{"step": 0, "loss": 0.5386869311332703, "lr": 1e-05, "grad_norm": 2.047222137451172, "tokens": 157, "tok_per_s": 28.228840149849468, "t": 6.59} {"step": 50, "loss": 0.4027070701122284, "lr": 0.00051, "grad_norm": 0.6417726278305054, "tokens": 8616, "tok_per_s": 68.34210160186906, "t": 127.1} {"step": 100, "loss": 0.388218492269516, "lr": 0.001, "grad_norm": 0.24445490539073944, "tokens": 16635, "tok_per_s": 68.62855923767432, "t": 243.41} {"step": 150, "loss": 0.5297861695289612, "lr": 0.0009993400362330058, "grad_norm": 0.09588441252708435, "tokens": 25042, "tok_per_s": 69.75453698680255, "t": 360.02} {"step": 200, "loss": 0.4561012089252472, "lr": 0.0009973620807194618, "grad_norm": 0.026288924738764763, "tokens": 33661, "tok_per_s": 71.32438513252636, "t": 472.96} {"step": 250, "loss": 0.4421783685684204, "lr": 0.000994071935143687, "grad_norm": 0.12178074568510056, "tokens": 42410, "tok_per_s": 74.62962531642891, "t": 569.29} {"step": 300, "loss": 0.39179977774620056, "lr": 0.000989479250069539, "grad_norm": 0.14763128757476807, "tokens": 50771, "tok_per_s": 76.4380470849866, "t": 665.23} {"step": 350, "loss": 0.4274325370788574, "lr": 0.0009835974966336503, "grad_norm": 0.0800609216094017, "tokens": 59133, "tok_per_s": 77.8408447120687, "t": 760.68} {"step": 400, "loss": 0.5898694396018982, "lr": 0.0009764439270322612, "grad_norm": 0.036028724163770676, "tokens": 67299, "tok_per_s": 78.70071363024259, "t": 856.14}