{ "config": { "epochs": 50, "lr": 0.0003, "warmup_steps": 500, "grad_accum_steps": 8, "grad_clip": 1.0 }, "train": [ { "epoch": 1, "avg_loss": 5.556523, "lr": 0.00029999464494118047, "step": 576, "time_s": 2699.1 }, { "epoch": 2, "avg_loss": 4.122129, "lr": 0.0002996060467905289, "step": 1152, "time_s": 2699.7 }, { "epoch": 3, "avg_loss": 3.51989, "lr": 0.00029860407648151763, "step": 1728, "time_s": 2700.1 }, { "epoch": 4, "avg_loss": 3.177908, "lr": 0.0002969928420078021, "step": 2304, "time_s": 2700.8 }, { "epoch": 5, "avg_loss": 2.955399, "lr": 0.0002947789492946639, "step": 2880, "time_s": 2701.4 }, { "epoch": 6, "avg_loss": 2.745658, "lr": 0.00029197147511527473, "step": 3456, "time_s": 2702.1 }, { "epoch": 7, "avg_loss": 2.583645, "lr": 0.000288581929876693, "step": 4032, "time_s": 2699.6 }, { "epoch": 8, "avg_loss": 2.457457, "lr": 0.0002846242104281667, "step": 4608, "time_s": 2703.2 }, { "epoch": 9, "avg_loss": 2.354412, "lr": 0.00028011454308522526, "step": 5184, "time_s": 2709.7 }, { "epoch": 10, "avg_loss": 2.266502, "lr": 0.0002750714171031559, "step": 5760, "time_s": 2708.4 }, { "epoch": 11, "avg_loss": 2.18637, "lr": 0.0002695155088726183, "step": 6336, "time_s": 2703.1 }, { "epoch": 12, "avg_loss": 2.112232, "lr": 0.00026346959714818914, "step": 6912, "time_s": 2705.5 } ], "val": [ { "epoch": 1, "avg_loss": 4.572999 }, { "epoch": 2, "avg_loss": 3.741502 }, { "epoch": 3, "avg_loss": 3.292827 }, { "epoch": 4, "avg_loss": 3.067192 }, { "epoch": 5, "avg_loss": 2.852124 }, { "epoch": 6, "avg_loss": 2.683327 }, { "epoch": 7, "avg_loss": 2.568512 }, { "epoch": 8, "avg_loss": 2.506051 }, { "epoch": 9, "avg_loss": 2.45621 }, { "epoch": 10, "avg_loss": 2.405939 }, { "epoch": 11, "avg_loss": 2.374089 }, { "epoch": 12, "avg_loss": 2.360602 } ], "sanity": [] }