Skip to content

Conversation

@charlifu
Copy link

This PR add bf16 output dtype to fp8 gemm.

  • Unify the fp8 gemm ops into single op, use the dtype of result tensor to determine the output type.
  • Change the interface of convert_fp8.

@charlifu charlifu merged commit 904b5b8 into main Jul 29, 2024
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