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

6 warnings

Python format

succeeded Feb 28, 2024 in 5m 4s