[diffusion] fix: only use fused qk_norm on nvidia gpu (#33994)
This commit is contained in:
@@ -440,6 +440,7 @@ def _apply_qk_norm(
|
||||
and q.stride(-2) == k.stride(-2) == head_dim
|
||||
and q_norm.eps == k_norm.eps
|
||||
and not torch.compiler.is_compiling()
|
||||
and current_platform.is_cuda()
|
||||
):
|
||||
fused_inplace_qknorm(
|
||||
q,
|
||||
|
||||
Reference in New Issue
Block a user