biondizzle ef9cd023a9 fix: unpack_ue4m3_u32 — uint32 lacks CUDA bitwise ops, use int32 ...

PyTorch doesn't implement bitwise_and/shift for UInt32 on CUDA.
Cast to int32 first, then extract bytes, then uint8 → view float8.

2026-05-14 13:44:42 +00:00

..

nvfp4_megamoe_kernel

fix: unpack_ue4m3_u32 — uint32 lacks CUDA bitwise ops, use int32

2026-05-14 13:44:42 +00:00

nvfp4_megamoe_kernel.egg-info

Implement TileLang NVFP4 mega_moe L1/L2 kernels

2026-05-13 22:36:58 +00:00