Skip to content

Conversation

@fmassa
Copy link
Contributor

@fmassa fmassa commented Aug 4, 2025

The solver assumes all ops (including placeholder and factory ops) have an input spec. So let's just respect this condition in the code.

Split from #29

The solver assumes all ops (including placeholder and factory ops) have an input spec. So let's just respect this condition in the code
@fmassa fmassa requested a review from wconstab August 4, 2025 11:38
@meta-cla meta-cla bot added the CLA Signed This label is managed by the Meta Open Source bot. label Aug 4, 2025
Copy link
Contributor

@wconstab wconstab left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems ok to me to unblock. It feels in principle cleaner to make the solver aware of the factory ops, but it might not be worth the trouble.

@fmassa
Copy link
Contributor Author

fmassa commented Aug 4, 2025

Yeah, we can try seeing if it would be feasible to handle it in the solver, but I'm not sure as we currently do num_input_placements * num_output_placements options for a given node, and if num_input_placements == 0 everything falls apart.

@fmassa fmassa merged commit d458c8a into main Aug 4, 2025
5 of 6 checks passed
@fmassa fmassa deleted the fmassa/fix_factory_ops branch August 4, 2025 14:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Meta Open Source bot.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants