1{"step": 0, "loss": 0.8655945062637329, "lr": 1e-05, "grad_norm": 12.048591613769531, "tokens": 139, "tok_per_s": 42.131988327797835, "t": 4.25}2{"step": 50, "loss": 0.3399727940559387, "lr": 0.00051, "grad_norm": 0.7269741296768188, "tokens": 8357, "tok_per_s": 69.32700399913416, "t": 121.5}3{"step": 100, "loss": 0.3171934187412262, "lr": 0.001, "grad_norm": 0.19586065411567688, "tokens": 16659, "tok_per_s": 66.53606252646155, "t": 251.33}4{"step": 150, "loss": 0.4541836082935333, "lr": 0.0009993400362330058, "grad_norm": 0.0716216191649437, "tokens": 25169, "tok_per_s": 65.85193566819649, "t": 383.16}5{"step": 200, "loss": 0.480942040681839, "lr": 0.0009973620807194618, "grad_norm": 0.07926613092422485, "tokens": 33572, "tok_per_s": 65.41533157047647, "t": 514.17}6{"step": 250, "loss": 0.4172719120979309, "lr": 0.000994071935143687, "grad_norm": 0.058588795363903046, "tokens": 41743, "tok_per_s": 66.36525449217173, "t": 629.94}7{"step": 300, "loss": 0.39874881505966187, "lr": 0.000989479250069539, "grad_norm": 0.027894267812371254, "tokens": 49922, "tok_per_s": 66.92025942184642, "t": 746.95}8{"step": 350, "loss": 0.3876688778400421, "lr": 0.0009835974966336503, "grad_norm": 0.06466042995452881, "tokens": 58442, "tok_per_s": 69.27683611152624, "t": 844.55}9