| {"current_steps": 5, "total_steps": 1505, "loss": 3.8116, "learning_rate": 4.999863832700438e-05, "epoch": 0.11594202898550725, "percentage": 0.33, "elapsed_time": "0:01:41", "remaining_time": "8:27:54", "throughput": "1052.64", "total_tokens": 106929} | |
| {"current_steps": 10, "total_steps": 1505, "loss": 3.6928, "learning_rate": 4.999455345634978e-05, "epoch": 0.2318840579710145, "percentage": 0.66, "elapsed_time": "0:03:33", "remaining_time": "8:51:03", "throughput": "1060.20", "total_tokens": 225964} | |
| {"current_steps": 15, "total_steps": 1505, "loss": 3.6227, "learning_rate": 4.9987745833016855e-05, "epoch": 0.34782608695652173, "percentage": 1.0, "elapsed_time": "0:05:53", "remaining_time": "9:45:33", "throughput": "1024.22", "total_tokens": 362264} | |
| {"current_steps": 20, "total_steps": 1505, "loss": 3.601, "learning_rate": 4.9978216198586135e-05, "epoch": 0.463768115942029, "percentage": 1.33, "elapsed_time": "0:07:42", "remaining_time": "9:32:27", "throughput": "1032.88", "total_tokens": 477807} | |
| {"current_steps": 25, "total_steps": 1505, "loss": 3.539, "learning_rate": 4.996596559115731e-05, "epoch": 0.5797101449275363, "percentage": 1.66, "elapsed_time": "0:09:25", "remaining_time": "9:18:14", "throughput": "1040.84", "total_tokens": 588900} | |
| {"current_steps": 30, "total_steps": 1505, "loss": 3.4956, "learning_rate": 4.995099534523607e-05, "epoch": 0.6956521739130435, "percentage": 1.99, "elapsed_time": "0:11:18", "remaining_time": "9:16:16", "throughput": "1040.12", "total_tokens": 706077} | |
| {"current_steps": 35, "total_steps": 1505, "loss": 3.5044, "learning_rate": 4.9933307091588796e-05, "epoch": 0.8115942028985508, "percentage": 2.33, "elapsed_time": "0:13:53", "remaining_time": "9:43:08", "throughput": "1024.54", "total_tokens": 853504} | |
| {"current_steps": 40, "total_steps": 1505, "loss": 3.4324, "learning_rate": 4.991290275706486e-05, "epoch": 0.927536231884058, "percentage": 2.66, "elapsed_time": "0:16:08", "remaining_time": "9:51:29", "throughput": "1022.18", "total_tokens": 990472} | |
| {"current_steps": 45, "total_steps": 1505, "loss": 3.2542, "learning_rate": 4.988978456438678e-05, "epoch": 1.0434782608695652, "percentage": 2.99, "elapsed_time": "0:18:22", "remaining_time": "9:56:13", "throughput": "1021.10", "total_tokens": 1125870} | |
| {"current_steps": 50, "total_steps": 1505, "loss": 2.9024, "learning_rate": 4.986395503190805e-05, "epoch": 1.1594202898550725, "percentage": 3.32, "elapsed_time": "0:20:21", "remaining_time": "9:52:16", "throughput": "1023.49", "total_tokens": 1249877} | |
| {"current_steps": 55, "total_steps": 1505, "loss": 2.8069, "learning_rate": 4.983541697333881e-05, "epoch": 1.2753623188405796, "percentage": 3.65, "elapsed_time": "0:22:24", "remaining_time": "9:50:52", "throughput": "1022.63", "total_tokens": 1375193} | |
| {"current_steps": 60, "total_steps": 1505, "loss": 2.75, "learning_rate": 4.980417349743936e-05, "epoch": 1.391304347826087, "percentage": 3.99, "elapsed_time": "0:24:19", "remaining_time": "9:45:47", "throughput": "1020.77", "total_tokens": 1489716} | |
| {"current_steps": 65, "total_steps": 1505, "loss": 2.7329, "learning_rate": 4.9770228007681494e-05, "epoch": 1.5072463768115942, "percentage": 4.32, "elapsed_time": "0:26:02", "remaining_time": "9:37:00", "throughput": "1024.15", "total_tokens": 1600483} | |
| {"current_steps": 70, "total_steps": 1505, "loss": 2.8212, "learning_rate": 4.973358420187776e-05, "epoch": 1.6231884057971016, "percentage": 4.65, "elapsed_time": "0:28:23", "remaining_time": "9:41:54", "throughput": "1016.55", "total_tokens": 1731315} | |
| {"current_steps": 75, "total_steps": 1505, "loss": 2.7935, "learning_rate": 4.9694246071778604e-05, "epoch": 1.7391304347826086, "percentage": 4.98, "elapsed_time": "0:30:28", "remaining_time": "9:40:59", "throughput": "1016.39", "total_tokens": 1858269} | |
| {"current_steps": 80, "total_steps": 1505, "loss": 2.7305, "learning_rate": 4.9652217902637596e-05, "epoch": 1.855072463768116, "percentage": 5.32, "elapsed_time": "0:32:31", "remaining_time": "9:39:17", "throughput": "1017.05", "total_tokens": 1984587} | |
| {"current_steps": 85, "total_steps": 1505, "loss": 2.6482, "learning_rate": 4.9607504272744575e-05, "epoch": 1.971014492753623, "percentage": 5.65, "elapsed_time": "0:34:48", "remaining_time": "9:41:24", "throughput": "1010.18", "total_tokens": 2109391} | |
| {"current_steps": 90, "total_steps": 1505, "loss": 2.4292, "learning_rate": 4.956011005292692e-05, "epoch": 2.0869565217391304, "percentage": 5.98, "elapsed_time": "0:37:08", "remaining_time": "9:43:59", "throughput": "1007.97", "total_tokens": 2246413} | |
| {"current_steps": 95, "total_steps": 1505, "loss": 2.1416, "learning_rate": 4.951004040601898e-05, "epoch": 2.2028985507246377, "percentage": 6.31, "elapsed_time": "0:39:34", "remaining_time": "9:47:21", "throughput": "1005.24", "total_tokens": 2386890} | |
| {"current_steps": 100, "total_steps": 1505, "loss": 2.2847, "learning_rate": 4.945730078629964e-05, "epoch": 2.318840579710145, "percentage": 6.64, "elapsed_time": "0:41:53", "remaining_time": "9:48:28", "throughput": "1003.68", "total_tokens": 2522302} | |
| {"current_steps": 105, "total_steps": 1505, "loss": 2.0944, "learning_rate": 4.9401896938898185e-05, "epoch": 2.4347826086956523, "percentage": 6.98, "elapsed_time": "0:44:27", "remaining_time": "9:52:47", "throughput": "990.48", "total_tokens": 2642208} | |
| {"current_steps": 110, "total_steps": 1505, "loss": 2.2862, "learning_rate": 4.934383489916843e-05, "epoch": 2.550724637681159, "percentage": 7.31, "elapsed_time": "0:46:51", "remaining_time": "9:54:19", "throughput": "988.89", "total_tokens": 2780587} | |
| {"current_steps": 115, "total_steps": 1505, "loss": 2.105, "learning_rate": 4.928312099203131e-05, "epoch": 2.6666666666666665, "percentage": 7.64, "elapsed_time": "0:48:26", "remaining_time": "9:45:36", "throughput": "992.54", "total_tokens": 2885320} | |
| {"current_steps": 120, "total_steps": 1505, "loss": 2.0287, "learning_rate": 4.921976183128585e-05, "epoch": 2.782608695652174, "percentage": 7.97, "elapsed_time": "0:50:19", "remaining_time": "9:40:48", "throughput": "992.56", "total_tokens": 2996923} | |
| {"current_steps": 125, "total_steps": 1505, "loss": 2.0162, "learning_rate": 4.9153764318888706e-05, "epoch": 2.898550724637681, "percentage": 8.31, "elapsed_time": "0:51:54", "remaining_time": "9:33:09", "throughput": "995.97", "total_tokens": 3102391} | |
| {"current_steps": 130, "total_steps": 1505, "loss": 2.2464, "learning_rate": 4.908513564420231e-05, "epoch": 3.0144927536231885, "percentage": 8.64, "elapsed_time": "0:54:01", "remaining_time": "9:31:24", "throughput": "997.54", "total_tokens": 3233443} | |
| {"current_steps": 135, "total_steps": 1505, "loss": 1.6561, "learning_rate": 4.90138832832117e-05, "epoch": 3.130434782608696, "percentage": 8.97, "elapsed_time": "0:56:00", "remaining_time": "9:28:23", "throughput": "999.52", "total_tokens": 3358966} | |
| {"current_steps": 140, "total_steps": 1505, "loss": 1.6113, "learning_rate": 4.894001499771015e-05, "epoch": 3.246376811594203, "percentage": 9.3, "elapsed_time": "0:58:09", "remaining_time": "9:27:05", "throughput": "1000.08", "total_tokens": 3490069} | |
| {"current_steps": 145, "total_steps": 1505, "loss": 1.6235, "learning_rate": 4.886353883445363e-05, "epoch": 3.36231884057971, "percentage": 9.63, "elapsed_time": "1:00:15", "remaining_time": "9:25:14", "throughput": "998.34", "total_tokens": 3609842} | |
| {"current_steps": 150, "total_steps": 1505, "loss": 1.7873, "learning_rate": 4.878446312428424e-05, "epoch": 3.4782608695652173, "percentage": 9.97, "elapsed_time": "1:02:40", "remaining_time": "9:26:06", "throughput": "997.72", "total_tokens": 3751570} | |
| {"current_steps": 155, "total_steps": 1505, "loss": 1.3723, "learning_rate": 4.8702796481222714e-05, "epoch": 3.5942028985507246, "percentage": 10.3, "elapsed_time": "1:04:42", "remaining_time": "9:23:34", "throughput": "995.58", "total_tokens": 3865303} | |
| {"current_steps": 160, "total_steps": 1505, "loss": 1.6512, "learning_rate": 4.861854780153004e-05, "epoch": 3.710144927536232, "percentage": 10.63, "elapsed_time": "1:06:42", "remaining_time": "9:20:49", "throughput": "997.10", "total_tokens": 3991347} | |
| {"current_steps": 165, "total_steps": 1505, "loss": 1.5524, "learning_rate": 4.853172626273841e-05, "epoch": 3.8260869565217392, "percentage": 10.96, "elapsed_time": "1:08:43", "remaining_time": "9:18:06", "throughput": "997.64", "total_tokens": 4113654} | |
| {"current_steps": 170, "total_steps": 1505, "loss": 1.5954, "learning_rate": 4.8442341322651385e-05, "epoch": 3.942028985507246, "percentage": 11.3, "elapsed_time": "1:10:40", "remaining_time": "9:14:57", "throughput": "999.10", "total_tokens": 4236348} | |
| {"current_steps": 175, "total_steps": 1505, "loss": 1.1652, "learning_rate": 4.83504027183137e-05, "epoch": 4.057971014492754, "percentage": 11.63, "elapsed_time": "1:12:15", "remaining_time": "9:09:10", "throughput": "1001.11", "total_tokens": 4340378} | |
| {"current_steps": 180, "total_steps": 1505, "loss": 1.1995, "learning_rate": 4.825592046495054e-05, "epoch": 4.173913043478261, "percentage": 11.96, "elapsed_time": "1:14:31", "remaining_time": "9:08:33", "throughput": "1000.51", "total_tokens": 4473601} | |
| {"current_steps": 185, "total_steps": 1505, "loss": 0.9431, "learning_rate": 4.8158904854876555e-05, "epoch": 4.2898550724637685, "percentage": 12.29, "elapsed_time": "1:16:28", "remaining_time": "9:05:42", "throughput": "999.58", "total_tokens": 4586911} | |
| {"current_steps": 190, "total_steps": 1505, "loss": 1.1136, "learning_rate": 4.805936645637463e-05, "epoch": 4.405797101449275, "percentage": 12.62, "elapsed_time": "1:18:30", "remaining_time": "9:03:20", "throughput": "998.32", "total_tokens": 4702445} | |
| {"current_steps": 195, "total_steps": 1505, "loss": 1.1509, "learning_rate": 4.795731611254473e-05, "epoch": 4.521739130434782, "percentage": 12.96, "elapsed_time": "1:20:34", "remaining_time": "9:01:18", "throughput": "999.32", "total_tokens": 4831301} | |
| {"current_steps": 200, "total_steps": 1505, "loss": 0.9962, "learning_rate": 4.785276494012263e-05, "epoch": 4.63768115942029, "percentage": 13.29, "elapsed_time": "1:22:18", "remaining_time": "8:57:05", "throughput": "1000.59", "total_tokens": 4941656} | |
| {"current_steps": 205, "total_steps": 1505, "loss": 1.2377, "learning_rate": 4.7745724328269e-05, "epoch": 4.753623188405797, "percentage": 13.62, "elapsed_time": "1:25:08", "remaining_time": "8:59:53", "throughput": "996.13", "total_tokens": 5088437} | |
| {"current_steps": 210, "total_steps": 1505, "loss": 1.234, "learning_rate": 4.763620593732867e-05, "epoch": 4.869565217391305, "percentage": 13.95, "elapsed_time": "1:27:18", "remaining_time": "8:58:21", "throughput": "996.52", "total_tokens": 5219806} | |
| {"current_steps": 215, "total_steps": 1505, "loss": 1.1453, "learning_rate": 4.752422169756048e-05, "epoch": 4.9855072463768115, "percentage": 14.29, "elapsed_time": "1:29:10", "remaining_time": "8:55:03", "throughput": "998.06", "total_tokens": 5340222} | |
| {"current_steps": 220, "total_steps": 1505, "loss": 0.9056, "learning_rate": 4.740978380783765e-05, "epoch": 5.101449275362318, "percentage": 14.62, "elapsed_time": "1:31:31", "remaining_time": "8:54:32", "throughput": "997.31", "total_tokens": 5476315} | |
| {"current_steps": 225, "total_steps": 1505, "loss": 0.7349, "learning_rate": 4.7292904734318924e-05, "epoch": 5.217391304347826, "percentage": 14.95, "elapsed_time": "1:33:21", "remaining_time": "8:51:03", "throughput": "998.01", "total_tokens": 5589951} | |
| {"current_steps": 230, "total_steps": 1505, "loss": 0.6968, "learning_rate": 4.7173597209090534e-05, "epoch": 5.333333333333333, "percentage": 15.28, "elapsed_time": "1:35:17", "remaining_time": "8:48:13", "throughput": "999.00", "total_tokens": 5711449} | |
| {"current_steps": 235, "total_steps": 1505, "loss": 0.851, "learning_rate": 4.70518742287793e-05, "epoch": 5.449275362318841, "percentage": 15.61, "elapsed_time": "1:37:41", "remaining_time": "8:47:58", "throughput": "998.45", "total_tokens": 5852650} | |
| {"current_steps": 240, "total_steps": 1505, "loss": 0.716, "learning_rate": 4.6927749053136866e-05, "epoch": 5.565217391304348, "percentage": 15.95, "elapsed_time": "1:39:37", "remaining_time": "8:45:03", "throughput": "999.21", "total_tokens": 5972289} | |
| {"current_steps": 245, "total_steps": 1505, "loss": 0.6384, "learning_rate": 4.6801235203595195e-05, "epoch": 5.681159420289855, "percentage": 16.28, "elapsed_time": "1:41:26", "remaining_time": "8:41:40", "throughput": "1000.40", "total_tokens": 6088707} | |
| {"current_steps": 250, "total_steps": 1505, "loss": 0.7799, "learning_rate": 4.667234646179368e-05, "epoch": 5.797101449275362, "percentage": 16.61, "elapsed_time": "1:43:47", "remaining_time": "8:41:00", "throughput": "998.13", "total_tokens": 6215471} | |
| {"current_steps": 255, "total_steps": 1505, "loss": 0.7923, "learning_rate": 4.654109686807787e-05, "epoch": 5.913043478260869, "percentage": 16.94, "elapsed_time": "1:45:48", "remaining_time": "8:38:42", "throughput": "997.95", "total_tokens": 6335935} | |
| {"current_steps": 260, "total_steps": 1505, "loss": 0.7452, "learning_rate": 4.640750071996995e-05, "epoch": 6.028985507246377, "percentage": 17.28, "elapsed_time": "1:47:52", "remaining_time": "8:36:31", "throughput": "998.69", "total_tokens": 6463689} | |
| {"current_steps": 265, "total_steps": 1505, "loss": 0.4085, "learning_rate": 4.6271572570611296e-05, "epoch": 6.144927536231884, "percentage": 17.61, "elapsed_time": "1:49:37", "remaining_time": "8:32:55", "throughput": "999.98", "total_tokens": 6576954} | |
| {"current_steps": 270, "total_steps": 1505, "loss": 0.5777, "learning_rate": 4.613332722717714e-05, "epoch": 6.260869565217392, "percentage": 17.94, "elapsed_time": "1:51:51", "remaining_time": "8:31:39", "throughput": "1000.40", "total_tokens": 6714404} | |
| {"current_steps": 275, "total_steps": 1505, "loss": 0.4718, "learning_rate": 4.5992779749263546e-05, "epoch": 6.3768115942028984, "percentage": 18.27, "elapsed_time": "1:53:54", "remaining_time": "8:29:28", "throughput": "1000.87", "total_tokens": 6840385} | |
| {"current_steps": 280, "total_steps": 1505, "loss": 0.3723, "learning_rate": 4.584994544724695e-05, "epoch": 6.492753623188406, "percentage": 18.6, "elapsed_time": "1:55:39", "remaining_time": "8:26:01", "throughput": "1002.09", "total_tokens": 6954269} | |
| {"current_steps": 285, "total_steps": 1505, "loss": 0.4453, "learning_rate": 4.5704839880616296e-05, "epoch": 6.608695652173913, "percentage": 18.94, "elapsed_time": "1:57:50", "remaining_time": "8:24:25", "throughput": "1000.84", "total_tokens": 7076143} | |
| {"current_steps": 290, "total_steps": 1505, "loss": 0.5742, "learning_rate": 4.5557478856278114e-05, "epoch": 6.72463768115942, "percentage": 19.27, "elapsed_time": "1:59:51", "remaining_time": "8:22:11", "throughput": "1001.38", "total_tokens": 7201833} | |
| {"current_steps": 295, "total_steps": 1505, "loss": 0.5291, "learning_rate": 4.5407878426834596e-05, "epoch": 6.840579710144928, "percentage": 19.6, "elapsed_time": "2:01:56", "remaining_time": "8:20:09", "throughput": "1001.95", "total_tokens": 7330479} | |
| {"current_steps": 300, "total_steps": 1505, "loss": 0.4968, "learning_rate": 4.5256054888834934e-05, "epoch": 6.956521739130435, "percentage": 19.93, "elapsed_time": "2:03:49", "remaining_time": "8:17:23", "throughput": "1002.60", "total_tokens": 7449244} | |
| {"current_steps": 305, "total_steps": 1505, "loss": 0.421, "learning_rate": 4.5102024781000077e-05, "epoch": 7.072463768115942, "percentage": 20.27, "elapsed_time": "2:06:19", "remaining_time": "8:17:00", "throughput": "999.94", "total_tokens": 7578947} | |
| {"current_steps": 310, "total_steps": 1505, "loss": 0.2767, "learning_rate": 4.4945804882421086e-05, "epoch": 7.188405797101449, "percentage": 20.6, "elapsed_time": "2:08:26", "remaining_time": "8:15:08", "throughput": "998.07", "total_tokens": 7691948} | |
| {"current_steps": 315, "total_steps": 1505, "loss": 0.2922, "learning_rate": 4.478741221073136e-05, "epoch": 7.304347826086957, "percentage": 20.93, "elapsed_time": "2:10:25", "remaining_time": "8:12:42", "throughput": "998.79", "total_tokens": 7815786} | |
| {"current_steps": 320, "total_steps": 1505, "loss": 0.2768, "learning_rate": 4.4626864020252774e-05, "epoch": 7.420289855072464, "percentage": 21.26, "elapsed_time": "2:12:16", "remaining_time": "8:09:49", "throughput": "998.57", "total_tokens": 7925106} | |
| {"current_steps": 325, "total_steps": 1505, "loss": 0.3281, "learning_rate": 4.446417780011618e-05, "epoch": 7.536231884057971, "percentage": 21.59, "elapsed_time": "2:14:49", "remaining_time": "8:09:31", "throughput": "995.98", "total_tokens": 8057202} | |
| {"current_steps": 330, "total_steps": 1505, "loss": 0.3374, "learning_rate": 4.42993712723562e-05, "epoch": 7.6521739130434785, "percentage": 21.93, "elapsed_time": "2:17:00", "remaining_time": "8:07:51", "throughput": "995.98", "total_tokens": 8187865} | |
| {"current_steps": 335, "total_steps": 1505, "loss": 0.2491, "learning_rate": 4.413246238998069e-05, "epoch": 7.768115942028985, "percentage": 22.26, "elapsed_time": "2:18:55", "remaining_time": "8:05:11", "throughput": "996.31", "total_tokens": 8304605} | |
| {"current_steps": 340, "total_steps": 1505, "loss": 0.3893, "learning_rate": 4.3963469335015085e-05, "epoch": 7.884057971014493, "percentage": 22.59, "elapsed_time": "2:21:09", "remaining_time": "8:03:40", "throughput": "996.21", "total_tokens": 8437319} | |
| {"current_steps": 345, "total_steps": 1505, "loss": 0.3761, "learning_rate": 4.379241051652174e-05, "epoch": 8.0, "percentage": 22.92, "elapsed_time": "2:23:26", "remaining_time": "8:02:17", "throughput": "996.14", "total_tokens": 8573080} | |
| {"current_steps": 350, "total_steps": 1505, "loss": 0.236, "learning_rate": 4.361930456859456e-05, "epoch": 8.115942028985508, "percentage": 23.26, "elapsed_time": "2:25:38", "remaining_time": "8:00:37", "throughput": "996.45", "total_tokens": 8707741} | |
| {"current_steps": 355, "total_steps": 1505, "loss": 0.1584, "learning_rate": 4.34441703483291e-05, "epoch": 8.231884057971014, "percentage": 23.59, "elapsed_time": "2:27:37", "remaining_time": "7:58:12", "throughput": "996.43", "total_tokens": 8825774} | |
| {"current_steps": 360, "total_steps": 1505, "loss": 0.1481, "learning_rate": 4.326702693376844e-05, "epoch": 8.347826086956522, "percentage": 23.92, "elapsed_time": "2:29:16", "remaining_time": "7:54:47", "throughput": "997.27", "total_tokens": 8932249} | |
| {"current_steps": 365, "total_steps": 1505, "loss": 0.1743, "learning_rate": 4.308789362182492e-05, "epoch": 8.46376811594203, "percentage": 24.25, "elapsed_time": "2:31:20", "remaining_time": "7:52:40", "throughput": "996.84", "total_tokens": 9051548} | |
| {"current_steps": 370, "total_steps": 1505, "loss": 0.3162, "learning_rate": 4.290678992617798e-05, "epoch": 8.579710144927537, "percentage": 24.58, "elapsed_time": "2:33:45", "remaining_time": "7:51:39", "throughput": "996.95", "total_tokens": 9197232} | |
| {"current_steps": 375, "total_steps": 1505, "loss": 0.2235, "learning_rate": 4.272373557514858e-05, "epoch": 8.695652173913043, "percentage": 24.92, "elapsed_time": "2:35:41", "remaining_time": "7:49:10", "throughput": "997.40", "total_tokens": 9317650} | |
| {"current_steps": 380, "total_steps": 1505, "loss": 0.2504, "learning_rate": 4.2538750509550054e-05, "epoch": 8.81159420289855, "percentage": 25.25, "elapsed_time": "2:37:57", "remaining_time": "7:47:39", "throughput": "997.16", "total_tokens": 9450765} | |
| {"current_steps": 385, "total_steps": 1505, "loss": 0.2136, "learning_rate": 4.235185488051585e-05, "epoch": 8.927536231884059, "percentage": 25.58, "elapsed_time": "2:40:03", "remaining_time": "7:45:37", "throughput": "997.85", "total_tokens": 9582961} | |
| {"current_steps": 390, "total_steps": 1505, "loss": 0.1047, "learning_rate": 4.216306904730447e-05, "epoch": 9.043478260869565, "percentage": 25.91, "elapsed_time": "2:41:35", "remaining_time": "7:41:58", "throughput": "998.27", "total_tokens": 9678616} | |
| {"current_steps": 395, "total_steps": 1505, "loss": 0.1015, "learning_rate": 4.1972413575081595e-05, "epoch": 9.159420289855072, "percentage": 26.25, "elapsed_time": "2:43:18", "remaining_time": "7:38:54", "throughput": "999.00", "total_tokens": 9788512} | |
| {"current_steps": 400, "total_steps": 1505, "loss": 0.1505, "learning_rate": 4.177990923267986e-05, "epoch": 9.27536231884058, "percentage": 26.58, "elapsed_time": "2:45:26", "remaining_time": "7:37:03", "throughput": "998.92", "total_tokens": 9916229} | |
| {"current_steps": 405, "total_steps": 1505, "loss": 0.1135, "learning_rate": 4.158557699033644e-05, "epoch": 9.391304347826088, "percentage": 26.91, "elapsed_time": "2:47:51", "remaining_time": "7:35:55", "throughput": "997.11", "total_tokens": 10042697} | |
| {"current_steps": 410, "total_steps": 1505, "loss": 0.1832, "learning_rate": 4.138943801740865e-05, "epoch": 9.507246376811594, "percentage": 27.24, "elapsed_time": "2:50:00", "remaining_time": "7:34:03", "throughput": "997.17", "total_tokens": 10171849} | |
| {"current_steps": 415, "total_steps": 1505, "loss": 0.1178, "learning_rate": 4.119151368006793e-05, "epoch": 9.623188405797102, "percentage": 27.57, "elapsed_time": "2:51:42", "remaining_time": "7:30:58", "throughput": "998.04", "total_tokens": 10281924} | |
| {"current_steps": 420, "total_steps": 1505, "loss": 0.1426, "learning_rate": 4.099182553897229e-05, "epoch": 9.73913043478261, "percentage": 27.91, "elapsed_time": "2:54:17", "remaining_time": "7:30:14", "throughput": "996.34", "total_tokens": 10418758} | |
| {"current_steps": 425, "total_steps": 1505, "loss": 0.1603, "learning_rate": 4.079039534691767e-05, "epoch": 9.855072463768115, "percentage": 28.24, "elapsed_time": "2:56:32", "remaining_time": "7:28:36", "throughput": "996.80", "total_tokens": 10558322} | |
| {"current_steps": 430, "total_steps": 1505, "loss": 0.1548, "learning_rate": 4.058724504646834e-05, "epoch": 9.971014492753623, "percentage": 28.57, "elapsed_time": "2:58:42", "remaining_time": "7:26:45", "throughput": "996.01", "total_tokens": 10679536} | |
| {"current_steps": 435, "total_steps": 1505, "loss": 0.1407, "learning_rate": 4.0382396767566536e-05, "epoch": 10.08695652173913, "percentage": 28.9, "elapsed_time": "3:01:04", "remaining_time": "7:25:25", "throughput": "995.97", "total_tokens": 10821076} | |
| {"current_steps": 440, "total_steps": 1505, "loss": 0.0791, "learning_rate": 4.017587282512181e-05, "epoch": 10.202898550724637, "percentage": 29.24, "elapsed_time": "3:03:15", "remaining_time": "7:23:34", "throughput": "995.82", "total_tokens": 10949771} | |
| {"current_steps": 445, "total_steps": 1505, "loss": 0.0722, "learning_rate": 3.9967695716580224e-05, "epoch": 10.318840579710145, "percentage": 29.57, "elapsed_time": "3:05:26", "remaining_time": "7:21:43", "throughput": "995.11", "total_tokens": 11072044} | |
| {"current_steps": 450, "total_steps": 1505, "loss": 0.0655, "learning_rate": 3.975788811947351e-05, "epoch": 10.434782608695652, "percentage": 29.9, "elapsed_time": "3:07:09", "remaining_time": "7:18:47", "throughput": "995.82", "total_tokens": 11182627} | |
| {"current_steps": 455, "total_steps": 1505, "loss": 0.0723, "learning_rate": 3.954647288894883e-05, "epoch": 10.55072463768116, "percentage": 30.23, "elapsed_time": "3:09:02", "remaining_time": "7:16:15", "throughput": "996.49", "total_tokens": 11303028} | |
| {"current_steps": 460, "total_steps": 1505, "loss": 0.0655, "learning_rate": 3.933347305527898e-05, "epoch": 10.666666666666666, "percentage": 30.56, "elapsed_time": "3:10:46", "remaining_time": "7:13:23", "throughput": "997.33", "total_tokens": 11415868} | |
| {"current_steps": 465, "total_steps": 1505, "loss": 0.1534, "learning_rate": 3.911891182135371e-05, "epoch": 10.782608695652174, "percentage": 30.9, "elapsed_time": "3:13:15", "remaining_time": "7:12:13", "throughput": "996.60", "total_tokens": 11555653} | |
| {"current_steps": 470, "total_steps": 1505, "loss": 0.0947, "learning_rate": 3.8902812560152066e-05, "epoch": 10.898550724637682, "percentage": 31.23, "elapsed_time": "3:15:17", "remaining_time": "7:10:03", "throughput": "996.89", "total_tokens": 11681065} | |
| {"current_steps": 475, "total_steps": 1505, "loss": 0.0868, "learning_rate": 3.868519881219631e-05, "epoch": 11.014492753623188, "percentage": 31.56, "elapsed_time": "3:17:22", "remaining_time": "7:07:58", "throughput": "997.29", "total_tokens": 11809957} | |
| {"current_steps": 480, "total_steps": 1505, "loss": 0.0467, "learning_rate": 3.846609428298757e-05, "epoch": 11.130434782608695, "percentage": 31.89, "elapsed_time": "3:19:27", "remaining_time": "7:05:55", "throughput": "997.52", "total_tokens": 11937881} | |
| {"current_steps": 485, "total_steps": 1505, "loss": 0.0521, "learning_rate": 3.824552284042351e-05, "epoch": 11.246376811594203, "percentage": 32.23, "elapsed_time": "3:21:11", "remaining_time": "7:03:06", "throughput": "998.15", "total_tokens": 12048905} | |
| {"current_steps": 490, "total_steps": 1505, "loss": 0.051, "learning_rate": 3.8023508512198256e-05, "epoch": 11.36231884057971, "percentage": 32.56, "elapsed_time": "3:23:21", "remaining_time": "7:01:15", "throughput": "998.66", "total_tokens": 12185453} | |
| {"current_steps": 495, "total_steps": 1505, "loss": 0.0753, "learning_rate": 3.780007548318507e-05, "epoch": 11.478260869565217, "percentage": 32.89, "elapsed_time": "3:25:21", "remaining_time": "6:59:01", "throughput": "999.10", "total_tokens": 12310911} | |
| {"current_steps": 500, "total_steps": 1505, "loss": 0.0601, "learning_rate": 3.7575248092801686e-05, "epoch": 11.594202898550725, "percentage": 33.22, "elapsed_time": "3:27:25", "remaining_time": "6:56:54", "throughput": "999.57", "total_tokens": 12439708} | |
| {"current_steps": 505, "total_steps": 1505, "loss": 0.0533, "learning_rate": 3.734905083235901e-05, "epoch": 11.710144927536232, "percentage": 33.55, "elapsed_time": "3:29:31", "remaining_time": "6:54:54", "throughput": "998.61", "total_tokens": 12554467} | |
| {"current_steps": 510, "total_steps": 1505, "loss": 0.064, "learning_rate": 3.712150834239313e-05, "epoch": 11.826086956521738, "percentage": 33.89, "elapsed_time": "3:31:33", "remaining_time": "6:52:45", "throughput": "999.10", "total_tokens": 12682329} | |
| {"current_steps": 515, "total_steps": 1505, "loss": 0.0755, "learning_rate": 3.689264540998116e-05, "epoch": 11.942028985507246, "percentage": 34.22, "elapsed_time": "3:33:26", "remaining_time": "6:50:17", "throughput": "999.59", "total_tokens": 12800852} | |
| {"current_steps": 520, "total_steps": 1505, "loss": 0.0553, "learning_rate": 3.66624869660411e-05, "epoch": 12.057971014492754, "percentage": 34.55, "elapsed_time": "3:35:16", "remaining_time": "6:47:47", "throughput": "1000.07", "total_tokens": 12917527} | |
| {"current_steps": 525, "total_steps": 1505, "loss": 0.0355, "learning_rate": 3.6431058082615964e-05, "epoch": 12.173913043478262, "percentage": 34.88, "elapsed_time": "3:37:30", "remaining_time": "6:46:00", "throughput": "999.57", "total_tokens": 13044774} | |
| {"current_steps": 530, "total_steps": 1505, "loss": 0.0413, "learning_rate": 3.619838397014263e-05, "epoch": 12.289855072463768, "percentage": 35.22, "elapsed_time": "3:39:37", "remaining_time": "6:44:01", "throughput": "999.88", "total_tokens": 13175692} | |
| {"current_steps": 535, "total_steps": 1505, "loss": 0.0596, "learning_rate": 3.5964489974705553e-05, "epoch": 12.405797101449275, "percentage": 35.55, "elapsed_time": "3:41:28", "remaining_time": "6:41:32", "throughput": "1000.36", "total_tokens": 13293164} | |
| {"current_steps": 540, "total_steps": 1505, "loss": 0.0479, "learning_rate": 3.572940157527572e-05, "epoch": 12.521739130434783, "percentage": 35.88, "elapsed_time": "3:43:28", "remaining_time": "6:39:21", "throughput": "1000.70", "total_tokens": 13417894} | |
| {"current_steps": 545, "total_steps": 1505, "loss": 0.047, "learning_rate": 3.549314438093515e-05, "epoch": 12.63768115942029, "percentage": 36.21, "elapsed_time": "3:45:42", "remaining_time": "6:37:33", "throughput": "1000.73", "total_tokens": 13551913} | |
| {"current_steps": 550, "total_steps": 1505, "loss": 0.0492, "learning_rate": 3.525574412808717e-05, "epoch": 12.753623188405797, "percentage": 36.54, "elapsed_time": "3:47:41", "remaining_time": "6:35:20", "throughput": "1001.04", "total_tokens": 13675309} | |
| {"current_steps": 555, "total_steps": 1505, "loss": 0.0471, "learning_rate": 3.501722667765286e-05, "epoch": 12.869565217391305, "percentage": 36.88, "elapsed_time": "3:49:38", "remaining_time": "6:33:05", "throughput": "1001.36", "total_tokens": 13797691} | |
| {"current_steps": 560, "total_steps": 1505, "loss": 0.1041, "learning_rate": 3.47776180122539e-05, "epoch": 12.985507246376812, "percentage": 37.21, "elapsed_time": "3:51:48", "remaining_time": "6:31:10", "throughput": "1000.82", "total_tokens": 13919770} | |
| {"current_steps": 565, "total_steps": 1505, "loss": 0.0282, "learning_rate": 3.453694423338225e-05, "epoch": 13.101449275362318, "percentage": 37.54, "elapsed_time": "3:53:36", "remaining_time": "6:28:39", "throughput": "1001.50", "total_tokens": 14037673} | |
| {"current_steps": 570, "total_steps": 1505, "loss": 0.0272, "learning_rate": 3.4295231558556715e-05, "epoch": 13.217391304347826, "percentage": 37.87, "elapsed_time": "3:55:59", "remaining_time": "6:27:07", "throughput": "1000.51", "total_tokens": 14167090} | |
| {"current_steps": 575, "total_steps": 1505, "loss": 0.0342, "learning_rate": 3.4052506318467084e-05, "epoch": 13.333333333333334, "percentage": 38.21, "elapsed_time": "3:58:23", "remaining_time": "6:25:33", "throughput": "1000.60", "total_tokens": 14311710} | |
| {"current_steps": 580, "total_steps": 1505, "loss": 0.0855, "learning_rate": 3.3808794954105716e-05, "epoch": 13.44927536231884, "percentage": 38.54, "elapsed_time": "3:59:53", "remaining_time": "6:22:35", "throughput": "1000.75", "total_tokens": 14404322} | |
| {"current_steps": 585, "total_steps": 1505, "loss": 0.0378, "learning_rate": 3.356412401388732e-05, "epoch": 13.565217391304348, "percentage": 38.87, "elapsed_time": "4:01:55", "remaining_time": "6:20:28", "throughput": "1001.03", "total_tokens": 14530794} | |
| {"current_steps": 590, "total_steps": 1505, "loss": 0.0457, "learning_rate": 3.3318520150756846e-05, "epoch": 13.681159420289855, "percentage": 39.2, "elapsed_time": "4:03:33", "remaining_time": "6:17:42", "throughput": "1001.65", "total_tokens": 14637342} | |
| {"current_steps": 595, "total_steps": 1505, "loss": 0.0453, "learning_rate": 3.307201011928616e-05, "epoch": 13.797101449275363, "percentage": 39.53, "elapsed_time": "4:06:05", "remaining_time": "6:16:21", "throughput": "1001.53", "total_tokens": 14787534} | |
| {"current_steps": 600, "total_steps": 1505, "loss": 0.0378, "learning_rate": 3.282462077275947e-05, "epoch": 13.91304347826087, "percentage": 39.87, "elapsed_time": "4:08:03", "remaining_time": "6:14:09", "throughput": "1001.72", "total_tokens": 14909175} | |
| {"current_steps": 605, "total_steps": 1505, "loss": 0.0296, "learning_rate": 3.257637906024822e-05, "epoch": 14.028985507246377, "percentage": 40.2, "elapsed_time": "4:10:31", "remaining_time": "6:12:40", "throughput": "999.94", "total_tokens": 15030530} | |
| {"current_steps": 610, "total_steps": 1505, "loss": 0.0216, "learning_rate": 3.2327312023675287e-05, "epoch": 14.144927536231885, "percentage": 40.53, "elapsed_time": "4:12:39", "remaining_time": "6:10:42", "throughput": "999.25", "total_tokens": 15148359} | |
| {"current_steps": 615, "total_steps": 1505, "loss": 0.0299, "learning_rate": 3.2077446794869295e-05, "epoch": 14.26086956521739, "percentage": 40.86, "elapsed_time": "4:14:57", "remaining_time": "6:08:58", "throughput": "998.89", "total_tokens": 15280749} | |
| {"current_steps": 620, "total_steps": 1505, "loss": 0.0247, "learning_rate": 3.1826810592609036e-05, "epoch": 14.376811594202898, "percentage": 41.2, "elapsed_time": "4:16:54", "remaining_time": "6:06:42", "throughput": "998.88", "total_tokens": 15397167} | |
| {"current_steps": 625, "total_steps": 1505, "loss": 0.0455, "learning_rate": 3.157543071965835e-05, "epoch": 14.492753623188406, "percentage": 41.53, "elapsed_time": "4:19:23", "remaining_time": "6:05:13", "throughput": "997.37", "total_tokens": 15522794} | |
| {"current_steps": 630, "total_steps": 1505, "loss": 0.0262, "learning_rate": 3.132333455979202e-05, "epoch": 14.608695652173914, "percentage": 41.86, "elapsed_time": "4:21:11", "remaining_time": "6:02:46", "throughput": "997.84", "total_tokens": 15637987} | |
| {"current_steps": 635, "total_steps": 1505, "loss": 0.0281, "learning_rate": 3.107054957481271e-05, "epoch": 14.72463768115942, "percentage": 42.19, "elapsed_time": "4:23:21", "remaining_time": "6:00:49", "throughput": "998.19", "total_tokens": 15773163} | |
| {"current_steps": 640, "total_steps": 1505, "loss": 0.0294, "learning_rate": 3.081710330155942e-05, "epoch": 14.840579710144928, "percentage": 42.52, "elapsed_time": "4:25:14", "remaining_time": "5:58:29", "throughput": "998.63", "total_tokens": 15892659} | |
| {"current_steps": 645, "total_steps": 1505, "loss": 0.0291, "learning_rate": 3.056302334890786e-05, "epoch": 14.956521739130435, "percentage": 42.86, "elapsed_time": "4:27:21", "remaining_time": "5:56:28", "throughput": "998.97", "total_tokens": 16024576} | |
| {"current_steps": 650, "total_steps": 1505, "loss": 0.0216, "learning_rate": 3.030833739476285e-05, "epoch": 15.072463768115941, "percentage": 43.19, "elapsed_time": "4:29:27", "remaining_time": "5:54:27", "throughput": "999.01", "total_tokens": 16151987} | |
| {"current_steps": 655, "total_steps": 1505, "loss": 0.0218, "learning_rate": 3.0053073183043256e-05, "epoch": 15.18840579710145, "percentage": 43.52, "elapsed_time": "4:31:54", "remaining_time": "5:52:51", "throughput": "997.79", "total_tokens": 16278639} | |
| {"current_steps": 660, "total_steps": 1505, "loss": 0.0283, "learning_rate": 2.979725852065981e-05, "epoch": 15.304347826086957, "percentage": 43.85, "elapsed_time": "4:34:13", "remaining_time": "5:51:05", "throughput": "997.64", "total_tokens": 16414743} | |
| {"current_steps": 665, "total_steps": 1505, "loss": 0.0259, "learning_rate": 2.954092127448591e-05, "epoch": 15.420289855072463, "percentage": 44.19, "elapsed_time": "4:36:02", "remaining_time": "5:48:41", "throughput": "997.98", "total_tokens": 16529298} | |
| {"current_steps": 670, "total_steps": 1505, "loss": 0.0716, "learning_rate": 2.9284089368322045e-05, "epoch": 15.53623188405797, "percentage": 44.52, "elapsed_time": "4:38:05", "remaining_time": "5:46:34", "throughput": "998.25", "total_tokens": 16655909} | |
| {"current_steps": 675, "total_steps": 1505, "loss": 0.025, "learning_rate": 2.9026790779853874e-05, "epoch": 15.652173913043478, "percentage": 44.85, "elapsed_time": "4:40:27", "remaining_time": "5:44:51", "throughput": "998.26", "total_tokens": 16798263} | |
| {"current_steps": 680, "total_steps": 1505, "loss": 0.0218, "learning_rate": 2.876905353760459e-05, "epoch": 15.768115942028986, "percentage": 45.18, "elapsed_time": "4:42:22", "remaining_time": "5:42:35", "throughput": "998.47", "total_tokens": 16916827} | |
| {"current_steps": 685, "total_steps": 1505, "loss": 0.0231, "learning_rate": 2.8510905717881614e-05, "epoch": 15.884057971014492, "percentage": 45.51, "elapsed_time": "4:44:20", "remaining_time": "5:40:23", "throughput": "998.80", "total_tokens": 17040247} | |
| {"current_steps": 690, "total_steps": 1505, "loss": 0.0228, "learning_rate": 2.8252375441718137e-05, "epoch": 16.0, "percentage": 45.85, "elapsed_time": "4:45:58", "remaining_time": "5:37:46", "throughput": "999.30", "total_tokens": 17146160} | |
| {"current_steps": 695, "total_steps": 1505, "loss": 0.029, "learning_rate": 2.7993490871809808e-05, "epoch": 16.115942028985508, "percentage": 46.18, "elapsed_time": "4:48:18", "remaining_time": "5:36:01", "throughput": "999.19", "total_tokens": 17284643} | |
| {"current_steps": 700, "total_steps": 1505, "loss": 0.0199, "learning_rate": 2.7734280209446865e-05, "epoch": 16.231884057971016, "percentage": 46.51, "elapsed_time": "4:50:40", "remaining_time": "5:34:16", "throughput": "999.18", "total_tokens": 17426644} | |
| {"current_steps": 705, "total_steps": 1505, "loss": 0.0259, "learning_rate": 2.7474771691442018e-05, "epoch": 16.347826086956523, "percentage": 46.84, "elapsed_time": "4:53:00", "remaining_time": "5:32:29", "throughput": "997.81", "total_tokens": 17541812} | |
| {"current_steps": 710, "total_steps": 1505, "loss": 0.021, "learning_rate": 2.721499358705458e-05, "epoch": 16.463768115942027, "percentage": 47.18, "elapsed_time": "4:55:06", "remaining_time": "5:30:25", "throughput": "997.83", "total_tokens": 17667755} | |
| {"current_steps": 715, "total_steps": 1505, "loss": 0.0199, "learning_rate": 2.6954974194910888e-05, "epoch": 16.579710144927535, "percentage": 47.51, "elapsed_time": "4:56:59", "remaining_time": "5:28:08", "throughput": "998.23", "total_tokens": 17788162} | |
| {"current_steps": 720, "total_steps": 1505, "loss": 0.0189, "learning_rate": 2.6694741839921732e-05, "epoch": 16.695652173913043, "percentage": 47.84, "elapsed_time": "4:58:57", "remaining_time": "5:25:57", "throughput": "998.55", "total_tokens": 17911718} | |
| {"current_steps": 725, "total_steps": 1505, "loss": 0.0169, "learning_rate": 2.6434324870196748e-05, "epoch": 16.81159420289855, "percentage": 48.17, "elapsed_time": "5:00:37", "remaining_time": "5:23:26", "throughput": "998.93", "total_tokens": 18018729} | |
| {"current_steps": 730, "total_steps": 1505, "loss": 0.0209, "learning_rate": 2.617375165395634e-05, "epoch": 16.92753623188406, "percentage": 48.5, "elapsed_time": "5:02:31", "remaining_time": "5:21:10", "throughput": "999.34", "total_tokens": 18139681} | |
| {"current_steps": 735, "total_steps": 1505, "loss": 0.0201, "learning_rate": 2.5913050576441477e-05, "epoch": 17.043478260869566, "percentage": 48.84, "elapsed_time": "5:04:46", "remaining_time": "5:19:16", "throughput": "999.59", "total_tokens": 18278544} | |
| {"current_steps": 740, "total_steps": 1505, "loss": 0.017, "learning_rate": 2.5652250036821523e-05, "epoch": 17.159420289855074, "percentage": 49.17, "elapsed_time": "5:06:53", "remaining_time": "5:17:15", "throughput": "999.09", "total_tokens": 18396700} | |
| {"current_steps": 745, "total_steps": 1505, "loss": 0.0187, "learning_rate": 2.5391378445100644e-05, "epoch": 17.27536231884058, "percentage": 49.5, "elapsed_time": "5:08:38", "remaining_time": "5:14:50", "throughput": "999.36", "total_tokens": 18506229} | |
| {"current_steps": 750, "total_steps": 1505, "loss": 0.0242, "learning_rate": 2.5130464219022992e-05, "epoch": 17.391304347826086, "percentage": 49.83, "elapsed_time": "5:10:26", "remaining_time": "5:12:31", "throughput": "999.72", "total_tokens": 18621580} | |
| {"current_steps": 755, "total_steps": 1505, "loss": 0.0153, "learning_rate": 2.486953578097702e-05, "epoch": 17.507246376811594, "percentage": 50.17, "elapsed_time": "5:12:27", "remaining_time": "5:10:23", "throughput": "1000.04", "total_tokens": 18748382} | |
| {"current_steps": 760, "total_steps": 1505, "loss": 0.0182, "learning_rate": 2.4608621554899362e-05, "epoch": 17.6231884057971, "percentage": 50.5, "elapsed_time": "5:14:42", "remaining_time": "5:08:29", "throughput": "1000.12", "total_tokens": 18884730} | |
| {"current_steps": 765, "total_steps": 1505, "loss": 0.0143, "learning_rate": 2.4347749963178486e-05, "epoch": 17.73913043478261, "percentage": 50.83, "elapsed_time": "5:16:37", "remaining_time": "5:06:16", "throughput": "1000.34", "total_tokens": 19003589} | |
| {"current_steps": 770, "total_steps": 1505, "loss": 0.0164, "learning_rate": 2.4086949423558526e-05, "epoch": 17.855072463768117, "percentage": 51.16, "elapsed_time": "5:18:47", "remaining_time": "5:04:18", "throughput": "1000.46", "total_tokens": 19136411} | |
| {"current_steps": 775, "total_steps": 1505, "loss": 0.0157, "learning_rate": 2.3826248346043663e-05, "epoch": 17.971014492753625, "percentage": 51.5, "elapsed_time": "5:20:52", "remaining_time": "5:02:14", "throughput": "1000.43", "total_tokens": 19260436} | |
| {"current_steps": 780, "total_steps": 1505, "loss": 0.0304, "learning_rate": 2.356567512980326e-05, "epoch": 18.08695652173913, "percentage": 51.83, "elapsed_time": "5:23:02", "remaining_time": "5:00:15", "throughput": "1000.34", "total_tokens": 19388733} | |
| {"current_steps": 785, "total_steps": 1505, "loss": 0.009, "learning_rate": 2.3305258160078274e-05, "epoch": 18.202898550724637, "percentage": 52.16, "elapsed_time": "5:25:22", "remaining_time": "4:58:25", "throughput": "1000.45", "total_tokens": 19531204} | |
| {"current_steps": 790, "total_steps": 1505, "loss": 0.0105, "learning_rate": 2.3045025805089118e-05, "epoch": 18.318840579710145, "percentage": 52.49, "elapsed_time": "5:26:47", "remaining_time": "4:55:45", "throughput": "1000.89", "total_tokens": 19624608} | |
| {"current_steps": 795, "total_steps": 1505, "loss": 0.0104, "learning_rate": 2.278500641294543e-05, "epoch": 18.434782608695652, "percentage": 52.82, "elapsed_time": "5:29:00", "remaining_time": "4:53:49", "throughput": "1000.55", "total_tokens": 19751062} | |
| {"current_steps": 800, "total_steps": 1505, "loss": 0.0103, "learning_rate": 2.252522830855798e-05, "epoch": 18.55072463768116, "percentage": 53.16, "elapsed_time": "5:31:08", "remaining_time": "4:51:49", "throughput": "1000.55", "total_tokens": 19879837} | |
| {"current_steps": 805, "total_steps": 1505, "loss": 0.0107, "learning_rate": 2.2265719790553147e-05, "epoch": 18.666666666666668, "percentage": 53.49, "elapsed_time": "5:33:50", "remaining_time": "4:50:17", "throughput": "999.47", "total_tokens": 20019385} | |
| {"current_steps": 810, "total_steps": 1505, "loss": 0.0269, "learning_rate": 2.2006509128190195e-05, "epoch": 18.782608695652176, "percentage": 53.82, "elapsed_time": "5:35:48", "remaining_time": "4:48:07", "throughput": "999.49", "total_tokens": 20138003} | |
| {"current_steps": 815, "total_steps": 1505, "loss": 0.0086, "learning_rate": 2.174762455828187e-05, "epoch": 18.89855072463768, "percentage": 54.15, "elapsed_time": "5:37:46", "remaining_time": "4:45:58", "throughput": "999.70", "total_tokens": 20260523} | |
| {"current_steps": 820, "total_steps": 1505, "loss": 0.0133, "learning_rate": 2.1489094282118395e-05, "epoch": 19.014492753623188, "percentage": 54.49, "elapsed_time": "5:39:36", "remaining_time": "4:43:41", "throughput": "999.96", "total_tokens": 20375322} | |
| {"current_steps": 825, "total_steps": 1505, "loss": 0.0114, "learning_rate": 2.123094646239541e-05, "epoch": 19.130434782608695, "percentage": 54.82, "elapsed_time": "5:41:09", "remaining_time": "4:41:11", "throughput": "1000.41", "total_tokens": 20477407} | |
| {"current_steps": 830, "total_steps": 1505, "loss": 0.007, "learning_rate": 2.0973209220146135e-05, "epoch": 19.246376811594203, "percentage": 55.15, "elapsed_time": "5:43:12", "remaining_time": "4:39:06", "throughput": "1000.64", "total_tokens": 20605728} | |
| {"current_steps": 835, "total_steps": 1505, "loss": 0.0088, "learning_rate": 2.0715910631677968e-05, "epoch": 19.36231884057971, "percentage": 55.48, "elapsed_time": "5:45:09", "remaining_time": "4:36:57", "throughput": "1000.77", "total_tokens": 20725799} | |
| {"current_steps": 840, "total_steps": 1505, "loss": 0.007, "learning_rate": 2.0459078725514092e-05, "epoch": 19.47826086956522, "percentage": 55.81, "elapsed_time": "5:47:35", "remaining_time": "4:35:10", "throughput": "1000.49", "total_tokens": 20865534} | |
| {"current_steps": 845, "total_steps": 1505, "loss": 0.0059, "learning_rate": 2.020274147934019e-05, "epoch": 19.594202898550726, "percentage": 56.15, "elapsed_time": "5:49:21", "remaining_time": "4:32:52", "throughput": "1000.78", "total_tokens": 20977913} | |
| {"current_steps": 850, "total_steps": 1505, "loss": 0.0069, "learning_rate": 1.9946926816956743e-05, "epoch": 19.71014492753623, "percentage": 56.48, "elapsed_time": "5:51:21", "remaining_time": "4:30:45", "throughput": "1000.99", "total_tokens": 21102848} | |
| {"current_steps": 855, "total_steps": 1505, "loss": 0.008, "learning_rate": 1.9691662605237166e-05, "epoch": 19.82608695652174, "percentage": 56.81, "elapsed_time": "5:53:40", "remaining_time": "4:28:52", "throughput": "1001.10", "total_tokens": 21243679} | |
| {"current_steps": 860, "total_steps": 1505, "loss": 0.0127, "learning_rate": 1.9436976651092144e-05, "epoch": 19.942028985507246, "percentage": 57.14, "elapsed_time": "5:55:37", "remaining_time": "4:26:43", "throughput": "1001.23", "total_tokens": 21364202} | |
| {"current_steps": 865, "total_steps": 1505, "loss": 0.0059, "learning_rate": 1.9182896698440584e-05, "epoch": 20.057971014492754, "percentage": 57.48, "elapsed_time": "5:57:53", "remaining_time": "4:24:47", "throughput": "1001.07", "total_tokens": 21496089} | |
| {"current_steps": 870, "total_steps": 1505, "loss": 0.0046, "learning_rate": 1.89294504251873e-05, "epoch": 20.17391304347826, "percentage": 57.81, "elapsed_time": "5:59:40", "remaining_time": "4:22:30", "throughput": "1001.07", "total_tokens": 21603193} | |
| {"current_steps": 875, "total_steps": 1505, "loss": 0.0058, "learning_rate": 1.867666544020798e-05, "epoch": 20.28985507246377, "percentage": 58.14, "elapsed_time": "6:01:57", "remaining_time": "4:20:36", "throughput": "1001.15", "total_tokens": 21742062} | |
| {"current_steps": 880, "total_steps": 1505, "loss": 0.0082, "learning_rate": 1.8424569280341653e-05, "epoch": 20.405797101449274, "percentage": 58.47, "elapsed_time": "6:04:10", "remaining_time": "4:18:38", "throughput": "1000.86", "total_tokens": 21869307} | |
| {"current_steps": 885, "total_steps": 1505, "loss": 0.0148, "learning_rate": 1.817318940739098e-05, "epoch": 20.52173913043478, "percentage": 58.8, "elapsed_time": "6:06:08", "remaining_time": "4:16:30", "throughput": "1001.09", "total_tokens": 21992573} | |
| {"current_steps": 890, "total_steps": 1505, "loss": 0.0064, "learning_rate": 1.7922553205130707e-05, "epoch": 20.63768115942029, "percentage": 59.14, "elapsed_time": "6:07:50", "remaining_time": "4:14:10", "throughput": "1001.43", "total_tokens": 22101845} | |
| {"current_steps": 895, "total_steps": 1505, "loss": 0.008, "learning_rate": 1.767268797632472e-05, "epoch": 20.753623188405797, "percentage": 59.47, "elapsed_time": "6:09:59", "remaining_time": "4:12:10", "throughput": "1001.37", "total_tokens": 22230253} | |
| {"current_steps": 900, "total_steps": 1505, "loss": 0.0053, "learning_rate": 1.7423620939751788e-05, "epoch": 20.869565217391305, "percentage": 59.8, "elapsed_time": "6:12:22", "remaining_time": "4:10:19", "throughput": "1001.39", "total_tokens": 22373454} | |
| {"current_steps": 905, "total_steps": 1505, "loss": 0.0054, "learning_rate": 1.7175379227240523e-05, "epoch": 20.985507246376812, "percentage": 60.13, "elapsed_time": "6:14:33", "remaining_time": "4:08:19", "throughput": "1000.86", "total_tokens": 22493123} | |
| {"current_steps": 910, "total_steps": 1505, "loss": 0.0044, "learning_rate": 1.692798988071385e-05, "epoch": 21.10144927536232, "percentage": 60.47, "elapsed_time": "6:16:48", "remaining_time": "4:06:22", "throughput": "1000.89", "total_tokens": 22629005} | |
| {"current_steps": 915, "total_steps": 1505, "loss": 0.0043, "learning_rate": 1.6681479849243153e-05, "epoch": 21.217391304347824, "percentage": 60.8, "elapsed_time": "6:18:57", "remaining_time": "4:04:21", "throughput": "1000.65", "total_tokens": 22752358} | |
| {"current_steps": 920, "total_steps": 1505, "loss": 0.0035, "learning_rate": 1.6435875986112685e-05, "epoch": 21.333333333333332, "percentage": 61.13, "elapsed_time": "6:21:01", "remaining_time": "4:02:17", "throughput": "1000.81", "total_tokens": 22880349} | |
| {"current_steps": 925, "total_steps": 1505, "loss": 0.0044, "learning_rate": 1.6191205045894283e-05, "epoch": 21.44927536231884, "percentage": 61.46, "elapsed_time": "6:22:39", "remaining_time": "3:59:56", "throughput": "1001.21", "total_tokens": 22987343} | |
| {"current_steps": 930, "total_steps": 1505, "loss": 0.0178, "learning_rate": 1.594749368153292e-05, "epoch": 21.565217391304348, "percentage": 61.79, "elapsed_time": "6:24:39", "remaining_time": "3:57:49", "throughput": "1001.45", "total_tokens": 23113462} | |
| {"current_steps": 935, "total_steps": 1505, "loss": 0.0089, "learning_rate": 1.570476844144329e-05, "epoch": 21.681159420289855, "percentage": 62.13, "elapsed_time": "6:26:20", "remaining_time": "3:55:31", "throughput": "1001.76", "total_tokens": 23221714} | |
| {"current_steps": 940, "total_steps": 1505, "loss": 0.004, "learning_rate": 1.546305576661776e-05, "epoch": 21.797101449275363, "percentage": 62.46, "elapsed_time": "6:29:03", "remaining_time": "3:53:50", "throughput": "1001.09", "total_tokens": 23368857} | |
| {"current_steps": 945, "total_steps": 1505, "loss": 0.004, "learning_rate": 1.5222381987746104e-05, "epoch": 21.91304347826087, "percentage": 62.79, "elapsed_time": "6:31:12", "remaining_time": "3:51:49", "throughput": "1000.93", "total_tokens": 23494483} | |
| {"current_steps": 950, "total_steps": 1505, "loss": 0.0034, "learning_rate": 1.4982773322347144e-05, "epoch": 22.028985507246375, "percentage": 63.12, "elapsed_time": "6:33:01", "remaining_time": "3:49:36", "throughput": "1001.01", "total_tokens": 23605463} | |
| {"current_steps": 955, "total_steps": 1505, "loss": 0.0066, "learning_rate": 1.4744255871912823e-05, "epoch": 22.144927536231883, "percentage": 63.46, "elapsed_time": "6:34:42", "remaining_time": "3:47:19", "throughput": "1001.40", "total_tokens": 23715776} | |
| {"current_steps": 960, "total_steps": 1505, "loss": 0.0034, "learning_rate": 1.4506855619064846e-05, "epoch": 22.26086956521739, "percentage": 63.79, "elapsed_time": "6:36:44", "remaining_time": "3:45:14", "throughput": "1001.54", "total_tokens": 23841669} | |
| {"current_steps": 965, "total_steps": 1505, "loss": 0.0032, "learning_rate": 1.4270598424724292e-05, "epoch": 22.3768115942029, "percentage": 64.12, "elapsed_time": "6:38:36", "remaining_time": "3:43:03", "throughput": "1001.84", "total_tokens": 23960567} | |
| {"current_steps": 970, "total_steps": 1505, "loss": 0.0124, "learning_rate": 1.4035510025294462e-05, "epoch": 22.492753623188406, "percentage": 64.45, "elapsed_time": "6:40:23", "remaining_time": "3:40:50", "throughput": "1002.12", "total_tokens": 24074628} | |
| {"current_steps": 975, "total_steps": 1505, "loss": 0.0027, "learning_rate": 1.3801616029857378e-05, "epoch": 22.608695652173914, "percentage": 64.78, "elapsed_time": "6:42:53", "remaining_time": "3:39:00", "throughput": "1001.67", "total_tokens": 24214324} | |
| {"current_steps": 980, "total_steps": 1505, "loss": 0.0037, "learning_rate": 1.3568941917384036e-05, "epoch": 22.72463768115942, "percentage": 65.12, "elapsed_time": "6:44:40", "remaining_time": "3:36:47", "throughput": "1001.90", "total_tokens": 24326727} | |
| {"current_steps": 985, "total_steps": 1505, "loss": 0.0029, "learning_rate": 1.3337513033958904e-05, "epoch": 22.840579710144926, "percentage": 65.45, "elapsed_time": "6:46:51", "remaining_time": "3:34:47", "throughput": "1001.88", "total_tokens": 24456961} | |
| {"current_steps": 990, "total_steps": 1505, "loss": 0.0035, "learning_rate": 1.310735459001884e-05, "epoch": 22.956521739130434, "percentage": 65.78, "elapsed_time": "6:49:42", "remaining_time": "3:33:07", "throughput": "1001.00", "total_tokens": 24606652} | |
| {"current_steps": 995, "total_steps": 1505, "loss": 0.002, "learning_rate": 1.2878491657606872e-05, "epoch": 23.07246376811594, "percentage": 66.11, "elapsed_time": "6:51:17", "remaining_time": "3:30:48", "throughput": "1001.32", "total_tokens": 24710410} | |
| {"current_steps": 1000, "total_steps": 1505, "loss": 0.0023, "learning_rate": 1.2650949167640993e-05, "epoch": 23.18840579710145, "percentage": 66.45, "elapsed_time": "6:53:12", "remaining_time": "3:28:40", "throughput": "1001.57", "total_tokens": 24831908} | |
| {"current_steps": 1005, "total_steps": 1505, "loss": 0.0031, "learning_rate": 1.2424751907198312e-05, "epoch": 23.304347826086957, "percentage": 66.78, "elapsed_time": "6:55:34", "remaining_time": "3:26:45", "throughput": "1000.67", "total_tokens": 24951342} | |
| {"current_steps": 1010, "total_steps": 1505, "loss": 0.0027, "learning_rate": 1.2199924516814939e-05, "epoch": 23.420289855072465, "percentage": 67.11, "elapsed_time": "6:57:55", "remaining_time": "3:24:49", "throughput": "1000.50", "total_tokens": 25088309} | |
| {"current_steps": 1015, "total_steps": 1505, "loss": 0.0124, "learning_rate": 1.1976491487801748e-05, "epoch": 23.536231884057973, "percentage": 67.44, "elapsed_time": "6:59:58", "remaining_time": "3:22:44", "throughput": "1000.70", "total_tokens": 25216080} | |
| {"current_steps": 1020, "total_steps": 1505, "loss": 0.0023, "learning_rate": 1.1754477159576499e-05, "epoch": 23.652173913043477, "percentage": 67.77, "elapsed_time": "7:01:40", "remaining_time": "3:20:30", "throughput": "1001.02", "total_tokens": 25326581} | |
| {"current_steps": 1025, "total_steps": 1505, "loss": 0.0027, "learning_rate": 1.1533905717012428e-05, "epoch": 23.768115942028984, "percentage": 68.11, "elapsed_time": "7:04:12", "remaining_time": "3:18:39", "throughput": "1000.96", "total_tokens": 25477500} | |
| {"current_steps": 1030, "total_steps": 1505, "loss": 0.0041, "learning_rate": 1.1314801187803686e-05, "epoch": 23.884057971014492, "percentage": 68.44, "elapsed_time": "7:06:11", "remaining_time": "3:16:32", "throughput": "1001.18", "total_tokens": 25601354} | |
| {"current_steps": 1035, "total_steps": 1505, "loss": 0.0021, "learning_rate": 1.1097187439847939e-05, "epoch": 24.0, "percentage": 68.77, "elapsed_time": "7:08:03", "remaining_time": "3:14:23", "throughput": "1001.39", "total_tokens": 25719240} | |
| {"current_steps": 1040, "total_steps": 1505, "loss": 0.0039, "learning_rate": 1.088108817864629e-05, "epoch": 24.115942028985508, "percentage": 69.1, "elapsed_time": "7:09:52", "remaining_time": "3:12:12", "throughput": "1001.66", "total_tokens": 25834910} | |
| {"current_steps": 1045, "total_steps": 1505, "loss": 0.0025, "learning_rate": 1.0666526944721016e-05, "epoch": 24.231884057971016, "percentage": 69.44, "elapsed_time": "7:12:13", "remaining_time": "3:10:15", "throughput": "1001.60", "total_tokens": 25974530} | |
| {"current_steps": 1050, "total_steps": 1505, "loss": 0.002, "learning_rate": 1.0453527111051184e-05, "epoch": 24.347826086956523, "percentage": 69.77, "elapsed_time": "7:14:17", "remaining_time": "3:08:11", "throughput": "1001.79", "total_tokens": 26104464} | |
| {"current_steps": 1055, "total_steps": 1505, "loss": 0.0024, "learning_rate": 1.0242111880526495e-05, "epoch": 24.463768115942027, "percentage": 70.1, "elapsed_time": "7:16:46", "remaining_time": "3:06:18", "throughput": "1001.71", "total_tokens": 26251334} | |
| {"current_steps": 1060, "total_steps": 1505, "loss": 0.0031, "learning_rate": 1.003230428341979e-05, "epoch": 24.579710144927535, "percentage": 70.43, "elapsed_time": "7:18:35", "remaining_time": "3:04:07", "throughput": "1001.93", "total_tokens": 26366561} | |
| {"current_steps": 1065, "total_steps": 1505, "loss": 0.0022, "learning_rate": 9.824127174878195e-06, "epoch": 24.695652173913043, "percentage": 70.76, "elapsed_time": "7:20:30", "remaining_time": "3:01:59", "throughput": "1002.12", "total_tokens": 26486437} | |
| {"current_steps": 1070, "total_steps": 1505, "loss": 0.0022, "learning_rate": 9.617603232433475e-06, "epoch": 24.81159420289855, "percentage": 71.1, "elapsed_time": "7:22:16", "remaining_time": "2:59:48", "throughput": "1002.46", "total_tokens": 26601526} | |
| {"current_steps": 1075, "total_steps": 1505, "loss": 0.0109, "learning_rate": 9.412754953531663e-06, "epoch": 24.92753623188406, "percentage": 71.43, "elapsed_time": "7:24:45", "remaining_time": "2:57:54", "throughput": "1001.58", "total_tokens": 26727922} | |
| {"current_steps": 1080, "total_steps": 1505, "loss": 0.0019, "learning_rate": 9.209604653082326e-06, "epoch": 25.043478260869566, "percentage": 71.76, "elapsed_time": "7:26:27", "remaining_time": "2:55:41", "throughput": "1001.78", "total_tokens": 26835621} | |
| {"current_steps": 1085, "total_steps": 1505, "loss": 0.0016, "learning_rate": 9.008174461027724e-06, "epoch": 25.159420289855074, "percentage": 72.09, "elapsed_time": "7:28:21", "remaining_time": "2:53:33", "throughput": "1002.01", "total_tokens": 26955101} | |
| {"current_steps": 1090, "total_steps": 1505, "loss": 0.002, "learning_rate": 8.808486319932083e-06, "epoch": 25.27536231884058, "percentage": 72.43, "elapsed_time": "7:30:19", "remaining_time": "2:51:27", "throughput": "1002.15", "total_tokens": 27077833} | |
| {"current_steps": 1095, "total_steps": 1505, "loss": 0.0018, "learning_rate": 8.610561982591357e-06, "epoch": 25.391304347826086, "percentage": 72.76, "elapsed_time": "7:32:08", "remaining_time": "2:49:17", "throughput": "1002.35", "total_tokens": 27192758} | |
| {"current_steps": 1100, "total_steps": 1505, "loss": 0.0028, "learning_rate": 8.414423009663563e-06, "epoch": 25.507246376811594, "percentage": 73.09, "elapsed_time": "7:34:18", "remaining_time": "2:47:16", "throughput": "1002.44", "total_tokens": 27324970} | |
| {"current_steps": 1105, "total_steps": 1505, "loss": 0.0021, "learning_rate": 8.220090767320137e-06, "epoch": 25.6231884057971, "percentage": 73.42, "elapsed_time": "7:38:48", "remaining_time": "2:46:05", "throughput": "998.15", "total_tokens": 27477531} | |
| {"current_steps": 1110, "total_steps": 1505, "loss": 0.0057, "learning_rate": 8.027586424918412e-06, "epoch": 25.73913043478261, "percentage": 73.75, "elapsed_time": "7:40:36", "remaining_time": "2:43:54", "throughput": "998.39", "total_tokens": 27592035} | |
| {"current_steps": 1115, "total_steps": 1505, "loss": 0.0067, "learning_rate": 7.836930952695533e-06, "epoch": 25.855072463768117, "percentage": 74.09, "elapsed_time": "7:42:31", "remaining_time": "2:41:46", "throughput": "998.59", "total_tokens": 27712377} | |
| {"current_steps": 1120, "total_steps": 1505, "loss": 0.002, "learning_rate": 7.648145119484153e-06, "epoch": 25.971014492753625, "percentage": 74.42, "elapsed_time": "7:44:31", "remaining_time": "2:39:40", "throughput": "998.67", "total_tokens": 27834613} | |
| {"current_steps": 1125, "total_steps": 1505, "loss": 0.0021, "learning_rate": 7.461249490449954e-06, "epoch": 26.08695652173913, "percentage": 74.75, "elapsed_time": "7:46:42", "remaining_time": "2:37:38", "throughput": "998.73", "total_tokens": 27966996} | |
| {"current_steps": 1130, "total_steps": 1505, "loss": 0.002, "learning_rate": 7.276264424851423e-06, "epoch": 26.202898550724637, "percentage": 75.08, "elapsed_time": "7:48:44", "remaining_time": "2:35:33", "throughput": "998.92", "total_tokens": 28093538} | |
| {"current_steps": 1135, "total_steps": 1505, "loss": 0.0017, "learning_rate": 7.0932100738220265e-06, "epoch": 26.318840579710145, "percentage": 75.42, "elapsed_time": "7:50:47", "remaining_time": "2:33:28", "throughput": "998.88", "total_tokens": 28215579} | |
| {"current_steps": 1140, "total_steps": 1505, "loss": 0.0014, "learning_rate": 6.912106378175098e-06, "epoch": 26.434782608695652, "percentage": 75.75, "elapsed_time": "7:52:59", "remaining_time": "2:31:26", "throughput": "998.77", "total_tokens": 28344144} | |
| {"current_steps": 1145, "total_steps": 1505, "loss": 0.0022, "learning_rate": 6.732973066231563e-06, "epoch": 26.55072463768116, "percentage": 76.08, "elapsed_time": "7:55:12", "remaining_time": "2:29:24", "throughput": "998.82", "total_tokens": 28478650} | |
| {"current_steps": 1150, "total_steps": 1505, "loss": 0.0023, "learning_rate": 6.555829651670911e-06, "epoch": 26.666666666666668, "percentage": 76.41, "elapsed_time": "7:57:16", "remaining_time": "2:27:19", "throughput": "998.48", "total_tokens": 28593004} | |
| {"current_steps": 1155, "total_steps": 1505, "loss": 0.0028, "learning_rate": 6.380695431405456e-06, "epoch": 26.782608695652176, "percentage": 76.74, "elapsed_time": "7:59:05", "remaining_time": "2:25:10", "throughput": "998.66", "total_tokens": 28707392} | |
| {"current_steps": 1160, "total_steps": 1505, "loss": 0.006, "learning_rate": 6.207589483478266e-06, "epoch": 26.89855072463768, "percentage": 77.08, "elapsed_time": "8:01:12", "remaining_time": "2:23:07", "throughput": "998.69", "total_tokens": 28834902} | |
| {"current_steps": 1165, "total_steps": 1505, "loss": 0.0045, "learning_rate": 6.0365306649849214e-06, "epoch": 27.014492753623188, "percentage": 77.41, "elapsed_time": "8:02:58", "remaining_time": "2:20:57", "throughput": "998.96", "total_tokens": 28948812} | |
| {"current_steps": 1170, "total_steps": 1505, "loss": 0.0019, "learning_rate": 5.867537610019317e-06, "epoch": 27.130434782608695, "percentage": 77.74, "elapsed_time": "8:05:02", "remaining_time": "2:18:52", "throughput": "999.15", "total_tokens": 29078309} | |
| {"current_steps": 1175, "total_steps": 1505, "loss": 0.002, "learning_rate": 5.700628727643806e-06, "epoch": 27.246376811594203, "percentage": 78.07, "elapsed_time": "8:07:18", "remaining_time": "2:16:51", "throughput": "999.08", "total_tokens": 29211503} | |
| {"current_steps": 1180, "total_steps": 1505, "loss": 0.0019, "learning_rate": 5.53582219988382e-06, "epoch": 27.36231884057971, "percentage": 78.41, "elapsed_time": "8:09:40", "remaining_time": "2:14:52", "throughput": "998.78", "total_tokens": 29344489} | |
| {"current_steps": 1185, "total_steps": 1505, "loss": 0.006, "learning_rate": 5.373135979747227e-06, "epoch": 27.47826086956522, "percentage": 78.74, "elapsed_time": "8:12:02", "remaining_time": "2:12:52", "throughput": "998.02", "total_tokens": 29464082} | |
| {"current_steps": 1190, "total_steps": 1505, "loss": 0.0043, "learning_rate": 5.2125877892686496e-06, "epoch": 27.594202898550726, "percentage": 79.07, "elapsed_time": "8:13:55", "remaining_time": "2:10:44", "throughput": "998.17", "total_tokens": 29581124} | |
| {"current_steps": 1195, "total_steps": 1505, "loss": 0.0019, "learning_rate": 5.054195117578914e-06, "epoch": 27.71014492753623, "percentage": 79.4, "elapsed_time": "8:15:43", "remaining_time": "2:08:35", "throughput": "998.41", "total_tokens": 29696346} | |
| {"current_steps": 1200, "total_steps": 1505, "loss": 0.002, "learning_rate": 4.897975218999926e-06, "epoch": 27.82608695652174, "percentage": 79.73, "elapsed_time": "8:17:35", "remaining_time": "2:06:28", "throughput": "998.66", "total_tokens": 29815117} | |
| {"current_steps": 1205, "total_steps": 1505, "loss": 0.0022, "learning_rate": 4.743945111165068e-06, "epoch": 27.942028985507246, "percentage": 80.07, "elapsed_time": "8:19:56", "remaining_time": "2:04:27", "throughput": "998.10", "total_tokens": 29939175} | |
| {"current_steps": 1210, "total_steps": 1505, "loss": 0.0016, "learning_rate": 4.592121573165414e-06, "epoch": 28.057971014492754, "percentage": 80.4, "elapsed_time": "8:22:34", "remaining_time": "2:02:31", "throughput": "997.51", "total_tokens": 30079840} | |
| {"current_steps": 1215, "total_steps": 1505, "loss": 0.0033, "learning_rate": 4.442521143721892e-06, "epoch": 28.17391304347826, "percentage": 80.73, "elapsed_time": "8:24:23", "remaining_time": "2:00:23", "throughput": "997.65", "total_tokens": 30192219} | |
| {"current_steps": 1220, "total_steps": 1505, "loss": 0.0018, "learning_rate": 4.295160119383712e-06, "epoch": 28.28985507246377, "percentage": 81.06, "elapsed_time": "8:26:48", "remaining_time": "1:58:23", "throughput": "997.44", "total_tokens": 30330969} | |
| {"current_steps": 1225, "total_steps": 1505, "loss": 0.0018, "learning_rate": 4.150054552753055e-06, "epoch": 28.405797101449274, "percentage": 81.4, "elapsed_time": "8:28:52", "remaining_time": "1:56:18", "throughput": "997.40", "total_tokens": 30453302} | |
| {"current_steps": 1230, "total_steps": 1505, "loss": 0.0078, "learning_rate": 4.007220250736454e-06, "epoch": 28.52173913043478, "percentage": 81.73, "elapsed_time": "8:30:43", "remaining_time": "1:54:11", "throughput": "997.56", "total_tokens": 30568943} | |
| {"current_steps": 1235, "total_steps": 1505, "loss": 0.0019, "learning_rate": 3.866672772822863e-06, "epoch": 28.63768115942029, "percentage": 82.06, "elapsed_time": "8:32:59", "remaining_time": "1:52:09", "throughput": "997.29", "total_tokens": 30696057} | |
| {"current_steps": 1240, "total_steps": 1505, "loss": 0.0019, "learning_rate": 3.7284274293887115e-06, "epoch": 28.753623188405797, "percentage": 82.39, "elapsed_time": "8:35:23", "remaining_time": "1:50:08", "throughput": "996.51", "total_tokens": 30815506} | |
| {"current_steps": 1245, "total_steps": 1505, "loss": 0.0027, "learning_rate": 3.592499280030057e-06, "epoch": 28.869565217391305, "percentage": 82.72, "elapsed_time": "8:36:55", "remaining_time": "1:47:57", "throughput": "996.81", "total_tokens": 30916446} | |
| {"current_steps": 1250, "total_steps": 1505, "loss": 0.0023, "learning_rate": 3.458903131922134e-06, "epoch": 28.985507246376812, "percentage": 83.06, "elapsed_time": "8:39:08", "remaining_time": "1:45:54", "throughput": "996.97", "total_tokens": 31054242} | |
| {"current_steps": 1255, "total_steps": 1505, "loss": 0.0029, "learning_rate": 3.3276535382063213e-06, "epoch": 29.10144927536232, "percentage": 83.39, "elapsed_time": "8:41:22", "remaining_time": "1:43:51", "throughput": "997.00", "total_tokens": 31189078} | |
| {"current_steps": 1260, "total_steps": 1505, "loss": 0.0018, "learning_rate": 3.198764796404807e-06, "epoch": 29.217391304347824, "percentage": 83.72, "elapsed_time": "8:43:21", "remaining_time": "1:41:45", "throughput": "997.14", "total_tokens": 31311374} | |
| {"current_steps": 1265, "total_steps": 1505, "loss": 0.0018, "learning_rate": 3.0722509468631392e-06, "epoch": 29.333333333333332, "percentage": 84.05, "elapsed_time": "8:45:30", "remaining_time": "1:39:42", "throughput": "997.27", "total_tokens": 31444681} | |
| {"current_steps": 1270, "total_steps": 1505, "loss": 0.0018, "learning_rate": 2.948125771220697e-06, "epoch": 29.44927536231884, "percentage": 84.39, "elapsed_time": "8:47:28", "remaining_time": "1:37:36", "throughput": "997.43", "total_tokens": 31567569} | |
| {"current_steps": 1275, "total_steps": 1505, "loss": 0.0019, "learning_rate": 2.8264027909094715e-06, "epoch": 29.565217391304348, "percentage": 84.72, "elapsed_time": "8:49:51", "remaining_time": "1:35:34", "throughput": "997.05", "total_tokens": 31697338} | |
| {"current_steps": 1280, "total_steps": 1505, "loss": 0.0018, "learning_rate": 2.707095265681081e-06, "epoch": 29.681159420289855, "percentage": 85.05, "elapsed_time": "8:51:55", "remaining_time": "1:33:30", "throughput": "997.23", "total_tokens": 31826661} | |
| {"current_steps": 1285, "total_steps": 1505, "loss": 0.0023, "learning_rate": 2.5902161921623454e-06, "epoch": 29.797101449275363, "percentage": 85.38, "elapsed_time": "8:53:53", "remaining_time": "1:31:24", "throughput": "997.24", "total_tokens": 31944680} | |
| {"current_steps": 1290, "total_steps": 1505, "loss": 0.0078, "learning_rate": 2.475778302439524e-06, "epoch": 29.91304347826087, "percentage": 85.71, "elapsed_time": "8:55:51", "remaining_time": "1:29:18", "throughput": "997.37", "total_tokens": 32067106} | |
| {"current_steps": 1295, "total_steps": 1505, "loss": 0.0018, "learning_rate": 2.3637940626713346e-06, "epoch": 30.028985507246375, "percentage": 86.05, "elapsed_time": "8:57:49", "remaining_time": "1:27:12", "throughput": "997.36", "total_tokens": 32184526} | |
| {"current_steps": 1300, "total_steps": 1505, "loss": 0.0017, "learning_rate": 2.254275671731007e-06, "epoch": 30.144927536231883, "percentage": 86.38, "elapsed_time": "8:59:48", "remaining_time": "1:25:07", "throughput": "997.56", "total_tokens": 32309423} | |
| {"current_steps": 1305, "total_steps": 1505, "loss": 0.0071, "learning_rate": 2.14723505987737e-06, "epoch": 30.26086956521739, "percentage": 86.71, "elapsed_time": "9:02:06", "remaining_time": "1:23:04", "throughput": "997.03", "total_tokens": 32429445} | |
| {"current_steps": 1310, "total_steps": 1505, "loss": 0.0016, "learning_rate": 2.0426838874552714e-06, "epoch": 30.3768115942029, "percentage": 87.04, "elapsed_time": "9:03:48", "remaining_time": "1:20:56", "throughput": "997.31", "total_tokens": 32540571} | |
| {"current_steps": 1315, "total_steps": 1505, "loss": 0.0018, "learning_rate": 1.9406335436253724e-06, "epoch": 30.492753623188406, "percentage": 87.38, "elapsed_time": "9:05:50", "remaining_time": "1:18:52", "throughput": "997.40", "total_tokens": 32665528} | |
| {"current_steps": 1320, "total_steps": 1505, "loss": 0.0017, "learning_rate": 1.8410951451234533e-06, "epoch": 30.608695652173914, "percentage": 87.71, "elapsed_time": "9:08:19", "remaining_time": "1:16:50", "throughput": "997.00", "total_tokens": 32800773} | |
| {"current_steps": 1325, "total_steps": 1505, "loss": 0.0017, "learning_rate": 1.7440795350494588e-06, "epoch": 30.72463768115942, "percentage": 88.04, "elapsed_time": "9:10:29", "remaining_time": "1:14:47", "throughput": "996.94", "total_tokens": 32928397} | |
| {"current_steps": 1330, "total_steps": 1505, "loss": 0.0019, "learning_rate": 1.649597281686302e-06, "epoch": 30.840579710144926, "percentage": 88.37, "elapsed_time": "9:12:31", "remaining_time": "1:12:41", "throughput": "997.09", "total_tokens": 33054819} | |
| {"current_steps": 1335, "total_steps": 1505, "loss": 0.0018, "learning_rate": 1.5576586773486195e-06, "epoch": 30.956521739130434, "percentage": 88.7, "elapsed_time": "9:14:42", "remaining_time": "1:10:38", "throughput": "996.94", "total_tokens": 33180616} | |
| {"current_steps": 1340, "total_steps": 1505, "loss": 0.0038, "learning_rate": 1.4682737372615967e-06, "epoch": 31.07246376811594, "percentage": 89.04, "elapsed_time": "9:16:33", "remaining_time": "1:08:31", "throughput": "997.13", "total_tokens": 33298041} | |
| {"current_steps": 1345, "total_steps": 1505, "loss": 0.0052, "learning_rate": 1.3814521984699596e-06, "epoch": 31.18840579710145, "percentage": 89.37, "elapsed_time": "9:18:18", "remaining_time": "1:06:24", "throughput": "997.30", "total_tokens": 33408343} | |
| {"current_steps": 1350, "total_steps": 1505, "loss": 0.0018, "learning_rate": 1.297203518777293e-06, "epoch": 31.304347826086957, "percentage": 89.7, "elapsed_time": "9:20:36", "remaining_time": "1:04:21", "throughput": "997.30", "total_tokens": 33545364} | |
| {"current_steps": 1355, "total_steps": 1505, "loss": 0.0019, "learning_rate": 1.2155368757157643e-06, "epoch": 31.420289855072465, "percentage": 90.03, "elapsed_time": "9:22:15", "remaining_time": "1:02:14", "throughput": "997.55", "total_tokens": 33652900} | |
| {"current_steps": 1360, "total_steps": 1505, "loss": 0.0019, "learning_rate": 1.1364611655463736e-06, "epoch": 31.536231884057973, "percentage": 90.37, "elapsed_time": "9:24:02", "remaining_time": "1:00:08", "throughput": "997.81", "total_tokens": 33768791} | |
| {"current_steps": 1365, "total_steps": 1505, "loss": 0.0017, "learning_rate": 1.0599850022898539e-06, "epoch": 31.652173913043477, "percentage": 90.7, "elapsed_time": "9:26:18", "remaining_time": "0:58:04", "throughput": "997.48", "total_tokens": 33892837} | |
| {"current_steps": 1370, "total_steps": 1505, "loss": 0.0022, "learning_rate": 9.861167167883046e-07, "epoch": 31.768115942028984, "percentage": 91.03, "elapsed_time": "9:28:18", "remaining_time": "0:56:00", "throughput": "997.57", "total_tokens": 34015288} | |
| {"current_steps": 1375, "total_steps": 1505, "loss": 0.0037, "learning_rate": 9.148643557976955e-07, "epoch": 31.884057971014492, "percentage": 91.36, "elapsed_time": "9:30:49", "remaining_time": "0:53:58", "throughput": "997.25", "total_tokens": 34154884} | |
| {"current_steps": 1380, "total_steps": 1505, "loss": 0.0019, "learning_rate": 8.462356811112987e-07, "epoch": 32.0, "percentage": 91.69, "elapsed_time": "9:33:07", "remaining_time": "0:51:54", "throughput": "997.23", "total_tokens": 34292320} | |
| {"current_steps": 1385, "total_steps": 1505, "loss": 0.0017, "learning_rate": 7.802381687141535e-07, "epoch": 32.11594202898551, "percentage": 92.03, "elapsed_time": "9:35:06", "remaining_time": "0:49:49", "throughput": "997.32", "total_tokens": 34413850} | |
| {"current_steps": 1390, "total_steps": 1505, "loss": 0.0018, "learning_rate": 7.168790079686932e-07, "epoch": 32.231884057971016, "percentage": 92.36, "elapsed_time": "9:37:22", "remaining_time": "0:47:46", "throughput": "997.25", "total_tokens": 34547127} | |
| {"current_steps": 1395, "total_steps": 1505, "loss": 0.0035, "learning_rate": 6.561651008315738e-07, "epoch": 32.34782608695652, "percentage": 92.69, "elapsed_time": "9:39:51", "remaining_time": "0:45:43", "throughput": "996.95", "total_tokens": 34685112} | |
| {"current_steps": 1400, "total_steps": 1505, "loss": 0.0063, "learning_rate": 5.981030611018234e-07, "epoch": 32.46376811594203, "percentage": 93.02, "elapsed_time": "9:41:58", "remaining_time": "0:43:38", "throughput": "996.92", "total_tokens": 34810484} | |
| {"current_steps": 1405, "total_steps": 1505, "loss": 0.0018, "learning_rate": 5.426992137003622e-07, "epoch": 32.57971014492754, "percentage": 93.36, "elapsed_time": "9:43:53", "remaining_time": "0:41:33", "throughput": "996.79", "total_tokens": 34920531} | |
| {"current_steps": 1410, "total_steps": 1505, "loss": 0.002, "learning_rate": 4.899595939810236e-07, "epoch": 32.69565217391305, "percentage": 93.69, "elapsed_time": "9:45:40", "remaining_time": "0:39:27", "throughput": "997.02", "total_tokens": 35035657} | |
| {"current_steps": 1415, "total_steps": 1505, "loss": 0.0017, "learning_rate": 4.398899470730827e-07, "epoch": 32.81159420289855, "percentage": 94.02, "elapsed_time": "9:47:49", "remaining_time": "0:37:23", "throughput": "997.12", "total_tokens": 35167466} | |
| {"current_steps": 1420, "total_steps": 1505, "loss": 0.0016, "learning_rate": 3.9249572725543196e-07, "epoch": 32.927536231884055, "percentage": 94.35, "elapsed_time": "9:49:57", "remaining_time": "0:35:18", "throughput": "997.16", "total_tokens": 35296818} | |
| {"current_steps": 1425, "total_steps": 1505, "loss": 0.0015, "learning_rate": 3.477820973624063e-07, "epoch": 33.04347826086956, "percentage": 94.68, "elapsed_time": "9:52:14", "remaining_time": "0:33:14", "throughput": "997.07", "total_tokens": 35430399} | |
| {"current_steps": 1430, "total_steps": 1505, "loss": 0.0057, "learning_rate": 3.0575392822139726e-07, "epoch": 33.15942028985507, "percentage": 95.02, "elapsed_time": "9:54:09", "remaining_time": "0:31:09", "throughput": "997.25", "total_tokens": 35551540} | |
| {"current_steps": 1435, "total_steps": 1505, "loss": 0.0016, "learning_rate": 2.664157981222437e-07, "epoch": 33.27536231884058, "percentage": 95.35, "elapsed_time": "9:56:30", "remaining_time": "0:29:05", "throughput": "996.81", "total_tokens": 35676077} | |
| {"current_steps": 1440, "total_steps": 1505, "loss": 0.0016, "learning_rate": 2.297719923185032e-07, "epoch": 33.391304347826086, "percentage": 95.68, "elapsed_time": "9:58:09", "remaining_time": "0:27:00", "throughput": "997.10", "total_tokens": 35785127} | |
| {"current_steps": 1445, "total_steps": 1505, "loss": 0.0019, "learning_rate": 1.9582650256064205e-07, "epoch": 33.507246376811594, "percentage": 96.01, "elapsed_time": "10:00:15", "remaining_time": "0:24:55", "throughput": "997.13", "total_tokens": 35911682} | |
| {"current_steps": 1450, "total_steps": 1505, "loss": 0.0017, "learning_rate": 1.645830266611914e-07, "epoch": 33.6231884057971, "percentage": 96.35, "elapsed_time": "10:02:13", "remaining_time": "0:22:50", "throughput": "997.16", "total_tokens": 36030754} | |
| {"current_steps": 1455, "total_steps": 1505, "loss": 0.0042, "learning_rate": 1.3604496809195288e-07, "epoch": 33.73913043478261, "percentage": 96.68, "elapsed_time": "10:04:04", "remaining_time": "0:20:45", "throughput": "997.29", "total_tokens": 36146749} | |
| {"current_steps": 1460, "total_steps": 1505, "loss": 0.0017, "learning_rate": 1.1021543561322012e-07, "epoch": 33.85507246376812, "percentage": 97.01, "elapsed_time": "10:06:10", "remaining_time": "0:18:41", "throughput": "997.47", "total_tokens": 36278454} | |
| {"current_steps": 1465, "total_steps": 1505, "loss": 0.0017, "learning_rate": 8.709724293513854e-08, "epoch": 33.971014492753625, "percentage": 97.34, "elapsed_time": "10:08:27", "remaining_time": "0:16:36", "throughput": "997.29", "total_tokens": 36408834} | |
| {"current_steps": 1470, "total_steps": 1505, "loss": 0.0015, "learning_rate": 6.66929084112089e-08, "epoch": 34.08695652173913, "percentage": 97.67, "elapsed_time": "10:10:58", "remaining_time": "0:14:32", "throughput": "997.05", "total_tokens": 36550538} | |
| {"current_steps": 1475, "total_steps": 1505, "loss": 0.0018, "learning_rate": 4.900465476393168e-08, "epoch": 34.20289855072464, "percentage": 98.01, "elapsed_time": "10:12:25", "remaining_time": "0:12:27", "throughput": "997.33", "total_tokens": 36647436} | |
| {"current_steps": 1480, "total_steps": 1505, "loss": 0.0024, "learning_rate": 3.403440884269526e-08, "epoch": 34.31884057971015, "percentage": 98.34, "elapsed_time": "10:14:46", "remaining_time": "0:10:23", "throughput": "997.25", "total_tokens": 36785387} | |
| {"current_steps": 1485, "total_steps": 1505, "loss": 0.0021, "learning_rate": 2.1783801413866046e-08, "epoch": 34.43478260869565, "percentage": 98.67, "elapsed_time": "10:16:51", "remaining_time": "0:08:18", "throughput": "997.41", "total_tokens": 36915606} | |
| {"current_steps": 1490, "total_steps": 1505, "loss": 0.0035, "learning_rate": 1.2254166983152737e-08, "epoch": 34.55072463768116, "percentage": 99.0, "elapsed_time": "10:18:46", "remaining_time": "0:06:13", "throughput": "997.57", "total_tokens": 37036117} | |
| {"current_steps": 1495, "total_steps": 1505, "loss": 0.0016, "learning_rate": 5.446543650219904e-09, "epoch": 34.666666666666664, "percentage": 99.34, "elapsed_time": "10:20:49", "remaining_time": "0:04:09", "throughput": "997.75", "total_tokens": 37165587} | |
| {"current_steps": 1500, "total_steps": 1505, "loss": 0.0015, "learning_rate": 1.3616729956228425e-09, "epoch": 34.78260869565217, "percentage": 99.67, "elapsed_time": "10:22:54", "remaining_time": "0:02:04", "throughput": "997.77", "total_tokens": 37290827} | |
| {"current_steps": 1505, "total_steps": 1505, "loss": 0.0053, "learning_rate": 0.0, "epoch": 34.89855072463768, "percentage": 100.0, "elapsed_time": "10:25:10", "remaining_time": "0:00:00", "throughput": "997.38", "total_tokens": 37412688} | |
| {"current_steps": 1505, "total_steps": 1505, "epoch": 34.89855072463768, "percentage": 100.0, "elapsed_time": "10:25:10", "remaining_time": "0:00:00", "throughput": "997.38", "total_tokens": 37412688} | |