PyTorch torch.promote_types Function
Pytorch torch Reference Manual
torch.promote_typesIt is a function in PyTorch used for type promotion. It accepts multiple data types as input and returns the common type (the most suitable type) that can accommodate all input types.
Function Definition
torch.promote_types(type1, type2)
Parameter Description
type1: First data typetype2: Second data type
Usage Example
Example
import torch
# Type promotion example
dtype1 = torch.float32
dtype2 = torch.float64
# Get common type
result_dtype = torch.promote_types(dtype1, dtype2)
print("Common type of float32 and float64:", result_dtype)
# Integer and float types
dtype3 = torch.int32
dtype4 = torch.float32
result_dtype2 = torch.promote_types(dtype3, dtype4)
print("Common type of int32 and float32:", result_dtype2)
# Type promotion example
dtype1 = torch.float32
dtype2 = torch.float64
# Get common type
result_dtype = torch.promote_types(dtype1, dtype2)
print("Common type of float32 and float64:", result_dtype)
# Integer and float types
dtype3 = torch.int32
dtype4 = torch.float32
result_dtype2 = torch.promote_types(dtype3, dtype4)
print("Common type of int32 and float32:", result_dtype2)
The output result is:
float32 和 float64 的公共类型: torch.float64 int32 和 float32 的公共类型: torch.float32
Other Extensions