search

eth-easl / modyn

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

Enable finetuning on `RHO-LOSS` selection strategy when `Twin` strategy is used to produce the irreducible loss model #563

Open XianzheMa opened 4 days ago

XianzheMa commented 4 days ago

Per the title

codecov[bot] commented 4 days ago

Codecov Report

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

Project coverage is 82.99%. Comparing base (c8f8235) to head (d2b9395). Report is 1 commits behind head on main.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## main #563 +/- ## ========================================== + Coverage 82.98% 82.99% +0.01% ========================================== Files 226 226 Lines 10506 10509 +3 ========================================== + Hits 8718 8722 +4 + Misses 1788 1787 -1 ```

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

github-actions[bot] commented 4 days ago

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

XianzheMa commented 4 days ago

Did some sanity check by running a small-scale pipeline. Everything is expected. image

XianzheMa commented 2 days ago

Please hold off on review, as I realized an issue not to be dealt with.

XianzheMa commented 2 days ago

Please hold off on review, as I realized an issue not to be dealt with.

fixed