Training in progress, step 40000
Browse files- trainer_log.jsonl +42 -0
trainer_log.jsonl
CHANGED
|
@@ -8157,3 +8157,45 @@
|
|
| 8157 |
{"current_steps": 39795, "total_steps": 40000, "loss": 0.1813, "lr": 3.2720109079037443e-09, "epoch": 6.4918835141528675, "percentage": 99.49, "elapsed_time": "7:51:40", "remaining_time": "0:02:25", "throughput": 2705.63, "total_tokens": 76570480}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.1381, "lr": 3.1151063006468193e-09, "epoch": 6.492699241373685, "percentage": 99.5, "elapsed_time": "7:51:42", "remaining_time": "0:02:22", "throughput": 2705.84, "total_tokens": 76581104}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 0.16232000291347504, "epoch": 6.492699241373685, "percentage": 99.5, "elapsed_time": "7:52:53", "remaining_time": "0:02:22", "throughput": 2699.03, "total_tokens": 76581104}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8157 |
{"current_steps": 39795, "total_steps": 40000, "loss": 0.1813, "lr": 3.2720109079037443e-09, "epoch": 6.4918835141528675, "percentage": 99.49, "elapsed_time": "7:51:40", "remaining_time": "0:02:25", "throughput": 2705.63, "total_tokens": 76570480}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.1381, "lr": 3.1151063006468193e-09, "epoch": 6.492699241373685, "percentage": 99.5, "elapsed_time": "7:51:42", "remaining_time": "0:02:22", "throughput": 2705.84, "total_tokens": 76581104}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 0.16232000291347504, "epoch": 6.492699241373685, "percentage": 99.5, "elapsed_time": "7:52:53", "remaining_time": "0:02:22", "throughput": 2699.03, "total_tokens": 76581104}
|
| 8160 |
+
{"current_steps": 39805, "total_steps": 40000, "loss": 0.161, "lr": 2.962056527169854e-09, "epoch": 6.493514968594502, "percentage": 99.51, "elapsed_time": "7:52:56", "remaining_time": "0:02:19", "throughput": 2699.06, "total_tokens": 76590896}
|
| 8161 |
+
{"current_steps": 39810, "total_steps": 40000, "loss": 0.2357, "lr": 2.8128616110761898e-09, "epoch": 6.49433069581532, "percentage": 99.52, "elapsed_time": "7:52:58", "remaining_time": "0:02:15", "throughput": 2699.23, "total_tokens": 76600464}
|
| 8162 |
+
{"current_steps": 39815, "total_steps": 40000, "loss": 0.1595, "lr": 2.6675215753724223e-09, "epoch": 6.495146423036137, "percentage": 99.54, "elapsed_time": "7:53:00", "remaining_time": "0:02:11", "throughput": 2699.43, "total_tokens": 76610864}
|
| 8163 |
+
{"current_steps": 39820, "total_steps": 40000, "loss": 0.2045, "lr": 2.5260364424739557e-09, "epoch": 6.495962150256954, "percentage": 99.55, "elapsed_time": "7:53:02", "remaining_time": "0:02:08", "throughput": 2699.61, "total_tokens": 76620688}
|
| 8164 |
+
{"current_steps": 39825, "total_steps": 40000, "loss": 0.1752, "lr": 2.3884062341994475e-09, "epoch": 6.496777877477771, "percentage": 99.56, "elapsed_time": "7:53:03", "remaining_time": "0:02:04", "throughput": 2699.78, "total_tokens": 76630128}
|
| 8165 |
+
{"current_steps": 39830, "total_steps": 40000, "loss": 0.1455, "lr": 2.25463097177081e-09, "epoch": 6.497593604698589, "percentage": 99.58, "elapsed_time": "7:53:05", "remaining_time": "0:02:01", "throughput": 2699.9, "total_tokens": 76638128}
|
| 8166 |
+
{"current_steps": 39835, "total_steps": 40000, "loss": 0.1419, "lr": 2.1247106758215397e-09, "epoch": 6.498409331919406, "percentage": 99.59, "elapsed_time": "7:53:07", "remaining_time": "0:01:57", "throughput": 2700.01, "total_tokens": 76646032}
|
| 8167 |
+
{"current_steps": 39840, "total_steps": 40000, "loss": 0.1466, "lr": 1.998645366382834e-09, "epoch": 6.499225059140223, "percentage": 99.6, "elapsed_time": "7:53:09", "remaining_time": "0:01:54", "throughput": 2700.23, "total_tokens": 76656768}
|
| 8168 |
+
{"current_steps": 39845, "total_steps": 40000, "loss": 0.1273, "lr": 1.876435062897475e-09, "epoch": 6.5000407863610405, "percentage": 99.61, "elapsed_time": "7:53:10", "remaining_time": "0:01:50", "throughput": 2700.46, "total_tokens": 76667984}
|
| 8169 |
+
{"current_steps": 39850, "total_steps": 40000, "loss": 0.2136, "lr": 1.758079784211497e-09, "epoch": 6.500856513581859, "percentage": 99.62, "elapsed_time": "7:53:12", "remaining_time": "0:01:46", "throughput": 2700.62, "total_tokens": 76677392}
|
| 8170 |
+
{"current_steps": 39855, "total_steps": 40000, "loss": 0.1214, "lr": 1.6435795485797434e-09, "epoch": 6.501672240802676, "percentage": 99.64, "elapsed_time": "7:53:14", "remaining_time": "0:01:43", "throughput": 2700.8, "total_tokens": 76687072}
|
| 8171 |
+
{"current_steps": 39860, "total_steps": 40000, "loss": 0.1411, "lr": 1.5329343736547596e-09, "epoch": 6.502487968023493, "percentage": 99.65, "elapsed_time": "7:53:15", "remaining_time": "0:01:39", "throughput": 2700.98, "total_tokens": 76696976}
|
| 8172 |
+
{"current_steps": 39865, "total_steps": 40000, "loss": 0.0593, "lr": 1.4261442765006739e-09, "epoch": 6.50330369524431, "percentage": 99.66, "elapsed_time": "7:53:17", "remaining_time": "0:01:36", "throughput": 2701.15, "total_tokens": 76706512}
|
| 8173 |
+
{"current_steps": 39870, "total_steps": 40000, "loss": 0.149, "lr": 1.3232092735876445e-09, "epoch": 6.504119422465128, "percentage": 99.67, "elapsed_time": "7:53:19", "remaining_time": "0:01:32", "throughput": 2701.34, "total_tokens": 76716496}
|
| 8174 |
+
{"current_steps": 39875, "total_steps": 40000, "loss": 0.1022, "lr": 1.2241293807918607e-09, "epoch": 6.504935149685945, "percentage": 99.69, "elapsed_time": "7:53:21", "remaining_time": "0:01:29", "throughput": 2701.5, "total_tokens": 76725744}
|
| 8175 |
+
{"current_steps": 39880, "total_steps": 40000, "loss": 0.1798, "lr": 1.128904613387216e-09, "epoch": 6.505750876906762, "percentage": 99.7, "elapsed_time": "7:53:22", "remaining_time": "0:01:25", "throughput": 2701.71, "total_tokens": 76736368}
|
| 8176 |
+
{"current_steps": 39885, "total_steps": 40000, "loss": 0.1883, "lr": 1.0375349860591853e-09, "epoch": 6.506566604127579, "percentage": 99.71, "elapsed_time": "7:53:24", "remaining_time": "0:01:21", "throughput": 2701.92, "total_tokens": 76747024}
|
| 8177 |
+
{"current_steps": 39890, "total_steps": 40000, "loss": 0.1374, "lr": 9.5002051290205e-10, "epoch": 6.507382331348397, "percentage": 99.72, "elapsed_time": "7:53:26", "remaining_time": "0:01:18", "throughput": 2702.03, "total_tokens": 76754944}
|
| 8178 |
+
{"current_steps": 39895, "total_steps": 40000, "loss": 0.2752, "lr": 8.663612074077954e-10, "epoch": 6.5081980585692145, "percentage": 99.74, "elapsed_time": "7:53:28", "remaining_time": "0:01:14", "throughput": 2702.2, "total_tokens": 76764368}
|
| 8179 |
+
{"current_steps": 39900, "total_steps": 40000, "loss": 0.1515, "lr": 7.865570824799884e-10, "epoch": 6.509013785790032, "percentage": 99.75, "elapsed_time": "7:53:29", "remaining_time": "0:01:11", "throughput": 2702.39, "total_tokens": 76774384}
|
| 8180 |
+
{"current_steps": 39905, "total_steps": 40000, "loss": 0.1695, "lr": 7.106081504254514e-10, "epoch": 6.50982951301085, "percentage": 99.76, "elapsed_time": "7:53:31", "remaining_time": "0:01:07", "throughput": 2702.57, "total_tokens": 76784208}
|
| 8181 |
+
{"current_steps": 39910, "total_steps": 40000, "loss": 0.1174, "lr": 6.385144229570372e-10, "epoch": 6.510645240231667, "percentage": 99.78, "elapsed_time": "7:53:33", "remaining_time": "0:01:04", "throughput": 2702.73, "total_tokens": 76793600}
|
| 8182 |
+
{"current_steps": 39915, "total_steps": 40000, "loss": 0.1251, "lr": 5.70275911190854e-10, "epoch": 6.511460967452484, "percentage": 99.79, "elapsed_time": "7:53:35", "remaining_time": "0:01:00", "throughput": 2702.87, "total_tokens": 76802192}
|
| 8183 |
+
{"current_steps": 39920, "total_steps": 40000, "loss": 0.1637, "lr": 5.058926256490403e-10, "epoch": 6.512276694673301, "percentage": 99.8, "elapsed_time": "7:53:36", "remaining_time": "0:00:56", "throughput": 2703.03, "total_tokens": 76811264}
|
| 8184 |
+
{"current_steps": 39925, "total_steps": 40000, "loss": 0.1125, "lr": 4.4536457626254134e-10, "epoch": 6.513092421894118, "percentage": 99.81, "elapsed_time": "7:53:38", "remaining_time": "0:00:53", "throughput": 2703.18, "total_tokens": 76820224}
|
| 8185 |
+
{"current_steps": 39930, "total_steps": 40000, "loss": 0.1956, "lr": 3.88691772365557e-10, "epoch": 6.513908149114936, "percentage": 99.83, "elapsed_time": "7:53:40", "remaining_time": "0:00:49", "throughput": 2703.35, "total_tokens": 76829824}
|
| 8186 |
+
{"current_steps": 39935, "total_steps": 40000, "loss": 0.1771, "lr": 3.358742226955425e-10, "epoch": 6.514723876335753, "percentage": 99.84, "elapsed_time": "7:53:41", "remaining_time": "0:00:46", "throughput": 2703.51, "total_tokens": 76838992}
|
| 8187 |
+
{"current_steps": 39940, "total_steps": 40000, "loss": 0.1913, "lr": 2.8691193539875925e-10, "epoch": 6.5155396035565705, "percentage": 99.85, "elapsed_time": "7:53:43", "remaining_time": "0:00:42", "throughput": 2703.68, "total_tokens": 76848464}
|
| 8188 |
+
{"current_steps": 39945, "total_steps": 40000, "loss": 0.1127, "lr": 2.418049180274995e-10, "epoch": 6.5163553307773885, "percentage": 99.86, "elapsed_time": "7:53:45", "remaining_time": "0:00:39", "throughput": 2703.86, "total_tokens": 76858288}
|
| 8189 |
+
{"current_steps": 39950, "total_steps": 40000, "loss": 0.1249, "lr": 2.005531775373104e-10, "epoch": 6.517171057998206, "percentage": 99.88, "elapsed_time": "7:53:47", "remaining_time": "0:00:35", "throughput": 2704.01, "total_tokens": 76867280}
|
| 8190 |
+
{"current_steps": 39955, "total_steps": 40000, "loss": 0.1537, "lr": 1.6315672028699435e-10, "epoch": 6.517986785219023, "percentage": 99.89, "elapsed_time": "7:53:48", "remaining_time": "0:00:32", "throughput": 2704.19, "total_tokens": 76876992}
|
| 8191 |
+
{"current_steps": 39960, "total_steps": 40000, "loss": 0.1407, "lr": 1.2961555204693555e-10, "epoch": 6.51880251243984, "percentage": 99.9, "elapsed_time": "7:53:50", "remaining_time": "0:00:28", "throughput": 2704.37, "total_tokens": 76887008}
|
| 8192 |
+
{"current_steps": 39965, "total_steps": 40000, "loss": 0.2046, "lr": 9.992967798799768e-11, "epoch": 6.519618239660657, "percentage": 99.91, "elapsed_time": "7:53:52", "remaining_time": "0:00:24", "throughput": 2704.53, "total_tokens": 76896288}
|
| 8193 |
+
{"current_steps": 39970, "total_steps": 40000, "loss": 0.0977, "lr": 7.409910268707521e-11, "epoch": 6.520433966881475, "percentage": 99.92, "elapsed_time": "7:53:54", "remaining_time": "0:00:21", "throughput": 2704.72, "total_tokens": 76906288}
|
| 8194 |
+
{"current_steps": 39975, "total_steps": 40000, "loss": 0.1358, "lr": 5.212383012986877e-11, "epoch": 6.521249694102292, "percentage": 99.94, "elapsed_time": "7:53:55", "remaining_time": "0:00:17", "throughput": 2704.83, "total_tokens": 76914000}
|
| 8195 |
+
{"current_steps": 39980, "total_steps": 40000, "loss": 0.129, "lr": 3.400386370533415e-11, "epoch": 6.522065421323109, "percentage": 99.95, "elapsed_time": "7:53:57", "remaining_time": "0:00:14", "throughput": 2704.98, "total_tokens": 76923104}
|
| 8196 |
+
{"current_steps": 39985, "total_steps": 40000, "loss": 0.1822, "lr": 1.9739206205682258e-11, "epoch": 6.522881148543927, "percentage": 99.96, "elapsed_time": "7:53:59", "remaining_time": "0:00:10", "throughput": 2705.18, "total_tokens": 76933424}
|
| 8197 |
+
{"current_steps": 39990, "total_steps": 40000, "loss": 0.1782, "lr": 9.329859829154685e-12, "epoch": 6.523696875764744, "percentage": 99.98, "elapsed_time": "7:54:01", "remaining_time": "0:00:07", "throughput": 2705.38, "total_tokens": 76943616}
|
| 8198 |
+
{"current_steps": 39995, "total_steps": 40000, "loss": 0.1818, "lr": 2.7758261855748148e-12, "epoch": 6.524512602985562, "percentage": 99.99, "elapsed_time": "7:54:02", "remaining_time": "0:00:03", "throughput": 2705.55, "total_tokens": 76953296}
|
| 8199 |
+
{"current_steps": 40000, "total_steps": 40000, "loss": 0.1757, "lr": 7.710628524559838e-14, "epoch": 6.525328330206379, "percentage": 100.0, "elapsed_time": "7:54:04", "remaining_time": "0:00:00", "throughput": 2705.73, "total_tokens": 76963024}
|
| 8200 |
+
{"current_steps": 40000, "total_steps": 40000, "eval_loss": 0.16232000291347504, "epoch": 6.525328330206379, "percentage": 100.0, "elapsed_time": "7:55:15", "remaining_time": "0:00:00", "throughput": 2698.95, "total_tokens": 76963024}
|
| 8201 |
+
{"current_steps": 40000, "total_steps": 40000, "epoch": 6.525328330206379, "percentage": 100.0, "elapsed_time": "7:55:17", "remaining_time": "0:00:00", "throughput": 2698.77, "total_tokens": 76963024}
|