Thomas B. Brunner
Thomas B. Brunner
## Problem Description The current PPO with LSTM script [ppo_atari_lstm.py](https://github.com/vwxyzjn/cleanrl/blob/master/cleanrl/ppo_atari_lstm.py) uses sequential stepping through the LSTM, i.e. each step of the sequence in processed individually: ```python for h, d in...
## Description Addresses some of the issues documented in https://github.com/pytorch/rl/issues/2441. These are: 1. _Specs are only modified if the key is in `in_keys`. Keys specified in `in_keys_inv` don't lead to...
## Describe the bug I think that there are a couple of bugs in the `RenameTransform` implementation. Let's through them go one-by-one: 1. Specs are only modified if the key...
[BUG] `RenameTransform` of `ParallelEnv` is not the same as `ParallelEnv` of transformed environment
## Describe the bug In short: `transform(ParallelEnv(base_env)) != ParallelEnv(transform(base_env))` I'm aware that this is not supported in [some cases](https://github.com/pytorch/rl/blob/e294c68ca8ac1794b19398b07a1cc42cca586ea1/torchrl/envs/utils.py#L71), but I'd expect that this would work for the `RenameTransform`. This...
## Description These are minor fixes to comments and docstrings that I've been gathering for the past weeks. ## Types of changes What types of changes does your code introduce?...
## Motivation In the current implementation of the VMAS environment, the done fields are only available on the root of the tensordict. However, for training, it is useful to have...
Currently the plugin will log a warning message for each xdist process. These changes prevent this by only logging in the main process. This behavior was introduced in my earlier...
Would it be possible to release a new version of `multi-agent-ale-py` with the latest fixes from the master branch? More specifically, I'm looking into this fix: https://github.com/Farama-Foundation/Multi-Agent-ALE/commit/92d22b8b98c2e46bca7170929e5b873bf0472c3f The latest package...