DeepSpeed icon indicating copy to clipboard operation
DeepSpeed copied to clipboard

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

Open bmedishe opened this issue 2 years ago • 5 comments

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)

bmedishe avatar Apr 20 '23 19:04 bmedishe

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

mrwyattii avatar Apr 21 '23 22:04 mrwyattii

@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 avatar Apr 23 '23 03:04 bmedishe

I am facing the same issue, any updates?

abhijitpal1247 avatar Apr 25 '23 06:04 abhijitpal1247

Same here. This fix hangs on Nvidia RTX 3090.

DEQDON avatar Apr 25 '23 20:04 DEQDON

@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)

https://github.com/microsoft/DeepSpeed/blob/d92539509b1e9a6178cfdb921d5080e76f690bce/requirements/requirements-sd.txt#L2

mrwyattii avatar Apr 25 '23 21:04 mrwyattii