Commits

Commits on Feb 28, 2025

[torch.compile] Fix RMSNorm + quant fusion in the non-cutlass-fp8 case, rename RedundantReshapesPass to NoopEliminationPass (#10902)
ProExpertProg
authored

Commits on Dec 13, 2024

[torch.compile] Dynamic fp8 + rms_norm fusion (#10906)

ProExpertProg
and
Varun Sundar Rabindranath
authored

Commits on Nov 21, 2024

[torch.compile] Inductor code caching fix (#10273)
ProExpertProg
authored

Commits on Nov 15, 2024

[bugfix] Fix static asymmetric quantization case (#10334)

ProExpertProg
and
danieldk
authored

Commits on Nov 8, 2024

Commits on Oct 23, 2024

[Build] Fix `FetchContent` multiple build issue (#9596)
ProExpertProg
authored

Commits on Oct 17, 2024

[torch.compile] Fine-grained CustomOp enabling mechanism (#9300)
ProExpertProg
authored

Commits on Sep 27, 2024

[Kernel] AQ AZP 4/4: Integrate asymmetric quantization to linear method (#7271)
ProExpertProg
authored

Commits on Sep 23, 2024

[Bugfix] Fix CPU CMake build (#8723)

ProExpertProg
and
zhouyuan
authored

Commits on Sep 21, 2024

[Kernel] Build flash-attn from source (#8245)
ProExpertProg
authored

Commits on Sep 16, 2024

[Kernel] AQ AZP 3/4: Asymmetric quantization kernels (#7270)
ProExpertProg
authored

Commits on Aug 22, 2024

[Kernel] Replaced `blockReduce[...]` functions with `cub::BlockReduce` (#7233)

ProExpertProg
and
mgoin
authored

Commits on Aug 8, 2024

[Bugfix][Kernel] Increased atol to fix failing tests (#7305)
ProExpertProg
authored

Commits on Aug 6, 2024

[Kernel] Add per-tensor and per-token AZP epilogues (#5941)

ProExpertProg
and
tlrmchlsmth
authored

Commits on Jun 26, 2024

[Kernel] Adding bias epilogue support for `cutlass_scaled_mm` (#5560)

authored