dan_the_3rd 9b8166e3f0 fMHA: Add backward pass (#844) ...

* fMHA: Add backward pass

* Better checks for strides/alignments

* Remove fb-internal URL

* torch.Tensor.untyped_storage requires pytorch 2.0+

* minor changes

* make test

---------

Co-authored-by: danthe3rd <danthe3rd>
Co-authored-by: Haicheng Wu <haichengw@nvidia.com>

2023-04-06 20:44:58 -04:00

..

library

fMHA: Add backward pass (#844)

2023-04-06 20:44:58 -04:00

profiler

fix split_k_mode and add reduction kernel for f16 input/accum/output (#896)

2023-03-30 15:31:08 -04:00

util

CUTLASS 3.0 Hopper GEMMs are GETTs in disguise (#897)

2023-03-29 10:42:40 -04:00

CMakeLists.txt

New updates for 2.11 (#775)

2023-01-20 16:32:57 -05:00