Skip to content
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.

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

tl.dot updated https://github.com/openai/triton/issues/1098 #3325

Closed
wants to merge 3 commits into from

Conversation

bmedishe
Copy link
Contributor

I got this error when trying to run stable_diffusion inference with deepspeed
TypeError: dot() got an unexpected keyword argument 'trans_b'

Issue found on triton with the same error : TypeError: dot() got an unexpected keyword argument 'trans_b'
solution suggested was to use qk += tl.dot(q, tl.trans(k)) , inplace of qk += tl.dot(q, k, trans_b=True)

@mrwyattii
Copy link
Contributor

@bmedishe have you tested this with a stable diffusion model? It causes DeepSpeed to hang for me.

@bmedishe
Copy link
Contributor Author

bmedishe commented Apr 23, 2023

@bmedishe have you tested this with a stable diffusion model? It causes DeepSpeed to hang for me.

@mrwyattii Hi , I see a hang on AMD GPU too, what version of triton should I be using to have working
qk += tl.dot(q, k, trans_b=True)
After this PR Merge triton-mlir branch - Complete rewrite of the backend from scratch qk += tl.dot(q, k, trans_b=True) throws TypeError

@bmedishe bmedishe marked this pull request as draft April 24, 2023 16:59
@abhijitpal1247
Copy link

I am facing the same issue, any updates?

@DEQDON
Copy link

DEQDON commented Apr 25, 2023

Same here. This fix hangs on Nvidia RTX 3090.

@mrwyattii
Copy link
Contributor

@mrwyattii Hi , I see a hang on AMD GPU too, what version of triton should I be using to have working qk += tl.dot(q, k, trans_b=True)

triton==2.0.0.dev20221202

@bmedishe bmedishe closed this Oct 19, 2023
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.

6 participants