ElizaWszolaandGitHub 9fb2d22032 [Performance] Performance improvements in non-blockwise fp8 CUTLASS MoE (#20762) ...

Signed-off-by: ElizaWszola <ewszola@redhat.com>

2025-07-17 09:56:44 -04:00

..

marlin_moe_wna16

remove unused variables in marlin_template.h (#20236)

2025-07-02 00:51:52 +00:00

permute_unpermute_kernels

[Refactor] Remove unused variables in moe_permute_unpermute_kernel.inl (#19573)

2025-06-13 06:12:15 -07:00

moe_align_sum_kernels.cu

Fix numel() downcast in vllm/csrc/moe/moe_align_sum_kernels.cu +2 (#17082)

2025-07-01 06:48:10 +00:00

moe_ops.h

[Perf] Optimize moe_align_block_size CUDA kernel (#19572)

2025-06-17 11:49:26 -07:00

moe_permute_unpermute_op.cu

[Performance] Performance improvements in non-blockwise fp8 CUTLASS MoE (#20762)

2025-07-17 09:56:44 -04:00

moe_wna16_utils.h

pre-commit autoupdate (#17380)

2025-04-29 06:46:55 -07:00

moe_wna16.cu

[BugFix] Accuracy fix for llama4 int4 - improperly casted scales (#16801)

2025-04-17 22:13:29 -07:00

topk_softmax_kernels.cu

Fix numel() downcast in vllm/csrc/moe/moe_align_sum_kernels.cu +2 (#17082)

2025-07-01 06:48:10 +00:00

torch_bindings.cpp

[Perf] Optimize moe_align_block_size CUDA kernel (#19572)

2025-06-17 11:49:26 -07:00