PyTorch torch.fmax Function
Pytorch torch Reference Manual
torch.fmaxIt is a function in PyTorch for element-wise maximum (ignoring NaN). When the input contains NaN, torch.maximum returns NaN, while torch.fmax ignores NaN and returns non-NaN values.
Function Definition
torch.fmax(input, other)
Usage Example
Example
import torch
# Create a tensor containing NaN
x = torch.tensor([3.0, float('nan'), 4.0])
y = torch.tensor([1.0, 5.0, 2.0])
# Element-wise maximum (ignoring NaN)
result = torch.fmax(x, y)
print(result)
# Create a tensor containing NaN
x = torch.tensor([3.0, float('nan'), 4.0])
y = torch.tensor([1.0, 5.0, 2.0])
# Element-wise maximum (ignoring NaN)
result = torch.fmax(x, y)
print(result)
The output result is:
tensor([3., 5., 4.])
Other Extensions