Luca Wehrstedt b8ea5b01c9 [fp8 rowwise] Allocate workspace as a PyTorch Tensor (#134110) ... This makes us pass through the CUDA caching allocator which is safer e.g. in case of CUDA graphs. Pull Request resolved: https://github.com/pytorch/pytorch/pull/134110 Approved by: https://github.com/drisspg		2024-08-22 11:07:26 +00:00
..
conda
src	[fp8 rowwise] Allocate workspace as a PyTorch Tensor (#134110)	2024-08-22 11:07:26 +00:00
tools	[BE][CI] Get rid of duplicated code (#131406)	2024-07-23 04:01:13 +00:00
CMakeLists.txt