| {"current_steps": 20, "total_steps": 3100, "loss": 2.8022, "learning_rate": 3.064516129032258e-06, "epoch": 0.32, "percentage": 0.65, "elapsed_time": "0:01:15", "remaining_time": "3:13:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 40, "total_steps": 3100, "loss": 2.6463, "learning_rate": 6.290322580645161e-06, "epoch": 0.64, "percentage": 1.29, "elapsed_time": "0:02:28", "remaining_time": "3:09:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 60, "total_steps": 3100, "loss": 2.4026, "learning_rate": 9.516129032258064e-06, "epoch": 0.96, "percentage": 1.94, "elapsed_time": "0:03:42", "remaining_time": "3:07:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 80, "total_steps": 3100, "loss": 1.9639, "learning_rate": 1.2741935483870968e-05, "epoch": 1.28, "percentage": 2.58, "elapsed_time": "0:04:54", "remaining_time": "3:05:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 100, "total_steps": 3100, "loss": 1.7115, "learning_rate": 1.596774193548387e-05, "epoch": 1.6, "percentage": 3.23, "elapsed_time": "0:06:08", "remaining_time": "3:04:08", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 120, "total_steps": 3100, "loss": 1.7534, "learning_rate": 1.9193548387096774e-05, "epoch": 1.92, "percentage": 3.87, "elapsed_time": "0:07:23", "remaining_time": "3:03:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 140, "total_steps": 3100, "loss": 1.3422, "learning_rate": 2.2419354838709678e-05, "epoch": 2.24, "percentage": 4.52, "elapsed_time": "0:08:37", "remaining_time": "3:02:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 160, "total_steps": 3100, "loss": 1.1582, "learning_rate": 2.5645161290322582e-05, "epoch": 2.56, "percentage": 5.16, "elapsed_time": "0:09:48", "remaining_time": "3:00:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 180, "total_steps": 3100, "loss": 1.1084, "learning_rate": 2.8870967741935483e-05, "epoch": 2.88, "percentage": 5.81, "elapsed_time": "0:10:59", "remaining_time": "2:58:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 200, "total_steps": 3100, "loss": 0.844, "learning_rate": 3.2096774193548393e-05, "epoch": 3.2, "percentage": 6.45, "elapsed_time": "0:12:12", "remaining_time": "2:56:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 220, "total_steps": 3100, "loss": 0.6051, "learning_rate": 3.532258064516129e-05, "epoch": 3.52, "percentage": 7.1, "elapsed_time": "0:13:22", "remaining_time": "2:55:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 240, "total_steps": 3100, "loss": 0.7587, "learning_rate": 3.8548387096774195e-05, "epoch": 3.84, "percentage": 7.74, "elapsed_time": "0:14:34", "remaining_time": "2:53:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 260, "total_steps": 3100, "loss": 0.537, "learning_rate": 4.17741935483871e-05, "epoch": 4.16, "percentage": 8.39, "elapsed_time": "0:15:46", "remaining_time": "2:52:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 280, "total_steps": 3100, "loss": 0.4063, "learning_rate": 4.5e-05, "epoch": 4.48, "percentage": 9.03, "elapsed_time": "0:16:59", "remaining_time": "2:51:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 300, "total_steps": 3100, "loss": 0.4817, "learning_rate": 4.822580645161291e-05, "epoch": 4.8, "percentage": 9.68, "elapsed_time": "0:18:11", "remaining_time": "2:49:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 320, "total_steps": 3100, "loss": 0.3556, "learning_rate": 4.9998716243505096e-05, "epoch": 5.12, "percentage": 10.32, "elapsed_time": "0:19:23", "remaining_time": "2:48:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 340, "total_steps": 3100, "loss": 0.2663, "learning_rate": 4.9986672191133314e-05, "epoch": 5.44, "percentage": 10.97, "elapsed_time": "0:20:35", "remaining_time": "2:47:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 360, "total_steps": 3100, "loss": 0.3273, "learning_rate": 4.9961956248762694e-05, "epoch": 5.76, "percentage": 11.61, "elapsed_time": "0:21:46", "remaining_time": "2:45:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 380, "total_steps": 3100, "loss": 0.2174, "learning_rate": 4.992458095098368e-05, "epoch": 6.08, "percentage": 12.26, "elapsed_time": "0:22:57", "remaining_time": "2:44:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 400, "total_steps": 3100, "loss": 0.1886, "learning_rate": 4.9874565252527765e-05, "epoch": 6.4, "percentage": 12.9, "elapsed_time": "0:24:11", "remaining_time": "2:43:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 420, "total_steps": 3100, "loss": 0.2278, "learning_rate": 4.981193451865465e-05, "epoch": 6.72, "percentage": 13.55, "elapsed_time": "0:25:24", "remaining_time": "2:42:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 440, "total_steps": 3100, "loss": 0.168, "learning_rate": 4.9736720512288334e-05, "epoch": 7.04, "percentage": 14.19, "elapsed_time": "0:26:37", "remaining_time": "2:40:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 460, "total_steps": 3100, "loss": 0.1227, "learning_rate": 4.964896137790873e-05, "epoch": 7.36, "percentage": 14.84, "elapsed_time": "0:27:47", "remaining_time": "2:39:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 480, "total_steps": 3100, "loss": 0.1261, "learning_rate": 4.954870162220679e-05, "epoch": 7.68, "percentage": 15.48, "elapsed_time": "0:28:59", "remaining_time": "2:38:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 500, "total_steps": 3100, "loss": 0.1167, "learning_rate": 4.943599209151314e-05, "epoch": 8.0, "percentage": 16.13, "elapsed_time": "0:30:11", "remaining_time": "2:36:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 520, "total_steps": 3100, "loss": 0.1049, "learning_rate": 4.931088994601157e-05, "epoch": 8.32, "percentage": 16.77, "elapsed_time": "0:31:20", "remaining_time": "2:35:32", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 540, "total_steps": 3100, "loss": 0.1016, "learning_rate": 4.917345863075048e-05, "epoch": 8.64, "percentage": 17.42, "elapsed_time": "0:32:32", "remaining_time": "2:34:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 560, "total_steps": 3100, "loss": 0.1157, "learning_rate": 4.902376784346697e-05, "epoch": 8.96, "percentage": 18.06, "elapsed_time": "0:33:48", "remaining_time": "2:33:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 580, "total_steps": 3100, "loss": 0.073, "learning_rate": 4.886189349923992e-05, "epoch": 9.28, "percentage": 18.71, "elapsed_time": "0:35:01", "remaining_time": "2:32:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 600, "total_steps": 3100, "loss": 0.0908, "learning_rate": 4.868791769198995e-05, "epoch": 9.6, "percentage": 19.35, "elapsed_time": "0:36:13", "remaining_time": "2:30:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 620, "total_steps": 3100, "loss": 0.0557, "learning_rate": 4.8501928652845854e-05, "epoch": 9.92, "percentage": 20.0, "elapsed_time": "0:37:22", "remaining_time": "2:29:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 640, "total_steps": 3100, "loss": 0.0779, "learning_rate": 4.83040207053985e-05, "epoch": 10.24, "percentage": 20.65, "elapsed_time": "0:38:33", "remaining_time": "2:28:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 660, "total_steps": 3100, "loss": 0.048, "learning_rate": 4.809429421786502e-05, "epoch": 10.56, "percentage": 21.29, "elapsed_time": "0:39:43", "remaining_time": "2:26:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 680, "total_steps": 3100, "loss": 0.0747, "learning_rate": 4.787285555218748e-05, "epoch": 10.88, "percentage": 21.94, "elapsed_time": "0:40:54", "remaining_time": "2:25:35", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 700, "total_steps": 3100, "loss": 0.0629, "learning_rate": 4.763981701009184e-05, "epoch": 11.2, "percentage": 22.58, "elapsed_time": "0:42:06", "remaining_time": "2:24:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 720, "total_steps": 3100, "loss": 0.051, "learning_rate": 4.739529677613456e-05, "epoch": 11.52, "percentage": 23.23, "elapsed_time": "0:43:21", "remaining_time": "2:23:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 740, "total_steps": 3100, "loss": 0.0699, "learning_rate": 4.713941885776586e-05, "epoch": 11.84, "percentage": 23.87, "elapsed_time": "0:44:34", "remaining_time": "2:22:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 760, "total_steps": 3100, "loss": 0.0526, "learning_rate": 4.687231302243975e-05, "epoch": 12.16, "percentage": 24.52, "elapsed_time": "0:45:46", "remaining_time": "2:20:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 780, "total_steps": 3100, "loss": 0.0412, "learning_rate": 4.659411473180304e-05, "epoch": 12.48, "percentage": 25.16, "elapsed_time": "0:46:59", "remaining_time": "2:19:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 800, "total_steps": 3100, "loss": 0.0495, "learning_rate": 4.6304965072996495e-05, "epoch": 12.8, "percentage": 25.81, "elapsed_time": "0:48:10", "remaining_time": "2:18:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 820, "total_steps": 3100, "loss": 0.063, "learning_rate": 4.6005010687103076e-05, "epoch": 13.12, "percentage": 26.45, "elapsed_time": "0:49:22", "remaining_time": "2:17:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 840, "total_steps": 3100, "loss": 0.0425, "learning_rate": 4.569440369477951e-05, "epoch": 13.44, "percentage": 27.1, "elapsed_time": "0:50:34", "remaining_time": "2:16:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 860, "total_steps": 3100, "loss": 0.0451, "learning_rate": 4.5373301619108854e-05, "epoch": 13.76, "percentage": 27.74, "elapsed_time": "0:51:47", "remaining_time": "2:14:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 880, "total_steps": 3100, "loss": 0.0445, "learning_rate": 4.5041867305713384e-05, "epoch": 14.08, "percentage": 28.39, "elapsed_time": "0:53:00", "remaining_time": "2:13:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 900, "total_steps": 3100, "loss": 0.0214, "learning_rate": 4.4700268840168045e-05, "epoch": 14.4, "percentage": 29.03, "elapsed_time": "0:54:14", "remaining_time": "2:12:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 920, "total_steps": 3100, "loss": 0.0552, "learning_rate": 4.4348679462756556e-05, "epoch": 14.72, "percentage": 29.68, "elapsed_time": "0:55:29", "remaining_time": "2:11:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 940, "total_steps": 3100, "loss": 0.0524, "learning_rate": 4.398727748061324e-05, "epoch": 15.04, "percentage": 30.32, "elapsed_time": "0:56:42", "remaining_time": "2:10:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 960, "total_steps": 3100, "loss": 0.0318, "learning_rate": 4.361624617729536e-05, "epoch": 15.36, "percentage": 30.97, "elapsed_time": "0:57:55", "remaining_time": "2:09:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 980, "total_steps": 3100, "loss": 0.0347, "learning_rate": 4.323577371983155e-05, "epoch": 15.68, "percentage": 31.61, "elapsed_time": "0:59:10", "remaining_time": "2:08:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1000, "total_steps": 3100, "loss": 0.0541, "learning_rate": 4.28460530632937e-05, "epoch": 16.0, "percentage": 32.26, "elapsed_time": "1:00:25", "remaining_time": "2:06:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1020, "total_steps": 3100, "loss": 0.0327, "learning_rate": 4.2447281852940525e-05, "epoch": 16.32, "percentage": 32.9, "elapsed_time": "1:01:40", "remaining_time": "2:05:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1040, "total_steps": 3100, "loss": 0.0251, "learning_rate": 4.203966232398261e-05, "epoch": 16.64, "percentage": 33.55, "elapsed_time": "1:02:52", "remaining_time": "2:04:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1060, "total_steps": 3100, "loss": 0.0451, "learning_rate": 4.162340119901961e-05, "epoch": 16.96, "percentage": 34.19, "elapsed_time": "1:04:05", "remaining_time": "2:03:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1080, "total_steps": 3100, "loss": 0.0272, "learning_rate": 4.1198709583201754e-05, "epoch": 17.28, "percentage": 34.84, "elapsed_time": "1:05:16", "remaining_time": "2:02:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1100, "total_steps": 3100, "loss": 0.0517, "learning_rate": 4.0765802857168687e-05, "epoch": 17.6, "percentage": 35.48, "elapsed_time": "1:06:29", "remaining_time": "2:00:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1120, "total_steps": 3100, "loss": 0.0286, "learning_rate": 4.0324900567820046e-05, "epoch": 17.92, "percentage": 36.13, "elapsed_time": "1:07:45", "remaining_time": "1:59:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1140, "total_steps": 3100, "loss": 0.0258, "learning_rate": 3.987622631697316e-05, "epoch": 18.24, "percentage": 36.77, "elapsed_time": "1:08:59", "remaining_time": "1:58:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1160, "total_steps": 3100, "loss": 0.0289, "learning_rate": 3.942000764796427e-05, "epoch": 18.56, "percentage": 37.42, "elapsed_time": "1:10:14", "remaining_time": "1:57:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1180, "total_steps": 3100, "loss": 0.0457, "learning_rate": 3.895647593025088e-05, "epoch": 18.88, "percentage": 38.06, "elapsed_time": "1:11:29", "remaining_time": "1:56:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1200, "total_steps": 3100, "loss": 0.0316, "learning_rate": 3.8485866242073584e-05, "epoch": 19.2, "percentage": 38.71, "elapsed_time": "1:12:42", "remaining_time": "1:55:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1220, "total_steps": 3100, "loss": 0.0326, "learning_rate": 3.80084172512372e-05, "epoch": 19.52, "percentage": 39.35, "elapsed_time": "1:13:55", "remaining_time": "1:53:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1240, "total_steps": 3100, "loss": 0.0238, "learning_rate": 3.7524371094071266e-05, "epoch": 19.84, "percentage": 40.0, "elapsed_time": "1:15:07", "remaining_time": "1:52:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1260, "total_steps": 3100, "loss": 0.0286, "learning_rate": 3.703397325263162e-05, "epoch": 20.16, "percentage": 40.65, "elapsed_time": "1:16:19", "remaining_time": "1:51:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1280, "total_steps": 3100, "loss": 0.0294, "learning_rate": 3.653747243020515e-05, "epoch": 20.48, "percentage": 41.29, "elapsed_time": "1:17:32", "remaining_time": "1:50:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1300, "total_steps": 3100, "loss": 0.0364, "learning_rate": 3.603512042518093e-05, "epoch": 20.8, "percentage": 41.94, "elapsed_time": "1:18:45", "remaining_time": "1:49:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1320, "total_steps": 3100, "loss": 0.0265, "learning_rate": 3.552717200335171e-05, "epoch": 21.12, "percentage": 42.58, "elapsed_time": "1:19:57", "remaining_time": "1:47:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1340, "total_steps": 3100, "loss": 0.0319, "learning_rate": 3.501388476871039e-05, "epoch": 21.44, "percentage": 43.23, "elapsed_time": "1:21:08", "remaining_time": "1:46:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1360, "total_steps": 3100, "loss": 0.0137, "learning_rate": 3.449551903280729e-05, "epoch": 21.76, "percentage": 43.87, "elapsed_time": "1:22:19", "remaining_time": "1:45:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1380, "total_steps": 3100, "loss": 0.0416, "learning_rate": 3.397233768273415e-05, "epoch": 22.08, "percentage": 44.52, "elapsed_time": "1:23:30", "remaining_time": "1:44:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1400, "total_steps": 3100, "loss": 0.0179, "learning_rate": 3.344460604780202e-05, "epoch": 22.4, "percentage": 45.16, "elapsed_time": "1:24:42", "remaining_time": "1:42:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1420, "total_steps": 3100, "loss": 0.0276, "learning_rate": 3.291259176498052e-05, "epoch": 22.72, "percentage": 45.81, "elapsed_time": "1:25:52", "remaining_time": "1:41:35", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1440, "total_steps": 3100, "loss": 0.0352, "learning_rate": 3.237656464316693e-05, "epoch": 23.04, "percentage": 46.45, "elapsed_time": "1:27:01", "remaining_time": "1:40:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1460, "total_steps": 3100, "loss": 0.0212, "learning_rate": 3.183679652635357e-05, "epoch": 23.36, "percentage": 47.1, "elapsed_time": "1:28:11", "remaining_time": "1:39:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1480, "total_steps": 3100, "loss": 0.0338, "learning_rate": 3.129356115576332e-05, "epoch": 23.68, "percentage": 47.74, "elapsed_time": "1:29:24", "remaining_time": "1:37:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1500, "total_steps": 3100, "loss": 0.0295, "learning_rate": 3.074713403102284e-05, "epoch": 24.0, "percentage": 48.39, "elapsed_time": "1:30:35", "remaining_time": "1:36:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1520, "total_steps": 3100, "loss": 0.0185, "learning_rate": 3.0197792270443982e-05, "epoch": 24.32, "percentage": 49.03, "elapsed_time": "1:31:46", "remaining_time": "1:35:24", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1540, "total_steps": 3100, "loss": 0.0328, "learning_rate": 2.9645814470484452e-05, "epoch": 24.64, "percentage": 49.68, "elapsed_time": "1:32:59", "remaining_time": "1:34:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1560, "total_steps": 3100, "loss": 0.025, "learning_rate": 2.9091480564458666e-05, "epoch": 24.96, "percentage": 50.32, "elapsed_time": "1:34:12", "remaining_time": "1:32:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1580, "total_steps": 3100, "loss": 0.0294, "learning_rate": 2.8535071680570734e-05, "epoch": 25.28, "percentage": 50.97, "elapsed_time": "1:35:23", "remaining_time": "1:31:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1600, "total_steps": 3100, "loss": 0.0282, "learning_rate": 2.7976869999341426e-05, "epoch": 25.6, "percentage": 51.61, "elapsed_time": "1:36:38", "remaining_time": "1:30:35", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1620, "total_steps": 3100, "loss": 0.0294, "learning_rate": 2.741715861050143e-05, "epoch": 25.92, "percentage": 52.26, "elapsed_time": "1:37:51", "remaining_time": "1:29:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1640, "total_steps": 3100, "loss": 0.0354, "learning_rate": 2.685622136942359e-05, "epoch": 26.24, "percentage": 52.9, "elapsed_time": "1:39:04", "remaining_time": "1:28:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1660, "total_steps": 3100, "loss": 0.0162, "learning_rate": 2.629434275316673e-05, "epoch": 26.56, "percentage": 53.55, "elapsed_time": "1:40:19", "remaining_time": "1:27:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1680, "total_steps": 3100, "loss": 0.0205, "learning_rate": 2.573180771620432e-05, "epoch": 26.88, "percentage": 54.19, "elapsed_time": "1:41:32", "remaining_time": "1:25:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1700, "total_steps": 3100, "loss": 0.0129, "learning_rate": 2.516890154591095e-05, "epoch": 27.2, "percentage": 54.84, "elapsed_time": "1:42:47", "remaining_time": "1:24:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1720, "total_steps": 3100, "loss": 0.0333, "learning_rate": 2.4605909717879964e-05, "epoch": 27.52, "percentage": 55.48, "elapsed_time": "1:44:01", "remaining_time": "1:23:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1740, "total_steps": 3100, "loss": 0.0261, "learning_rate": 2.4043117751145694e-05, "epoch": 27.84, "percentage": 56.13, "elapsed_time": "1:45:14", "remaining_time": "1:22:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1760, "total_steps": 3100, "loss": 0.0418, "learning_rate": 2.34808110633836e-05, "epoch": 28.16, "percentage": 56.77, "elapsed_time": "1:46:27", "remaining_time": "1:21:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1780, "total_steps": 3100, "loss": 0.0111, "learning_rate": 2.291927482616191e-05, "epoch": 28.48, "percentage": 57.42, "elapsed_time": "1:47:39", "remaining_time": "1:19:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1800, "total_steps": 3100, "loss": 0.0263, "learning_rate": 2.235879382031794e-05, "epoch": 28.8, "percentage": 58.06, "elapsed_time": "1:48:52", "remaining_time": "1:18:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1820, "total_steps": 3100, "loss": 0.0299, "learning_rate": 2.179965229153265e-05, "epoch": 29.12, "percentage": 58.71, "elapsed_time": "1:50:05", "remaining_time": "1:17:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1840, "total_steps": 3100, "loss": 0.0267, "learning_rate": 2.1242133806176667e-05, "epoch": 29.44, "percentage": 59.35, "elapsed_time": "1:51:16", "remaining_time": "1:16:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1860, "total_steps": 3100, "loss": 0.0204, "learning_rate": 2.0686521107500638e-05, "epoch": 29.76, "percentage": 60.0, "elapsed_time": "1:52:29", "remaining_time": "1:14:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1880, "total_steps": 3100, "loss": 0.0308, "learning_rate": 2.0133095972243233e-05, "epoch": 30.08, "percentage": 60.65, "elapsed_time": "1:53:42", "remaining_time": "1:13:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1900, "total_steps": 3100, "loss": 0.024, "learning_rate": 1.9582139067729117e-05, "epoch": 30.4, "percentage": 61.29, "elapsed_time": "1:54:52", "remaining_time": "1:12:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1920, "total_steps": 3100, "loss": 0.0113, "learning_rate": 1.90339298095297e-05, "epoch": 30.72, "percentage": 61.94, "elapsed_time": "1:56:03", "remaining_time": "1:11:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1940, "total_steps": 3100, "loss": 0.0301, "learning_rate": 1.8488746219758674e-05, "epoch": 31.04, "percentage": 62.58, "elapsed_time": "1:57:14", "remaining_time": "1:10:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1960, "total_steps": 3100, "loss": 0.0293, "learning_rate": 1.7946864786074165e-05, "epoch": 31.36, "percentage": 63.23, "elapsed_time": "1:58:26", "remaining_time": "1:08:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 1980, "total_steps": 3100, "loss": 0.0242, "learning_rate": 1.740856032145917e-05, "epoch": 31.68, "percentage": 63.87, "elapsed_time": "1:59:35", "remaining_time": "1:07:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2000, "total_steps": 3100, "loss": 0.022, "learning_rate": 1.6874105824851267e-05, "epoch": 32.0, "percentage": 64.52, "elapsed_time": "2:00:48", "remaining_time": "1:06:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2020, "total_steps": 3100, "loss": 0.0264, "learning_rate": 1.634377234269226e-05, "epoch": 32.32, "percentage": 65.16, "elapsed_time": "2:02:02", "remaining_time": "1:05:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2040, "total_steps": 3100, "loss": 0.0155, "learning_rate": 1.5817828831468144e-05, "epoch": 32.64, "percentage": 65.81, "elapsed_time": "2:03:17", "remaining_time": "1:04:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2060, "total_steps": 3100, "loss": 0.0208, "learning_rate": 1.5296542021308825e-05, "epoch": 32.96, "percentage": 66.45, "elapsed_time": "2:04:32", "remaining_time": "1:02:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2080, "total_steps": 3100, "loss": 0.0264, "learning_rate": 1.478017628071706e-05, "epoch": 33.28, "percentage": 67.1, "elapsed_time": "2:05:48", "remaining_time": "1:01:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2100, "total_steps": 3100, "loss": 0.0174, "learning_rate": 1.4268993482495055e-05, "epoch": 33.6, "percentage": 67.74, "elapsed_time": "2:07:02", "remaining_time": "1:00:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2120, "total_steps": 3100, "loss": 0.0214, "learning_rate": 1.3763252870936649e-05, "epoch": 33.92, "percentage": 68.39, "elapsed_time": "2:08:20", "remaining_time": "0:59:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2140, "total_steps": 3100, "loss": 0.0306, "learning_rate": 1.3263210930352737e-05, "epoch": 34.24, "percentage": 69.03, "elapsed_time": "2:09:32", "remaining_time": "0:58:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2160, "total_steps": 3100, "loss": 0.0146, "learning_rate": 1.2769121254996159e-05, "epoch": 34.56, "percentage": 69.68, "elapsed_time": "2:10:44", "remaining_time": "0:56:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2180, "total_steps": 3100, "loss": 0.0254, "learning_rate": 1.228123442045249e-05, "epoch": 34.88, "percentage": 70.32, "elapsed_time": "2:12:00", "remaining_time": "0:55:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2200, "total_steps": 3100, "loss": 0.0176, "learning_rate": 1.1799797856561606e-05, "epoch": 35.2, "percentage": 70.97, "elapsed_time": "2:13:13", "remaining_time": "0:54:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2220, "total_steps": 3100, "loss": 0.0205, "learning_rate": 1.1325055721934637e-05, "epoch": 35.52, "percentage": 71.61, "elapsed_time": "2:14:27", "remaining_time": "0:53:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2240, "total_steps": 3100, "loss": 0.0153, "learning_rate": 1.0857248780129928e-05, "epoch": 35.84, "percentage": 72.26, "elapsed_time": "2:15:41", "remaining_time": "0:52:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2260, "total_steps": 3100, "loss": 0.0251, "learning_rate": 1.0396614277550752e-05, "epoch": 36.16, "percentage": 72.9, "elapsed_time": "2:16:56", "remaining_time": "0:50:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2280, "total_steps": 3100, "loss": 0.0224, "learning_rate": 9.943385823126775e-06, "epoch": 36.48, "percentage": 73.55, "elapsed_time": "2:18:09", "remaining_time": "0:49:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2300, "total_steps": 3100, "loss": 0.0219, "learning_rate": 9.497793269840211e-06, "epoch": 36.8, "percentage": 74.19, "elapsed_time": "2:19:24", "remaining_time": "0:48:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2320, "total_steps": 3100, "loss": 0.021, "learning_rate": 9.06006259815683e-06, "epoch": 37.12, "percentage": 74.84, "elapsed_time": "2:20:38", "remaining_time": "0:47:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2340, "total_steps": 3100, "loss": 0.0236, "learning_rate": 8.630415801420835e-06, "epoch": 37.44, "percentage": 75.48, "elapsed_time": "2:21:52", "remaining_time": "0:46:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2360, "total_steps": 3100, "loss": 0.0216, "learning_rate": 8.209070773271894e-06, "epoch": 37.76, "percentage": 76.13, "elapsed_time": "2:23:05", "remaining_time": "0:44:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2380, "total_steps": 3100, "loss": 0.0268, "learning_rate": 7.79624119714121e-06, "epoch": 38.08, "percentage": 76.77, "elapsed_time": "2:24:15", "remaining_time": "0:43:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2400, "total_steps": 3100, "loss": 0.0306, "learning_rate": 7.392136437882855e-06, "epoch": 38.4, "percentage": 77.42, "elapsed_time": "2:25:28", "remaining_time": "0:42:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2420, "total_steps": 3100, "loss": 0.0224, "learning_rate": 6.996961435595223e-06, "epoch": 38.72, "percentage": 78.06, "elapsed_time": "2:26:44", "remaining_time": "0:41:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2440, "total_steps": 3100, "loss": 0.0138, "learning_rate": 6.610916601686481e-06, "epoch": 39.04, "percentage": 78.71, "elapsed_time": "2:27:57", "remaining_time": "0:40:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2460, "total_steps": 3100, "loss": 0.0234, "learning_rate": 6.234197717236742e-06, "epoch": 39.36, "percentage": 79.35, "elapsed_time": "2:29:10", "remaining_time": "0:38:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2480, "total_steps": 3100, "loss": 0.0164, "learning_rate": 5.866995833708464e-06, "epoch": 39.68, "percentage": 80.0, "elapsed_time": "2:30:25", "remaining_time": "0:37:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2500, "total_steps": 3100, "loss": 0.0169, "learning_rate": 5.509497176055492e-06, "epoch": 40.0, "percentage": 80.65, "elapsed_time": "2:31:36", "remaining_time": "0:36:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2520, "total_steps": 3100, "loss": 0.0116, "learning_rate": 5.161883048279817e-06, "epoch": 40.32, "percentage": 81.29, "elapsed_time": "2:32:48", "remaining_time": "0:35:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2540, "total_steps": 3100, "loss": 0.0219, "learning_rate": 4.824329741483949e-06, "epoch": 40.64, "percentage": 81.94, "elapsed_time": "2:33:57", "remaining_time": "0:33:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2560, "total_steps": 3100, "loss": 0.0366, "learning_rate": 4.497008444465681e-06, "epoch": 40.96, "percentage": 82.58, "elapsed_time": "2:35:09", "remaining_time": "0:32:43", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2580, "total_steps": 3100, "loss": 0.0228, "learning_rate": 4.180085156900274e-06, "epoch": 41.28, "percentage": 83.23, "elapsed_time": "2:36:21", "remaining_time": "0:31:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2600, "total_steps": 3100, "loss": 0.0135, "learning_rate": 3.873720605154468e-06, "epoch": 41.6, "percentage": 83.87, "elapsed_time": "2:37:34", "remaining_time": "0:30:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2620, "total_steps": 3100, "loss": 0.0267, "learning_rate": 3.578070160774724e-06, "epoch": 41.92, "percentage": 84.52, "elapsed_time": "2:38:46", "remaining_time": "0:29:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2640, "total_steps": 3100, "loss": 0.0218, "learning_rate": 3.293283761691182e-06, "epoch": 42.24, "percentage": 85.16, "elapsed_time": "2:40:00", "remaining_time": "0:27:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2660, "total_steps": 3100, "loss": 0.0315, "learning_rate": 3.0195058361772277e-06, "epoch": 42.56, "percentage": 85.81, "elapsed_time": "2:41:16", "remaining_time": "0:26:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2680, "total_steps": 3100, "loss": 0.022, "learning_rate": 2.756875229603295e-06, "epoch": 42.88, "percentage": 86.45, "elapsed_time": "2:42:28", "remaining_time": "0:25:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2700, "total_steps": 3100, "loss": 0.016, "learning_rate": 2.5055251340219855e-06, "epoch": 43.2, "percentage": 87.1, "elapsed_time": "2:43:40", "remaining_time": "0:24:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2720, "total_steps": 3100, "loss": 0.0267, "learning_rate": 2.2655830206202655e-06, "epoch": 43.52, "percentage": 87.74, "elapsed_time": "2:44:52", "remaining_time": "0:23:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2740, "total_steps": 3100, "loss": 0.0167, "learning_rate": 2.037170575072944e-06, "epoch": 43.84, "percentage": 88.39, "elapsed_time": "2:46:03", "remaining_time": "0:21:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2760, "total_steps": 3100, "loss": 0.0492, "learning_rate": 1.8204036358303173e-06, "epoch": 44.16, "percentage": 89.03, "elapsed_time": "2:47:16", "remaining_time": "0:20:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2780, "total_steps": 3100, "loss": 0.0254, "learning_rate": 1.615392135371116e-06, "epoch": 44.48, "percentage": 89.68, "elapsed_time": "2:48:27", "remaining_time": "0:19:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2800, "total_steps": 3100, "loss": 0.0136, "learning_rate": 1.4222400444507318e-06, "epoch": 44.8, "percentage": 90.32, "elapsed_time": "2:49:38", "remaining_time": "0:18:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2820, "total_steps": 3100, "loss": 0.0116, "learning_rate": 1.2410453193728493e-06, "epoch": 45.12, "percentage": 90.97, "elapsed_time": "2:50:51", "remaining_time": "0:16:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2840, "total_steps": 3100, "loss": 0.0311, "learning_rate": 1.0718998523113004e-06, "epoch": 45.44, "percentage": 91.61, "elapsed_time": "2:52:02", "remaining_time": "0:15:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2860, "total_steps": 3100, "loss": 0.0283, "learning_rate": 9.148894247073298e-07, "epoch": 45.76, "percentage": 92.26, "elapsed_time": "2:53:15", "remaining_time": "0:14:32", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2880, "total_steps": 3100, "loss": 0.0186, "learning_rate": 7.700936637658779e-07, "epoch": 46.08, "percentage": 92.9, "elapsed_time": "2:54:29", "remaining_time": "0:13:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2900, "total_steps": 3100, "loss": 0.0229, "learning_rate": 6.375860020729541e-07, "epoch": 46.4, "percentage": 93.55, "elapsed_time": "2:55:42", "remaining_time": "0:12:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2920, "total_steps": 3100, "loss": 0.0268, "learning_rate": 5.174336403546226e-07, "epoch": 46.72, "percentage": 94.19, "elapsed_time": "2:56:56", "remaining_time": "0:10:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2940, "total_steps": 3100, "loss": 0.0161, "learning_rate": 4.096975133963954e-07, "epoch": 47.04, "percentage": 94.84, "elapsed_time": "2:58:07", "remaining_time": "0:09:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2960, "total_steps": 3100, "loss": 0.0113, "learning_rate": 3.144322591404292e-07, "epoch": 47.36, "percentage": 95.48, "elapsed_time": "2:59:19", "remaining_time": "0:08:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2980, "total_steps": 3100, "loss": 0.0222, "learning_rate": 2.316861909760909e-07, "epoch": 47.68, "percentage": 96.13, "elapsed_time": "3:00:30", "remaining_time": "0:07:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3000, "total_steps": 3100, "loss": 0.028, "learning_rate": 1.6150127323803222e-07, "epoch": 48.0, "percentage": 96.77, "elapsed_time": "3:01:48", "remaining_time": "0:06:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3020, "total_steps": 3100, "loss": 0.0217, "learning_rate": 1.0391309992413833e-07, "epoch": 48.32, "percentage": 97.42, "elapsed_time": "3:03:02", "remaining_time": "0:04:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3040, "total_steps": 3100, "loss": 0.0191, "learning_rate": 5.895087664417876e-08, "epoch": 48.64, "percentage": 98.06, "elapsed_time": "3:04:13", "remaining_time": "0:03:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3060, "total_steps": 3100, "loss": 0.0162, "learning_rate": 2.6637405808302428e-08, "epoch": 48.96, "percentage": 98.71, "elapsed_time": "3:05:27", "remaining_time": "0:02:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3080, "total_steps": 3100, "loss": 0.0146, "learning_rate": 6.989075062879824e-09, "epoch": 49.28, "percentage": 99.35, "elapsed_time": "3:06:41", "remaining_time": "0:01:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3100, "total_steps": 3100, "loss": 0.0217, "learning_rate": 1.584897958428755e-11, "epoch": 49.6, "percentage": 100.0, "elapsed_time": "3:07:54", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3100, "total_steps": 3100, "epoch": 49.6, "percentage": 100.0, "elapsed_time": "3:07:54", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0} | |