Training in progress, step 1800
Browse files- trainer_log.jsonl +15 -0
trainer_log.jsonl
CHANGED
|
@@ -364,3 +364,18 @@
|
|
| 364 |
{"current_steps": 1725, "total_steps": 1800, "loss": 0.3313, "lr": 2.71032038404323e-07, "epoch": 19.166666666666668, "percentage": 95.83, "elapsed_time": "0:04:04", "remaining_time": "0:00:10", "throughput": 2206.28, "total_tokens": 539872}
|
| 365 |
{"current_steps": 1730, "total_steps": 1800, "loss": 0.291, "lr": 2.365975347105448e-07, "epoch": 19.22222222222222, "percentage": 96.11, "elapsed_time": "0:04:05", "remaining_time": "0:00:09", "throughput": 2207.16, "total_tokens": 541472}
|
| 366 |
{"current_steps": 1735, "total_steps": 1800, "loss": 0.3669, "lr": 2.0449121146845774e-07, "epoch": 19.27777777777778, "percentage": 96.39, "elapsed_time": "0:04:06", "remaining_time": "0:00:09", "throughput": 2206.99, "total_tokens": 543008}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 364 |
{"current_steps": 1725, "total_steps": 1800, "loss": 0.3313, "lr": 2.71032038404323e-07, "epoch": 19.166666666666668, "percentage": 95.83, "elapsed_time": "0:04:04", "remaining_time": "0:00:10", "throughput": 2206.28, "total_tokens": 539872}
|
| 365 |
{"current_steps": 1730, "total_steps": 1800, "loss": 0.291, "lr": 2.365975347105448e-07, "epoch": 19.22222222222222, "percentage": 96.11, "elapsed_time": "0:04:05", "remaining_time": "0:00:09", "throughput": 2207.16, "total_tokens": 541472}
|
| 366 |
{"current_steps": 1735, "total_steps": 1800, "loss": 0.3669, "lr": 2.0449121146845774e-07, "epoch": 19.27777777777778, "percentage": 96.39, "elapsed_time": "0:04:06", "remaining_time": "0:00:09", "throughput": 2206.99, "total_tokens": 543008}
|
| 367 |
+
{"current_steps": 1740, "total_steps": 1800, "loss": 0.2722, "lr": 1.747160872177883e-07, "epoch": 19.333333333333332, "percentage": 96.67, "elapsed_time": "0:04:06", "remaining_time": "0:00:08", "throughput": 2207.86, "total_tokens": 544608}
|
| 368 |
+
{"current_steps": 1745, "total_steps": 1800, "loss": 0.4585, "lr": 1.472749613259661e-07, "epoch": 19.38888888888889, "percentage": 96.94, "elapsed_time": "0:04:07", "remaining_time": "0:00:07", "throughput": 2208.36, "total_tokens": 546112}
|
| 369 |
+
{"current_steps": 1750, "total_steps": 1800, "loss": 0.6482, "lr": 1.22170413724923e-07, "epoch": 19.444444444444443, "percentage": 97.22, "elapsed_time": "0:04:07", "remaining_time": "0:00:07", "throughput": 2208.87, "total_tokens": 547616}
|
| 370 |
+
{"current_steps": 1755, "total_steps": 1800, "loss": 0.3946, "lr": 9.940480466855417e-08, "epoch": 19.5, "percentage": 97.5, "elapsed_time": "0:04:08", "remaining_time": "0:00:06", "throughput": 2209.76, "total_tokens": 549216}
|
| 371 |
+
{"current_steps": 1760, "total_steps": 1800, "loss": 0.3057, "lr": 7.898027451078982e-08, "epoch": 19.555555555555557, "percentage": 97.78, "elapsed_time": "0:04:09", "remaining_time": "0:00:05", "throughput": 2210.39, "total_tokens": 550752}
|
| 372 |
+
{"current_steps": 1765, "total_steps": 1800, "loss": 0.3053, "lr": 6.089874350439506e-08, "epoch": 19.61111111111111, "percentage": 98.06, "elapsed_time": "0:04:09", "remaining_time": "0:00:04", "throughput": 2211.28, "total_tokens": 552352}
|
| 373 |
+
{"current_steps": 1770, "total_steps": 1800, "loss": 0.4263, "lr": 4.516191162040051e-08, "epoch": 19.666666666666668, "percentage": 98.33, "elapsed_time": "0:04:10", "remaining_time": "0:00:04", "throughput": 2211.98, "total_tokens": 553920}
|
| 374 |
+
{"current_steps": 1775, "total_steps": 1800, "loss": 0.3279, "lr": 3.177125838830786e-08, "epoch": 19.72222222222222, "percentage": 98.61, "elapsed_time": "0:04:11", "remaining_time": "0:00:03", "throughput": 2212.93, "total_tokens": 555552}
|
| 375 |
+
{"current_steps": 1780, "total_steps": 1800, "loss": 0.2175, "lr": 2.0728042756967824e-08, "epoch": 19.77777777777778, "percentage": 98.89, "elapsed_time": "0:04:11", "remaining_time": "0:00:02", "throughput": 2213.41, "total_tokens": 557056}
|
| 376 |
+
{"current_steps": 1785, "total_steps": 1800, "loss": 0.3091, "lr": 1.2033302976222071e-08, "epoch": 19.833333333333332, "percentage": 99.17, "elapsed_time": "0:04:12", "remaining_time": "0:00:02", "throughput": 2214.37, "total_tokens": 558688}
|
| 377 |
+
{"current_steps": 1790, "total_steps": 1800, "loss": 0.364, "lr": 5.687856499297928e-09, "epoch": 19.88888888888889, "percentage": 99.44, "elapsed_time": "0:04:12", "remaining_time": "0:00:01", "throughput": 2214.95, "total_tokens": 560224}
|
| 378 |
+
{"current_steps": 1795, "total_steps": 1800, "loss": 0.2603, "lr": 1.692299905944883e-09, "epoch": 19.944444444444443, "percentage": 99.72, "elapsed_time": "0:04:13", "remaining_time": "0:00:00", "throughput": 2215.44, "total_tokens": 561728}
|
| 379 |
+
{"current_steps": 1800, "total_steps": 1800, "loss": 0.3311, "lr": 4.700884634611935e-11, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:14", "remaining_time": "0:00:00", "throughput": 2215.96, "total_tokens": 563328}
|
| 380 |
+
{"current_steps": 1800, "total_steps": 1800, "eval_loss": 0.39043864607810974, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:14", "remaining_time": "0:00:00", "throughput": 2211.56, "total_tokens": 563328}
|
| 381 |
+
{"current_steps": 1800, "total_steps": 1800, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:15", "remaining_time": "0:00:00", "throughput": 2204.92, "total_tokens": 563328}
|