Skander Moalla
Skander Moalla
## Describe the bug When the batched env device is `cuda` the step count on the batched env seems completely off from what it should be. When the batches env...
## Describe the bug When you get a tensordict rollout of shape `(N_envs, N_steps, C, H, W)` out of a collector and you want to apply an advantage module that...
## To Reproduce ```python from torchrl.envs import EnvCreator, ParallelEnv from torchrl.envs.libs.gym import GymEnv def run(from_pixels): env = ParallelEnv( 2, EnvCreator(lambda: GymEnv("CartPole-v1", from_pixels=from_pixels)) ) print(env.reset()) env.close() if __name__ == "__main__": #...
Aren't operations like this blocking? causing performance issues due to GPU synchronization? https://github.com/pytorch/rl/blob/5b3dd988ad7f0f91a2f61f77a8b3380eae3131fd/torchrl/objectives/ppo.py#L443
## Motivation An advantage of `TensorDict` modules, and in particular the `TensorDictSequential` is that they can write intermediate keys during their forward pass, which can then be reused. Say for...
installing the project with `pip install -e .` goes through a build phase and needs to download and install build dependencies. Sometimes this is not needed if the project is...
RCP will disable root on running containers but the template runs an SSH server with root for some IDEs (PyCharm necessarily, VSCode optionally) to connect to the container. Make the...
## Describe the bug I'm following the `CONTRIBUTING.md` guide and installing the package with `python setup.py develop` yields lots of deprecation warnings: ``` ******************************************************************************** The license_file parameter is deprecated, use...