sdar_4b_random_mask-final / trainer_log.jsonl
autoprogrammer's picture
Upload sdar_4b_random_mask-final
d95f204 verified
{"current_steps": 5, "total_steps": 351, "loss": 0.2805, "lr": 3.6363636363636366e-06, "epoch": 0.042735042735042736, "percentage": 1.42, "elapsed_time": "0:00:55", "remaining_time": "1:03:53"}
{"current_steps": 10, "total_steps": 351, "loss": 0.2371, "lr": 8.181818181818183e-06, "epoch": 0.08547008547008547, "percentage": 2.85, "elapsed_time": "0:00:58", "remaining_time": "0:33:23"}
{"current_steps": 15, "total_steps": 351, "loss": 0.243, "lr": 9.998079135987437e-06, "epoch": 0.1282051282051282, "percentage": 4.27, "elapsed_time": "0:01:02", "remaining_time": "0:23:12"}
{"current_steps": 20, "total_steps": 351, "loss": 0.2425, "lr": 9.98634586692894e-06, "epoch": 0.17094017094017094, "percentage": 5.7, "elapsed_time": "0:01:05", "remaining_time": "0:18:07"}
{"current_steps": 25, "total_steps": 351, "loss": 0.2579, "lr": 9.963971484502247e-06, "epoch": 0.21367521367521367, "percentage": 7.12, "elapsed_time": "0:01:08", "remaining_time": "0:14:59"}
{"current_steps": 30, "total_steps": 351, "loss": 0.241, "lr": 9.931003736767013e-06, "epoch": 0.2564102564102564, "percentage": 8.55, "elapsed_time": "0:01:12", "remaining_time": "0:12:53"}
{"current_steps": 35, "total_steps": 351, "loss": 0.2511, "lr": 9.887512978558329e-06, "epoch": 0.29914529914529914, "percentage": 9.97, "elapsed_time": "0:01:15", "remaining_time": "0:11:21"}
{"current_steps": 40, "total_steps": 351, "loss": 0.2389, "lr": 9.833592021345938e-06, "epoch": 0.3418803418803419, "percentage": 11.4, "elapsed_time": "0:01:18", "remaining_time": "0:10:14"}
{"current_steps": 45, "total_steps": 351, "loss": 0.258, "lr": 9.76935593516989e-06, "epoch": 0.38461538461538464, "percentage": 12.82, "elapsed_time": "0:01:22", "remaining_time": "0:09:19"}
{"current_steps": 50, "total_steps": 351, "loss": 0.2323, "lr": 9.694941803075285e-06, "epoch": 0.42735042735042733, "percentage": 14.25, "elapsed_time": "0:01:25", "remaining_time": "0:08:34"}
{"current_steps": 55, "total_steps": 351, "loss": 0.2685, "lr": 9.610508428570122e-06, "epoch": 0.4700854700854701, "percentage": 15.67, "elapsed_time": "0:01:28", "remaining_time": "0:07:57"}
{"current_steps": 60, "total_steps": 351, "loss": 0.2455, "lr": 9.516235996730645e-06, "epoch": 0.5128205128205128, "percentage": 17.09, "elapsed_time": "0:01:31", "remaining_time": "0:07:26"}
{"current_steps": 65, "total_steps": 351, "loss": 0.2509, "lr": 9.41232568967728e-06, "epoch": 0.5555555555555556, "percentage": 18.52, "elapsed_time": "0:01:35", "remaining_time": "0:06:59"}
{"current_steps": 70, "total_steps": 351, "loss": 0.2322, "lr": 9.298999257241862e-06, "epoch": 0.5982905982905983, "percentage": 19.94, "elapsed_time": "0:01:38", "remaining_time": "0:06:35"}
{"current_steps": 75, "total_steps": 351, "loss": 0.2334, "lr": 9.176498543742328e-06, "epoch": 0.6410256410256411, "percentage": 21.37, "elapsed_time": "0:01:41", "remaining_time": "0:06:14"}
{"current_steps": 80, "total_steps": 351, "loss": 0.2331, "lr": 9.045084971874738e-06, "epoch": 0.6837606837606838, "percentage": 22.79, "elapsed_time": "0:01:44", "remaining_time": "0:05:55"}
{"current_steps": 85, "total_steps": 351, "loss": 0.252, "lr": 8.905038984824079e-06, "epoch": 0.7264957264957265, "percentage": 24.22, "elapsed_time": "0:01:48", "remaining_time": "0:05:39"}
{"current_steps": 90, "total_steps": 351, "loss": 0.2558, "lr": 8.756659447784367e-06, "epoch": 0.7692307692307693, "percentage": 25.64, "elapsed_time": "0:01:51", "remaining_time": "0:05:24"}
{"current_steps": 95, "total_steps": 351, "loss": 0.2455, "lr": 8.600263010165275e-06, "epoch": 0.811965811965812, "percentage": 27.07, "elapsed_time": "0:01:54", "remaining_time": "0:05:09"}
{"current_steps": 100, "total_steps": 351, "loss": 0.2539, "lr": 8.436183429846314e-06, "epoch": 0.8547008547008547, "percentage": 28.49, "elapsed_time": "0:01:58", "remaining_time": "0:04:56"}
{"current_steps": 105, "total_steps": 351, "loss": 0.2604, "lr": 8.264770860920722e-06, "epoch": 0.8974358974358975, "percentage": 29.91, "elapsed_time": "0:02:01", "remaining_time": "0:04:44"}
{"current_steps": 110, "total_steps": 351, "loss": 0.2432, "lr": 8.086391106448965e-06, "epoch": 0.9401709401709402, "percentage": 31.34, "elapsed_time": "0:02:04", "remaining_time": "0:04:33"}
{"current_steps": 115, "total_steps": 351, "loss": 0.2617, "lr": 7.90142483781658e-06, "epoch": 0.9829059829059829, "percentage": 32.76, "elapsed_time": "0:02:08", "remaining_time": "0:04:22"}
{"current_steps": 120, "total_steps": 351, "loss": 0.1536, "lr": 7.710266782362248e-06, "epoch": 1.0256410256410255, "percentage": 34.19, "elapsed_time": "0:03:12", "remaining_time": "0:06:10"}
{"current_steps": 125, "total_steps": 351, "loss": 0.0775, "lr": 7.513324881009769e-06, "epoch": 1.0683760683760684, "percentage": 35.61, "elapsed_time": "0:03:15", "remaining_time": "0:05:54"}
{"current_steps": 130, "total_steps": 351, "loss": 0.0733, "lr": 7.311019417701567e-06, "epoch": 1.1111111111111112, "percentage": 37.04, "elapsed_time": "0:03:19", "remaining_time": "0:05:38"}
{"current_steps": 135, "total_steps": 351, "loss": 0.0884, "lr": 7.103782122491577e-06, "epoch": 1.1538461538461537, "percentage": 38.46, "elapsed_time": "0:03:22", "remaining_time": "0:05:24"}
{"current_steps": 140, "total_steps": 351, "loss": 0.0797, "lr": 6.892055250211552e-06, "epoch": 1.1965811965811965, "percentage": 39.89, "elapsed_time": "0:03:25", "remaining_time": "0:05:10"}
{"current_steps": 145, "total_steps": 351, "loss": 0.0911, "lr": 6.67629063667697e-06, "epoch": 1.2393162393162394, "percentage": 41.31, "elapsed_time": "0:03:29", "remaining_time": "0:04:57"}
{"current_steps": 150, "total_steps": 351, "loss": 0.0856, "lr": 6.456948734446624e-06, "epoch": 1.282051282051282, "percentage": 42.74, "elapsed_time": "0:03:32", "remaining_time": "0:04:44"}
{"current_steps": 155, "total_steps": 351, "loss": 0.0841, "lr": 6.234497630193666e-06, "epoch": 1.3247863247863247, "percentage": 44.16, "elapsed_time": "0:03:35", "remaining_time": "0:04:32"}
{"current_steps": 160, "total_steps": 351, "loss": 0.0795, "lr": 6.009412045785051e-06, "epoch": 1.3675213675213675, "percentage": 45.58, "elapsed_time": "0:03:38", "remaining_time": "0:04:21"}
{"current_steps": 165, "total_steps": 351, "loss": 0.0769, "lr": 5.782172325201155e-06, "epoch": 1.4102564102564101, "percentage": 47.01, "elapsed_time": "0:03:41", "remaining_time": "0:04:10"}
{"current_steps": 170, "total_steps": 351, "loss": 0.0756, "lr": 5.553263409457504e-06, "epoch": 1.452991452991453, "percentage": 48.43, "elapsed_time": "0:03:45", "remaining_time": "0:03:59"}
{"current_steps": 175, "total_steps": 351, "loss": 0.0852, "lr": 5.323173801716222e-06, "epoch": 1.4957264957264957, "percentage": 49.86, "elapsed_time": "0:03:48", "remaining_time": "0:03:49"}
{"current_steps": 180, "total_steps": 351, "loss": 0.0762, "lr": 5.09239452479565e-06, "epoch": 1.5384615384615383, "percentage": 51.28, "elapsed_time": "0:03:51", "remaining_time": "0:03:40"}
{"current_steps": 185, "total_steps": 351, "loss": 0.0762, "lr": 4.861418073302919e-06, "epoch": 1.5811965811965814, "percentage": 52.71, "elapsed_time": "0:03:54", "remaining_time": "0:03:30"}
{"current_steps": 190, "total_steps": 351, "loss": 0.0763, "lr": 4.630737362625631e-06, "epoch": 1.623931623931624, "percentage": 54.13, "elapsed_time": "0:03:58", "remaining_time": "0:03:21"}
{"current_steps": 195, "total_steps": 351, "loss": 0.0718, "lr": 4.400844677025585e-06, "epoch": 1.6666666666666665, "percentage": 55.56, "elapsed_time": "0:04:01", "remaining_time": "0:03:13"}
{"current_steps": 200, "total_steps": 351, "loss": 0.0733, "lr": 4.17223061907935e-06, "epoch": 1.7094017094017095, "percentage": 56.98, "elapsed_time": "0:04:04", "remaining_time": "0:03:04"}
{"current_steps": 205, "total_steps": 351, "loss": 0.0732, "lr": 3.945383062707652e-06, "epoch": 1.7521367521367521, "percentage": 58.4, "elapsed_time": "0:04:07", "remaining_time": "0:02:56"}
{"current_steps": 210, "total_steps": 351, "loss": 0.0739, "lr": 3.720786112027822e-06, "epoch": 1.7948717948717947, "percentage": 59.83, "elapsed_time": "0:04:11", "remaining_time": "0:02:48"}
{"current_steps": 215, "total_steps": 351, "loss": 0.0726, "lr": 3.498919068251237e-06, "epoch": 1.8376068376068377, "percentage": 61.25, "elapsed_time": "0:04:14", "remaining_time": "0:02:40"}
{"current_steps": 220, "total_steps": 351, "loss": 0.0678, "lr": 3.2802554068303595e-06, "epoch": 1.8803418803418803, "percentage": 62.68, "elapsed_time": "0:04:17", "remaining_time": "0:02:33"}
{"current_steps": 225, "total_steps": 351, "loss": 0.0652, "lr": 3.0652617670382745e-06, "epoch": 1.9230769230769231, "percentage": 64.1, "elapsed_time": "0:04:21", "remaining_time": "0:02:26"}
{"current_steps": 230, "total_steps": 351, "loss": 0.0704, "lr": 2.8543969561369556e-06, "epoch": 1.965811965811966, "percentage": 65.53, "elapsed_time": "0:04:24", "remaining_time": "0:02:19"}
{"current_steps": 235, "total_steps": 351, "loss": 0.0586, "lr": 2.648110970259454e-06, "epoch": 2.0085470085470085, "percentage": 66.95, "elapsed_time": "0:05:32", "remaining_time": "0:02:44"}
{"current_steps": 240, "total_steps": 351, "loss": 0.0223, "lr": 2.4468440340954664e-06, "epoch": 2.051282051282051, "percentage": 68.38, "elapsed_time": "0:05:35", "remaining_time": "0:02:35"}
{"current_steps": 245, "total_steps": 351, "loss": 0.0203, "lr": 2.2510256614296638e-06, "epoch": 2.094017094017094, "percentage": 69.8, "elapsed_time": "0:05:38", "remaining_time": "0:02:26"}
{"current_steps": 250, "total_steps": 351, "loss": 0.0193, "lr": 2.061073738537635e-06, "epoch": 2.1367521367521367, "percentage": 71.23, "elapsed_time": "0:05:42", "remaining_time": "0:02:18"}
{"current_steps": 255, "total_steps": 351, "loss": 0.0198, "lr": 1.8773936323955055e-06, "epoch": 2.1794871794871793, "percentage": 72.65, "elapsed_time": "0:05:45", "remaining_time": "0:02:09"}
{"current_steps": 260, "total_steps": 351, "loss": 0.0193, "lr": 1.7003773256063882e-06, "epoch": 2.2222222222222223, "percentage": 74.07, "elapsed_time": "0:05:48", "remaining_time": "0:02:02"}
{"current_steps": 265, "total_steps": 351, "loss": 0.0174, "lr": 1.5304025798897521e-06, "epoch": 2.264957264957265, "percentage": 75.5, "elapsed_time": "0:05:52", "remaining_time": "0:01:54"}
{"current_steps": 270, "total_steps": 351, "loss": 0.0159, "lr": 1.3678321299188802e-06, "epoch": 2.3076923076923075, "percentage": 76.92, "elapsed_time": "0:05:55", "remaining_time": "0:01:46"}
{"current_steps": 275, "total_steps": 351, "loss": 0.0198, "lr": 1.213012909226786e-06, "epoch": 2.3504273504273505, "percentage": 78.35, "elapsed_time": "0:05:58", "remaining_time": "0:01:39"}
{"current_steps": 280, "total_steps": 351, "loss": 0.0165, "lr": 1.066275309832584e-06, "epoch": 2.393162393162393, "percentage": 79.77, "elapsed_time": "0:06:01", "remaining_time": "0:01:31"}
{"current_steps": 285, "total_steps": 351, "loss": 0.0173, "lr": 9.279324771682586e-07, "epoch": 2.435897435897436, "percentage": 81.2, "elapsed_time": "0:06:05", "remaining_time": "0:01:24"}
{"current_steps": 290, "total_steps": 351, "loss": 0.0191, "lr": 7.98279641810537e-07, "epoch": 2.4786324786324787, "percentage": 82.62, "elapsed_time": "0:06:09", "remaining_time": "0:01:17"}
{"current_steps": 295, "total_steps": 351, "loss": 0.0184, "lr": 6.775934894439606e-07, "epoch": 2.5213675213675213, "percentage": 84.05, "elapsed_time": "0:06:12", "remaining_time": "0:01:10"}
{"current_steps": 300, "total_steps": 351, "loss": 0.0161, "lr": 5.661315703996905e-07, "epoch": 2.564102564102564, "percentage": 85.47, "elapsed_time": "0:06:15", "remaining_time": "0:01:03"}
{"current_steps": 305, "total_steps": 351, "loss": 0.0132, "lr": 4.641317500301173e-07, "epoch": 2.606837606837607, "percentage": 86.89, "elapsed_time": "0:06:18", "remaining_time": "0:00:57"}
{"current_steps": 310, "total_steps": 351, "loss": 0.018, "lr": 3.71811701092219e-07, "epoch": 2.6495726495726495, "percentage": 88.32, "elapsed_time": "0:06:22", "remaining_time": "0:00:50"}
{"current_steps": 315, "total_steps": 351, "loss": 0.0179, "lr": 2.893684392229185e-07, "epoch": 2.6923076923076925, "percentage": 89.74, "elapsed_time": "0:06:25", "remaining_time": "0:00:44"}
{"current_steps": 320, "total_steps": 351, "loss": 0.0186, "lr": 2.1697790249779638e-07, "epoch": 2.735042735042735, "percentage": 91.17, "elapsed_time": "0:06:28", "remaining_time": "0:00:37"}
{"current_steps": 325, "total_steps": 351, "loss": 0.0204, "lr": 1.547945759703623e-07, "epoch": 2.7777777777777777, "percentage": 92.59, "elapsed_time": "0:06:31", "remaining_time": "0:00:31"}
{"current_steps": 330, "total_steps": 351, "loss": 0.0213, "lr": 1.0295116199317057e-07, "epoch": 2.8205128205128203, "percentage": 94.02, "elapsed_time": "0:06:35", "remaining_time": "0:00:25"}
{"current_steps": 335, "total_steps": 351, "loss": 0.0196, "lr": 6.15582970243117e-08, "epoch": 2.8632478632478633, "percentage": 95.44, "elapsed_time": "0:06:38", "remaining_time": "0:00:19"}
{"current_steps": 340, "total_steps": 351, "loss": 0.0178, "lr": 3.0704315523631956e-08, "epoch": 2.905982905982906, "percentage": 96.87, "elapsed_time": "0:06:41", "remaining_time": "0:00:12"}
{"current_steps": 345, "total_steps": 351, "loss": 0.0226, "lr": 1.0455061442548597e-08, "epoch": 2.948717948717949, "percentage": 98.29, "elapsed_time": "0:06:44", "remaining_time": "0:00:07"}
{"current_steps": 350, "total_steps": 351, "loss": 0.0168, "lr": 8.537477097364522e-10, "epoch": 2.9914529914529915, "percentage": 99.72, "elapsed_time": "0:06:48", "remaining_time": "0:00:01"}
{"current_steps": 351, "total_steps": 351, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "0:07:51", "remaining_time": "0:00:00"}