verl-project/verl
[Bug][CI] FDSP2 test in `model_rmpad` job seems unstable
Aperta
#1388 aperta il 4 mag 2025
bugcall for contributiongood first issue
Metriche repository
- Star
- (23.038 stelle)
- Metriche merge PR
- (Merge medio 3g 6h) (125 PR mergiate in 30 g)
Descrizione
Motivation
https://github.com/volcengine/verl/actions/workflows/model.yml shows that:
- the FDSP2 test in
model_rmpadworkflow fails sometimes; - but can also pass sometimes.
Plan
- Find a setup that can reproduce the error steadily (possibly using the test container)
- Locate the root cause
- Fix the bug
Additional Info.
- Related PR: https://github.com/volcengine/verl/pull/1026
- cc: @lxg2015 @PeterSH6