Skip to content

Second attempt at issue #995#1385

Open
arn97 wants to merge 1 commit intoSciML:masterfrom
arn97:master
Open

Second attempt at issue #995#1385
arn97 wants to merge 1 commit intoSciML:masterfrom
arn97:master

Conversation

@arn97
Copy link

@arn97 arn97 commented Mar 7, 2026

Checklist

  • Appropriate tests were added
  • [x ] Any code changes were done in a way that does not break public API
  • All documentation related to code changes were updated
  • The new code follows the
    contributor guidelines, in particular the SciML Style Guide and
    COLPRAC.
  • Any new documentation only uses public API

Additional context

When p === nothing, tunables becomes SciMLBase.NullParameters().
allocate_vjp(::SciMLBase.NullParameters) previously returned NullParameters(), which caused SavingIntegrandSumAffect(..., ::NullParameters, ::Nothing) to be constructed inside IntegratingSumCallback, producing a MethodError. Returning nothing instead ensures the callback prototype/cache types match and restores correct behavior. This change fixes Zygote.gradient(loss2, nothing) while leaving the existing test suite results unchanged.

@arn97
Copy link
Author

arn97 commented Mar 13, 2026

I checked the failing tests and, from what I can tell, they appear to be preexisting rather than introduced by this PR. I also see similar failures on the current main branch.

Locally, the issue from #995 is fixed by this change: the original reproduction now succeeds, and I also reran Pkg.test() to compare against the baseline behavior.

If helpful, I can try to isolate whether any of the current CI failures are actually specific to this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant