Jinzhen Lin 1d0c9d6b2d

[Kernel] some optimizations for dense marlin and moe marlin (#16850) ...

Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>

2025-05-05 09:39:30 -07:00

..

marlin_kernels

pre-commit autoupdate (#17380)

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

marlin_moe_wna16

[Kernel] some optimizations for dense marlin and moe marlin (#16850)

2025-05-05 09:39:30 -07:00

permute_unpermute_kernels

permute/unpermute kernel for moe optimization (#14568)

2025-05-02 11:31:55 -07:00

marlin_moe_ops.cu

[Bugfix] Fix support for dimension like integers and ScalarType (#9299)

2024-10-17 19:08:34 +00:00

moe_align_sum_kernels.cu

Optimize moe_align_block_size for deepseek_v3 (#12850)

2025-02-13 18:43:37 -05:00

moe_ops.h

[ROCm][Bugfix] Ensure that the moe_wna16_gemm kernel is not built on ROCm platforms. (#14629)

2025-03-12 08:00:28 -04:00

moe_permute_unpermute_op.cu

permute/unpermute kernel for moe optimization (#14568)

2025-05-02 11:31:55 -07:00

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

[Kernel][Misc] Use TORCH_LIBRARY instead of PYBIND11_MODULE for custom ops (#5047)

2024-06-09 16:23:30 -04:00

torch_bindings.cpp

permute/unpermute kernel for moe optimization (#14568)

2025-05-02 11:31:55 -07:00