Hi,

First of all, thank you for opening this library to the public, personally, I find it a very convenient framework.
When running PPO with the lstm option enabled, I get the following error, no matter which env I'm running:
File "/mnt/data/usr/code/TSRL/ppo.py", line 174, in <module> main(args) File "/mnt/data/usr/code/TSRL/ppo.py", line 168, in main trainer.train() File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/trainers/trainers.py", line 363, in train self._post_steps_log_hook(batch) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/trainers/trainers.py", line 340, in _post_steps_log_hook result = op(batch, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context return func(*args, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/trainers/trainers.py", line 849, in __call__ td_record = self.recorder.rollout( File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/envs/common.py", line 451, in rollout tensordict = policy(tensordict) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/td_module/probabilistic.py", line 253, in forward dist, tensordict_out = self.get_dist( File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/td_module/probabilistic.py", line 221, in get_dist tensordict_out = self._call_module( File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/td_module/probabilistic.py", line 193, in _call_module return self.module(tensordict, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/td_module/common.py", line 346, in forward tensors = self._call_module(tensors, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/td_module/common.py", line 336, in _call_module out = self.module(*tensors, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/distributions/continuous.py", line 143, in forward net_output = self.operator(*tensors) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/models/models.py", line 967, in forward return self._lstm(input, hidden0_in, hidden1_in) File "/home/usr/anaconda3/envs/torch_rl/lib/python3.9/site-packages/torchrl-0.1-py3.9-linux-x86_64.egg/torchrl/modules/models/models.py", line 910, in _lstm batch, steps = input.shape[:2] ValueError: not enough values to unpack (expected 2, got 1)