Tyler Michael Smith d7982daff5 [Bugfix] Fix fused MoE IMA (sans chunking) by using int64 for strides (#34279) ...

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-11 05:15:52 +00:00

..

configs

glm 4.6 fused tuned inference config for B200 (#32958)

2026-02-08 18:55:47 +00:00

oracle

[XPU][7/N] enable xpu fp8 moe (#34202)

2026-02-11 03:33:59 +00:00

router

Fix RoutingMethodType logic (#33919)

2026-02-06 14:03:34 -08:00

runner

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

__init__.py

[XPU][7/N] enable xpu fp8 moe (#34202)

2026-02-11 03:33:59 +00:00

all2all_utils.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

batched_deep_gemm_moe.py

[MoE Refactor] Oracle Select FP8+NVFP4 Kernels In Priority (#32414)

2026-01-21 08:22:33 -05:00

config.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

cpu_fused_moe.py

[Bugfix] Fix _CPU_MOE_ACT AssertionError when vLLM config not set (#32777)

2026-01-23 08:22:37 +00:00

cutlass_moe.py

[Moe Refactor] Make Inplace Flag for FusedMoEModularKernel part of the constructor (#33375)

2026-02-05 18:07:18 +00:00

deep_gemm_moe.py

[Models] Step-3.5-Flash (#33523)

2026-02-02 10:21:18 +08:00

deep_gemm_utils.py

[Perf] Optimize deepgemm experts initialization, 3.9% TTFT improvement (#30494)

2025-12-11 17:28:34 -05:00

deepep_ht_prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

deepep_ll_prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

fallback.py

[MoE Refactor] Oracle Select FP8+NVFP4 Kernels In Priority (#32414)

2026-01-21 08:22:33 -05:00

flashinfer_a2a_prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

flashinfer_cutedsl_moe.py

fix: Add SM120 (RTX Blackwell) support for FlashInfer CUTLASS NVFP4 MoE kernels (#33417)

2026-01-31 14:06:42 -08:00

flashinfer_cutlass_moe.py

[Bugfix] Fix shared expert input for latent MoE in EP+DP (Nemotron-H) (#34087)

2026-02-09 16:44:18 +00:00

flashinfer_trtllm_moe.py

Fix RoutingMethodType logic (#33919)

2026-02-06 14:03:34 -08:00

fused_batched_moe.py

[Bugfix] Enable Triton MoE for FP8 per-tensor dynamic (#33300)

2026-01-29 12:15:17 -08:00

fused_marlin_moe.py

[BugFix] Change support no act and mul for marlin (#34088)

2026-02-08 17:18:22 +00:00

fused_moe_method_base.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

fused_moe_modular_method.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

fused_moe.py

[Bugfix] Fix fused MoE IMA (sans chunking) by using int64 for strides (#34279)

2026-02-11 05:15:52 +00:00

gpt_oss_triton_kernels_moe.py

[Bugfix][ROCm][GPT-OSS] Use old triton_kernels implementation on ROCm if the new API is not available (#34153)

2026-02-09 17:38:54 -06:00

layer.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

modular_kernel.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

moe_align_block_size.py

[Kernel][MoE] optimize moe_align_block_size (#29642)

2025-12-07 01:58:47 -08:00

moe_permute_unpermute.py

[Refactor] Remove align block size logic in moe_permute (#33449)

2026-02-06 10:57:06 -08:00

mori_prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

pplx_prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

prepare_finalize.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

rocm_aiter_fused_moe.py

[MoE Refactor] Integrate Naive Prepare Finalize into MK (#32567)

2026-01-27 01:28:02 +00:00

routed_experts_capturer.py

[Bugfix] Refactor to support DP parallel in R3 (#32306)

2026-01-16 15:13:58 +08:00

shared_fused_moe.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

topk_weight_and_reduce.py

Update Optional[x] -> x | None and Union[x, y] to x | y (#26633)

2025-10-12 09:51:31 -07:00

triton_cutlass_moe.py

[MoE Refactor] Oracle Select FP8+NVFP4 Kernels In Priority (#32414)

2026-01-21 08:22:33 -05:00

triton_deep_gemm_moe.py

[MoE Refactor] Oracle Select FP8+NVFP4 Kernels In Priority (#32414)

2026-01-21 08:22:33 -05:00

trtllm_moe.py

Bump Flashinfer to v0.6.1 (#30993)

2026-01-21 08:49:50 -08:00

unquantized_fused_moe_method.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344)

2026-02-10 19:51:07 -05:00

utils.py

[ROCm][Quantization] GPT_OSS in amd-quark format model loading and emulations (#29008)

2026-02-10 10:08:05 -05:00

xpu_fused_moe.py

[XPU][7/N] enable xpu fp8 moe (#34202)

2026-02-11 03:33:59 +00:00

zero_expert_fused_moe.py

[MoE Refactor][5/N] Isolate zero expert to LongCatFlash (#28891)

2025-12-20 18:22:04 +00:00