{"step": 0, "loss": 0.347229927778244, "lr": 1e-05, "grad_norm": 1.322141408920288, "tokens": 148, "tok_per_s": 37.75254109902415, "t": 5.5} {"step": 50, "loss": 0.10128622502088547, "lr": 0.00051, "grad_norm": 0.4043615758419037, "tokens": 7949, "tok_per_s": 192.99241679614582, "t": 42.77} {"step": 99, "val_loss_b8": 0.7306965569442118, "t": 90.8} {"step": 100, "loss": 0.016504138708114624, "lr": 0.001, "grad_norm": 0.1634058803319931, "tokens": 15744, "tok_per_s": 174.6667163690665, "t": 91.71} {"step": 150, "loss": 0.005306841805577278, "lr": 0.0009779754323328191, "grad_norm": 0.016647804528474808, "tokens": 23618, "tok_per_s": 172.09258657480268, "t": 138.82} {"step": 199, "val_loss_b8": 1.1004879354152708, "t": 173.76} {"step": 200, "loss": 0.00241711363196373, "lr": 0.0009140576474687263, "grad_norm": 0.008768348954617977, "tokens": 31693, "tok_per_s": 183.42497445922888, "t": 174.36} {"step": 250, "loss": 0.0009894538670778275, "lr": 0.000814503363531613, "grad_norm": 0.0031006645876914263, "tokens": 39738, "tok_per_s": 193.07667895946577, "t": 207.39} {"step": 299, "val_loss_b8": 1.3594950490487114, "t": 254.77} {"step": 300, "loss": 0.0008304077200591564, "lr": 0.0006890576474687264, "grad_norm": 0.001891698339022696, "tokens": 47878, "tok_per_s": 188.51353850802386, "t": 255.56} {"step": 350, "loss": 0.00040095686563290656, "lr": 0.00055, "grad_norm": 0.0012410919880494475, "tokens": 55733, "tok_per_s": 185.6831282288836, "t": 301.73} {"step": 399, "val_loss_b8": 1.499051700142222, "t": 347.61} {"step": 400, "loss": 0.0004196925729047507, "lr": 0.0004109423525312738, "grad_norm": 0.0009135301224887371, "tokens": 63742, "tok_per_s": 183.67223822194268, "t": 348.62} {"step": 450, "loss": 0.0003603784425649792, "lr": 0.0002854966364683872, "grad_norm": 0.0007731085061095655, "tokens": 71807, "tok_per_s": 183.6377874356163, "t": 392.61} {"step": 499, "val_loss_b8": 1.6038084745772216, "t": 438.01} {"step": 500, "loss": 0.0006298388470895588, "lr": 0.00018594235253127373, "grad_norm": 0.0006691015441901982, "tokens": 79967, "tok_per_s": 182.82307496171111, "t": 438.98} {"step": 550, "loss": 0.000337488716468215, "lr": 0.00012202456766718093, "grad_norm": 0.0003788845206145197, "tokens": 87959, "tok_per_s": 183.06438353653638, "t": 482.06} {"step": 599, "loss": 0.0002673463022802025, "lr": 0.00010000888261473832, "grad_norm": 0.00025494094006717205, "tokens": 95773, "tok_per_s": 185.19145987663353, "t": 518.73} {"step": 599, "val_loss_b8": 1.6449561009545786, "t": 519.47} {"step": 599, "checkpoint_sha256": "bb9dc4e9f2ffa760b97cb1d16b164ff6dc35c9f7c3d1c4d1ee8e375ec7a99914", "t": 519.6}