[torchlib][tracking] Handle Gemm with a graph rewriting pass #1110

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Closed

justinchuby opened this issue Oct 24, 2023 · 0 comments · Fixed by #1113

Labels

module: torchlib

Collaborator

justinchuby commented Oct 24, 2023

Handle Gemm with a graph rewriting pass instead of hard coding the branching logic

justinchuby added the module: torchlib label

justinchuby mentioned this issue

Use Gemm to implement addmm | fix(torchlib) #1113

Merged

justinchuby closed this as not planned

justinchuby closed this as completed in #1113

justinchuby added a commit that referenced this issue


          Use Gemm to implement addmm | fix(torchlib) (#1113)

c68468e

When I looked at the test coverage for `addmm` (below), I realized mat1
and mat2 are always 2d tensors. So the rank check is redundant. `addmm`
is now fully mapped to `Gemm`, which should completely resolve
#1089

Closes #1110


![image](https://github.com/microsoft/onnxscript/assets/11205048/073347ca-d677-4c87-94fa-e40a13642569)

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