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": 2.3611, "lr": 3.2720109079037443e-09, "epoch": 18.762376237623762, "percentage": 99.49, "elapsed_time": "5:02:58", "remaining_time": "0:01:33", "throughput": 2030.17, "total_tokens": 36905440}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.4624, "lr": 3.1151063006468193e-09, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:03:00", "remaining_time": "0:01:31", "throughput": 2030.25, "total_tokens": 36910256}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 1.5271672010421753, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:03:24", "remaining_time": "0:01:31", "throughput": 2027.53, "total_tokens": 36910256}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8157 |
{"current_steps": 39795, "total_steps": 40000, "loss": 2.3611, "lr": 3.2720109079037443e-09, "epoch": 18.762376237623762, "percentage": 99.49, "elapsed_time": "5:02:58", "remaining_time": "0:01:33", "throughput": 2030.17, "total_tokens": 36905440}
|
| 8158 |
{"current_steps": 39800, "total_steps": 40000, "loss": 0.4624, "lr": 3.1151063006468193e-09, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:03:00", "remaining_time": "0:01:31", "throughput": 2030.25, "total_tokens": 36910256}
|
| 8159 |
{"current_steps": 39800, "total_steps": 40000, "eval_loss": 1.5271672010421753, "epoch": 18.764733616218766, "percentage": 99.5, "elapsed_time": "5:03:24", "remaining_time": "0:01:31", "throughput": 2027.53, "total_tokens": 36910256}
|
| 8160 |
+
{"current_steps": 39805, "total_steps": 40000, "loss": 0.8698, "lr": 2.962056527169854e-09, "epoch": 18.767090994813767, "percentage": 99.51, "elapsed_time": "5:03:27", "remaining_time": "0:01:29", "throughput": 2027.48, "total_tokens": 36914608}
|
| 8161 |
+
{"current_steps": 39810, "total_steps": 40000, "loss": 0.8919, "lr": 2.8128616110761898e-09, "epoch": 18.76944837340877, "percentage": 99.52, "elapsed_time": "5:03:28", "remaining_time": "0:01:26", "throughput": 2027.54, "total_tokens": 36918928}
|
| 8162 |
+
{"current_steps": 39815, "total_steps": 40000, "loss": 1.6939, "lr": 2.6675215753724223e-09, "epoch": 18.77180575200377, "percentage": 99.54, "elapsed_time": "5:03:30", "remaining_time": "0:01:24", "throughput": 2027.61, "total_tokens": 36923520}
|
| 8163 |
+
{"current_steps": 39820, "total_steps": 40000, "loss": 1.0347, "lr": 2.5260364424739557e-09, "epoch": 18.774163130598772, "percentage": 99.55, "elapsed_time": "5:03:32", "remaining_time": "0:01:22", "throughput": 2027.74, "total_tokens": 36929904}
|
| 8164 |
+
{"current_steps": 39825, "total_steps": 40000, "loss": 2.0484, "lr": 2.3884062341994475e-09, "epoch": 18.776520509193777, "percentage": 99.56, "elapsed_time": "5:03:33", "remaining_time": "0:01:20", "throughput": 2027.79, "total_tokens": 36933840}
|
| 8165 |
+
{"current_steps": 39830, "total_steps": 40000, "loss": 2.8191, "lr": 2.25463097177081e-09, "epoch": 18.778877887788777, "percentage": 99.58, "elapsed_time": "5:03:35", "remaining_time": "0:01:17", "throughput": 2027.86, "total_tokens": 36938416}
|
| 8166 |
+
{"current_steps": 39835, "total_steps": 40000, "loss": 1.0839, "lr": 2.1247106758215397e-09, "epoch": 18.78123526638378, "percentage": 99.59, "elapsed_time": "5:03:37", "remaining_time": "0:01:15", "throughput": 2027.93, "total_tokens": 36943136}
|
| 8167 |
+
{"current_steps": 39840, "total_steps": 40000, "loss": 0.9837, "lr": 1.998645366382834e-09, "epoch": 18.783592644978782, "percentage": 99.6, "elapsed_time": "5:03:38", "remaining_time": "0:01:13", "throughput": 2028.01, "total_tokens": 36948224}
|
| 8168 |
+
{"current_steps": 39845, "total_steps": 40000, "loss": 1.6198, "lr": 1.876435062897475e-09, "epoch": 18.785950023573786, "percentage": 99.61, "elapsed_time": "5:03:40", "remaining_time": "0:01:10", "throughput": 2028.11, "total_tokens": 36953872}
|
| 8169 |
+
{"current_steps": 39850, "total_steps": 40000, "loss": 1.5032, "lr": 1.758079784211497e-09, "epoch": 18.788307402168787, "percentage": 99.62, "elapsed_time": "5:03:42", "remaining_time": "0:01:08", "throughput": 2028.17, "total_tokens": 36958032}
|
| 8170 |
+
{"current_steps": 39855, "total_steps": 40000, "loss": 1.5903, "lr": 1.6435795485797434e-09, "epoch": 18.79066478076379, "percentage": 99.64, "elapsed_time": "5:03:43", "remaining_time": "0:01:06", "throughput": 2028.21, "total_tokens": 36961760}
|
| 8171 |
+
{"current_steps": 39860, "total_steps": 40000, "loss": 0.9671, "lr": 1.5329343736547596e-09, "epoch": 18.79302215935879, "percentage": 99.65, "elapsed_time": "5:03:45", "remaining_time": "0:01:04", "throughput": 2028.24, "total_tokens": 36965136}
|
| 8172 |
+
{"current_steps": 39865, "total_steps": 40000, "loss": 0.9797, "lr": 1.4261442765006739e-09, "epoch": 18.795379537953796, "percentage": 99.66, "elapsed_time": "5:03:46", "remaining_time": "0:01:01", "throughput": 2028.32, "total_tokens": 36969840}
|
| 8173 |
+
{"current_steps": 39870, "total_steps": 40000, "loss": 1.8275, "lr": 1.3232092735876445e-09, "epoch": 18.797736916548796, "percentage": 99.67, "elapsed_time": "5:03:48", "remaining_time": "0:00:59", "throughput": 2028.38, "total_tokens": 36974336}
|
| 8174 |
+
{"current_steps": 39875, "total_steps": 40000, "loss": 2.8684, "lr": 1.2241293807918607e-09, "epoch": 18.8000942951438, "percentage": 99.69, "elapsed_time": "5:03:50", "remaining_time": "0:00:57", "throughput": 2028.44, "total_tokens": 36978752}
|
| 8175 |
+
{"current_steps": 39880, "total_steps": 40000, "loss": 0.8405, "lr": 1.128904613387216e-09, "epoch": 18.8024516737388, "percentage": 99.7, "elapsed_time": "5:03:52", "remaining_time": "0:00:54", "throughput": 2028.53, "total_tokens": 36984416}
|
| 8176 |
+
{"current_steps": 39885, "total_steps": 40000, "loss": 0.7931, "lr": 1.0375349860591853e-09, "epoch": 18.804809052333805, "percentage": 99.71, "elapsed_time": "5:03:53", "remaining_time": "0:00:52", "throughput": 2028.59, "total_tokens": 36988512}
|
| 8177 |
+
{"current_steps": 39890, "total_steps": 40000, "loss": 1.1178, "lr": 9.5002051290205e-10, "epoch": 18.807166430928806, "percentage": 99.72, "elapsed_time": "5:03:55", "remaining_time": "0:00:50", "throughput": 2028.65, "total_tokens": 36993024}
|
| 8178 |
+
{"current_steps": 39895, "total_steps": 40000, "loss": 1.5576, "lr": 8.663612074077954e-10, "epoch": 18.80952380952381, "percentage": 99.74, "elapsed_time": "5:03:57", "remaining_time": "0:00:47", "throughput": 2028.74, "total_tokens": 36998448}
|
| 8179 |
+
{"current_steps": 39900, "total_steps": 40000, "loss": 1.5329, "lr": 7.865570824799884e-10, "epoch": 18.81188118811881, "percentage": 99.75, "elapsed_time": "5:03:58", "remaining_time": "0:00:45", "throughput": 2028.82, "total_tokens": 37003600}
|
| 8180 |
+
{"current_steps": 39905, "total_steps": 40000, "loss": 1.4528, "lr": 7.106081504254514e-10, "epoch": 18.814238566713815, "percentage": 99.76, "elapsed_time": "5:04:00", "remaining_time": "0:00:43", "throughput": 2028.88, "total_tokens": 37008000}
|
| 8181 |
+
{"current_steps": 39910, "total_steps": 40000, "loss": 0.7531, "lr": 6.385144229570372e-10, "epoch": 18.816595945308816, "percentage": 99.78, "elapsed_time": "5:04:02", "remaining_time": "0:00:41", "throughput": 2028.95, "total_tokens": 37012608}
|
| 8182 |
+
{"current_steps": 39915, "total_steps": 40000, "loss": 0.8995, "lr": 5.70275911190854e-10, "epoch": 18.81895332390382, "percentage": 99.79, "elapsed_time": "5:04:03", "remaining_time": "0:00:38", "throughput": 2029.03, "total_tokens": 37017408}
|
| 8183 |
+
{"current_steps": 39920, "total_steps": 40000, "loss": 1.5928, "lr": 5.058926256490403e-10, "epoch": 18.82131070249882, "percentage": 99.8, "elapsed_time": "5:04:05", "remaining_time": "0:00:36", "throughput": 2029.09, "total_tokens": 37021616}
|
| 8184 |
+
{"current_steps": 39925, "total_steps": 40000, "loss": 1.5734, "lr": 4.4536457626254134e-10, "epoch": 18.823668081093825, "percentage": 99.81, "elapsed_time": "5:04:07", "remaining_time": "0:00:34", "throughput": 2029.17, "total_tokens": 37026480}
|
| 8185 |
+
{"current_steps": 39930, "total_steps": 40000, "loss": 0.7303, "lr": 3.88691772365557e-10, "epoch": 18.826025459688825, "percentage": 99.83, "elapsed_time": "5:04:08", "remaining_time": "0:00:31", "throughput": 2029.24, "total_tokens": 37031408}
|
| 8186 |
+
{"current_steps": 39935, "total_steps": 40000, "loss": 0.8245, "lr": 3.358742226955425e-10, "epoch": 18.82838283828383, "percentage": 99.84, "elapsed_time": "5:04:10", "remaining_time": "0:00:29", "throughput": 2029.3, "total_tokens": 37035920}
|
| 8187 |
+
{"current_steps": 39940, "total_steps": 40000, "loss": 1.8481, "lr": 2.8691193539875925e-10, "epoch": 18.83074021687883, "percentage": 99.85, "elapsed_time": "5:04:12", "remaining_time": "0:00:27", "throughput": 2029.34, "total_tokens": 37039696}
|
| 8188 |
+
{"current_steps": 39945, "total_steps": 40000, "loss": 1.2516, "lr": 2.418049180274995e-10, "epoch": 18.833097595473834, "percentage": 99.86, "elapsed_time": "5:04:13", "remaining_time": "0:00:25", "throughput": 2029.41, "total_tokens": 37044208}
|
| 8189 |
+
{"current_steps": 39950, "total_steps": 40000, "loss": 2.5772, "lr": 2.005531775373104e-10, "epoch": 18.835454974068835, "percentage": 99.88, "elapsed_time": "5:04:15", "remaining_time": "0:00:22", "throughput": 2029.48, "total_tokens": 37049168}
|
| 8190 |
+
{"current_steps": 39955, "total_steps": 40000, "loss": 1.0694, "lr": 1.6315672028699435e-10, "epoch": 18.83781235266384, "percentage": 99.89, "elapsed_time": "5:04:17", "remaining_time": "0:00:20", "throughput": 2029.55, "total_tokens": 37053712}
|
| 8191 |
+
{"current_steps": 39960, "total_steps": 40000, "loss": 2.0415, "lr": 1.2961555204693555e-10, "epoch": 18.84016973125884, "percentage": 99.9, "elapsed_time": "5:04:18", "remaining_time": "0:00:18", "throughput": 2029.63, "total_tokens": 37058592}
|
| 8192 |
+
{"current_steps": 39965, "total_steps": 40000, "loss": 1.6679, "lr": 9.992967798799768e-11, "epoch": 18.842527109853844, "percentage": 99.91, "elapsed_time": "5:04:20", "remaining_time": "0:00:15", "throughput": 2029.69, "total_tokens": 37063312}
|
| 8193 |
+
{"current_steps": 39970, "total_steps": 40000, "loss": 1.2771, "lr": 7.409910268707521e-11, "epoch": 18.844884488448844, "percentage": 99.92, "elapsed_time": "5:04:22", "remaining_time": "0:00:13", "throughput": 2029.77, "total_tokens": 37068192}
|
| 8194 |
+
{"current_steps": 39975, "total_steps": 40000, "loss": 0.8793, "lr": 5.212383012986877e-11, "epoch": 18.84724186704385, "percentage": 99.94, "elapsed_time": "5:04:24", "remaining_time": "0:00:11", "throughput": 2029.88, "total_tokens": 37074112}
|
| 8195 |
+
{"current_steps": 39980, "total_steps": 40000, "loss": 0.8617, "lr": 3.400386370533415e-11, "epoch": 18.84959924563885, "percentage": 99.95, "elapsed_time": "5:04:26", "remaining_time": "0:00:09", "throughput": 2029.97, "total_tokens": 37079376}
|
| 8196 |
+
{"current_steps": 39985, "total_steps": 40000, "loss": 1.8308, "lr": 1.9739206205682258e-11, "epoch": 18.851956624233853, "percentage": 99.96, "elapsed_time": "5:04:27", "remaining_time": "0:00:06", "throughput": 2030.03, "total_tokens": 37083760}
|
| 8197 |
+
{"current_steps": 39990, "total_steps": 40000, "loss": 1.0907, "lr": 9.329859829154685e-12, "epoch": 18.854314002828854, "percentage": 99.98, "elapsed_time": "5:04:29", "remaining_time": "0:00:04", "throughput": 2030.1, "total_tokens": 37088688}
|
| 8198 |
+
{"current_steps": 39995, "total_steps": 40000, "loss": 1.2159, "lr": 2.7758261855748148e-12, "epoch": 18.85667138142386, "percentage": 99.99, "elapsed_time": "5:04:30", "remaining_time": "0:00:02", "throughput": 2030.16, "total_tokens": 37092992}
|
| 8199 |
+
{"current_steps": 40000, "total_steps": 40000, "loss": 0.7265, "lr": 7.710628524559838e-14, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:04:32", "remaining_time": "0:00:00", "throughput": 2030.23, "total_tokens": 37097424}
|
| 8200 |
+
{"current_steps": 40000, "total_steps": 40000, "eval_loss": 1.5271672010421753, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:04:57", "remaining_time": "0:00:00", "throughput": 2027.5, "total_tokens": 37097424}
|
| 8201 |
+
{"current_steps": 40000, "total_steps": 40000, "epoch": 18.85902876001886, "percentage": 100.0, "elapsed_time": "5:04:57", "remaining_time": "0:00:00", "throughput": 2027.41, "total_tokens": 37097424}
|