When I am using torch.add(), I have encountered the following error
inputs = torch.add(inputs, -0.02, noise)
TypeError: add() takes exactly 2 arguments (3 given)
When I am using torch.add(), I have encountered the following error
inputs = torch.add(inputs, -0.02, noise)
TypeError: add() takes exactly 2 arguments (3 given)
The error should be quite self explanatory, you’ve given torch.add
too many arguments.
I have the same problem - and while it seems inded self-explanatory, it contradicts the documentation in: http://pytorch.org/docs/torch.html
Which says:
torch.add(input, value, out=None)
Adds the scalar value to each element of the input input and returns a new resulting tensor.
So - I will be using torch.mul, but the question is in place, and someone has to either implement the correct function or fix the documentation
I’ve tried to reproduce Shiyu’s error, but I cant.
Can one of you give me a repro:
In [1]: import torch
In [2]: a=torch.randn(10)
In [3]: torch.add(a, 0.1, a)
Out[3]:
0.9079
1.1069
-1.5801
-0.3657
-0.6019
-0.5571
0.5797
0.2054
-0.9112
-0.7749
[torch.FloatTensor of size 10]
I am not near pc now, but thing it - it worked when I used python in shell
and did the problem when I ran it from a file with some importa and so on.
Will be able to send details tomorrow.
Now it is working, but doesn’t do anything:
The code I wrote in a file:
import torch
q =torch.randn(10)
print "1 ",q
torch.add(q, 0.1, q)
print "2 ",q
exit()
The output I get:
1
1.6000
-0.5714
-2.1857
0.9774
2.3266
-2.2662
-1.2868
-0.7741
-0.3532
-0.0593
[torch.FloatTensor of size 10]
2
1.6000
-0.5714
-2.1857
0.9774
2.3266
-2.2662
-1.2868
-0.7741
-0.3532
-0.0593
[torch.FloatTensor of size 10]
you did a basic mistake. you should do: q = torch.add(q, 0.1, q)
OK, :embarrassed:, i was in a hurry - looks like it’s working now. Thanks.