Pytorch autograd failing when running bear/examples/sac.py

[ashishk88]

Hi,

I am using PyTorch version 1.6 to run this script: bear/examples/sac.py. The script fails with the following error:

Traceback (most recent call last):
  File "sac.py", line 111, in <module>
    experiment(variant)
  File "sac.py", line 78, in experiment
    algorithm.train()
  File "/home/ashish/d4rl_evaluations/bear/rlkit/core/rl_algorithm.py", line 46, in train
    self._train()
  File "/home/ashish/d4rl_evaluations/bear/rlkit/core/batch_rl_algorithm.py", line 172, in _train
    self.trainer.train(train_data)
  File "/home/ashish/d4rl_evaluations/bear/rlkit/torch/torch_rl_algorithm.py", line 40, in train
    self.train_from_torch(batch)
  File "/home/ashish/d4rl_evaluations/bear/rlkit/torch/sac/sac.py", line 144, in train_from_torch
    policy_loss.backward()
  File "/home/ashish/ve/py36/lib/python3.6/site-packages/torch/tensor.py", line 185, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph)
  File "/home/ashish/ve/py36/lib/python3.6/site-packages/torch/autograd/__init__.py", line 127, in backward
    allow_unreachable=True)  # allow_unreachable flag
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor [256, 1]], which is output 0 of TBackward, is at version 2; expected version 1 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True).

Does rlkit support pytorch 1.6 or is this a deeper issue?

[ashishk88]

Installing pytorch 1.4.0 solved the issue. Thanks!

[George-Chia]

It does not work for me. Is it related to other packages' versions?