Skip to content

Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM. #26701

Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM.

Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM. #26701

Annotations

1 error and 10 warnings

Lint C++

failed Feb 29, 2024 in 23m 10s