{"step": 0, "epoch": 0, "elapsed_seconds": 115.495145448, "eval_loss": 0.5684167742729187, "eval_model_preparation_time": 0.0304, "eval_runtime": 41.2126, "eval_samples_per_second": 3.543, "eval_steps_per_second": 3.543}
{"step": 5, "epoch": 0.03638017280582083, "elapsed_seconds": 251.60456207000001, "loss": 0.4317525863647461, "grad_norm": 24.30816078186035, "learning_rate": 2.857142857142857e-05}
{"step": 10, "epoch": 0.07276034561164166, "elapsed_seconds": 382.98394259500003, "loss": 0.12554776668548584, "grad_norm": 2.216491460800171, "learning_rate": 6.428571428571429e-05}
{"step": 15, "epoch": 0.10914051841746249, "elapsed_seconds": 523.1122485459999, "loss": 0.04251816570758819, "grad_norm": 0.0014739494072273374, "learning_rate": 0.0001}
{"step": 20, "epoch": 0.14552069122328332, "elapsed_seconds": 654.207221639, "loss": 0.25625, "grad_norm": 3.857721087019428e-11, "learning_rate": 9.991016468658499e-05}
{"step": 25, "epoch": 0.18190086402910413, "elapsed_seconds": 786.3623533809999, "loss": 0.0, "grad_norm": 2.001864993217417e-11, "learning_rate": 9.964098156168142e-05}
{"step": 25, "epoch": 0.18190086402910413, "elapsed_seconds": 829.06631217, "eval_loss": 0.0, "eval_model_preparation_time": 0.0304, "eval_runtime": 42.0163, "eval_samples_per_second": 3.475, "eval_steps_per_second": 3.475}
{"step": 30, "epoch": 0.21828103683492497, "elapsed_seconds": 967.551984536, "loss": 0.0, "grad_norm": 1.9943463547278384e-10, "learning_rate": 9.919341791130496e-05}
{"step": 35, "epoch": 0.2546612096407458, "elapsed_seconds": 1097.494034193, "loss": 0.0, "grad_norm": 1.46394452116283e-10, "learning_rate": 9.85690820162878e-05}
{"step": 40, "epoch": 0.29104138244656663, "elapsed_seconds": 1240.2977030739999, "loss": 0.0, "grad_norm": 1.4816532722949916e-10, "learning_rate": 9.777021737306214e-05}
{"step": 45, "epoch": 0.3274215552523875, "elapsed_seconds": 1378.815700288, "loss": 0.0, "grad_norm": 2.9816080160394165e-10, "learning_rate": 9.6799694631852e-05}
{"step": 50, "epoch": 0.36380172805820826, "elapsed_seconds": 1513.959694255, "loss": 0.0, "grad_norm": 3.788522828784835e-10, "learning_rate": 9.56610012812427e-05}
{"step": 50, "epoch": 0.36380172805820826, "elapsed_seconds": 1557.517019976, "eval_loss": 0.0, "eval_model_preparation_time": 0.0304, "eval_runtime": 42.7782, "eval_samples_per_second": 3.413, "eval_steps_per_second": 3.413}
{"step": 55, "epoch": 0.4001819008640291, "elapsed_seconds": 1694.631363275, "loss": 0.0, "grad_norm": 3.8055242290724323e-10, "learning_rate": 9.435822911619564e-05}
{"step": 60, "epoch": 0.43656207366984995, "elapsed_seconds": 1826.859244014, "loss": 0.0, "grad_norm": 5.153630300291923e-10, "learning_rate": 9.289605953454107e-05}
{"step": 65, "epoch": 0.47294224647567074, "elapsed_seconds": 1959.234216417, "loss": 0.0, "grad_norm": 4.723861302124988e-10, "learning_rate": 9.127974671478432e-05}
{"step": 70, "epoch": 0.5093224192814916, "elapsed_seconds": 2093.121699353, "loss": 0.0, "grad_norm": 3.3545297095649573e-10, "learning_rate": 8.951509873567499e-05}
{"step": 75, "epoch": 0.5457025920873124, "elapsed_seconds": 2223.87149162, "loss": 0.0, "grad_norm": 4.869515346506148e-10, "learning_rate": 8.760845670538387e-05}
{"step": 75, "epoch": 0.5457025920873124, "elapsed_seconds": 2266.9824363, "eval_loss": 0.0, "eval_model_preparation_time": 0.0304, "eval_runtime": 42.2002, "eval_samples_per_second": 3.46, "eval_steps_per_second": 3.46}
{"step": 80, "epoch": 0.5820827648931333, "elapsed_seconds": 2406.064981508, "loss": 0.0, "grad_norm": 6.026958376814662e-10, "learning_rate": 8.556667197528543e-05}
{"step": 85, "epoch": 0.6184629376989541, "elapsed_seconds": 2538.350284011, "loss": 0.0, "grad_norm": 7.610323482509784e-10, "learning_rate": 8.339708152022585e-05}
{"step": 90, "epoch": 0.654843110504775, "elapsed_seconds": 2677.335809601, "loss": 0.0, "grad_norm": 4.122030772268914e-10, "learning_rate": 8.110748157374565e-05}
{"step": 95, "epoch": 0.6912232833105957, "elapsed_seconds": 2809.3651115459998, "loss": 0.0, "grad_norm": 6.49666598295795e-10, "learning_rate": 7.870609961299627e-05}
{"step": 100, "epoch": 0.7276034561164165, "elapsed_seconds": 2940.535695941, "loss": 0.0, "grad_norm": 6.920322648262811e-10, "learning_rate": 7.620156479402066e-05}
{"step": 100, "epoch": 0.7276034561164165, "elapsed_seconds": 2983.927569312, "eval_loss": 0.0, "eval_model_preparation_time": 0.0304, "eval_runtime": 42.6124, "eval_samples_per_second": 3.426, "eval_steps_per_second": 3.426}
{"step": 105, "epoch": 0.7639836289222374, "elapsed_seconds": 3121.164387972, "loss": 0.0, "grad_norm": 5.246895695698583e-10, "learning_rate": 7.360287694363566e-05}
{"step": 110, "epoch": 0.8003638017280582, "elapsed_seconds": 3254.451429533, "loss": 0.0, "grad_norm": 5.858165619265776e-10, "learning_rate": 7.091937421934157e-05}
{"step": 115, "epoch": 0.836743974533879, "elapsed_seconds": 3384.57085055, "loss": 0.0, "grad_norm": 2.461385817387196e-10, "learning_rate": 6.816069955346985e-05}
{"step": 120, "epoch": 0.8731241473396999, "elapsed_seconds": 3512.622943047, "loss": 0.0, "grad_norm": 2.6043153744659264e-10, "learning_rate": 6.533676600214928e-05}
{"step": 125, "epoch": 0.9095043201455207, "elapsed_seconds": 3642.251300354, "loss": 0.0, "grad_norm": 3.8649006217639226e-10, "learning_rate": 6.245772112360568e-05}
{"step": 125, "epoch": 0.9095043201455207, "elapsed_seconds": 3685.246071066, "eval_loss": 0.0, "eval_model_preparation_time": 0.0304, "eval_runtime": 42.0667, "eval_samples_per_second": 3.471, "eval_steps_per_second": 3.471}
{"step": 130, "epoch": 0.9458844929513415, "elapsed_seconds": 3819.478919847, "loss": 0.27734487056732177, "grad_norm": 0.0006892899982631207, "learning_rate": 5.953391051379904e-05}
{"step": 135, "epoch": 0.9822646657571623, "elapsed_seconds": 3952.3443189189998, "loss": 0.0003256572876125574, "grad_norm": 0.008191271685063839, "learning_rate": 5.6575840630429286e-05}
{"step": 140, "epoch": 1.0145520691223284, "elapsed_seconds": 4067.849474322, "loss": 2.2244230785872786e-05, "grad_norm": 0.0006689362344332039, "learning_rate": 5.359414103889947e-05}
{"step": 145, "epoch": 1.050932241928149, "elapsed_seconds": 4199.778645104, "loss": 4.622311826096847e-06, "grad_norm": 0.0003005589824169874, "learning_rate": 5.059952621590216e-05}