GitLab

Because current nle module version is up to 0.8.x,

I got some issue with shape of model like below when I test with test_submission.py.

So, I think we need to set version to 0.7.3 like docker environment you provide.

Traceback (most recent call last):
  File "test_submission.py", line 36, in <module>
    evaluate()
  File "test_submission.py", line 25, in evaluate
    agent = Agent(num_envs, batched_env.num_actions)
  File "/home/jake/workspace/neurips-2021-the-nethack-challenge/agents/torchbeast_agent.py", line 20, in __init__
    self.model = load_model(MODEL_DIR, self.device)
  File "/home/jake/workspace/neurips-2021-the-nethack-challenge/nethack_baselines/torchbeast/models/__init__.py", line 55, in load_model
    model.load_state_dict(checkpoint_states["model_state_dict"])
  File "/home/jake/anaconda3/envs/nle_challenge/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1406, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for BaselineNet:
	size mismatch for policy_logits_mask: copying a param with shape torch.Size([113]) from checkpoint, the shape in current model is torch.Size([121]).
	size mismatch for policy.weight: copying a param with shape torch.Size([113, 256]) from checkpoint, the shape in current model is torch.Size([121, 256]).
	size mismatch for policy.bias: copying a param with shape torch.Size([113]) from checkpoint, the shape in current model is torch.Size([121]).