| {"current_steps": 1, "total_steps": 500, "loss": 1.4464, "learning_rate": 2.0000000000000003e-06, "epoch": 0.008888888888888889, "percentage": 0.2, "elapsed_time": "0:00:03", "remaining_time": "0:32:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 2, "total_steps": 500, "loss": 1.2646, "learning_rate": 4.000000000000001e-06, "epoch": 0.017777777777777778, "percentage": 0.4, "elapsed_time": "0:00:06", "remaining_time": "0:27:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 3, "total_steps": 500, "loss": 1.4738, "learning_rate": 6e-06, "epoch": 0.02666666666666667, "percentage": 0.6, "elapsed_time": "0:00:09", "remaining_time": "0:25:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 4, "total_steps": 500, "loss": 1.3848, "learning_rate": 8.000000000000001e-06, "epoch": 0.035555555555555556, "percentage": 0.8, "elapsed_time": "0:00:11", "remaining_time": "0:24:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 5, "total_steps": 500, "loss": 1.5257, "learning_rate": 1e-05, "epoch": 0.044444444444444446, "percentage": 1.0, "elapsed_time": "0:00:14", "remaining_time": "0:23:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 6, "total_steps": 500, "loss": 1.4105, "learning_rate": 1.2e-05, "epoch": 0.05333333333333334, "percentage": 1.2, "elapsed_time": "0:00:16", "remaining_time": "0:22:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 7, "total_steps": 500, "loss": 1.287, "learning_rate": 1.4000000000000001e-05, "epoch": 0.06222222222222222, "percentage": 1.4, "elapsed_time": "0:00:19", "remaining_time": "0:22:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 8, "total_steps": 500, "loss": 1.2099, "learning_rate": 1.6000000000000003e-05, "epoch": 0.07111111111111111, "percentage": 1.6, "elapsed_time": "0:00:22", "remaining_time": "0:23:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 9, "total_steps": 500, "loss": 1.311, "learning_rate": 1.8e-05, "epoch": 0.08, "percentage": 1.8, "elapsed_time": "0:00:25", "remaining_time": "0:23:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 10, "total_steps": 500, "loss": 2.4975, "learning_rate": 2e-05, "epoch": 0.08888888888888889, "percentage": 2.0, "elapsed_time": "0:00:28", "remaining_time": "0:23:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 11, "total_steps": 500, "loss": 1.2125, "learning_rate": 2.2000000000000003e-05, "epoch": 0.09777777777777778, "percentage": 2.2, "elapsed_time": "0:00:31", "remaining_time": "0:23:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 12, "total_steps": 500, "loss": 1.3688, "learning_rate": 2.4e-05, "epoch": 0.10666666666666667, "percentage": 2.4, "elapsed_time": "0:00:33", "remaining_time": "0:22:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 13, "total_steps": 500, "loss": 1.2372, "learning_rate": 2.6000000000000002e-05, "epoch": 0.11555555555555555, "percentage": 2.6, "elapsed_time": "0:00:36", "remaining_time": "0:22:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 14, "total_steps": 500, "loss": 1.3227, "learning_rate": 2.8000000000000003e-05, "epoch": 0.12444444444444444, "percentage": 2.8, "elapsed_time": "0:00:39", "remaining_time": "0:22:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 15, "total_steps": 500, "loss": 1.1693, "learning_rate": 3e-05, "epoch": 0.13333333333333333, "percentage": 3.0, "elapsed_time": "0:00:42", "remaining_time": "0:22:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 16, "total_steps": 500, "loss": 1.0922, "learning_rate": 3.2000000000000005e-05, "epoch": 0.14222222222222222, "percentage": 3.2, "elapsed_time": "0:00:44", "remaining_time": "0:22:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 17, "total_steps": 500, "loss": 1.0916, "learning_rate": 3.4000000000000007e-05, "epoch": 0.1511111111111111, "percentage": 3.4, "elapsed_time": "0:00:47", "remaining_time": "0:22:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 18, "total_steps": 500, "loss": 1.1219, "learning_rate": 3.6e-05, "epoch": 0.16, "percentage": 3.6, "elapsed_time": "0:00:50", "remaining_time": "0:22:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 19, "total_steps": 500, "loss": 1.2237, "learning_rate": 3.8e-05, "epoch": 0.1688888888888889, "percentage": 3.8, "elapsed_time": "0:00:52", "remaining_time": "0:22:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 20, "total_steps": 500, "loss": 1.0106, "learning_rate": 4e-05, "epoch": 0.17777777777777778, "percentage": 4.0, "elapsed_time": "0:00:55", "remaining_time": "0:22:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 21, "total_steps": 500, "loss": 1.0266, "learning_rate": 4.2e-05, "epoch": 0.18666666666666668, "percentage": 4.2, "elapsed_time": "0:00:58", "remaining_time": "0:22:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 22, "total_steps": 500, "loss": 0.9007, "learning_rate": 4.4000000000000006e-05, "epoch": 0.19555555555555557, "percentage": 4.4, "elapsed_time": "0:01:01", "remaining_time": "0:22:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 23, "total_steps": 500, "loss": 0.9784, "learning_rate": 4.600000000000001e-05, "epoch": 0.20444444444444446, "percentage": 4.6, "elapsed_time": "0:01:03", "remaining_time": "0:22:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 24, "total_steps": 500, "loss": 1.1803, "learning_rate": 4.8e-05, "epoch": 0.21333333333333335, "percentage": 4.8, "elapsed_time": "0:01:07", "remaining_time": "0:22:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 25, "total_steps": 500, "loss": 1.2705, "learning_rate": 5e-05, "epoch": 0.2222222222222222, "percentage": 5.0, "elapsed_time": "0:01:10", "remaining_time": "0:22:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 26, "total_steps": 500, "loss": 1.0653, "learning_rate": 5.2000000000000004e-05, "epoch": 0.2311111111111111, "percentage": 5.2, "elapsed_time": "0:01:12", "remaining_time": "0:22:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 27, "total_steps": 500, "loss": 0.988, "learning_rate": 5.4000000000000005e-05, "epoch": 0.24, "percentage": 5.4, "elapsed_time": "0:01:15", "remaining_time": "0:22:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 28, "total_steps": 500, "loss": 1.138, "learning_rate": 5.6000000000000006e-05, "epoch": 0.24888888888888888, "percentage": 5.6, "elapsed_time": "0:01:18", "remaining_time": "0:22:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 29, "total_steps": 500, "loss": 0.9095, "learning_rate": 5.8e-05, "epoch": 0.2577777777777778, "percentage": 5.8, "elapsed_time": "0:01:21", "remaining_time": "0:22:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 30, "total_steps": 500, "loss": 1.1132, "learning_rate": 6e-05, "epoch": 0.26666666666666666, "percentage": 6.0, "elapsed_time": "0:01:24", "remaining_time": "0:22:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 31, "total_steps": 500, "loss": 0.9967, "learning_rate": 6.2e-05, "epoch": 0.27555555555555555, "percentage": 6.2, "elapsed_time": "0:01:27", "remaining_time": "0:22:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 32, "total_steps": 500, "loss": 0.9615, "learning_rate": 6.400000000000001e-05, "epoch": 0.28444444444444444, "percentage": 6.4, "elapsed_time": "0:01:30", "remaining_time": "0:21:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 33, "total_steps": 500, "loss": 1.0561, "learning_rate": 6.6e-05, "epoch": 0.29333333333333333, "percentage": 6.6, "elapsed_time": "0:01:32", "remaining_time": "0:21:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 34, "total_steps": 500, "loss": 1.1543, "learning_rate": 6.800000000000001e-05, "epoch": 0.3022222222222222, "percentage": 6.8, "elapsed_time": "0:01:35", "remaining_time": "0:21:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 35, "total_steps": 500, "loss": 1.1004, "learning_rate": 7e-05, "epoch": 0.3111111111111111, "percentage": 7.0, "elapsed_time": "0:01:38", "remaining_time": "0:21:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 36, "total_steps": 500, "loss": 1.0279, "learning_rate": 7.2e-05, "epoch": 0.32, "percentage": 7.2, "elapsed_time": "0:01:40", "remaining_time": "0:21:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 37, "total_steps": 500, "loss": 0.9469, "learning_rate": 7.4e-05, "epoch": 0.3288888888888889, "percentage": 7.4, "elapsed_time": "0:01:42", "remaining_time": "0:21:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 38, "total_steps": 500, "loss": 1.0991, "learning_rate": 7.6e-05, "epoch": 0.3377777777777778, "percentage": 7.6, "elapsed_time": "0:01:45", "remaining_time": "0:21:24", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 39, "total_steps": 500, "loss": 1.0668, "learning_rate": 7.800000000000001e-05, "epoch": 0.3466666666666667, "percentage": 7.8, "elapsed_time": "0:01:48", "remaining_time": "0:21:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 40, "total_steps": 500, "loss": 0.9899, "learning_rate": 8e-05, "epoch": 0.35555555555555557, "percentage": 8.0, "elapsed_time": "0:01:51", "remaining_time": "0:21:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 41, "total_steps": 500, "loss": 1.0636, "learning_rate": 8.2e-05, "epoch": 0.36444444444444446, "percentage": 8.2, "elapsed_time": "0:01:53", "remaining_time": "0:21:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 42, "total_steps": 500, "loss": 1.0076, "learning_rate": 8.4e-05, "epoch": 0.37333333333333335, "percentage": 8.4, "elapsed_time": "0:01:57", "remaining_time": "0:21:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 43, "total_steps": 500, "loss": 0.9301, "learning_rate": 8.6e-05, "epoch": 0.38222222222222224, "percentage": 8.6, "elapsed_time": "0:02:00", "remaining_time": "0:21:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 44, "total_steps": 500, "loss": 1.171, "learning_rate": 8.800000000000001e-05, "epoch": 0.39111111111111113, "percentage": 8.8, "elapsed_time": "0:02:03", "remaining_time": "0:21:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 45, "total_steps": 500, "loss": 0.9985, "learning_rate": 9e-05, "epoch": 0.4, "percentage": 9.0, "elapsed_time": "0:02:06", "remaining_time": "0:21:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 46, "total_steps": 500, "loss": 1.034, "learning_rate": 9.200000000000001e-05, "epoch": 0.4088888888888889, "percentage": 9.2, "elapsed_time": "0:02:09", "remaining_time": "0:21:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 47, "total_steps": 500, "loss": 0.9552, "learning_rate": 9.4e-05, "epoch": 0.4177777777777778, "percentage": 9.4, "elapsed_time": "0:02:11", "remaining_time": "0:21:08", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 48, "total_steps": 500, "loss": 1.2027, "learning_rate": 9.6e-05, "epoch": 0.4266666666666667, "percentage": 9.6, "elapsed_time": "0:02:14", "remaining_time": "0:21:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 49, "total_steps": 500, "loss": 1.2453, "learning_rate": 9.8e-05, "epoch": 0.43555555555555553, "percentage": 9.8, "elapsed_time": "0:02:18", "remaining_time": "0:21:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 50, "total_steps": 500, "loss": 1.1497, "learning_rate": 0.0001, "epoch": 0.4444444444444444, "percentage": 10.0, "elapsed_time": "0:02:21", "remaining_time": "0:21:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 51, "total_steps": 500, "loss": 1.0698, "learning_rate": 9.999878153526974e-05, "epoch": 0.4533333333333333, "percentage": 10.2, "elapsed_time": "0:02:23", "remaining_time": "0:21:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 52, "total_steps": 500, "loss": 1.093, "learning_rate": 9.999512620046522e-05, "epoch": 0.4622222222222222, "percentage": 10.4, "elapsed_time": "0:02:26", "remaining_time": "0:21:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 53, "total_steps": 500, "loss": 1.0031, "learning_rate": 9.998903417374228e-05, "epoch": 0.4711111111111111, "percentage": 10.6, "elapsed_time": "0:02:28", "remaining_time": "0:20:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 54, "total_steps": 500, "loss": 1.2479, "learning_rate": 9.998050575201771e-05, "epoch": 0.48, "percentage": 10.8, "elapsed_time": "0:02:31", "remaining_time": "0:20:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 55, "total_steps": 500, "loss": 1.1337, "learning_rate": 9.99695413509548e-05, "epoch": 0.4888888888888889, "percentage": 11.0, "elapsed_time": "0:02:35", "remaining_time": "0:20:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 56, "total_steps": 500, "loss": 1.0747, "learning_rate": 9.995614150494293e-05, "epoch": 0.49777777777777776, "percentage": 11.2, "elapsed_time": "0:02:38", "remaining_time": "0:20:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 57, "total_steps": 500, "loss": 1.1863, "learning_rate": 9.99403068670717e-05, "epoch": 0.5066666666666667, "percentage": 11.4, "elapsed_time": "0:02:40", "remaining_time": "0:20:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 58, "total_steps": 500, "loss": 0.8643, "learning_rate": 9.992203820909906e-05, "epoch": 0.5155555555555555, "percentage": 11.6, "elapsed_time": "0:02:43", "remaining_time": "0:20:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 59, "total_steps": 500, "loss": 1.0411, "learning_rate": 9.990133642141359e-05, "epoch": 0.5244444444444445, "percentage": 11.8, "elapsed_time": "0:02:45", "remaining_time": "0:20:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 60, "total_steps": 500, "loss": 1.0888, "learning_rate": 9.987820251299122e-05, "epoch": 0.5333333333333333, "percentage": 12.0, "elapsed_time": "0:02:48", "remaining_time": "0:20:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 61, "total_steps": 500, "loss": 0.9589, "learning_rate": 9.985263761134602e-05, "epoch": 0.5422222222222223, "percentage": 12.2, "elapsed_time": "0:02:51", "remaining_time": "0:20:35", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 62, "total_steps": 500, "loss": 1.0835, "learning_rate": 9.982464296247522e-05, "epoch": 0.5511111111111111, "percentage": 12.4, "elapsed_time": "0:02:54", "remaining_time": "0:20:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 63, "total_steps": 500, "loss": 0.9976, "learning_rate": 9.979421993079852e-05, "epoch": 0.56, "percentage": 12.6, "elapsed_time": "0:02:57", "remaining_time": "0:20:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 64, "total_steps": 500, "loss": 1.0755, "learning_rate": 9.976136999909156e-05, "epoch": 0.5688888888888889, "percentage": 12.8, "elapsed_time": "0:03:00", "remaining_time": "0:20:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 65, "total_steps": 500, "loss": 1.0238, "learning_rate": 9.972609476841367e-05, "epoch": 0.5777777777777777, "percentage": 13.0, "elapsed_time": "0:03:03", "remaining_time": "0:20:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 66, "total_steps": 500, "loss": 0.9071, "learning_rate": 9.968839595802982e-05, "epoch": 0.5866666666666667, "percentage": 13.2, "elapsed_time": "0:03:05", "remaining_time": "0:20:22", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 67, "total_steps": 500, "loss": 0.9407, "learning_rate": 9.964827540532685e-05, "epoch": 0.5955555555555555, "percentage": 13.4, "elapsed_time": "0:03:08", "remaining_time": "0:20:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 68, "total_steps": 500, "loss": 1.1039, "learning_rate": 9.96057350657239e-05, "epoch": 0.6044444444444445, "percentage": 13.6, "elapsed_time": "0:03:11", "remaining_time": "0:20:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 69, "total_steps": 500, "loss": 1.0651, "learning_rate": 9.956077701257709e-05, "epoch": 0.6133333333333333, "percentage": 13.8, "elapsed_time": "0:03:13", "remaining_time": "0:20:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 70, "total_steps": 500, "loss": 1.0068, "learning_rate": 9.951340343707852e-05, "epoch": 0.6222222222222222, "percentage": 14.0, "elapsed_time": "0:03:16", "remaining_time": "0:20:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 71, "total_steps": 500, "loss": 0.9485, "learning_rate": 9.946361664814943e-05, "epoch": 0.6311111111111111, "percentage": 14.2, "elapsed_time": "0:03:19", "remaining_time": "0:20:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 72, "total_steps": 500, "loss": 1.247, "learning_rate": 9.941141907232765e-05, "epoch": 0.64, "percentage": 14.4, "elapsed_time": "0:03:22", "remaining_time": "0:20:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 73, "total_steps": 500, "loss": 1.0063, "learning_rate": 9.93568132536494e-05, "epoch": 0.6488888888888888, "percentage": 14.6, "elapsed_time": "0:03:25", "remaining_time": "0:19:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 74, "total_steps": 500, "loss": 0.9367, "learning_rate": 9.929980185352526e-05, "epoch": 0.6577777777777778, "percentage": 14.8, "elapsed_time": "0:03:27", "remaining_time": "0:19:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 75, "total_steps": 500, "loss": 1.05, "learning_rate": 9.924038765061042e-05, "epoch": 0.6666666666666666, "percentage": 15.0, "elapsed_time": "0:03:30", "remaining_time": "0:19:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 76, "total_steps": 500, "loss": 1.1569, "learning_rate": 9.917857354066931e-05, "epoch": 0.6755555555555556, "percentage": 15.2, "elapsed_time": "0:03:32", "remaining_time": "0:19:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 77, "total_steps": 500, "loss": 0.8825, "learning_rate": 9.911436253643445e-05, "epoch": 0.6844444444444444, "percentage": 15.4, "elapsed_time": "0:03:36", "remaining_time": "0:19:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 78, "total_steps": 500, "loss": 0.8632, "learning_rate": 9.904775776745958e-05, "epoch": 0.6933333333333334, "percentage": 15.6, "elapsed_time": "0:03:38", "remaining_time": "0:19:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 79, "total_steps": 500, "loss": 0.9715, "learning_rate": 9.89787624799672e-05, "epoch": 0.7022222222222222, "percentage": 15.8, "elapsed_time": "0:03:41", "remaining_time": "0:19:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 80, "total_steps": 500, "loss": 1.0032, "learning_rate": 9.890738003669029e-05, "epoch": 0.7111111111111111, "percentage": 16.0, "elapsed_time": "0:03:44", "remaining_time": "0:19:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 81, "total_steps": 500, "loss": 1.1563, "learning_rate": 9.88336139167084e-05, "epoch": 0.72, "percentage": 16.2, "elapsed_time": "0:03:47", "remaining_time": "0:19:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 82, "total_steps": 500, "loss": 1.0159, "learning_rate": 9.875746771527816e-05, "epoch": 0.7288888888888889, "percentage": 16.4, "elapsed_time": "0:03:49", "remaining_time": "0:19:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 83, "total_steps": 500, "loss": 1.1233, "learning_rate": 9.867894514365802e-05, "epoch": 0.7377777777777778, "percentage": 16.6, "elapsed_time": "0:03:52", "remaining_time": "0:19:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 84, "total_steps": 500, "loss": 1.0673, "learning_rate": 9.859805002892732e-05, "epoch": 0.7466666666666667, "percentage": 16.8, "elapsed_time": "0:03:55", "remaining_time": "0:19:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 85, "total_steps": 500, "loss": 0.9338, "learning_rate": 9.851478631379982e-05, "epoch": 0.7555555555555555, "percentage": 17.0, "elapsed_time": "0:03:57", "remaining_time": "0:19:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 86, "total_steps": 500, "loss": 1.0843, "learning_rate": 9.842915805643155e-05, "epoch": 0.7644444444444445, "percentage": 17.2, "elapsed_time": "0:04:00", "remaining_time": "0:19:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 87, "total_steps": 500, "loss": 0.9081, "learning_rate": 9.834116943022298e-05, "epoch": 0.7733333333333333, "percentage": 17.4, "elapsed_time": "0:04:03", "remaining_time": "0:19:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 88, "total_steps": 500, "loss": 0.9989, "learning_rate": 9.825082472361557e-05, "epoch": 0.7822222222222223, "percentage": 17.6, "elapsed_time": "0:04:05", "remaining_time": "0:19:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 89, "total_steps": 500, "loss": 0.8834, "learning_rate": 9.815812833988291e-05, "epoch": 0.7911111111111111, "percentage": 17.8, "elapsed_time": "0:04:08", "remaining_time": "0:19:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 90, "total_steps": 500, "loss": 0.9779, "learning_rate": 9.806308479691595e-05, "epoch": 0.8, "percentage": 18.0, "elapsed_time": "0:04:11", "remaining_time": "0:19:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 91, "total_steps": 500, "loss": 1.0189, "learning_rate": 9.796569872700288e-05, "epoch": 0.8088888888888889, "percentage": 18.2, "elapsed_time": "0:04:13", "remaining_time": "0:19:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 92, "total_steps": 500, "loss": 1.0696, "learning_rate": 9.786597487660337e-05, "epoch": 0.8177777777777778, "percentage": 18.4, "elapsed_time": "0:04:16", "remaining_time": "0:18:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 93, "total_steps": 500, "loss": 1.059, "learning_rate": 9.776391810611718e-05, "epoch": 0.8266666666666667, "percentage": 18.6, "elapsed_time": "0:04:19", "remaining_time": "0:18:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 94, "total_steps": 500, "loss": 0.8577, "learning_rate": 9.765953338964735e-05, "epoch": 0.8355555555555556, "percentage": 18.8, "elapsed_time": "0:04:22", "remaining_time": "0:18:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 95, "total_steps": 500, "loss": 0.9671, "learning_rate": 9.755282581475769e-05, "epoch": 0.8444444444444444, "percentage": 19.0, "elapsed_time": "0:04:24", "remaining_time": "0:18:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 96, "total_steps": 500, "loss": 1.0792, "learning_rate": 9.744380058222483e-05, "epoch": 0.8533333333333334, "percentage": 19.2, "elapsed_time": "0:04:27", "remaining_time": "0:18:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 97, "total_steps": 500, "loss": 1.0958, "learning_rate": 9.733246300578483e-05, "epoch": 0.8622222222222222, "percentage": 19.4, "elapsed_time": "0:04:30", "remaining_time": "0:18:43", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 98, "total_steps": 500, "loss": 0.966, "learning_rate": 9.721881851187406e-05, "epoch": 0.8711111111111111, "percentage": 19.6, "elapsed_time": "0:04:32", "remaining_time": "0:18:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 99, "total_steps": 500, "loss": 0.8868, "learning_rate": 9.710287263936484e-05, "epoch": 0.88, "percentage": 19.8, "elapsed_time": "0:04:35", "remaining_time": "0:18:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 100, "total_steps": 500, "loss": 1.0577, "learning_rate": 9.698463103929542e-05, "epoch": 0.8888888888888888, "percentage": 20.0, "elapsed_time": "0:04:38", "remaining_time": "0:18:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 101, "total_steps": 500, "loss": 0.7904, "learning_rate": 9.686409947459458e-05, "epoch": 0.8977777777777778, "percentage": 20.2, "elapsed_time": "0:04:40", "remaining_time": "0:18:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 102, "total_steps": 500, "loss": 0.8927, "learning_rate": 9.674128381980072e-05, "epoch": 0.9066666666666666, "percentage": 20.4, "elapsed_time": "0:04:43", "remaining_time": "0:18:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 103, "total_steps": 500, "loss": 1.0477, "learning_rate": 9.661619006077561e-05, "epoch": 0.9155555555555556, "percentage": 20.6, "elapsed_time": "0:04:46", "remaining_time": "0:18:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 104, "total_steps": 500, "loss": 1.0107, "learning_rate": 9.648882429441257e-05, "epoch": 0.9244444444444444, "percentage": 20.8, "elapsed_time": "0:04:49", "remaining_time": "0:18:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 105, "total_steps": 500, "loss": 1.0277, "learning_rate": 9.635919272833938e-05, "epoch": 0.9333333333333333, "percentage": 21.0, "elapsed_time": "0:04:52", "remaining_time": "0:18:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 106, "total_steps": 500, "loss": 0.9528, "learning_rate": 9.622730168061567e-05, "epoch": 0.9422222222222222, "percentage": 21.2, "elapsed_time": "0:04:54", "remaining_time": "0:18:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 107, "total_steps": 500, "loss": 0.8783, "learning_rate": 9.609315757942503e-05, "epoch": 0.9511111111111111, "percentage": 21.4, "elapsed_time": "0:04:57", "remaining_time": "0:18:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 108, "total_steps": 500, "loss": 0.991, "learning_rate": 9.595676696276172e-05, "epoch": 0.96, "percentage": 21.6, "elapsed_time": "0:05:00", "remaining_time": "0:18:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 109, "total_steps": 500, "loss": 1.0017, "learning_rate": 9.581813647811198e-05, "epoch": 0.9688888888888889, "percentage": 21.8, "elapsed_time": "0:05:04", "remaining_time": "0:18:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 110, "total_steps": 500, "loss": 1.0257, "learning_rate": 9.567727288213005e-05, "epoch": 0.9777777777777777, "percentage": 22.0, "elapsed_time": "0:05:06", "remaining_time": "0:18:08", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 111, "total_steps": 500, "loss": 0.9087, "learning_rate": 9.553418304030886e-05, "epoch": 0.9866666666666667, "percentage": 22.2, "elapsed_time": "0:05:09", "remaining_time": "0:18:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 112, "total_steps": 500, "loss": 1.002, "learning_rate": 9.538887392664544e-05, "epoch": 0.9955555555555555, "percentage": 22.4, "elapsed_time": "0:05:11", "remaining_time": "0:18:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 113, "total_steps": 500, "loss": 0.7493, "learning_rate": 9.524135262330098e-05, "epoch": 1.0044444444444445, "percentage": 22.6, "elapsed_time": "0:05:15", "remaining_time": "0:18:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 114, "total_steps": 500, "loss": 0.8562, "learning_rate": 9.50916263202557e-05, "epoch": 1.0133333333333334, "percentage": 22.8, "elapsed_time": "0:05:18", "remaining_time": "0:17:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 115, "total_steps": 500, "loss": 0.9812, "learning_rate": 9.493970231495835e-05, "epoch": 1.0222222222222221, "percentage": 23.0, "elapsed_time": "0:05:21", "remaining_time": "0:17:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 116, "total_steps": 500, "loss": 0.8069, "learning_rate": 9.478558801197065e-05, "epoch": 1.031111111111111, "percentage": 23.2, "elapsed_time": "0:05:23", "remaining_time": "0:17:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 117, "total_steps": 500, "loss": 0.8336, "learning_rate": 9.462929092260628e-05, "epoch": 1.04, "percentage": 23.4, "elapsed_time": "0:05:26", "remaining_time": "0:17:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 118, "total_steps": 500, "loss": 0.9364, "learning_rate": 9.447081866456489e-05, "epoch": 1.048888888888889, "percentage": 23.6, "elapsed_time": "0:05:29", "remaining_time": "0:17:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 119, "total_steps": 500, "loss": 0.7365, "learning_rate": 9.431017896156074e-05, "epoch": 1.0577777777777777, "percentage": 23.8, "elapsed_time": "0:05:31", "remaining_time": "0:17:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 120, "total_steps": 500, "loss": 1.0947, "learning_rate": 9.414737964294636e-05, "epoch": 1.0666666666666667, "percentage": 24.0, "elapsed_time": "0:05:34", "remaining_time": "0:17:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 121, "total_steps": 500, "loss": 0.9642, "learning_rate": 9.398242864333083e-05, "epoch": 1.0755555555555556, "percentage": 24.2, "elapsed_time": "0:05:37", "remaining_time": "0:17:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 122, "total_steps": 500, "loss": 0.824, "learning_rate": 9.381533400219318e-05, "epoch": 1.0844444444444445, "percentage": 24.4, "elapsed_time": "0:05:40", "remaining_time": "0:17:35", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 123, "total_steps": 500, "loss": 0.814, "learning_rate": 9.364610386349049e-05, "epoch": 1.0933333333333333, "percentage": 24.6, "elapsed_time": "0:05:43", "remaining_time": "0:17:32", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 124, "total_steps": 500, "loss": 0.7891, "learning_rate": 9.347474647526095e-05, "epoch": 1.1022222222222222, "percentage": 24.8, "elapsed_time": "0:05:45", "remaining_time": "0:17:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 125, "total_steps": 500, "loss": 0.8481, "learning_rate": 9.330127018922194e-05, "epoch": 1.1111111111111112, "percentage": 25.0, "elapsed_time": "0:05:49", "remaining_time": "0:17:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 126, "total_steps": 500, "loss": 0.7348, "learning_rate": 9.312568346036288e-05, "epoch": 1.12, "percentage": 25.2, "elapsed_time": "0:05:52", "remaining_time": "0:17:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 127, "total_steps": 500, "loss": 1.1175, "learning_rate": 9.294799484653323e-05, "epoch": 1.1288888888888888, "percentage": 25.4, "elapsed_time": "0:05:55", "remaining_time": "0:17:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 128, "total_steps": 500, "loss": 1.1577, "learning_rate": 9.276821300802534e-05, "epoch": 1.1377777777777778, "percentage": 25.6, "elapsed_time": "0:05:58", "remaining_time": "0:17:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 129, "total_steps": 500, "loss": 0.8026, "learning_rate": 9.258634670715238e-05, "epoch": 1.1466666666666667, "percentage": 25.8, "elapsed_time": "0:06:00", "remaining_time": "0:17:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 130, "total_steps": 500, "loss": 0.8051, "learning_rate": 9.24024048078213e-05, "epoch": 1.1555555555555554, "percentage": 26.0, "elapsed_time": "0:06:03", "remaining_time": "0:17:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 131, "total_steps": 500, "loss": 1.0332, "learning_rate": 9.221639627510076e-05, "epoch": 1.1644444444444444, "percentage": 26.2, "elapsed_time": "0:06:07", "remaining_time": "0:17:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 132, "total_steps": 500, "loss": 0.9912, "learning_rate": 9.202833017478422e-05, "epoch": 1.1733333333333333, "percentage": 26.4, "elapsed_time": "0:06:09", "remaining_time": "0:17:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 133, "total_steps": 500, "loss": 0.7942, "learning_rate": 9.183821567294809e-05, "epoch": 1.1822222222222223, "percentage": 26.6, "elapsed_time": "0:06:12", "remaining_time": "0:17:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 134, "total_steps": 500, "loss": 0.8423, "learning_rate": 9.164606203550497e-05, "epoch": 1.1911111111111112, "percentage": 26.8, "elapsed_time": "0:06:15", "remaining_time": "0:17:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 135, "total_steps": 500, "loss": 0.7237, "learning_rate": 9.145187862775209e-05, "epoch": 1.2, "percentage": 27.0, "elapsed_time": "0:06:17", "remaining_time": "0:17:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 136, "total_steps": 500, "loss": 0.8881, "learning_rate": 9.125567491391476e-05, "epoch": 1.208888888888889, "percentage": 27.2, "elapsed_time": "0:06:20", "remaining_time": "0:16:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 137, "total_steps": 500, "loss": 0.8953, "learning_rate": 9.105746045668521e-05, "epoch": 1.2177777777777778, "percentage": 27.4, "elapsed_time": "0:06:22", "remaining_time": "0:16:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 138, "total_steps": 500, "loss": 1.0433, "learning_rate": 9.085724491675642e-05, "epoch": 1.2266666666666666, "percentage": 27.6, "elapsed_time": "0:06:25", "remaining_time": "0:16:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 139, "total_steps": 500, "loss": 0.9354, "learning_rate": 9.065503805235138e-05, "epoch": 1.2355555555555555, "percentage": 27.8, "elapsed_time": "0:06:28", "remaining_time": "0:16:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 140, "total_steps": 500, "loss": 0.9633, "learning_rate": 9.045084971874738e-05, "epoch": 1.2444444444444445, "percentage": 28.0, "elapsed_time": "0:06:31", "remaining_time": "0:16:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 141, "total_steps": 500, "loss": 1.0975, "learning_rate": 9.02446898677957e-05, "epoch": 1.2533333333333334, "percentage": 28.2, "elapsed_time": "0:06:33", "remaining_time": "0:16:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 142, "total_steps": 500, "loss": 0.8621, "learning_rate": 9.003656854743667e-05, "epoch": 1.2622222222222224, "percentage": 28.4, "elapsed_time": "0:06:36", "remaining_time": "0:16:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 143, "total_steps": 500, "loss": 0.8958, "learning_rate": 8.982649590120982e-05, "epoch": 1.271111111111111, "percentage": 28.6, "elapsed_time": "0:06:39", "remaining_time": "0:16:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 144, "total_steps": 500, "loss": 0.9975, "learning_rate": 8.961448216775954e-05, "epoch": 1.28, "percentage": 28.8, "elapsed_time": "0:06:41", "remaining_time": "0:16:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 145, "total_steps": 500, "loss": 0.9486, "learning_rate": 8.940053768033609e-05, "epoch": 1.2888888888888888, "percentage": 29.0, "elapsed_time": "0:06:44", "remaining_time": "0:16:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 146, "total_steps": 500, "loss": 0.876, "learning_rate": 8.9184672866292e-05, "epoch": 1.2977777777777777, "percentage": 29.2, "elapsed_time": "0:06:46", "remaining_time": "0:16:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 147, "total_steps": 500, "loss": 0.8772, "learning_rate": 8.896689824657372e-05, "epoch": 1.3066666666666666, "percentage": 29.4, "elapsed_time": "0:06:49", "remaining_time": "0:16:22", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 148, "total_steps": 500, "loss": 0.8023, "learning_rate": 8.874722443520899e-05, "epoch": 1.3155555555555556, "percentage": 29.6, "elapsed_time": "0:06:52", "remaining_time": "0:16:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 149, "total_steps": 500, "loss": 1.0432, "learning_rate": 8.852566213878947e-05, "epoch": 1.3244444444444445, "percentage": 29.8, "elapsed_time": "0:06:55", "remaining_time": "0:16:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 150, "total_steps": 500, "loss": 0.9147, "learning_rate": 8.83022221559489e-05, "epoch": 1.3333333333333333, "percentage": 30.0, "elapsed_time": "0:06:57", "remaining_time": "0:16:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 151, "total_steps": 500, "loss": 0.9798, "learning_rate": 8.807691537683685e-05, "epoch": 1.3422222222222222, "percentage": 30.2, "elapsed_time": "0:07:00", "remaining_time": "0:16:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 152, "total_steps": 500, "loss": 0.8849, "learning_rate": 8.784975278258783e-05, "epoch": 1.3511111111111112, "percentage": 30.4, "elapsed_time": "0:07:02", "remaining_time": "0:16:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 153, "total_steps": 500, "loss": 0.9281, "learning_rate": 8.762074544478623e-05, "epoch": 1.3599999999999999, "percentage": 30.6, "elapsed_time": "0:07:05", "remaining_time": "0:16:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 154, "total_steps": 500, "loss": 1.0482, "learning_rate": 8.73899045249266e-05, "epoch": 1.3688888888888888, "percentage": 30.8, "elapsed_time": "0:07:07", "remaining_time": "0:16:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 155, "total_steps": 500, "loss": 1.0076, "learning_rate": 8.715724127386972e-05, "epoch": 1.3777777777777778, "percentage": 31.0, "elapsed_time": "0:07:10", "remaining_time": "0:15:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 156, "total_steps": 500, "loss": 0.817, "learning_rate": 8.692276703129421e-05, "epoch": 1.3866666666666667, "percentage": 31.2, "elapsed_time": "0:07:12", "remaining_time": "0:15:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 157, "total_steps": 500, "loss": 1.0906, "learning_rate": 8.668649322514382e-05, "epoch": 1.3955555555555557, "percentage": 31.4, "elapsed_time": "0:07:15", "remaining_time": "0:15:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 158, "total_steps": 500, "loss": 0.9078, "learning_rate": 8.644843137107059e-05, "epoch": 1.4044444444444444, "percentage": 31.6, "elapsed_time": "0:07:17", "remaining_time": "0:15:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 159, "total_steps": 500, "loss": 0.9268, "learning_rate": 8.620859307187339e-05, "epoch": 1.4133333333333333, "percentage": 31.8, "elapsed_time": "0:07:20", "remaining_time": "0:15:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 160, "total_steps": 500, "loss": 0.8873, "learning_rate": 8.596699001693255e-05, "epoch": 1.4222222222222223, "percentage": 32.0, "elapsed_time": "0:07:22", "remaining_time": "0:15:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 161, "total_steps": 500, "loss": 0.8488, "learning_rate": 8.572363398164017e-05, "epoch": 1.431111111111111, "percentage": 32.2, "elapsed_time": "0:07:25", "remaining_time": "0:15:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 162, "total_steps": 500, "loss": 1.001, "learning_rate": 8.547853682682604e-05, "epoch": 1.44, "percentage": 32.4, "elapsed_time": "0:07:27", "remaining_time": "0:15:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 163, "total_steps": 500, "loss": 1.0562, "learning_rate": 8.523171049817974e-05, "epoch": 1.448888888888889, "percentage": 32.6, "elapsed_time": "0:07:30", "remaining_time": "0:15:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 164, "total_steps": 500, "loss": 0.845, "learning_rate": 8.498316702566828e-05, "epoch": 1.4577777777777778, "percentage": 32.8, "elapsed_time": "0:07:33", "remaining_time": "0:15:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 165, "total_steps": 500, "loss": 0.7172, "learning_rate": 8.473291852294987e-05, "epoch": 1.4666666666666668, "percentage": 33.0, "elapsed_time": "0:07:35", "remaining_time": "0:15:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 166, "total_steps": 500, "loss": 0.9064, "learning_rate": 8.44809771867835e-05, "epoch": 1.4755555555555555, "percentage": 33.2, "elapsed_time": "0:07:38", "remaining_time": "0:15:22", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 167, "total_steps": 500, "loss": 0.888, "learning_rate": 8.422735529643444e-05, "epoch": 1.4844444444444445, "percentage": 33.4, "elapsed_time": "0:07:41", "remaining_time": "0:15:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 168, "total_steps": 500, "loss": 0.8158, "learning_rate": 8.397206521307584e-05, "epoch": 1.4933333333333334, "percentage": 33.6, "elapsed_time": "0:07:43", "remaining_time": "0:15:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 169, "total_steps": 500, "loss": 0.8498, "learning_rate": 8.371511937918616e-05, "epoch": 1.5022222222222221, "percentage": 33.8, "elapsed_time": "0:07:46", "remaining_time": "0:15:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 170, "total_steps": 500, "loss": 0.93, "learning_rate": 8.345653031794292e-05, "epoch": 1.511111111111111, "percentage": 34.0, "elapsed_time": "0:07:49", "remaining_time": "0:15:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 171, "total_steps": 500, "loss": 1.2189, "learning_rate": 8.319631063261209e-05, "epoch": 1.52, "percentage": 34.2, "elapsed_time": "0:07:51", "remaining_time": "0:15:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 172, "total_steps": 500, "loss": 1.0049, "learning_rate": 8.293447300593402e-05, "epoch": 1.528888888888889, "percentage": 34.4, "elapsed_time": "0:07:53", "remaining_time": "0:15:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 173, "total_steps": 500, "loss": 0.9254, "learning_rate": 8.267103019950529e-05, "epoch": 1.537777777777778, "percentage": 34.6, "elapsed_time": "0:07:56", "remaining_time": "0:14:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 174, "total_steps": 500, "loss": 0.8603, "learning_rate": 8.240599505315655e-05, "epoch": 1.5466666666666666, "percentage": 34.8, "elapsed_time": "0:07:58", "remaining_time": "0:14:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 175, "total_steps": 500, "loss": 0.9305, "learning_rate": 8.213938048432697e-05, "epoch": 1.5555555555555556, "percentage": 35.0, "elapsed_time": "0:08:01", "remaining_time": "0:14:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 176, "total_steps": 500, "loss": 1.0136, "learning_rate": 8.18711994874345e-05, "epoch": 1.5644444444444443, "percentage": 35.2, "elapsed_time": "0:08:04", "remaining_time": "0:14:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 177, "total_steps": 500, "loss": 1.0119, "learning_rate": 8.160146513324254e-05, "epoch": 1.5733333333333333, "percentage": 35.4, "elapsed_time": "0:08:06", "remaining_time": "0:14:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 178, "total_steps": 500, "loss": 0.8887, "learning_rate": 8.133019056822304e-05, "epoch": 1.5822222222222222, "percentage": 35.6, "elapsed_time": "0:08:08", "remaining_time": "0:14:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 179, "total_steps": 500, "loss": 1.0235, "learning_rate": 8.105738901391552e-05, "epoch": 1.5911111111111111, "percentage": 35.8, "elapsed_time": "0:08:11", "remaining_time": "0:14:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 180, "total_steps": 500, "loss": 0.8074, "learning_rate": 8.07830737662829e-05, "epoch": 1.6, "percentage": 36.0, "elapsed_time": "0:08:13", "remaining_time": "0:14:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 181, "total_steps": 500, "loss": 0.861, "learning_rate": 8.05072581950634e-05, "epoch": 1.608888888888889, "percentage": 36.2, "elapsed_time": "0:08:16", "remaining_time": "0:14:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 182, "total_steps": 500, "loss": 0.9879, "learning_rate": 8.022995574311876e-05, "epoch": 1.6177777777777778, "percentage": 36.4, "elapsed_time": "0:08:19", "remaining_time": "0:14:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 183, "total_steps": 500, "loss": 0.7969, "learning_rate": 7.99511799257793e-05, "epoch": 1.6266666666666667, "percentage": 36.6, "elapsed_time": "0:08:22", "remaining_time": "0:14:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 184, "total_steps": 500, "loss": 0.85, "learning_rate": 7.967094433018508e-05, "epoch": 1.6355555555555554, "percentage": 36.8, "elapsed_time": "0:08:25", "remaining_time": "0:14:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 185, "total_steps": 500, "loss": 0.765, "learning_rate": 7.938926261462366e-05, "epoch": 1.6444444444444444, "percentage": 37.0, "elapsed_time": "0:08:27", "remaining_time": "0:14:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 186, "total_steps": 500, "loss": 1.0234, "learning_rate": 7.910614850786448e-05, "epoch": 1.6533333333333333, "percentage": 37.2, "elapsed_time": "0:08:29", "remaining_time": "0:14:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 187, "total_steps": 500, "loss": 0.9385, "learning_rate": 7.882161580848967e-05, "epoch": 1.6622222222222223, "percentage": 37.4, "elapsed_time": "0:08:32", "remaining_time": "0:14:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 188, "total_steps": 500, "loss": 0.8661, "learning_rate": 7.85356783842216e-05, "epoch": 1.6711111111111112, "percentage": 37.6, "elapsed_time": "0:08:34", "remaining_time": "0:14:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 189, "total_steps": 500, "loss": 1.1284, "learning_rate": 7.82483501712469e-05, "epoch": 1.6800000000000002, "percentage": 37.8, "elapsed_time": "0:08:37", "remaining_time": "0:14:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 190, "total_steps": 500, "loss": 0.9681, "learning_rate": 7.795964517353735e-05, "epoch": 1.6888888888888889, "percentage": 38.0, "elapsed_time": "0:08:39", "remaining_time": "0:14:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 191, "total_steps": 500, "loss": 0.8778, "learning_rate": 7.766957746216721e-05, "epoch": 1.6977777777777778, "percentage": 38.2, "elapsed_time": "0:08:42", "remaining_time": "0:14:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 192, "total_steps": 500, "loss": 0.9828, "learning_rate": 7.737816117462752e-05, "epoch": 1.7066666666666666, "percentage": 38.4, "elapsed_time": "0:08:45", "remaining_time": "0:14:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 193, "total_steps": 500, "loss": 0.9298, "learning_rate": 7.7085410514137e-05, "epoch": 1.7155555555555555, "percentage": 38.6, "elapsed_time": "0:08:47", "remaining_time": "0:13:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 194, "total_steps": 500, "loss": 0.9032, "learning_rate": 7.679133974894983e-05, "epoch": 1.7244444444444444, "percentage": 38.8, "elapsed_time": "0:08:51", "remaining_time": "0:13:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 195, "total_steps": 500, "loss": 1.0169, "learning_rate": 7.649596321166024e-05, "epoch": 1.7333333333333334, "percentage": 39.0, "elapsed_time": "0:08:53", "remaining_time": "0:13:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 196, "total_steps": 500, "loss": 0.8967, "learning_rate": 7.619929529850397e-05, "epoch": 1.7422222222222223, "percentage": 39.2, "elapsed_time": "0:08:56", "remaining_time": "0:13:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 197, "total_steps": 500, "loss": 0.926, "learning_rate": 7.590135046865651e-05, "epoch": 1.751111111111111, "percentage": 39.4, "elapsed_time": "0:08:59", "remaining_time": "0:13:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 198, "total_steps": 500, "loss": 1.0398, "learning_rate": 7.560214324352858e-05, "epoch": 1.76, "percentage": 39.6, "elapsed_time": "0:09:01", "remaining_time": "0:13:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 199, "total_steps": 500, "loss": 0.9307, "learning_rate": 7.530168820605818e-05, "epoch": 1.7688888888888887, "percentage": 39.8, "elapsed_time": "0:09:04", "remaining_time": "0:13:43", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 200, "total_steps": 500, "loss": 0.8868, "learning_rate": 7.500000000000001e-05, "epoch": 1.7777777777777777, "percentage": 40.0, "elapsed_time": "0:09:07", "remaining_time": "0:13:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 201, "total_steps": 500, "loss": 0.8817, "learning_rate": 7.469709332921155e-05, "epoch": 1.7866666666666666, "percentage": 40.2, "elapsed_time": "0:09:09", "remaining_time": "0:13:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 202, "total_steps": 500, "loss": 1.1957, "learning_rate": 7.439298295693665e-05, "epoch": 1.7955555555555556, "percentage": 40.4, "elapsed_time": "0:09:12", "remaining_time": "0:13:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 203, "total_steps": 500, "loss": 1.087, "learning_rate": 7.408768370508576e-05, "epoch": 1.8044444444444445, "percentage": 40.6, "elapsed_time": "0:09:14", "remaining_time": "0:13:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 204, "total_steps": 500, "loss": 0.9296, "learning_rate": 7.378121045351378e-05, "epoch": 1.8133333333333335, "percentage": 40.8, "elapsed_time": "0:09:17", "remaining_time": "0:13:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 205, "total_steps": 500, "loss": 1.0053, "learning_rate": 7.347357813929454e-05, "epoch": 1.8222222222222222, "percentage": 41.0, "elapsed_time": "0:09:19", "remaining_time": "0:13:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 206, "total_steps": 500, "loss": 0.8588, "learning_rate": 7.316480175599309e-05, "epoch": 1.8311111111111111, "percentage": 41.2, "elapsed_time": "0:09:22", "remaining_time": "0:13:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 207, "total_steps": 500, "loss": 0.8839, "learning_rate": 7.285489635293472e-05, "epoch": 1.8399999999999999, "percentage": 41.4, "elapsed_time": "0:09:26", "remaining_time": "0:13:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 208, "total_steps": 500, "loss": 0.9149, "learning_rate": 7.254387703447154e-05, "epoch": 1.8488888888888888, "percentage": 41.6, "elapsed_time": "0:09:29", "remaining_time": "0:13:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 209, "total_steps": 500, "loss": 0.8972, "learning_rate": 7.223175895924638e-05, "epoch": 1.8577777777777778, "percentage": 41.8, "elapsed_time": "0:09:31", "remaining_time": "0:13:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 210, "total_steps": 500, "loss": 1.0323, "learning_rate": 7.191855733945387e-05, "epoch": 1.8666666666666667, "percentage": 42.0, "elapsed_time": "0:09:34", "remaining_time": "0:13:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 211, "total_steps": 500, "loss": 1.0817, "learning_rate": 7.160428744009912e-05, "epoch": 1.8755555555555556, "percentage": 42.2, "elapsed_time": "0:09:37", "remaining_time": "0:13:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 212, "total_steps": 500, "loss": 1.0261, "learning_rate": 7.128896457825364e-05, "epoch": 1.8844444444444446, "percentage": 42.4, "elapsed_time": "0:09:39", "remaining_time": "0:13:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 213, "total_steps": 500, "loss": 0.9723, "learning_rate": 7.097260412230886e-05, "epoch": 1.8933333333333333, "percentage": 42.6, "elapsed_time": "0:09:42", "remaining_time": "0:13:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 214, "total_steps": 500, "loss": 1.1117, "learning_rate": 7.06552214912271e-05, "epoch": 1.9022222222222223, "percentage": 42.8, "elapsed_time": "0:09:45", "remaining_time": "0:13:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 215, "total_steps": 500, "loss": 0.8511, "learning_rate": 7.033683215379002e-05, "epoch": 1.911111111111111, "percentage": 43.0, "elapsed_time": "0:09:48", "remaining_time": "0:13:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 216, "total_steps": 500, "loss": 0.9363, "learning_rate": 7.001745162784477e-05, "epoch": 1.92, "percentage": 43.2, "elapsed_time": "0:09:51", "remaining_time": "0:12:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 217, "total_steps": 500, "loss": 0.9447, "learning_rate": 6.969709547954756e-05, "epoch": 1.9288888888888889, "percentage": 43.4, "elapsed_time": "0:09:54", "remaining_time": "0:12:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 218, "total_steps": 500, "loss": 0.7898, "learning_rate": 6.937577932260515e-05, "epoch": 1.9377777777777778, "percentage": 43.6, "elapsed_time": "0:09:56", "remaining_time": "0:12:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 219, "total_steps": 500, "loss": 0.8289, "learning_rate": 6.905351881751372e-05, "epoch": 1.9466666666666668, "percentage": 43.8, "elapsed_time": "0:09:58", "remaining_time": "0:12:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 220, "total_steps": 500, "loss": 0.8569, "learning_rate": 6.873032967079561e-05, "epoch": 1.9555555555555557, "percentage": 44.0, "elapsed_time": "0:10:01", "remaining_time": "0:12:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 221, "total_steps": 500, "loss": 0.8994, "learning_rate": 6.840622763423391e-05, "epoch": 1.9644444444444444, "percentage": 44.2, "elapsed_time": "0:10:04", "remaining_time": "0:12:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 222, "total_steps": 500, "loss": 0.8751, "learning_rate": 6.808122850410461e-05, "epoch": 1.9733333333333334, "percentage": 44.4, "elapsed_time": "0:10:06", "remaining_time": "0:12:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 223, "total_steps": 500, "loss": 1.0663, "learning_rate": 6.775534812040685e-05, "epoch": 1.982222222222222, "percentage": 44.6, "elapsed_time": "0:10:08", "remaining_time": "0:12:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 224, "total_steps": 500, "loss": 1.1425, "learning_rate": 6.742860236609077e-05, "epoch": 1.991111111111111, "percentage": 44.8, "elapsed_time": "0:10:11", "remaining_time": "0:12:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 225, "total_steps": 500, "loss": 0.9115, "learning_rate": 6.710100716628344e-05, "epoch": 2.0, "percentage": 45.0, "elapsed_time": "0:10:13", "remaining_time": "0:12:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 226, "total_steps": 500, "loss": 0.9296, "learning_rate": 6.677257848751277e-05, "epoch": 2.008888888888889, "percentage": 45.2, "elapsed_time": "0:10:16", "remaining_time": "0:12:27", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 227, "total_steps": 500, "loss": 1.0117, "learning_rate": 6.644333233692916e-05, "epoch": 2.017777777777778, "percentage": 45.4, "elapsed_time": "0:10:19", "remaining_time": "0:12:24", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 228, "total_steps": 500, "loss": 0.9275, "learning_rate": 6.611328476152557e-05, "epoch": 2.026666666666667, "percentage": 45.6, "elapsed_time": "0:10:21", "remaining_time": "0:12:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 229, "total_steps": 500, "loss": 0.9839, "learning_rate": 6.578245184735513e-05, "epoch": 2.0355555555555553, "percentage": 45.8, "elapsed_time": "0:10:23", "remaining_time": "0:12:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 230, "total_steps": 500, "loss": 0.6496, "learning_rate": 6.545084971874738e-05, "epoch": 2.0444444444444443, "percentage": 46.0, "elapsed_time": "0:10:26", "remaining_time": "0:12:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 231, "total_steps": 500, "loss": 0.903, "learning_rate": 6.511849453752223e-05, "epoch": 2.0533333333333332, "percentage": 46.2, "elapsed_time": "0:10:28", "remaining_time": "0:12:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 232, "total_steps": 500, "loss": 0.9134, "learning_rate": 6.478540250220234e-05, "epoch": 2.062222222222222, "percentage": 46.4, "elapsed_time": "0:10:31", "remaining_time": "0:12:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 233, "total_steps": 500, "loss": 0.7132, "learning_rate": 6.445158984722358e-05, "epoch": 2.071111111111111, "percentage": 46.6, "elapsed_time": "0:10:34", "remaining_time": "0:12:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 234, "total_steps": 500, "loss": 0.7823, "learning_rate": 6.411707284214384e-05, "epoch": 2.08, "percentage": 46.8, "elapsed_time": "0:10:36", "remaining_time": "0:12:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 235, "total_steps": 500, "loss": 0.8642, "learning_rate": 6.378186779084995e-05, "epoch": 2.088888888888889, "percentage": 47.0, "elapsed_time": "0:10:38", "remaining_time": "0:12:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 236, "total_steps": 500, "loss": 0.8489, "learning_rate": 6.344599103076329e-05, "epoch": 2.097777777777778, "percentage": 47.2, "elapsed_time": "0:10:41", "remaining_time": "0:11:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 237, "total_steps": 500, "loss": 0.8013, "learning_rate": 6.310945893204324e-05, "epoch": 2.1066666666666665, "percentage": 47.4, "elapsed_time": "0:10:43", "remaining_time": "0:11:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 238, "total_steps": 500, "loss": 0.8753, "learning_rate": 6.277228789678953e-05, "epoch": 2.1155555555555554, "percentage": 47.6, "elapsed_time": "0:10:46", "remaining_time": "0:11:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 239, "total_steps": 500, "loss": 0.8545, "learning_rate": 6.243449435824276e-05, "epoch": 2.1244444444444444, "percentage": 47.8, "elapsed_time": "0:10:49", "remaining_time": "0:11:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 240, "total_steps": 500, "loss": 0.7864, "learning_rate": 6.209609477998338e-05, "epoch": 2.1333333333333333, "percentage": 48.0, "elapsed_time": "0:10:51", "remaining_time": "0:11:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 241, "total_steps": 500, "loss": 0.9452, "learning_rate": 6.17571056551295e-05, "epoch": 2.1422222222222222, "percentage": 48.2, "elapsed_time": "0:10:54", "remaining_time": "0:11:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 242, "total_steps": 500, "loss": 0.8245, "learning_rate": 6.141754350553279e-05, "epoch": 2.151111111111111, "percentage": 48.4, "elapsed_time": "0:10:56", "remaining_time": "0:11:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 243, "total_steps": 500, "loss": 0.8336, "learning_rate": 6.107742488097338e-05, "epoch": 2.16, "percentage": 48.6, "elapsed_time": "0:10:59", "remaining_time": "0:11:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 244, "total_steps": 500, "loss": 0.729, "learning_rate": 6.073676635835317e-05, "epoch": 2.168888888888889, "percentage": 48.8, "elapsed_time": "0:11:01", "remaining_time": "0:11:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 245, "total_steps": 500, "loss": 0.7696, "learning_rate": 6.0395584540887963e-05, "epoch": 2.1777777777777776, "percentage": 49.0, "elapsed_time": "0:11:04", "remaining_time": "0:11:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 246, "total_steps": 500, "loss": 0.754, "learning_rate": 6.005389605729824e-05, "epoch": 2.1866666666666665, "percentage": 49.2, "elapsed_time": "0:11:07", "remaining_time": "0:11:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 247, "total_steps": 500, "loss": 0.8451, "learning_rate": 5.97117175609986e-05, "epoch": 2.1955555555555555, "percentage": 49.4, "elapsed_time": "0:11:09", "remaining_time": "0:11:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 248, "total_steps": 500, "loss": 0.9647, "learning_rate": 5.9369065729286245e-05, "epoch": 2.2044444444444444, "percentage": 49.6, "elapsed_time": "0:11:12", "remaining_time": "0:11:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 249, "total_steps": 500, "loss": 0.9317, "learning_rate": 5.902595726252801e-05, "epoch": 2.2133333333333334, "percentage": 49.8, "elapsed_time": "0:11:15", "remaining_time": "0:11:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 250, "total_steps": 500, "loss": 1.1104, "learning_rate": 5.868240888334653e-05, "epoch": 2.2222222222222223, "percentage": 50.0, "elapsed_time": "0:11:17", "remaining_time": "0:11:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 251, "total_steps": 500, "loss": 0.8364, "learning_rate": 5.833843733580512e-05, "epoch": 2.2311111111111113, "percentage": 50.2, "elapsed_time": "0:11:20", "remaining_time": "0:11:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 252, "total_steps": 500, "loss": 0.9288, "learning_rate": 5.799405938459175e-05, "epoch": 2.24, "percentage": 50.4, "elapsed_time": "0:11:22", "remaining_time": "0:11:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 253, "total_steps": 500, "loss": 0.7471, "learning_rate": 5.764929181420191e-05, "epoch": 2.2488888888888887, "percentage": 50.6, "elapsed_time": "0:11:25", "remaining_time": "0:11:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 254, "total_steps": 500, "loss": 0.826, "learning_rate": 5.730415142812059e-05, "epoch": 2.2577777777777777, "percentage": 50.8, "elapsed_time": "0:11:27", "remaining_time": "0:11:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 255, "total_steps": 500, "loss": 0.9825, "learning_rate": 5.695865504800327e-05, "epoch": 2.2666666666666666, "percentage": 51.0, "elapsed_time": "0:11:30", "remaining_time": "0:11:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 256, "total_steps": 500, "loss": 0.8868, "learning_rate": 5.661281951285613e-05, "epoch": 2.2755555555555556, "percentage": 51.2, "elapsed_time": "0:11:33", "remaining_time": "0:11:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 257, "total_steps": 500, "loss": 0.8671, "learning_rate": 5.6266661678215216e-05, "epoch": 2.2844444444444445, "percentage": 51.4, "elapsed_time": "0:11:35", "remaining_time": "0:10:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 258, "total_steps": 500, "loss": 0.6903, "learning_rate": 5.5920198415325064e-05, "epoch": 2.2933333333333334, "percentage": 51.6, "elapsed_time": "0:11:37", "remaining_time": "0:10:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 259, "total_steps": 500, "loss": 0.7786, "learning_rate": 5.557344661031627e-05, "epoch": 2.3022222222222224, "percentage": 51.8, "elapsed_time": "0:11:40", "remaining_time": "0:10:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 260, "total_steps": 500, "loss": 1.0231, "learning_rate": 5.522642316338268e-05, "epoch": 2.311111111111111, "percentage": 52.0, "elapsed_time": "0:11:43", "remaining_time": "0:10:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 261, "total_steps": 500, "loss": 0.7585, "learning_rate": 5.487914498795747e-05, "epoch": 2.32, "percentage": 52.2, "elapsed_time": "0:11:45", "remaining_time": "0:10:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 262, "total_steps": 500, "loss": 0.9553, "learning_rate": 5.453162900988902e-05, "epoch": 2.328888888888889, "percentage": 52.4, "elapsed_time": "0:11:48", "remaining_time": "0:10:43", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 263, "total_steps": 500, "loss": 0.7379, "learning_rate": 5.418389216661579e-05, "epoch": 2.3377777777777777, "percentage": 52.6, "elapsed_time": "0:11:51", "remaining_time": "0:10:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 264, "total_steps": 500, "loss": 0.7197, "learning_rate": 5.383595140634093e-05, "epoch": 2.3466666666666667, "percentage": 52.8, "elapsed_time": "0:11:53", "remaining_time": "0:10:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 265, "total_steps": 500, "loss": 0.9332, "learning_rate": 5.348782368720626e-05, "epoch": 2.3555555555555556, "percentage": 53.0, "elapsed_time": "0:11:55", "remaining_time": "0:10:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 266, "total_steps": 500, "loss": 0.8006, "learning_rate": 5.313952597646568e-05, "epoch": 2.3644444444444446, "percentage": 53.2, "elapsed_time": "0:11:58", "remaining_time": "0:10:32", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 267, "total_steps": 500, "loss": 0.8995, "learning_rate": 5.279107524965819e-05, "epoch": 2.3733333333333335, "percentage": 53.4, "elapsed_time": "0:12:01", "remaining_time": "0:10:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 268, "total_steps": 500, "loss": 1.009, "learning_rate": 5.244248848978067e-05, "epoch": 2.3822222222222225, "percentage": 53.6, "elapsed_time": "0:12:03", "remaining_time": "0:10:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 269, "total_steps": 500, "loss": 1.0106, "learning_rate": 5.209378268645998e-05, "epoch": 2.391111111111111, "percentage": 53.8, "elapsed_time": "0:12:07", "remaining_time": "0:10:24", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 270, "total_steps": 500, "loss": 0.8872, "learning_rate": 5.174497483512506e-05, "epoch": 2.4, "percentage": 54.0, "elapsed_time": "0:12:10", "remaining_time": "0:10:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 271, "total_steps": 500, "loss": 0.7844, "learning_rate": 5.139608193617845e-05, "epoch": 2.408888888888889, "percentage": 54.2, "elapsed_time": "0:12:12", "remaining_time": "0:10:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 272, "total_steps": 500, "loss": 0.9682, "learning_rate": 5.104712099416785e-05, "epoch": 2.417777777777778, "percentage": 54.4, "elapsed_time": "0:12:15", "remaining_time": "0:10:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 273, "total_steps": 500, "loss": 0.7677, "learning_rate": 5.0698109016957274e-05, "epoch": 2.4266666666666667, "percentage": 54.6, "elapsed_time": "0:12:17", "remaining_time": "0:10:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 274, "total_steps": 500, "loss": 0.807, "learning_rate": 5.034906301489808e-05, "epoch": 2.4355555555555557, "percentage": 54.8, "elapsed_time": "0:12:20", "remaining_time": "0:10:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 275, "total_steps": 500, "loss": 0.8616, "learning_rate": 5e-05, "epoch": 2.4444444444444446, "percentage": 55.0, "elapsed_time": "0:12:22", "remaining_time": "0:10:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 276, "total_steps": 500, "loss": 0.8555, "learning_rate": 4.965093698510193e-05, "epoch": 2.453333333333333, "percentage": 55.2, "elapsed_time": "0:12:25", "remaining_time": "0:10:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 277, "total_steps": 500, "loss": 0.9175, "learning_rate": 4.9301890983042744e-05, "epoch": 2.462222222222222, "percentage": 55.4, "elapsed_time": "0:12:28", "remaining_time": "0:10:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 278, "total_steps": 500, "loss": 0.8347, "learning_rate": 4.895287900583216e-05, "epoch": 2.471111111111111, "percentage": 55.6, "elapsed_time": "0:12:30", "remaining_time": "0:09:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 279, "total_steps": 500, "loss": 0.6997, "learning_rate": 4.860391806382157e-05, "epoch": 2.48, "percentage": 55.8, "elapsed_time": "0:12:32", "remaining_time": "0:09:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 280, "total_steps": 500, "loss": 0.7535, "learning_rate": 4.825502516487497e-05, "epoch": 2.488888888888889, "percentage": 56.0, "elapsed_time": "0:12:35", "remaining_time": "0:09:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 281, "total_steps": 500, "loss": 0.6498, "learning_rate": 4.790621731354003e-05, "epoch": 2.497777777777778, "percentage": 56.2, "elapsed_time": "0:12:38", "remaining_time": "0:09:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 282, "total_steps": 500, "loss": 0.8289, "learning_rate": 4.755751151021934e-05, "epoch": 2.506666666666667, "percentage": 56.4, "elapsed_time": "0:12:40", "remaining_time": "0:09:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 283, "total_steps": 500, "loss": 0.7781, "learning_rate": 4.720892475034181e-05, "epoch": 2.5155555555555553, "percentage": 56.6, "elapsed_time": "0:12:42", "remaining_time": "0:09:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 284, "total_steps": 500, "loss": 0.887, "learning_rate": 4.6860474023534335e-05, "epoch": 2.5244444444444447, "percentage": 56.8, "elapsed_time": "0:12:45", "remaining_time": "0:09:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 285, "total_steps": 500, "loss": 0.9711, "learning_rate": 4.6512176312793736e-05, "epoch": 2.533333333333333, "percentage": 57.0, "elapsed_time": "0:12:48", "remaining_time": "0:09:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 286, "total_steps": 500, "loss": 0.7761, "learning_rate": 4.616404859365907e-05, "epoch": 2.542222222222222, "percentage": 57.2, "elapsed_time": "0:12:50", "remaining_time": "0:09:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 287, "total_steps": 500, "loss": 0.8821, "learning_rate": 4.5816107833384234e-05, "epoch": 2.551111111111111, "percentage": 57.4, "elapsed_time": "0:12:53", "remaining_time": "0:09:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 288, "total_steps": 500, "loss": 0.9093, "learning_rate": 4.5468370990111006e-05, "epoch": 2.56, "percentage": 57.6, "elapsed_time": "0:12:56", "remaining_time": "0:09:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 289, "total_steps": 500, "loss": 0.8367, "learning_rate": 4.512085501204253e-05, "epoch": 2.568888888888889, "percentage": 57.8, "elapsed_time": "0:12:59", "remaining_time": "0:09:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 290, "total_steps": 500, "loss": 0.8892, "learning_rate": 4.477357683661734e-05, "epoch": 2.5777777777777775, "percentage": 58.0, "elapsed_time": "0:13:01", "remaining_time": "0:09:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 291, "total_steps": 500, "loss": 0.8095, "learning_rate": 4.442655338968373e-05, "epoch": 2.586666666666667, "percentage": 58.2, "elapsed_time": "0:13:04", "remaining_time": "0:09:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 292, "total_steps": 500, "loss": 0.6999, "learning_rate": 4.407980158467495e-05, "epoch": 2.5955555555555554, "percentage": 58.4, "elapsed_time": "0:13:06", "remaining_time": "0:09:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 293, "total_steps": 500, "loss": 0.9533, "learning_rate": 4.373333832178478e-05, "epoch": 2.6044444444444443, "percentage": 58.6, "elapsed_time": "0:13:08", "remaining_time": "0:09:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 294, "total_steps": 500, "loss": 1.0029, "learning_rate": 4.3387180487143876e-05, "epoch": 2.6133333333333333, "percentage": 58.8, "elapsed_time": "0:13:11", "remaining_time": "0:09:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 295, "total_steps": 500, "loss": 0.8176, "learning_rate": 4.3041344951996746e-05, "epoch": 2.6222222222222222, "percentage": 59.0, "elapsed_time": "0:13:14", "remaining_time": "0:09:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 296, "total_steps": 500, "loss": 0.7422, "learning_rate": 4.269584857187943e-05, "epoch": 2.631111111111111, "percentage": 59.2, "elapsed_time": "0:13:16", "remaining_time": "0:09:08", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 297, "total_steps": 500, "loss": 0.9981, "learning_rate": 4.23507081857981e-05, "epoch": 2.64, "percentage": 59.4, "elapsed_time": "0:13:19", "remaining_time": "0:09:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 298, "total_steps": 500, "loss": 0.7285, "learning_rate": 4.2005940615408264e-05, "epoch": 2.648888888888889, "percentage": 59.6, "elapsed_time": "0:13:21", "remaining_time": "0:09:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 299, "total_steps": 500, "loss": 0.7904, "learning_rate": 4.166156266419489e-05, "epoch": 2.6577777777777776, "percentage": 59.8, "elapsed_time": "0:13:24", "remaining_time": "0:09:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 300, "total_steps": 500, "loss": 0.8762, "learning_rate": 4.131759111665349e-05, "epoch": 2.6666666666666665, "percentage": 60.0, "elapsed_time": "0:13:26", "remaining_time": "0:08:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 301, "total_steps": 500, "loss": 0.7147, "learning_rate": 4.0974042737472006e-05, "epoch": 2.6755555555555555, "percentage": 60.2, "elapsed_time": "0:13:28", "remaining_time": "0:08:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 302, "total_steps": 500, "loss": 0.8428, "learning_rate": 4.063093427071376e-05, "epoch": 2.6844444444444444, "percentage": 60.4, "elapsed_time": "0:13:31", "remaining_time": "0:08:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 303, "total_steps": 500, "loss": 0.9927, "learning_rate": 4.028828243900141e-05, "epoch": 2.6933333333333334, "percentage": 60.6, "elapsed_time": "0:13:33", "remaining_time": "0:08:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 304, "total_steps": 500, "loss": 0.882, "learning_rate": 3.9946103942701777e-05, "epoch": 2.7022222222222223, "percentage": 60.8, "elapsed_time": "0:13:35", "remaining_time": "0:08:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 305, "total_steps": 500, "loss": 0.9207, "learning_rate": 3.960441545911204e-05, "epoch": 2.7111111111111112, "percentage": 61.0, "elapsed_time": "0:13:38", "remaining_time": "0:08:43", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 306, "total_steps": 500, "loss": 0.9289, "learning_rate": 3.926323364164684e-05, "epoch": 2.7199999999999998, "percentage": 61.2, "elapsed_time": "0:13:40", "remaining_time": "0:08:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 307, "total_steps": 500, "loss": 0.972, "learning_rate": 3.892257511902664e-05, "epoch": 2.728888888888889, "percentage": 61.4, "elapsed_time": "0:13:43", "remaining_time": "0:08:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 308, "total_steps": 500, "loss": 0.8648, "learning_rate": 3.858245649446721e-05, "epoch": 2.7377777777777776, "percentage": 61.6, "elapsed_time": "0:13:45", "remaining_time": "0:08:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 309, "total_steps": 500, "loss": 0.7915, "learning_rate": 3.82428943448705e-05, "epoch": 2.7466666666666666, "percentage": 61.8, "elapsed_time": "0:13:48", "remaining_time": "0:08:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 310, "total_steps": 500, "loss": 0.9658, "learning_rate": 3.790390522001662e-05, "epoch": 2.7555555555555555, "percentage": 62.0, "elapsed_time": "0:13:50", "remaining_time": "0:08:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 311, "total_steps": 500, "loss": 0.9002, "learning_rate": 3.756550564175727e-05, "epoch": 2.7644444444444445, "percentage": 62.2, "elapsed_time": "0:13:53", "remaining_time": "0:08:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 312, "total_steps": 500, "loss": 0.671, "learning_rate": 3.7227712103210486e-05, "epoch": 2.7733333333333334, "percentage": 62.4, "elapsed_time": "0:13:55", "remaining_time": "0:08:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 313, "total_steps": 500, "loss": 0.8574, "learning_rate": 3.6890541067956776e-05, "epoch": 2.7822222222222224, "percentage": 62.6, "elapsed_time": "0:13:58", "remaining_time": "0:08:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 314, "total_steps": 500, "loss": 1.0365, "learning_rate": 3.655400896923672e-05, "epoch": 2.7911111111111113, "percentage": 62.8, "elapsed_time": "0:14:00", "remaining_time": "0:08:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 315, "total_steps": 500, "loss": 0.8598, "learning_rate": 3.6218132209150045e-05, "epoch": 2.8, "percentage": 63.0, "elapsed_time": "0:14:03", "remaining_time": "0:08:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 316, "total_steps": 500, "loss": 0.8167, "learning_rate": 3.588292715785617e-05, "epoch": 2.8088888888888888, "percentage": 63.2, "elapsed_time": "0:14:05", "remaining_time": "0:08:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 317, "total_steps": 500, "loss": 0.8554, "learning_rate": 3.554841015277641e-05, "epoch": 2.8177777777777777, "percentage": 63.4, "elapsed_time": "0:14:07", "remaining_time": "0:08:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 318, "total_steps": 500, "loss": 0.9625, "learning_rate": 3.5214597497797684e-05, "epoch": 2.8266666666666667, "percentage": 63.6, "elapsed_time": "0:14:10", "remaining_time": "0:08:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 319, "total_steps": 500, "loss": 0.9035, "learning_rate": 3.488150546247778e-05, "epoch": 2.8355555555555556, "percentage": 63.8, "elapsed_time": "0:14:12", "remaining_time": "0:08:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 320, "total_steps": 500, "loss": 0.85, "learning_rate": 3.4549150281252636e-05, "epoch": 2.8444444444444446, "percentage": 64.0, "elapsed_time": "0:14:15", "remaining_time": "0:08:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 321, "total_steps": 500, "loss": 1.1274, "learning_rate": 3.4217548152644885e-05, "epoch": 2.8533333333333335, "percentage": 64.2, "elapsed_time": "0:14:17", "remaining_time": "0:07:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 322, "total_steps": 500, "loss": 0.7901, "learning_rate": 3.388671523847445e-05, "epoch": 2.862222222222222, "percentage": 64.4, "elapsed_time": "0:14:20", "remaining_time": "0:07:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 323, "total_steps": 500, "loss": 0.9324, "learning_rate": 3.355666766307084e-05, "epoch": 2.871111111111111, "percentage": 64.6, "elapsed_time": "0:14:23", "remaining_time": "0:07:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 324, "total_steps": 500, "loss": 0.8024, "learning_rate": 3.322742151248725e-05, "epoch": 2.88, "percentage": 64.8, "elapsed_time": "0:14:26", "remaining_time": "0:07:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 325, "total_steps": 500, "loss": 0.9827, "learning_rate": 3.289899283371657e-05, "epoch": 2.888888888888889, "percentage": 65.0, "elapsed_time": "0:14:29", "remaining_time": "0:07:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 326, "total_steps": 500, "loss": 0.8115, "learning_rate": 3.257139763390925e-05, "epoch": 2.897777777777778, "percentage": 65.2, "elapsed_time": "0:14:31", "remaining_time": "0:07:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 327, "total_steps": 500, "loss": 0.8628, "learning_rate": 3.224465187959316e-05, "epoch": 2.9066666666666667, "percentage": 65.4, "elapsed_time": "0:14:34", "remaining_time": "0:07:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 328, "total_steps": 500, "loss": 0.8138, "learning_rate": 3.1918771495895396e-05, "epoch": 2.9155555555555557, "percentage": 65.6, "elapsed_time": "0:14:36", "remaining_time": "0:07:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 329, "total_steps": 500, "loss": 0.7686, "learning_rate": 3.1593772365766105e-05, "epoch": 2.924444444444444, "percentage": 65.8, "elapsed_time": "0:14:39", "remaining_time": "0:07:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 330, "total_steps": 500, "loss": 0.861, "learning_rate": 3.12696703292044e-05, "epoch": 2.9333333333333336, "percentage": 66.0, "elapsed_time": "0:14:41", "remaining_time": "0:07:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 331, "total_steps": 500, "loss": 0.7815, "learning_rate": 3.09464811824863e-05, "epoch": 2.942222222222222, "percentage": 66.2, "elapsed_time": "0:14:44", "remaining_time": "0:07:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 332, "total_steps": 500, "loss": 0.8253, "learning_rate": 3.062422067739485e-05, "epoch": 2.951111111111111, "percentage": 66.4, "elapsed_time": "0:14:46", "remaining_time": "0:07:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 333, "total_steps": 500, "loss": 0.8101, "learning_rate": 3.0302904520452447e-05, "epoch": 2.96, "percentage": 66.6, "elapsed_time": "0:14:49", "remaining_time": "0:07:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 334, "total_steps": 500, "loss": 0.828, "learning_rate": 2.9982548372155263e-05, "epoch": 2.968888888888889, "percentage": 66.8, "elapsed_time": "0:14:52", "remaining_time": "0:07:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 335, "total_steps": 500, "loss": 0.7968, "learning_rate": 2.9663167846209998e-05, "epoch": 2.977777777777778, "percentage": 67.0, "elapsed_time": "0:14:54", "remaining_time": "0:07:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 336, "total_steps": 500, "loss": 1.0188, "learning_rate": 2.934477850877292e-05, "epoch": 2.986666666666667, "percentage": 67.2, "elapsed_time": "0:14:56", "remaining_time": "0:07:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 337, "total_steps": 500, "loss": 0.8255, "learning_rate": 2.9027395877691144e-05, "epoch": 2.9955555555555557, "percentage": 67.4, "elapsed_time": "0:14:58", "remaining_time": "0:07:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 338, "total_steps": 500, "loss": 0.6085, "learning_rate": 2.8711035421746367e-05, "epoch": 3.0044444444444443, "percentage": 67.6, "elapsed_time": "0:15:01", "remaining_time": "0:07:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 339, "total_steps": 500, "loss": 0.7823, "learning_rate": 2.8395712559900877e-05, "epoch": 3.013333333333333, "percentage": 67.8, "elapsed_time": "0:15:03", "remaining_time": "0:07:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 340, "total_steps": 500, "loss": 0.8947, "learning_rate": 2.8081442660546125e-05, "epoch": 3.022222222222222, "percentage": 68.0, "elapsed_time": "0:15:06", "remaining_time": "0:07:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 341, "total_steps": 500, "loss": 0.8595, "learning_rate": 2.776824104075364e-05, "epoch": 3.031111111111111, "percentage": 68.2, "elapsed_time": "0:15:09", "remaining_time": "0:07:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 342, "total_steps": 500, "loss": 0.829, "learning_rate": 2.7456122965528475e-05, "epoch": 3.04, "percentage": 68.4, "elapsed_time": "0:15:11", "remaining_time": "0:07:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 343, "total_steps": 500, "loss": 0.5761, "learning_rate": 2.7145103647065308e-05, "epoch": 3.048888888888889, "percentage": 68.6, "elapsed_time": "0:15:14", "remaining_time": "0:06:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 344, "total_steps": 500, "loss": 0.7939, "learning_rate": 2.6835198244006927e-05, "epoch": 3.057777777777778, "percentage": 68.8, "elapsed_time": "0:15:16", "remaining_time": "0:06:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 345, "total_steps": 500, "loss": 0.6288, "learning_rate": 2.6526421860705473e-05, "epoch": 3.066666666666667, "percentage": 69.0, "elapsed_time": "0:15:18", "remaining_time": "0:06:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 346, "total_steps": 500, "loss": 0.7359, "learning_rate": 2.6218789546486234e-05, "epoch": 3.0755555555555554, "percentage": 69.2, "elapsed_time": "0:15:21", "remaining_time": "0:06:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 347, "total_steps": 500, "loss": 0.7755, "learning_rate": 2.591231629491423e-05, "epoch": 3.0844444444444443, "percentage": 69.4, "elapsed_time": "0:15:23", "remaining_time": "0:06:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 348, "total_steps": 500, "loss": 0.6429, "learning_rate": 2.560701704306336e-05, "epoch": 3.0933333333333333, "percentage": 69.6, "elapsed_time": "0:15:26", "remaining_time": "0:06:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 349, "total_steps": 500, "loss": 0.7758, "learning_rate": 2.5302906670788462e-05, "epoch": 3.102222222222222, "percentage": 69.8, "elapsed_time": "0:15:28", "remaining_time": "0:06:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 350, "total_steps": 500, "loss": 0.6468, "learning_rate": 2.500000000000001e-05, "epoch": 3.111111111111111, "percentage": 70.0, "elapsed_time": "0:15:31", "remaining_time": "0:06:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 351, "total_steps": 500, "loss": 0.6939, "learning_rate": 2.469831179394182e-05, "epoch": 3.12, "percentage": 70.2, "elapsed_time": "0:15:33", "remaining_time": "0:06:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 352, "total_steps": 500, "loss": 0.8737, "learning_rate": 2.4397856756471432e-05, "epoch": 3.128888888888889, "percentage": 70.4, "elapsed_time": "0:15:36", "remaining_time": "0:06:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 353, "total_steps": 500, "loss": 0.6743, "learning_rate": 2.4098649531343497e-05, "epoch": 3.137777777777778, "percentage": 70.6, "elapsed_time": "0:15:38", "remaining_time": "0:06:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 354, "total_steps": 500, "loss": 0.7847, "learning_rate": 2.3800704701496053e-05, "epoch": 3.1466666666666665, "percentage": 70.8, "elapsed_time": "0:15:41", "remaining_time": "0:06:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 355, "total_steps": 500, "loss": 0.8001, "learning_rate": 2.350403678833976e-05, "epoch": 3.1555555555555554, "percentage": 71.0, "elapsed_time": "0:15:43", "remaining_time": "0:06:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 356, "total_steps": 500, "loss": 0.7351, "learning_rate": 2.3208660251050158e-05, "epoch": 3.1644444444444444, "percentage": 71.2, "elapsed_time": "0:15:46", "remaining_time": "0:06:22", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 357, "total_steps": 500, "loss": 0.7505, "learning_rate": 2.2914589485863014e-05, "epoch": 3.1733333333333333, "percentage": 71.4, "elapsed_time": "0:15:48", "remaining_time": "0:06:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 358, "total_steps": 500, "loss": 0.8125, "learning_rate": 2.2621838825372493e-05, "epoch": 3.1822222222222223, "percentage": 71.6, "elapsed_time": "0:15:50", "remaining_time": "0:06:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 359, "total_steps": 500, "loss": 0.8101, "learning_rate": 2.23304225378328e-05, "epoch": 3.1911111111111112, "percentage": 71.8, "elapsed_time": "0:15:53", "remaining_time": "0:06:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 360, "total_steps": 500, "loss": 0.8313, "learning_rate": 2.2040354826462668e-05, "epoch": 3.2, "percentage": 72.0, "elapsed_time": "0:15:56", "remaining_time": "0:06:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 361, "total_steps": 500, "loss": 0.6678, "learning_rate": 2.1751649828753106e-05, "epoch": 3.2088888888888887, "percentage": 72.2, "elapsed_time": "0:15:58", "remaining_time": "0:06:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 362, "total_steps": 500, "loss": 0.7392, "learning_rate": 2.1464321615778422e-05, "epoch": 3.2177777777777776, "percentage": 72.4, "elapsed_time": "0:16:00", "remaining_time": "0:06:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 363, "total_steps": 500, "loss": 0.6166, "learning_rate": 2.117838419151034e-05, "epoch": 3.2266666666666666, "percentage": 72.6, "elapsed_time": "0:16:03", "remaining_time": "0:06:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 364, "total_steps": 500, "loss": 0.5738, "learning_rate": 2.0893851492135537e-05, "epoch": 3.2355555555555555, "percentage": 72.8, "elapsed_time": "0:16:05", "remaining_time": "0:06:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 365, "total_steps": 500, "loss": 0.6822, "learning_rate": 2.061073738537635e-05, "epoch": 3.2444444444444445, "percentage": 73.0, "elapsed_time": "0:16:08", "remaining_time": "0:05:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 366, "total_steps": 500, "loss": 0.9337, "learning_rate": 2.0329055669814934e-05, "epoch": 3.2533333333333334, "percentage": 73.2, "elapsed_time": "0:16:11", "remaining_time": "0:05:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 367, "total_steps": 500, "loss": 0.8167, "learning_rate": 2.0048820074220715e-05, "epoch": 3.2622222222222224, "percentage": 73.4, "elapsed_time": "0:16:13", "remaining_time": "0:05:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 368, "total_steps": 500, "loss": 0.7508, "learning_rate": 1.977004425688126e-05, "epoch": 3.2711111111111113, "percentage": 73.6, "elapsed_time": "0:16:16", "remaining_time": "0:05:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 369, "total_steps": 500, "loss": 0.745, "learning_rate": 1.9492741804936622e-05, "epoch": 3.2800000000000002, "percentage": 73.8, "elapsed_time": "0:16:18", "remaining_time": "0:05:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 370, "total_steps": 500, "loss": 0.8252, "learning_rate": 1.9216926233717085e-05, "epoch": 3.2888888888888888, "percentage": 74.0, "elapsed_time": "0:16:21", "remaining_time": "0:05:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 371, "total_steps": 500, "loss": 0.6201, "learning_rate": 1.8942610986084486e-05, "epoch": 3.2977777777777777, "percentage": 74.2, "elapsed_time": "0:16:23", "remaining_time": "0:05:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 372, "total_steps": 500, "loss": 0.6965, "learning_rate": 1.866980943177699e-05, "epoch": 3.3066666666666666, "percentage": 74.4, "elapsed_time": "0:16:26", "remaining_time": "0:05:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 373, "total_steps": 500, "loss": 0.7522, "learning_rate": 1.8398534866757454e-05, "epoch": 3.3155555555555556, "percentage": 74.6, "elapsed_time": "0:16:28", "remaining_time": "0:05:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 374, "total_steps": 500, "loss": 0.7584, "learning_rate": 1.8128800512565513e-05, "epoch": 3.3244444444444445, "percentage": 74.8, "elapsed_time": "0:16:30", "remaining_time": "0:05:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 375, "total_steps": 500, "loss": 0.6047, "learning_rate": 1.7860619515673033e-05, "epoch": 3.3333333333333335, "percentage": 75.0, "elapsed_time": "0:16:32", "remaining_time": "0:05:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 376, "total_steps": 500, "loss": 0.77, "learning_rate": 1.7594004946843456e-05, "epoch": 3.3422222222222224, "percentage": 75.2, "elapsed_time": "0:16:34", "remaining_time": "0:05:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 377, "total_steps": 500, "loss": 0.7815, "learning_rate": 1.7328969800494726e-05, "epoch": 3.351111111111111, "percentage": 75.4, "elapsed_time": "0:16:37", "remaining_time": "0:05:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 378, "total_steps": 500, "loss": 0.7838, "learning_rate": 1.7065526994065973e-05, "epoch": 3.36, "percentage": 75.6, "elapsed_time": "0:16:39", "remaining_time": "0:05:22", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 379, "total_steps": 500, "loss": 0.7368, "learning_rate": 1.680368936738792e-05, "epoch": 3.368888888888889, "percentage": 75.8, "elapsed_time": "0:16:41", "remaining_time": "0:05:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 380, "total_steps": 500, "loss": 0.8079, "learning_rate": 1.6543469682057106e-05, "epoch": 3.3777777777777778, "percentage": 76.0, "elapsed_time": "0:16:44", "remaining_time": "0:05:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 381, "total_steps": 500, "loss": 0.9025, "learning_rate": 1.6284880620813848e-05, "epoch": 3.3866666666666667, "percentage": 76.2, "elapsed_time": "0:16:47", "remaining_time": "0:05:14", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 382, "total_steps": 500, "loss": 0.8089, "learning_rate": 1.602793478692419e-05, "epoch": 3.3955555555555557, "percentage": 76.4, "elapsed_time": "0:16:49", "remaining_time": "0:05:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 383, "total_steps": 500, "loss": 0.6786, "learning_rate": 1.5772644703565565e-05, "epoch": 3.4044444444444446, "percentage": 76.6, "elapsed_time": "0:16:52", "remaining_time": "0:05:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 384, "total_steps": 500, "loss": 0.6152, "learning_rate": 1.551902281321651e-05, "epoch": 3.413333333333333, "percentage": 76.8, "elapsed_time": "0:16:54", "remaining_time": "0:05:06", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 385, "total_steps": 500, "loss": 0.5888, "learning_rate": 1.526708147705013e-05, "epoch": 3.422222222222222, "percentage": 77.0, "elapsed_time": "0:16:56", "remaining_time": "0:05:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 386, "total_steps": 500, "loss": 0.8046, "learning_rate": 1.5016832974331724e-05, "epoch": 3.431111111111111, "percentage": 77.2, "elapsed_time": "0:16:59", "remaining_time": "0:05:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 387, "total_steps": 500, "loss": 0.7949, "learning_rate": 1.4768289501820265e-05, "epoch": 3.44, "percentage": 77.4, "elapsed_time": "0:17:01", "remaining_time": "0:04:58", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 388, "total_steps": 500, "loss": 0.6952, "learning_rate": 1.4521463173173965e-05, "epoch": 3.448888888888889, "percentage": 77.6, "elapsed_time": "0:17:03", "remaining_time": "0:04:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 389, "total_steps": 500, "loss": 0.7142, "learning_rate": 1.4276366018359844e-05, "epoch": 3.457777777777778, "percentage": 77.8, "elapsed_time": "0:17:06", "remaining_time": "0:04:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 390, "total_steps": 500, "loss": 0.8104, "learning_rate": 1.4033009983067452e-05, "epoch": 3.466666666666667, "percentage": 78.0, "elapsed_time": "0:17:08", "remaining_time": "0:04:50", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 391, "total_steps": 500, "loss": 0.8418, "learning_rate": 1.3791406928126638e-05, "epoch": 3.4755555555555557, "percentage": 78.2, "elapsed_time": "0:17:11", "remaining_time": "0:04:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 392, "total_steps": 500, "loss": 0.6546, "learning_rate": 1.3551568628929434e-05, "epoch": 3.4844444444444447, "percentage": 78.4, "elapsed_time": "0:17:13", "remaining_time": "0:04:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 393, "total_steps": 500, "loss": 0.8296, "learning_rate": 1.3313506774856177e-05, "epoch": 3.493333333333333, "percentage": 78.6, "elapsed_time": "0:17:16", "remaining_time": "0:04:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 394, "total_steps": 500, "loss": 0.7203, "learning_rate": 1.3077232968705805e-05, "epoch": 3.502222222222222, "percentage": 78.8, "elapsed_time": "0:17:18", "remaining_time": "0:04:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 395, "total_steps": 500, "loss": 0.6301, "learning_rate": 1.2842758726130283e-05, "epoch": 3.511111111111111, "percentage": 79.0, "elapsed_time": "0:17:21", "remaining_time": "0:04:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 396, "total_steps": 500, "loss": 0.7751, "learning_rate": 1.2610095475073414e-05, "epoch": 3.52, "percentage": 79.2, "elapsed_time": "0:17:23", "remaining_time": "0:04:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 397, "total_steps": 500, "loss": 0.7448, "learning_rate": 1.2379254555213788e-05, "epoch": 3.528888888888889, "percentage": 79.4, "elapsed_time": "0:17:25", "remaining_time": "0:04:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 398, "total_steps": 500, "loss": 0.6768, "learning_rate": 1.2150247217412186e-05, "epoch": 3.537777777777778, "percentage": 79.6, "elapsed_time": "0:17:27", "remaining_time": "0:04:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 399, "total_steps": 500, "loss": 0.7193, "learning_rate": 1.1923084623163172e-05, "epoch": 3.546666666666667, "percentage": 79.8, "elapsed_time": "0:17:30", "remaining_time": "0:04:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 400, "total_steps": 500, "loss": 0.6328, "learning_rate": 1.1697777844051105e-05, "epoch": 3.5555555555555554, "percentage": 80.0, "elapsed_time": "0:17:33", "remaining_time": "0:04:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 401, "total_steps": 500, "loss": 0.727, "learning_rate": 1.1474337861210543e-05, "epoch": 3.5644444444444443, "percentage": 80.2, "elapsed_time": "0:17:35", "remaining_time": "0:04:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 402, "total_steps": 500, "loss": 0.7467, "learning_rate": 1.1252775564791024e-05, "epoch": 3.5733333333333333, "percentage": 80.4, "elapsed_time": "0:17:38", "remaining_time": "0:04:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 403, "total_steps": 500, "loss": 0.7672, "learning_rate": 1.1033101753426283e-05, "epoch": 3.582222222222222, "percentage": 80.6, "elapsed_time": "0:17:40", "remaining_time": "0:04:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 404, "total_steps": 500, "loss": 0.7215, "learning_rate": 1.0815327133708015e-05, "epoch": 3.591111111111111, "percentage": 80.8, "elapsed_time": "0:17:43", "remaining_time": "0:04:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 405, "total_steps": 500, "loss": 0.6078, "learning_rate": 1.0599462319663905e-05, "epoch": 3.6, "percentage": 81.0, "elapsed_time": "0:17:45", "remaining_time": "0:04:09", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 406, "total_steps": 500, "loss": 0.7894, "learning_rate": 1.0385517832240471e-05, "epoch": 3.608888888888889, "percentage": 81.2, "elapsed_time": "0:17:47", "remaining_time": "0:04:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 407, "total_steps": 500, "loss": 0.7391, "learning_rate": 1.0173504098790187e-05, "epoch": 3.6177777777777775, "percentage": 81.4, "elapsed_time": "0:17:49", "remaining_time": "0:04:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 408, "total_steps": 500, "loss": 0.6962, "learning_rate": 9.963431452563332e-06, "epoch": 3.626666666666667, "percentage": 81.6, "elapsed_time": "0:17:52", "remaining_time": "0:04:01", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 409, "total_steps": 500, "loss": 0.8193, "learning_rate": 9.755310132204298e-06, "epoch": 3.6355555555555554, "percentage": 81.8, "elapsed_time": "0:17:54", "remaining_time": "0:03:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 410, "total_steps": 500, "loss": 0.7433, "learning_rate": 9.549150281252633e-06, "epoch": 3.6444444444444444, "percentage": 82.0, "elapsed_time": "0:17:57", "remaining_time": "0:03:56", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 411, "total_steps": 500, "loss": 0.5545, "learning_rate": 9.344961947648623e-06, "epoch": 3.6533333333333333, "percentage": 82.2, "elapsed_time": "0:17:59", "remaining_time": "0:03:53", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 412, "total_steps": 500, "loss": 0.938, "learning_rate": 9.142755083243576e-06, "epoch": 3.6622222222222223, "percentage": 82.4, "elapsed_time": "0:18:01", "remaining_time": "0:03:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 413, "total_steps": 500, "loss": 0.8629, "learning_rate": 8.9425395433148e-06, "epoch": 3.671111111111111, "percentage": 82.6, "elapsed_time": "0:18:04", "remaining_time": "0:03:48", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 414, "total_steps": 500, "loss": 0.8981, "learning_rate": 8.744325086085248e-06, "epoch": 3.68, "percentage": 82.8, "elapsed_time": "0:18:06", "remaining_time": "0:03:45", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 415, "total_steps": 500, "loss": 0.7164, "learning_rate": 8.548121372247918e-06, "epoch": 3.688888888888889, "percentage": 83.0, "elapsed_time": "0:18:08", "remaining_time": "0:03:42", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 416, "total_steps": 500, "loss": 0.7385, "learning_rate": 8.353937964495029e-06, "epoch": 3.6977777777777776, "percentage": 83.2, "elapsed_time": "0:18:11", "remaining_time": "0:03:40", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 417, "total_steps": 500, "loss": 0.6595, "learning_rate": 8.16178432705192e-06, "epoch": 3.7066666666666666, "percentage": 83.4, "elapsed_time": "0:18:13", "remaining_time": "0:03:37", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 418, "total_steps": 500, "loss": 0.7062, "learning_rate": 7.971669825215788e-06, "epoch": 3.7155555555555555, "percentage": 83.6, "elapsed_time": "0:18:15", "remaining_time": "0:03:34", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 419, "total_steps": 500, "loss": 0.8843, "learning_rate": 7.783603724899257e-06, "epoch": 3.7244444444444444, "percentage": 83.8, "elapsed_time": "0:18:17", "remaining_time": "0:03:32", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 420, "total_steps": 500, "loss": 0.6351, "learning_rate": 7.597595192178702e-06, "epoch": 3.7333333333333334, "percentage": 84.0, "elapsed_time": "0:18:20", "remaining_time": "0:03:29", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 421, "total_steps": 500, "loss": 0.7729, "learning_rate": 7.413653292847617e-06, "epoch": 3.7422222222222223, "percentage": 84.2, "elapsed_time": "0:18:23", "remaining_time": "0:03:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 422, "total_steps": 500, "loss": 0.8741, "learning_rate": 7.2317869919746705e-06, "epoch": 3.7511111111111113, "percentage": 84.4, "elapsed_time": "0:18:25", "remaining_time": "0:03:24", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 423, "total_steps": 500, "loss": 0.7252, "learning_rate": 7.052005153466779e-06, "epoch": 3.76, "percentage": 84.6, "elapsed_time": "0:18:28", "remaining_time": "0:03:21", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 424, "total_steps": 500, "loss": 0.6773, "learning_rate": 6.874316539637127e-06, "epoch": 3.7688888888888887, "percentage": 84.8, "elapsed_time": "0:18:30", "remaining_time": "0:03:19", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 425, "total_steps": 500, "loss": 0.6376, "learning_rate": 6.698729810778065e-06, "epoch": 3.7777777777777777, "percentage": 85.0, "elapsed_time": "0:18:32", "remaining_time": "0:03:16", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 426, "total_steps": 500, "loss": 0.7645, "learning_rate": 6.52525352473905e-06, "epoch": 3.7866666666666666, "percentage": 85.2, "elapsed_time": "0:18:35", "remaining_time": "0:03:13", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 427, "total_steps": 500, "loss": 0.797, "learning_rate": 6.353896136509524e-06, "epoch": 3.7955555555555556, "percentage": 85.4, "elapsed_time": "0:18:37", "remaining_time": "0:03:11", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 428, "total_steps": 500, "loss": 0.7175, "learning_rate": 6.184665997806832e-06, "epoch": 3.8044444444444445, "percentage": 85.6, "elapsed_time": "0:18:39", "remaining_time": "0:03:08", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 429, "total_steps": 500, "loss": 0.8561, "learning_rate": 6.017571356669183e-06, "epoch": 3.8133333333333335, "percentage": 85.8, "elapsed_time": "0:18:42", "remaining_time": "0:03:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 430, "total_steps": 500, "loss": 0.5715, "learning_rate": 5.852620357053651e-06, "epoch": 3.822222222222222, "percentage": 86.0, "elapsed_time": "0:18:44", "remaining_time": "0:03:03", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 431, "total_steps": 500, "loss": 0.68, "learning_rate": 5.689821038439263e-06, "epoch": 3.8311111111111114, "percentage": 86.2, "elapsed_time": "0:18:47", "remaining_time": "0:03:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 432, "total_steps": 500, "loss": 0.7089, "learning_rate": 5.529181335435124e-06, "epoch": 3.84, "percentage": 86.4, "elapsed_time": "0:18:49", "remaining_time": "0:02:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 433, "total_steps": 500, "loss": 0.7589, "learning_rate": 5.370709077393721e-06, "epoch": 3.848888888888889, "percentage": 86.6, "elapsed_time": "0:18:51", "remaining_time": "0:02:55", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 434, "total_steps": 500, "loss": 0.7617, "learning_rate": 5.214411988029355e-06, "epoch": 3.8577777777777778, "percentage": 86.8, "elapsed_time": "0:18:54", "remaining_time": "0:02:52", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 435, "total_steps": 500, "loss": 0.9029, "learning_rate": 5.060297685041659e-06, "epoch": 3.8666666666666667, "percentage": 87.0, "elapsed_time": "0:18:56", "remaining_time": "0:02:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 436, "total_steps": 500, "loss": 0.6779, "learning_rate": 4.908373679744316e-06, "epoch": 3.8755555555555556, "percentage": 87.2, "elapsed_time": "0:18:58", "remaining_time": "0:02:47", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 437, "total_steps": 500, "loss": 0.8606, "learning_rate": 4.758647376699032e-06, "epoch": 3.8844444444444446, "percentage": 87.4, "elapsed_time": "0:19:01", "remaining_time": "0:02:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 438, "total_steps": 500, "loss": 0.7648, "learning_rate": 4.611126073354571e-06, "epoch": 3.8933333333333335, "percentage": 87.6, "elapsed_time": "0:19:03", "remaining_time": "0:02:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 439, "total_steps": 500, "loss": 0.6418, "learning_rate": 4.465816959691149e-06, "epoch": 3.902222222222222, "percentage": 87.8, "elapsed_time": "0:19:06", "remaining_time": "0:02:39", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 440, "total_steps": 500, "loss": 0.7436, "learning_rate": 4.322727117869951e-06, "epoch": 3.911111111111111, "percentage": 88.0, "elapsed_time": "0:19:08", "remaining_time": "0:02:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 441, "total_steps": 500, "loss": 0.7485, "learning_rate": 4.181863521888019e-06, "epoch": 3.92, "percentage": 88.2, "elapsed_time": "0:19:10", "remaining_time": "0:02:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 442, "total_steps": 500, "loss": 0.7783, "learning_rate": 4.043233037238281e-06, "epoch": 3.928888888888889, "percentage": 88.4, "elapsed_time": "0:19:13", "remaining_time": "0:02:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 443, "total_steps": 500, "loss": 0.5753, "learning_rate": 3.90684242057498e-06, "epoch": 3.937777777777778, "percentage": 88.6, "elapsed_time": "0:19:15", "remaining_time": "0:02:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 444, "total_steps": 500, "loss": 0.6734, "learning_rate": 3.772698319384349e-06, "epoch": 3.9466666666666668, "percentage": 88.8, "elapsed_time": "0:19:18", "remaining_time": "0:02:26", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 445, "total_steps": 500, "loss": 0.7748, "learning_rate": 3.6408072716606346e-06, "epoch": 3.9555555555555557, "percentage": 89.0, "elapsed_time": "0:19:20", "remaining_time": "0:02:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 446, "total_steps": 500, "loss": 0.7126, "learning_rate": 3.511175705587433e-06, "epoch": 3.964444444444444, "percentage": 89.2, "elapsed_time": "0:19:23", "remaining_time": "0:02:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 447, "total_steps": 500, "loss": 0.7099, "learning_rate": 3.3838099392243916e-06, "epoch": 3.9733333333333336, "percentage": 89.4, "elapsed_time": "0:19:25", "remaining_time": "0:02:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 448, "total_steps": 500, "loss": 0.9139, "learning_rate": 3.258716180199278e-06, "epoch": 3.982222222222222, "percentage": 89.6, "elapsed_time": "0:19:27", "remaining_time": "0:02:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 449, "total_steps": 500, "loss": 0.7272, "learning_rate": 3.1359005254054273e-06, "epoch": 3.991111111111111, "percentage": 89.8, "elapsed_time": "0:19:30", "remaining_time": "0:02:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 450, "total_steps": 500, "loss": 0.8561, "learning_rate": 3.0153689607045845e-06, "epoch": 4.0, "percentage": 90.0, "elapsed_time": "0:19:33", "remaining_time": "0:02:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 451, "total_steps": 500, "loss": 0.7015, "learning_rate": 2.8971273606351658e-06, "epoch": 4.0088888888888885, "percentage": 90.2, "elapsed_time": "0:19:36", "remaining_time": "0:02:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 452, "total_steps": 500, "loss": 0.6025, "learning_rate": 2.7811814881259503e-06, "epoch": 4.017777777777778, "percentage": 90.4, "elapsed_time": "0:19:38", "remaining_time": "0:02:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 453, "total_steps": 500, "loss": 0.7018, "learning_rate": 2.667536994215186e-06, "epoch": 4.026666666666666, "percentage": 90.6, "elapsed_time": "0:19:40", "remaining_time": "0:02:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 454, "total_steps": 500, "loss": 0.6452, "learning_rate": 2.5561994177751737e-06, "epoch": 4.035555555555556, "percentage": 90.8, "elapsed_time": "0:19:43", "remaining_time": "0:01:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 455, "total_steps": 500, "loss": 0.573, "learning_rate": 2.4471741852423237e-06, "epoch": 4.044444444444444, "percentage": 91.0, "elapsed_time": "0:19:45", "remaining_time": "0:01:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 456, "total_steps": 500, "loss": 0.6177, "learning_rate": 2.340466610352654e-06, "epoch": 4.053333333333334, "percentage": 91.2, "elapsed_time": "0:19:47", "remaining_time": "0:01:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 457, "total_steps": 500, "loss": 0.6479, "learning_rate": 2.2360818938828187e-06, "epoch": 4.062222222222222, "percentage": 91.4, "elapsed_time": "0:19:49", "remaining_time": "0:01:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 458, "total_steps": 500, "loss": 0.6464, "learning_rate": 2.134025123396638e-06, "epoch": 4.071111111111111, "percentage": 91.6, "elapsed_time": "0:19:52", "remaining_time": "0:01:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 459, "total_steps": 500, "loss": 0.6545, "learning_rate": 2.0343012729971243e-06, "epoch": 4.08, "percentage": 91.8, "elapsed_time": "0:19:54", "remaining_time": "0:01:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 460, "total_steps": 500, "loss": 0.6401, "learning_rate": 1.9369152030840556e-06, "epoch": 4.088888888888889, "percentage": 92.0, "elapsed_time": "0:19:56", "remaining_time": "0:01:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 461, "total_steps": 500, "loss": 0.5654, "learning_rate": 1.841871660117095e-06, "epoch": 4.097777777777778, "percentage": 92.2, "elapsed_time": "0:19:59", "remaining_time": "0:01:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 462, "total_steps": 500, "loss": 0.6355, "learning_rate": 1.7491752763844293e-06, "epoch": 4.1066666666666665, "percentage": 92.4, "elapsed_time": "0:20:01", "remaining_time": "0:01:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 463, "total_steps": 500, "loss": 0.8252, "learning_rate": 1.6588305697770313e-06, "epoch": 4.115555555555556, "percentage": 92.6, "elapsed_time": "0:20:04", "remaining_time": "0:01:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 464, "total_steps": 500, "loss": 0.573, "learning_rate": 1.5708419435684462e-06, "epoch": 4.124444444444444, "percentage": 92.8, "elapsed_time": "0:20:06", "remaining_time": "0:01:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 465, "total_steps": 500, "loss": 0.6218, "learning_rate": 1.4852136862001764e-06, "epoch": 4.133333333333334, "percentage": 93.0, "elapsed_time": "0:20:08", "remaining_time": "0:01:30", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 466, "total_steps": 500, "loss": 0.5241, "learning_rate": 1.4019499710726913e-06, "epoch": 4.142222222222222, "percentage": 93.2, "elapsed_time": "0:20:10", "remaining_time": "0:01:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 467, "total_steps": 500, "loss": 0.6395, "learning_rate": 1.3210548563419856e-06, "epoch": 4.151111111111111, "percentage": 93.4, "elapsed_time": "0:20:13", "remaining_time": "0:01:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 468, "total_steps": 500, "loss": 0.7088, "learning_rate": 1.2425322847218368e-06, "epoch": 4.16, "percentage": 93.6, "elapsed_time": "0:20:15", "remaining_time": "0:01:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 469, "total_steps": 500, "loss": 0.7894, "learning_rate": 1.166386083291604e-06, "epoch": 4.168888888888889, "percentage": 93.8, "elapsed_time": "0:20:18", "remaining_time": "0:01:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 470, "total_steps": 500, "loss": 0.6565, "learning_rate": 1.0926199633097157e-06, "epoch": 4.177777777777778, "percentage": 94.0, "elapsed_time": "0:20:20", "remaining_time": "0:01:17", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 471, "total_steps": 500, "loss": 0.5306, "learning_rate": 1.0212375200327973e-06, "epoch": 4.1866666666666665, "percentage": 94.2, "elapsed_time": "0:20:22", "remaining_time": "0:01:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 472, "total_steps": 500, "loss": 0.6762, "learning_rate": 9.522422325404235e-07, "epoch": 4.195555555555556, "percentage": 94.4, "elapsed_time": "0:20:25", "remaining_time": "0:01:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 473, "total_steps": 500, "loss": 0.5428, "learning_rate": 8.856374635655695e-07, "epoch": 4.204444444444444, "percentage": 94.6, "elapsed_time": "0:20:28", "remaining_time": "0:01:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 474, "total_steps": 500, "loss": 0.6506, "learning_rate": 8.214264593307098e-07, "epoch": 4.213333333333333, "percentage": 94.8, "elapsed_time": "0:20:30", "remaining_time": "0:01:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 475, "total_steps": 500, "loss": 0.5546, "learning_rate": 7.596123493895991e-07, "epoch": 4.222222222222222, "percentage": 95.0, "elapsed_time": "0:20:32", "remaining_time": "0:01:04", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 476, "total_steps": 500, "loss": 0.6947, "learning_rate": 7.001981464747565e-07, "epoch": 4.231111111111111, "percentage": 95.2, "elapsed_time": "0:20:35", "remaining_time": "0:01:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 477, "total_steps": 500, "loss": 0.8005, "learning_rate": 6.431867463506048e-07, "epoch": 4.24, "percentage": 95.4, "elapsed_time": "0:20:38", "remaining_time": "0:00:59", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 478, "total_steps": 500, "loss": 0.7249, "learning_rate": 5.885809276723608e-07, "epoch": 4.248888888888889, "percentage": 95.6, "elapsed_time": "0:20:40", "remaining_time": "0:00:57", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 479, "total_steps": 500, "loss": 0.7854, "learning_rate": 5.363833518505834e-07, "epoch": 4.257777777777778, "percentage": 95.8, "elapsed_time": "0:20:43", "remaining_time": "0:00:54", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 480, "total_steps": 500, "loss": 0.762, "learning_rate": 4.865965629214819e-07, "epoch": 4.266666666666667, "percentage": 96.0, "elapsed_time": "0:20:45", "remaining_time": "0:00:51", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 481, "total_steps": 500, "loss": 0.7091, "learning_rate": 4.392229874229159e-07, "epoch": 4.275555555555556, "percentage": 96.2, "elapsed_time": "0:20:48", "remaining_time": "0:00:49", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 482, "total_steps": 500, "loss": 0.6231, "learning_rate": 3.9426493427611177e-07, "epoch": 4.2844444444444445, "percentage": 96.4, "elapsed_time": "0:20:50", "remaining_time": "0:00:46", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 483, "total_steps": 500, "loss": 0.7348, "learning_rate": 3.517245946731529e-07, "epoch": 4.293333333333333, "percentage": 96.6, "elapsed_time": "0:20:53", "remaining_time": "0:00:44", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 484, "total_steps": 500, "loss": 0.531, "learning_rate": 3.1160404197018154e-07, "epoch": 4.302222222222222, "percentage": 96.8, "elapsed_time": "0:20:55", "remaining_time": "0:00:41", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 485, "total_steps": 500, "loss": 0.5834, "learning_rate": 2.7390523158633554e-07, "epoch": 4.311111111111111, "percentage": 97.0, "elapsed_time": "0:20:57", "remaining_time": "0:00:38", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 486, "total_steps": 500, "loss": 0.5736, "learning_rate": 2.386300009084408e-07, "epoch": 4.32, "percentage": 97.2, "elapsed_time": "0:20:59", "remaining_time": "0:00:36", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 487, "total_steps": 500, "loss": 0.4758, "learning_rate": 2.057800692014833e-07, "epoch": 4.328888888888889, "percentage": 97.4, "elapsed_time": "0:21:01", "remaining_time": "0:00:33", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 488, "total_steps": 500, "loss": 0.4747, "learning_rate": 1.753570375247815e-07, "epoch": 4.337777777777778, "percentage": 97.6, "elapsed_time": "0:21:04", "remaining_time": "0:00:31", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 489, "total_steps": 500, "loss": 0.6023, "learning_rate": 1.4736238865398765e-07, "epoch": 4.346666666666667, "percentage": 97.8, "elapsed_time": "0:21:06", "remaining_time": "0:00:28", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 490, "total_steps": 500, "loss": 0.6376, "learning_rate": 1.2179748700879012e-07, "epoch": 4.355555555555555, "percentage": 98.0, "elapsed_time": "0:21:08", "remaining_time": "0:00:25", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 491, "total_steps": 500, "loss": 0.5588, "learning_rate": 9.866357858642205e-08, "epoch": 4.364444444444445, "percentage": 98.2, "elapsed_time": "0:21:11", "remaining_time": "0:00:23", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 492, "total_steps": 500, "loss": 0.6635, "learning_rate": 7.796179090094891e-08, "epoch": 4.373333333333333, "percentage": 98.4, "elapsed_time": "0:21:14", "remaining_time": "0:00:20", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 493, "total_steps": 500, "loss": 0.5795, "learning_rate": 5.969313292830125e-08, "epoch": 4.3822222222222225, "percentage": 98.6, "elapsed_time": "0:21:16", "remaining_time": "0:00:18", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 494, "total_steps": 500, "loss": 0.5087, "learning_rate": 4.385849505708084e-08, "epoch": 4.391111111111111, "percentage": 98.8, "elapsed_time": "0:21:18", "remaining_time": "0:00:15", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 495, "total_steps": 500, "loss": 0.5711, "learning_rate": 3.04586490452119e-08, "epoch": 4.4, "percentage": 99.0, "elapsed_time": "0:21:20", "remaining_time": "0:00:12", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 496, "total_steps": 500, "loss": 0.6284, "learning_rate": 1.949424798228239e-08, "epoch": 4.408888888888889, "percentage": 99.2, "elapsed_time": "0:21:23", "remaining_time": "0:00:10", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 497, "total_steps": 500, "loss": 0.7002, "learning_rate": 1.096582625772502e-08, "epoch": 4.417777777777777, "percentage": 99.4, "elapsed_time": "0:21:25", "remaining_time": "0:00:07", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 498, "total_steps": 500, "loss": 0.5831, "learning_rate": 4.873799534788059e-09, "epoch": 4.426666666666667, "percentage": 99.6, "elapsed_time": "0:21:27", "remaining_time": "0:00:05", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 499, "total_steps": 500, "loss": 0.5923, "learning_rate": 1.2184647302626583e-09, "epoch": 4.435555555555555, "percentage": 99.8, "elapsed_time": "0:21:29", "remaining_time": "0:00:02", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 500, "total_steps": 500, "loss": 0.6272, "learning_rate": 0.0, "epoch": 4.444444444444445, "percentage": 100.0, "elapsed_time": "0:21:31", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 500, "total_steps": 500, "eval_loss": 0.996656596660614, "epoch": 4.444444444444445, "percentage": 100.0, "elapsed_time": "0:21:42", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0} | |
| {"current_steps": 500, "total_steps": 500, "epoch": 4.444444444444445, "percentage": 100.0, "elapsed_time": "0:22:32", "remaining_time": "0:00:00", "throughput": "0.00", "total_tokens": 0} | |