Woosuk KwonandGitHub 9c61864bf8 [DeepSeek] Use torch.mm for bf16xbf16->fp32 gemm (#41300) ...

Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>

2026-04-30 16:28:57 -07:00

..

marlin_moe_wna16

[Kernel] Add MXFP8 to Marlin GEMM/MoE and refactor Mxfp8LinearOp (#34664)

2026-04-01 09:41:42 -07:00

mxfp8_moe

[Kernel] Integrate SM100 MXFP8 blockscaled grouped MM and quant kernels (#34448)

2026-03-01 23:31:19 -08:00

permute_unpermute_kernels

[Compile] Fix compile warning in moe_permute (#36529)

2026-03-16 10:16:14 -04:00

dsv3_router_gemm_bf16_out.cu

[ModelBash][DSV3] Add TRTLLM DSV3 Router GEMM kernel (6% B1 Speedup) (#34302)

2026-02-23 14:02:26 +00:00

dsv3_router_gemm_entry.cu

[Bugfix] Fix DSV3 kernels breaking _C and _moe_C on unsupported arches (#35123)

2026-02-23 17:11:27 -08:00

dsv3_router_gemm_float_out.cu

[ModelBash][DSV3] Add TRTLLM DSV3 Router GEMM kernel (6% B1 Speedup) (#34302)

2026-02-23 14:02:26 +00:00

dsv3_router_gemm_utils.h

[ModelBash][DSV3] Add TRTLLM DSV3 Router GEMM kernel (6% B1 Speedup) (#34302)

2026-02-23 14:02:26 +00:00

dynamic_4bit_int_moe_cpu.cpp

[CPU]Parallelize over tokens in int4 moe (#29600)

2025-12-02 06:21:39 +00:00

grouped_topk_kernels.cu

[Kernel] Optimize grouped topk kernel (#34206)

2026-02-20 01:34:45 -08:00

moe_align_sum_kernels.cu

[Bugfix][CI] fix typos (#34934)

2026-03-05 17:05:46 +00:00

moe_ops.h

[DeepSeek] Use torch.mm for bf16xbf16->fp32 gemm (#41300)

2026-04-30 16:28:57 -07:00

moe_permute_unpermute_op.cu

[Compile] Fix compile warning in moe_permute (#36529)

2026-03-16 10:16:14 -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

moeTopKFuncs.cuh

[Kernel] Optimize grouped topk kernel (#34206)

2026-02-20 01:34:45 -08:00

topk_softmax_kernels.cu

fix: clamp NaN/Inf in topk_softmax to prevent duplicate expert IDs (#39391)

2026-04-21 15:04:41 +04:00

topk_softplus_sqrt_kernels.cu

[Feat] DeepSeek V4 Rebased (#40860)

2026-04-26 18:31:08 -07:00

torch_bindings.cpp

[DeepSeek] Use torch.mm for bf16xbf16->fp32 gemm (#41300)

2026-04-30 16:28:57 -07:00