[torch.bind_symbolic_shape] Fix verifier for shapeSymbol detection

[meshtag]

The op can be valid with no attached shape symbols if they are not required by the corresponding affine map. Fix the verifier to consider number of arguments for both.

[stellaraccident]

Good catch. Thanks. Do you happen to have a minimal example that produces this behavior. It's not obvious to me and we may want to add a test.

[meshtag]

Thanks for the review.

Do you happen to have a minimal example that produces this behavior. It's not obvious to me and we may want to add a test.

I got this error in graph 2 (as partitioned by dynamo) of nano_llava model (model spec here). At the moment, I do not have a python spec which would directly lead to this (it is not obvious to me as well, tbh).

PS: I would require your help to land this (if we want to merge in this state). Thanks.

[stellaraccident]

Thanks for the review.

Do you happen to have a minimal example that produces this behavior. It's not obvious to me and we may want to add a test.

I got this error in graph 2 (as partitioned by dynamo) of nano_llava model (model spec here). At the moment, I do not have a python spec which would directly lead to this (it is not obvious to me as well, tbh).

PS: I would require your help to land this (if we want to merge in this state). Thanks.

Thanks. It's a reasonable thing to fix -- I just try to keep a mental tally on where things come up.