Skip to content

[PyTorch][torch.compile] Support for DotProductAttention on flash and unfused backends - #3286

Open
pggPL wants to merge 47 commits into
NVIDIA:mainfrom
pggPL:dpa_torch_compile
Open

[PyTorch][torch.compile] Support for DotProductAttention on flash and unfused backends#3286
pggPL wants to merge 47 commits into
NVIDIA:mainfrom
pggPL:dpa_torch_compile

Make undeclared non-contiguous q/k/v contiguous under compile

b3e9949
Select commit
Loading
Failed to load commit list.