-06, | |
| "loss": 0.7738, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.754259750094662e-06, | |
| "loss": 0.7071, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.716395304808785e-06, | |
| "loss": 0.7339, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.678530859522908e-06, | |
| "loss": 0.7506, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.640666414237032e-06, | |
| "loss": 0.7121, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.602801968951155e-06, | |
| "loss": 0.7018, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.564937523665279e-06, | |
| "loss": 0.7677, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.527073078379403e-06, | |
| "loss": 0.6925, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.489208633093526e-06, | |
| "loss": 0.7205, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.451344187807649e-06, | |
| "loss": 0.7163, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.413479742521772e-06, | |
| "loss": 0.752, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.3756152972358965e-06, | |
| "loss": 0.7706, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.33775085195002e-06, | |
| "loss": 0.7556, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.299886406664143e-06, | |
| "loss": 0.7626, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.262021961378266e-06, | |
| "loss": 0.7447, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.224157516092389e-06, | |
| "loss": 0.741, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.186293070806513e-06, | |
| "loss": 0.7567, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.1484286255206375e-06, | |
| "loss": 0.7364, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.110564180234761e-06, | |
| "loss": 0.739, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.072699734948884e-06, | |
| "loss": 0.7309, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.034835289663007e-06, | |
| "loss": 0.7385, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.9969708443771305e-06, | |
| "loss": 0.7588, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.959106399091254e-06, | |
| "loss": 0.7442, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.921241953805377e-06, | |
| "loss": 0.7871, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.883377508519501e-06, | |
| "loss": 0.7817, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.845513063233624e-06, | |
| "loss": 0.751, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.8076486179477474e-06, | |
| "loss": 0.7379, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.769784172661871e-06, | |
| "loss": 0.7151, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.731919727375994e-06, | |
| "loss": 0.7989, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.694055282090117e-06, | |
| "loss": 0.752, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.656190836804242e-06, | |
| "loss": 0.755, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.618326391518365e-06, | |
| "loss": 0.7355, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.5804619462324885e-06, | |
| "loss": 0.668, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.542597500946612e-06, | |
| "loss": 0.763, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.504733055660735e-06, | |
| "loss": 0.7369, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.466868610374858e-06, | |
| "loss": 0.7389, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.429004165088982e-06, | |
| "loss": 0.7435, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.3911397198031055e-06, | |
| "loss": 0.7103, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.353275274517229e-06, | |
| "loss": 0.7, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.315410829231352e-06, | |
| "loss": 0.7269, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.277546383945475e-06, | |
| "loss": 0.713, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.2396819386595984e-06, | |
| "loss": 0.7442, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.201817493373722e-06, | |
| "loss": 0.731, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.163953048087847e-06, | |
| "loss": 0.7389, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.12608860280197e-06, | |
| "loss": 0.7348, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.088224157516093e-06, | |
| "loss": 0.7607, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.050359712230216e-06, | |
| "loss": 0.7151, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.0124952669443395e-06, | |
| "loss": 0.7002, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.974630821658464e-06, | |
| "loss": 0.7734, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.936766376372587e-06, | |
| "loss": 0.7506, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.89890193108671e-06, | |
| "loss": 0.7369, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.861037485800833e-06, | |
| "loss": 0.7118, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.8231730405149565e-06, | |
| "loss": 0.7152, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.78530859522908e-06, | |
| "loss": 0.7366, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.747444149943204e-06, | |
| "loss": 0.7086, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.709579704657327e-06, | |
| "loss": 0.7306, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.67171525937145e-06, | |
| "loss": 0.7266, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.633850814085574e-06, | |
| "loss": 0.7241, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.596743657705415e-06, | |
| "loss": 0.708, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.558879212419539e-06, | |
| "loss": 0.7544, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.521014767133662e-06, | |
| "loss": 0.739, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.483150321847785e-06, | |
| "loss": 0.7492, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.4452858765619086e-06, | |
| "loss": 0.7323, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.407421431276032e-06, | |
| "loss": 0.7431, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.369556985990155e-06, | |
| "loss": 0.7302, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.331692540704279e-06, | |
| "loss": 0.7112, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.293828095418402e-06, | |
| "loss": 0.7359, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.2559636501325256e-06, | |
| "loss": 0.7274, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.21809920484665e-06, | |
| "loss": 0.7307, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.180234759560773e-06, | |
| "loss": 0.7339, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.142370314274896e-06, | |
| "loss": 0.7743, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.104505868989019e-06, | |
| "loss": 0.7424, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.066641423703143e-06, | |
| "loss": 0.7152, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.028776978417267e-06, | |
| "loss": 0.707, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.99091253313139e-06, | |
| "loss": 0.7538, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.953048087845514e-06, | |
| "loss": 0.7293, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.915183642559637e-06, | |
| "loss": 0.7524, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.87731919727376e-06, | |
| "loss": 0.7938, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.839454751987884e-06, | |
| "loss": 0.7266, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.8015903067020073e-06, | |
| "loss": 0.7566, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.7637258614161305e-06, | |
| "loss": 0.713, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.725861416130254e-06, | |
| "loss": 0.7546, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.6879969708443774e-06, | |
| "loss": 0.6955, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.6501325255585006e-06, | |
| "loss": 0.7084, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.6122680802726247e-06, | |
| "loss": 0.7447, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.574403634986748e-06, | |
| "loss": 0.7173, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.536539189700871e-06, | |
| "loss": 0.7324, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.498674744414995e-06, | |
| "loss": 0.7561, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.460810299129118e-06, | |
| "loss": 0.7505, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.4229458538432413e-06, | |
| "loss": 0.7337, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.3850814085573645e-06, | |
| "loss": 0.7753, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.3472169632714886e-06, | |
| "loss": 0.7703, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.309352517985612e-06, | |
| "loss": 0.7384, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.271488072699735e-06, | |
| "loss": 0.7759, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.2336236274138587e-06, | |
| "loss": 0.7446, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.195759182127982e-06, | |
| "loss": 0.7437, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.157894736842105e-06, | |
| "loss": 0.7817, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.1200302915562292e-06, | |
| "loss": 0.7396, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.0821658462703525e-06, | |
| "loss": 0.743, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.0443014009844757e-06, | |
| "loss": 0.7597, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.0064369556985994e-06, | |
| "loss": 0.6844, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.9685725104127226e-06, | |
| "loss": 0.7737, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.930708065126846e-06, | |
| "loss": 0.7024, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.89284361984097e-06, | |
| "loss": 0.7136, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.854979174555093e-06, | |
| "loss": 0.7482, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.8171147292692164e-06, | |
| "loss": 0.7608, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.7792502839833396e-06, | |
| "loss": 0.7442, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.7413858386974632e-06, | |
| "loss": 0.6687, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.703521393411587e-06, | |
| "loss": 0.7594, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.66565694812571e-06, | |
| "loss": 0.7042, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.6277925028398338e-06, | |
| "loss": 0.7556, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.589928057553957e-06, | |
| "loss": 0.7357, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.5520636122680802e-06, | |
| "loss": 0.7292, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.5141991669822043e-06, | |
| "loss": 0.6899, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.4763347216963275e-06, | |
| "loss": 0.729, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.4384702764104508e-06, | |
| "loss": 0.7473, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.400605831124574e-06, | |
| "loss": 0.7264, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.3627413858386977e-06, | |
| "loss": 0.7314, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.3256342294585385e-06, | |
| "loss": 0.6819, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.287769784172662e-06, | |
| "loss": 0.7637, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.2499053388867854e-06, | |
| "loss": 0.727, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.2120408936009086e-06, | |
| "loss": 0.7044, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.1741764483150323e-06, | |
| "loss": 0.6926, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.136312003029156e-06, | |
| "loss": 0.7406, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.098447557743279e-06, | |
| "loss": 0.7572, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.060583112457403e-06, | |
| "loss": 0.7239, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.022718667171526e-06, | |
| "loss": 0.7353, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9848542218856497e-06, | |
| "loss": 0.7287, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.946989776599773e-06, | |
| "loss": 0.7461, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.909125331313896e-06, | |
| "loss": 0.7489, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.8712608860280198e-06, | |
| "loss": 0.7156, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.8333964407421435e-06, | |
| "loss": 0.6765, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.7955319954562667e-06, | |
| "loss": 0.7496, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.7576675501703901e-06, | |
| "loss": 0.7269, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.7198031048845134e-06, | |
| "loss": 0.728, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.681938659598637e-06, | |
| "loss": 0.7914, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.6440742143127605e-06, | |
| "loss": 0.7419, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.6062097690268837e-06, | |
| "loss": 0.7509, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.5683453237410074e-06, | |
| "loss": 0.7372, | |
| "step": 24350 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.5304808784551308e-06, | |
| "loss": 0.7159, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.4926164331692542e-06, | |
| "loss": 0.7272, | |
| "step": 24450 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.4547519878833777e-06, | |
| "loss": 0.735, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.416887542597501e-06, | |
| "loss": 0.715, | |
| "step": 24550 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.3790230973116246e-06, | |
| "loss": 0.7507, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.341158652025748e-06, | |
| "loss": 0.7723, | |
| "step": 24650 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.3032942067398712e-06, | |
| "loss": 0.7658, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.2654297614539949e-06, | |
| "loss": 0.7727, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.2275653161681181e-06, | |
| "loss": 0.748, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1897008708822416e-06, | |
| "loss": 0.7135, | |
| "step": 24850 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1518364255963652e-06, | |
| "loss": 0.76, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1139719803104887e-06, | |
| "loss": 0.7439, | |
| "step": 24950 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.0761075350246119e-06, | |
| "loss": 0.7496, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.0382430897387353e-06, | |
| "loss": 0.7631, | |
| "step": 25050 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.000378644452859e-06, | |
| "loss": 0.7033, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.625141991669822e-07, | |
| "loss": 0.717, | |
| "step": 25150 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.246497538811057e-07, | |
| "loss": 0.7329, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.867853085952291e-07, | |
| "loss": 0.7065, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.489208633093526e-07, | |
| "loss": 0.702, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.110564180234761e-07, | |
| "loss": 0.7161, | |
| "step": 25350 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.731919727375994e-07, | |
| "loss": 0.725, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.353275274517229e-07, | |
| "loss": 0.7256, | |
| "step": 25450 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.974630821658464e-07, | |
| "loss": 0.7395, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.595986368799697e-07, | |
| "loss": 0.794, | |
| "step": 25550 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.217341915940932e-07, | |
| "loss": 0.7349, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.838697463082166e-07, | |
| "loss": 0.7394, | |
| "step": 25650 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.460053010223401e-07, | |
| "loss": 0.7345, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.081408557364635e-07, | |
| "loss": 0.6888, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.702764104505869e-07, | |
| "loss": 0.7347, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.324119651647104e-07, | |
| "loss": 0.7246, | |
| "step": 25850 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.945475198788338e-07, | |
| "loss": 0.7438, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.5668307459295723e-07, | |
| "loss": 0.7108, | |
| "step": 25950 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 3.1881862930708067e-07, | |
| "loss": 0.7208, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.809541840212041e-07, | |
| "loss": 0.7633, | |
| "step": 26050 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.4308973873532756e-07, | |
| "loss": 0.7707, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.05225293449451e-07, | |
| "loss": 0.7524, | |
| "step": 26150 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6736084816357444e-07, | |
| "loss": 0.707, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.2949640287769786e-07, | |
| "loss": 0.7751, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.163195759182128e-08, | |
| "loss": 0.7672, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.376751230594472e-08, | |
| "loss": 0.6998, | |
| "step": 26350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.5903067020068158e-08, | |
| "loss": 0.7069, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 26410, | |
| "total_flos": 6.728145843607406e+16, | |
| "train_loss": 0.7582937260732105, | |
| "train_runtime": 12885.6962, | |
| "train_samples_per_second": 65.589, | |
| "train_steps_per_second": 2.05 | |
| } | |
| ], | |
| "max_steps": 26410, | |
| "num_train_epochs": 2, | |
| "total_flos": 6.728145843607406e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |