Beh Chuen Yang

## Describe the bug Attempting to invoke `torch.compile` on any of the abovementioned classes results in similar errors (see below) ## To Reproduce Steps to reproduce the behavior. Please try...

[Bug] With `MultiSyncDataCollector`, `tensors` cannot be instantiated on CUDA in child processes.

2

## Describe the bug Despite applying the appropriate guards (`mp.set_start_method('spawn')`, `if __name__ == "__main__"`), using `MultiSyncDataCollector` with the `cuda` device causes program to freeze. ## To Reproduce ```python # BEFORE...

[BUG] Numerical Instability issues with `torchrl.modules.TanhNormal`

5

## Describe the bug When training on `PettingZoo/MultiWalker-v9` with `Multi-Agent Soft Actor-Critic`, **all** losses (`loss_actor`, `loss_qvalue`, `loss_alpha`) explode after ~1M environment steps at most. This phenomenon occurs regardless of (reasonable)...

Draft for better `write_video` documentation

1

Should address some of the suggestions pointed out in #8509 (and possibly resolve #8569). NOTE: I was unable to get the documentation to build likely due to some other bug...

[Documentation] Improve Clarity on `torchvision.io.write_video` options

1

### 📚 The doc issue ### tl;dr `options` parameter under `torchvision.io.write_video` is inadequately documented, challenging for end-users to tweak videos to their liking. same goes for `audio_options` In [`torchvision.io.write_video`](https://pytorch.org/vision/stable/generated/torchvision.io.write_video.html#torchvision.io.write_video), `options`...