pub fn dequantize(tensor: CubeTensor, dtype: DType) -> CubeTensorExpand description
Convert the tensor back to a higher precision data type.
ยงPanics
A storage-tiled tensor: the kernel reads its values as rows.
pub fn dequantize(tensor: CubeTensor, dtype: DType) -> CubeTensorConvert the tensor back to a higher precision data type.
A storage-tiled tensor: the kernel reads its values as rows.