timestamp	step	epoch	train_loss	grad_norm	learning_rate
2025-09-21T21:41:21.703893	10	0.32	903.355000	614.498596	1.00e-07
2025-09-21T21:41:27.856246	20	0.65	902.448600	617.974670	5.00e-07
2025-09-21T21:41:33.781063	30	0.97	886.335200	578.719543	1.00e-06
2025-09-21T21:41:39.650969	40	1.29	941.081400	763.962463	1.50e-06
2025-09-21T21:41:45.448703	50	1.61	878.512000	745.441284	2.00e-06
2025-09-21T21:41:46.303569	50	1.61	NA	NA	NA
2025-09-21T21:41:53.272236	60	1.94	870.239400	805.721558	2.50e-06
2025-09-21T21:41:58.923912	70	2.26	883.066600	810.874512	3.00e-06
2025-09-21T21:42:04.647459	80	2.58	858.037100	876.961365	3.50e-06
2025-09-21T21:42:10.439181	90	2.90	879.551700	879.198303	4.00e-06
2025-09-21T21:42:16.088489	100	3.23	838.324500	1011.899841	4.50e-06
2025-09-21T21:42:16.862886	100	3.23	NA	NA	NA
2025-09-21T21:42:24.087341	110	3.55	789.837500	1295.429565	5.00e-06
2025-09-21T21:42:29.817688	120	3.87	745.176600	1823.199463	5.50e-06
2025-09-21T21:42:35.473868	130	4.19	701.951900	1844.962524	6.00e-06
2025-09-21T21:42:41.363224	140	4.52	661.328600	1836.961670	6.50e-06
2025-09-21T21:42:47.037840	150	4.84	558.101000	1857.315308	7.00e-06
2025-09-21T21:42:47.769415	150	4.84	NA	NA	NA
2025-09-21T21:42:54.652980	160	5.16	499.938000	2131.156982	7.50e-06
2025-09-21T21:43:00.440291	170	5.48	450.506500	1810.863647	8.00e-06
2025-09-21T21:43:06.233803	180	5.81	390.488400	1685.968994	8.50e-06
2025-09-21T21:43:12.164436	190	6.13	324.997900	1826.587402	9.00e-06
2025-09-21T21:43:17.924886	200	6.45	280.306800	1618.513672	9.50e-06
2025-09-21T21:43:18.676823	200	6.45	NA	NA	NA
2025-09-21T21:43:20.259029	200	6.45	NA	NA	NA