metrics @ step 32

runs/metrics.jsonl

@@ -3,3 +3,5 @@
 {"type": "train", "step": 10, "epoch": 0.1739, "loss": 2.5477417839898004, "train_ppl": 12.778215206455922, "grad_norm": 28.139535903930664, "lr": 0.0, "elapsed_h": 0.053, "budget_h": 0.5}
 {"type": "train", "step": 20, "epoch": 0.3478, "loss": 1.5902856845781208, "train_ppl": 4.9051500539008766, "mtp_loss": null, "grad_norm": 555.1751235169178, "lr": 0.00014064175664315652, "tokens_per_sec": 297.7, "gpu_mem_gb": 9.71, "elapsed_h": 0.076, "budget_h": 0.5}
 {"type": "train", "step": 20, "epoch": 0.3478, "loss": 0.7900997161865234, "train_ppl": 2.203616151499862, "grad_norm": 12.04226303100586, "lr": 0.0001425726650926814, "elapsed_h": 0.076, "budget_h": 0.5}
+{"type": "train", "step": 30, "epoch": 0.5217, "loss": 0.1879760754434392, "train_ppl": 1.2068046427382915, "mtp_loss": null, "grad_norm": 527.797155470124, "lr": 0.00011127891653289297, "tokens_per_sec": 374.1, "gpu_mem_gb": 8.98, "elapsed_h": 0.099, "budget_h": 0.5}
+{"type": "train", "step": 30, "epoch": 0.5217, "loss": 0.15211989879608154, "train_ppl": 1.1642998262110626, "grad_norm": 3.499764919281006, "lr": 0.00011490240573865023, "elapsed_h": 0.099, "budget_h": 0.5}