{ "config": { "epochs": 50, "lr": 0.0003, "warmup_steps": 500, "grad_accum_steps": 8, "grad_clip": 1.0 }, "train": [ { "epoch": 1, "avg_loss": 6.387481, "lr": 0.00029999490289274813, "step": 574, "time_s": 2632.4 }, { "epoch": 2, "avg_loss": 4.254239, "lr": 0.00029960931729134234, "step": 1148, "time_s": 2630.6 }, { "epoch": 3, "avg_loss": 3.588417, "lr": 0.0002986121760707165, "step": 1722, "time_s": 2626.8 }, { "epoch": 4, "avg_loss": 3.170827, "lr": 0.00029700755522912847, "step": 2296, "time_s": 2631.8 }, { "epoch": 5, "avg_loss": 2.897759, "lr": 0.00029480201394959006, "step": 2870, "time_s": 2629.0 }, { "epoch": 6, "avg_loss": 2.720968, "lr": 0.00029200456778799946, "step": 3444, "time_s": 2626.1 }, { "epoch": 7, "avg_loss": 2.592565, "lr": 0.00028862665182039697, "step": 4018, "time_s": 2626.3 } ], "val": [ { "epoch": 1, "avg_loss": 4.748258 }, { "epoch": 2, "avg_loss": 3.860872 }, { "epoch": 3, "avg_loss": 3.437115 }, { "epoch": 4, "avg_loss": 3.278911 }, { "epoch": 5, "avg_loss": 3.129724 }, { "epoch": 6, "avg_loss": 3.133365 }, { "epoch": 7, "avg_loss": 3.135211 } ], "sanity": [] }