[XPU][MXFP4] add mxfp4 quant op for XPU (#39857)
Signed-off-by: Zhu, Zufang <zufang.zhu@intel.com>
This commit is contained in:
@@ -162,3 +162,7 @@ try:
|
||||
quant_dequant_mxfp4 = torch.ops.vllm.quant_dequant_mxfp4
|
||||
except AttributeError as error:
|
||||
raise error
|
||||
|
||||
|
||||
def xpu_mxfp4_quantize(x: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
|
||||
return torch.ops.vllm.xpu_mxfp4_quantize(x)
|
||||
|
||||
Reference in New Issue
Block a user