Closed fuzihaofzh closed 4 years ago
This code not support half precision. This is because it directly use float() instead of using type_as
This code not support half precision. This is because it directly use float() instead of using type_as