FP4 capabilities Blackwell GPUs #6161
Unanswered
Settek-git
asked this question in
Q&A
Replies: 1 comment
-
|
You don't do something like torch.float4_e2m1, you need pack 2 fp4 values into single torch.uint8 https://triton-lang.org/main/python-api/generated/triton.language.dot_scaled.html |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
How can we define FP4 datatype in python code that actually enables the FP4 capabilities of the Blackwell GPUs?
Beta Was this translation helpful? Give feedback.
All reactions