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

[Testing] Generalize matmul truncation numerical testing #1152

Open

newling wants to merge 1 commit into nod-ai:main from newling:trunci_testing_improvement

Contributor

newling commented Feb 28, 2025

Show the working in code (where does the 60 come from?), useful for future debugging.
Make numerical test more robust: it's better to not do numerical testing with tensors where all the elements have the same value unless there is no alternative -- this opens gaps where bugs can slip through numerical tests unnoticed. So in this PR I use random numpy tensors instead of np.ones


          first commit

672f2e3

Contributor Author

newling commented Feb 28, 2025

There's a problem on ukernel path for trunci. In this PR, the peano test passes numerically but the ukernel test doesn't.

(FYI @Abhishek-Varma @jtuyls )

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

Labels

None yet