Function _

bitsandbytes/backends/cuda/ops.py:79–81

(A: torch.Tensor, B: torch.Tensor)

@register_kernel("bitsandbytes::int8_linear_matmul", "cuda")
def _(A: torch.Tensor, B: torch.Tensor):
    out = torch.empty((*A.shape[:-1], B.shape[0]), device=A.device, dtype=torch.int32)
    return _int8_linear_matmul_impl(A, B, out)


@register_kernel("bitsandbytes::int8_linear_matmul.out", "cuda")