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": 0.2076, "lr": 3.2720109079037443e-09, "epoch": 18.762376237623762, "percentage": 99.49, "elapsed_time": "5:12:06", "remaining_time": "0:01:36", "throughput": 1810.41, "total_tokens": 33902720}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.0173, "lr": 3.1151063006468193e-09, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:12:08", "remaining_time": "0:01:34", "throughput": 1810.48, "total_tokens": 33907088}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 0.15617910027503967, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:12:33", "remaining_time": "0:01:34", "throughput": 1808.06, "total_tokens": 33907088}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8157 |
{"current_steps": 39795, "total_steps": 40000, "loss": 0.2076, "lr": 3.2720109079037443e-09, "epoch": 18.762376237623762, "percentage": 99.49, "elapsed_time": "5:12:06", "remaining_time": "0:01:36", "throughput": 1810.41, "total_tokens": 33902720}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.0173, "lr": 3.1151063006468193e-09, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:12:08", "remaining_time": "0:01:34", "throughput": 1810.48, "total_tokens": 33907088}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 0.15617910027503967, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:12:33", "remaining_time": "0:01:34", "throughput": 1808.06, "total_tokens": 33907088}
|
| 8160 |
+
{"current_steps": 39805, "total_steps": 40000, "loss": 0.0998, "lr": 2.962056527169854e-09, "epoch": 18.767090994813767, "percentage": 99.51, "elapsed_time": "5:12:36", "remaining_time": "0:01:31", "throughput": 1808.0, "total_tokens": 33911040}
|
| 8161 |
+
{"current_steps": 39810, "total_steps": 40000, "loss": 0.0838, "lr": 2.8128616110761898e-09, "epoch": 18.76944837340877, "percentage": 99.52, "elapsed_time": "5:12:37", "remaining_time": "0:01:29", "throughput": 1808.05, "total_tokens": 33915040}
|
| 8162 |
+
{"current_steps": 39815, "total_steps": 40000, "loss": 0.0695, "lr": 2.6675215753724223e-09, "epoch": 18.77180575200377, "percentage": 99.54, "elapsed_time": "5:12:39", "remaining_time": "0:01:27", "throughput": 1808.11, "total_tokens": 33919200}
|
| 8163 |
+
{"current_steps": 39820, "total_steps": 40000, "loss": 0.0816, "lr": 2.5260364424739557e-09, "epoch": 18.774163130598772, "percentage": 99.55, "elapsed_time": "5:12:41", "remaining_time": "0:01:24", "throughput": 1808.22, "total_tokens": 33925008}
|
| 8164 |
+
{"current_steps": 39825, "total_steps": 40000, "loss": 0.2194, "lr": 2.3884062341994475e-09, "epoch": 18.776520509193777, "percentage": 99.56, "elapsed_time": "5:12:43", "remaining_time": "0:01:22", "throughput": 1808.27, "total_tokens": 33928704}
|
| 8165 |
+
{"current_steps": 39830, "total_steps": 40000, "loss": 0.0637, "lr": 2.25463097177081e-09, "epoch": 18.778877887788777, "percentage": 99.58, "elapsed_time": "5:12:44", "remaining_time": "0:01:20", "throughput": 1808.33, "total_tokens": 33932992}
|
| 8166 |
+
{"current_steps": 39835, "total_steps": 40000, "loss": 0.0777, "lr": 2.1247106758215397e-09, "epoch": 18.78123526638378, "percentage": 99.59, "elapsed_time": "5:12:46", "remaining_time": "0:01:17", "throughput": 1808.4, "total_tokens": 33937392}
|
| 8167 |
+
{"current_steps": 39840, "total_steps": 40000, "loss": 0.0472, "lr": 1.998645366382834e-09, "epoch": 18.783592644978782, "percentage": 99.6, "elapsed_time": "5:12:48", "remaining_time": "0:01:15", "throughput": 1808.47, "total_tokens": 33941984}
|
| 8168 |
+
{"current_steps": 39845, "total_steps": 40000, "loss": 0.2683, "lr": 1.876435062897475e-09, "epoch": 18.785950023573786, "percentage": 99.61, "elapsed_time": "5:12:50", "remaining_time": "0:01:13", "throughput": 1808.55, "total_tokens": 33947040}
|
| 8169 |
+
{"current_steps": 39850, "total_steps": 40000, "loss": 0.0373, "lr": 1.758079784211497e-09, "epoch": 18.788307402168787, "percentage": 99.62, "elapsed_time": "5:12:51", "remaining_time": "0:01:10", "throughput": 1808.61, "total_tokens": 33950864}
|
| 8170 |
+
{"current_steps": 39855, "total_steps": 40000, "loss": 0.1031, "lr": 1.6435795485797434e-09, "epoch": 18.79066478076379, "percentage": 99.64, "elapsed_time": "5:12:53", "remaining_time": "0:01:08", "throughput": 1808.64, "total_tokens": 33954384}
|
| 8171 |
+
{"current_steps": 39860, "total_steps": 40000, "loss": 0.0484, "lr": 1.5329343736547596e-09, "epoch": 18.79302215935879, "percentage": 99.65, "elapsed_time": "5:12:54", "remaining_time": "0:01:05", "throughput": 1808.67, "total_tokens": 33957552}
|
| 8172 |
+
{"current_steps": 39865, "total_steps": 40000, "loss": 0.0182, "lr": 1.4261442765006739e-09, "epoch": 18.795379537953796, "percentage": 99.66, "elapsed_time": "5:12:56", "remaining_time": "0:01:03", "throughput": 1808.74, "total_tokens": 33961888}
|
| 8173 |
+
{"current_steps": 39870, "total_steps": 40000, "loss": 0.1488, "lr": 1.3232092735876445e-09, "epoch": 18.797736916548796, "percentage": 99.67, "elapsed_time": "5:12:58", "remaining_time": "0:01:01", "throughput": 1808.8, "total_tokens": 33966080}
|
| 8174 |
+
{"current_steps": 39875, "total_steps": 40000, "loss": 0.0341, "lr": 1.2241293807918607e-09, "epoch": 18.8000942951438, "percentage": 99.69, "elapsed_time": "5:12:59", "remaining_time": "0:00:58", "throughput": 1808.86, "total_tokens": 33970128}
|
| 8175 |
+
{"current_steps": 39880, "total_steps": 40000, "loss": 0.1442, "lr": 1.128904613387216e-09, "epoch": 18.8024516737388, "percentage": 99.7, "elapsed_time": "5:13:01", "remaining_time": "0:00:56", "throughput": 1808.94, "total_tokens": 33975248}
|
| 8176 |
+
{"current_steps": 39885, "total_steps": 40000, "loss": 0.0576, "lr": 1.0375349860591853e-09, "epoch": 18.804809052333805, "percentage": 99.71, "elapsed_time": "5:13:03", "remaining_time": "0:00:54", "throughput": 1808.99, "total_tokens": 33979104}
|
| 8177 |
+
{"current_steps": 39890, "total_steps": 40000, "loss": 0.1398, "lr": 9.5002051290205e-10, "epoch": 18.807166430928806, "percentage": 99.72, "elapsed_time": "5:13:05", "remaining_time": "0:00:51", "throughput": 1809.05, "total_tokens": 33983312}
|
| 8178 |
+
{"current_steps": 39895, "total_steps": 40000, "loss": 0.1206, "lr": 8.663612074077954e-10, "epoch": 18.80952380952381, "percentage": 99.74, "elapsed_time": "5:13:07", "remaining_time": "0:00:49", "throughput": 1809.13, "total_tokens": 33988256}
|
| 8179 |
+
{"current_steps": 39900, "total_steps": 40000, "loss": 0.2097, "lr": 7.865570824799884e-10, "epoch": 18.81188118811881, "percentage": 99.75, "elapsed_time": "5:13:08", "remaining_time": "0:00:47", "throughput": 1809.21, "total_tokens": 33992976}
|
| 8180 |
+
{"current_steps": 39905, "total_steps": 40000, "loss": 0.0678, "lr": 7.106081504254514e-10, "epoch": 18.814238566713815, "percentage": 99.76, "elapsed_time": "5:13:10", "remaining_time": "0:00:44", "throughput": 1809.26, "total_tokens": 33997024}
|
| 8181 |
+
{"current_steps": 39910, "total_steps": 40000, "loss": 0.0428, "lr": 6.385144229570372e-10, "epoch": 18.816595945308816, "percentage": 99.78, "elapsed_time": "5:13:12", "remaining_time": "0:00:42", "throughput": 1809.32, "total_tokens": 34001200}
|
| 8182 |
+
{"current_steps": 39915, "total_steps": 40000, "loss": 0.0345, "lr": 5.70275911190854e-10, "epoch": 18.81895332390382, "percentage": 99.79, "elapsed_time": "5:13:13", "remaining_time": "0:00:40", "throughput": 1809.39, "total_tokens": 34005616}
|
| 8183 |
+
{"current_steps": 39920, "total_steps": 40000, "loss": 0.2502, "lr": 5.058926256490403e-10, "epoch": 18.82131070249882, "percentage": 99.8, "elapsed_time": "5:13:15", "remaining_time": "0:00:37", "throughput": 1809.44, "total_tokens": 34009504}
|
| 8184 |
+
{"current_steps": 39925, "total_steps": 40000, "loss": 0.0856, "lr": 4.4536457626254134e-10, "epoch": 18.823668081093825, "percentage": 99.81, "elapsed_time": "5:13:17", "remaining_time": "0:00:35", "throughput": 1809.52, "total_tokens": 34014048}
|
| 8185 |
+
{"current_steps": 39930, "total_steps": 40000, "loss": 0.1808, "lr": 3.88691772365557e-10, "epoch": 18.826025459688825, "percentage": 99.83, "elapsed_time": "5:13:19", "remaining_time": "0:00:32", "throughput": 1809.58, "total_tokens": 34018560}
|
| 8186 |
+
{"current_steps": 39935, "total_steps": 40000, "loss": 0.0258, "lr": 3.358742226955425e-10, "epoch": 18.82838283828383, "percentage": 99.84, "elapsed_time": "5:13:20", "remaining_time": "0:00:30", "throughput": 1809.64, "total_tokens": 34022656}
|
| 8187 |
+
{"current_steps": 39940, "total_steps": 40000, "loss": 0.1615, "lr": 2.8691193539875925e-10, "epoch": 18.83074021687883, "percentage": 99.85, "elapsed_time": "5:13:22", "remaining_time": "0:00:28", "throughput": 1809.68, "total_tokens": 34026160}
|
| 8188 |
+
{"current_steps": 39945, "total_steps": 40000, "loss": 0.1471, "lr": 2.418049180274995e-10, "epoch": 18.833097595473834, "percentage": 99.86, "elapsed_time": "5:13:24", "remaining_time": "0:00:25", "throughput": 1809.74, "total_tokens": 34030320}
|
| 8189 |
+
{"current_steps": 39950, "total_steps": 40000, "loss": 0.0507, "lr": 2.005531775373104e-10, "epoch": 18.835454974068835, "percentage": 99.88, "elapsed_time": "5:13:25", "remaining_time": "0:00:23", "throughput": 1809.81, "total_tokens": 34034800}
|
| 8190 |
+
{"current_steps": 39955, "total_steps": 40000, "loss": 0.084, "lr": 1.6315672028699435e-10, "epoch": 18.83781235266384, "percentage": 99.89, "elapsed_time": "5:13:27", "remaining_time": "0:00:21", "throughput": 1809.87, "total_tokens": 34038960}
|
| 8191 |
+
{"current_steps": 39960, "total_steps": 40000, "loss": 0.2022, "lr": 1.2961555204693555e-10, "epoch": 18.84016973125884, "percentage": 99.9, "elapsed_time": "5:13:29", "remaining_time": "0:00:18", "throughput": 1809.94, "total_tokens": 34043456}
|
| 8192 |
+
{"current_steps": 39965, "total_steps": 40000, "loss": 0.0383, "lr": 9.992967798799768e-11, "epoch": 18.842527109853844, "percentage": 99.91, "elapsed_time": "5:13:30", "remaining_time": "0:00:16", "throughput": 1810.01, "total_tokens": 34047776}
|
| 8193 |
+
{"current_steps": 39970, "total_steps": 40000, "loss": 0.1178, "lr": 7.409910268707521e-11, "epoch": 18.844884488448844, "percentage": 99.92, "elapsed_time": "5:13:32", "remaining_time": "0:00:14", "throughput": 1810.08, "total_tokens": 34052224}
|
| 8194 |
+
{"current_steps": 39975, "total_steps": 40000, "loss": 0.1312, "lr": 5.212383012986877e-11, "epoch": 18.84724186704385, "percentage": 99.94, "elapsed_time": "5:13:34", "remaining_time": "0:00:11", "throughput": 1810.17, "total_tokens": 34057568}
|
| 8195 |
+
{"current_steps": 39980, "total_steps": 40000, "loss": 0.0748, "lr": 3.400386370533415e-11, "epoch": 18.84959924563885, "percentage": 99.95, "elapsed_time": "5:13:36", "remaining_time": "0:00:09", "throughput": 1810.26, "total_tokens": 34062400}
|
| 8196 |
+
{"current_steps": 39985, "total_steps": 40000, "loss": 0.1386, "lr": 1.9739206205682258e-11, "epoch": 18.851956624233853, "percentage": 99.96, "elapsed_time": "5:13:37", "remaining_time": "0:00:07", "throughput": 1810.32, "total_tokens": 34066448}
|
| 8197 |
+
{"current_steps": 39990, "total_steps": 40000, "loss": 0.1178, "lr": 9.329859829154685e-12, "epoch": 18.854314002828854, "percentage": 99.98, "elapsed_time": "5:13:39", "remaining_time": "0:00:04", "throughput": 1810.39, "total_tokens": 34071056}
|
| 8198 |
+
{"current_steps": 39995, "total_steps": 40000, "loss": 0.0979, "lr": 2.7758261855748148e-12, "epoch": 18.85667138142386, "percentage": 99.99, "elapsed_time": "5:13:41", "remaining_time": "0:00:02", "throughput": 1810.44, "total_tokens": 34074976}
|
| 8199 |
+
{"current_steps": 40000, "total_steps": 40000, "loss": 0.0138, "lr": 7.710628524559838e-14, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:13:42", "remaining_time": "0:00:00", "throughput": 1810.5, "total_tokens": 34078960}
|
| 8200 |
+
{"current_steps": 40000, "total_steps": 40000, "eval_loss": 0.15617910027503967, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:14:08", "remaining_time": "0:00:00", "throughput": 1808.08, "total_tokens": 34078960}
|
| 8201 |
+
{"current_steps": 40000, "total_steps": 40000, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:14:09", "remaining_time": "0:00:00", "throughput": 1807.97, "total_tokens": 34078960}
|