search

eth-easl / modyn

Modyn is a research-platform for training ML models on dynamic datasets.
MIT License
22 stars 3 forks source link

Add rho training log to the pipeline file #558

Closed XianzheMa closed 4 days ago

XianzheMa commented 5 days ago

Per the title, this PR actually adds the log of downsampling phase at selector to the selector log, hence further to the pipeline log. The intention here is to see how well the il model is trained with the record_loss_every.

github-actions[bot] commented 5 days ago

Line Coverage: -% ( % to main) Branch Coverage: -% ( % to main)

codecov[bot] commented 4 days ago

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 83.00%. Comparing base (bbd067a) to head (5f768ac).

Additional details and impacted files ```diff @@ Coverage Diff @@ ## main #558 +/- ## ========================================== + Coverage 82.98% 83.00% +0.01% ========================================== Files 226 226 Lines 10487 10499 +12 ========================================== + Hits 8703 8715 +12 Misses 1784 1784 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

XianzheMa commented 4 days ago

The unit test failed last time last night was due to flakiness, not something really wrong :)