Fail to visualize behaviors on Wipe

[NagisaZj]

I am trying to visualize robot's behavior on Wipe. I use the results provided by the repo, and the command is: python scripts/rollout.py --load_dir ./runs/Wipe-Sawyer-OSC-POSE-SEED83/Wipe_Sawyer_OSC_POSE_SEED83_2020_09_21_23_14_15_0000--s-0/ --horizon 200 --camera frontview

It reports the following error:

Policy loaded Simulating policy... Rollout episode 0 Creating window glfw Traceback (most recent call last): File "scripts/rollout.py", line 107, in use_gpu=args.gpu, File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/util/rlkit_utils.py", line 264, in simulate_policy video_writer=video_writer, File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/util/rlkit_custom.py", line 425, in rollout a, agent_info = agent.get_action(o) File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/rlkit/rlkit/torch/sac/policies.py", line 134, in get_action deterministic=True) File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/rlkit/rlkit/torch/sac/policies.py", line 63, in get_action actions = self.get_actions(obs_np[None], deterministic=deterministic) File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/rlkit/rlkit/torch/sac/policies.py", line 67, in get_actions return eval_np(self, obs_np, deterministic=deterministic)[0] File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/rlkit/rlkit/torch/core.py", line 18, in eval_np outputs = module(*torch_args, **torch_kwargs) File "/home/lthpc/anaconda3/envs/modular/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, **kwargs) File "/home/lthpc/Desktop/Research/meta-role/robosuite-benchmark/rlkit/rlkit/torch/sac/policies.py", line 83, in forward h = self.hidden_activation(fc(h)) File "/home/lthpc/anaconda3/envs/modular/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, **kwargs) File "/home/lthpc/anaconda3/envs/modular/lib/python3.6/site-packages/torch/nn/modules/linear.py", line 92, in forward return F.linear(input, self.weight, self.bias) File "/home/lthpc/anaconda3/envs/modular/lib/python3.6/site-packages/torch/nn/functional.py", line 1406, in linear ret = torch.addmm(bias, input, weight.t()) RuntimeError: size mismatch, m1: [1 x 37], m2: [379 x 256] at /pytorch/aten/src/THC/generic/THCTensorMathBlas.cu:268

The same error occurs for other robot arms and controller modes in Wipe. However, there seems to be no error running other environments like Door and Lift. Any idea for this? Thank you!