RuntimeError

[marooncn]

Hi, when I run
python3 trainer.py
There comes an error:

Traceback (most recent call last):
File "trainer.py", line 203, in
R, probs, actions, actions_idxs = model(bat)
File "/home/maroon/.local/lib/python3.5/site-packages/torch/nn/modules/module.py", line 357, in call
result = self.forward(*input, **kwargs)
File "/home/maroon/PycharmProjects/neural-combinatorial-rl-pytorch/neural_combinatorial_rl.py", line 539, in forward
R = self.objective_fn(actions, self.use_cuda)
File "/home/maroon/PycharmProjects/neural-combinatorial-rl-pytorch/sorting_task.py", line 49, in reward
current += res.float()
RuntimeError: invalid argument 1: the number of sizes provided must be greater or equal to the number of dimensions in the tensor at /pytorch/torch/lib/TH/generic/THTensor.c:299

How can I solve it? Thank you.

[maltedeckers]

Hi, I had the same issue, using PyTorch 0.4.0.
Maybe my solution in my fork (https://github.com/mteser/neural-combinatorial-rl-pytorch) works for you, too?

[pemami4911]

This is an issue due to using PyTorch 0.3 (previously, this code was only compatible with PyTorch 0.2). The master branch should now work with both 0.2 and 0.3.