Alexis DUBURCQ
Alexis DUBURCQ
Currently, in `SubprocVectorEnv`, there is a single [send](https://github.com/thu-ml/tianshou/blob/2336a7db1b7ed2d27ef09462f4084f5a45daa008/tianshou/env/worker/subproc.py#L189) method to do both `reset` and `step`. So, it relies on the action being [None](https://github.com/thu-ml/tianshou/blob/10d919052bb2a07fce9d1cf3d890a57d93209b96/tianshou/env/worker/subproc.py#L89) to call `reset` instead of `step`. Thus...
I would like to add the following extra term to the loss function,  where  is the action sampled by the distribution, and  for...
Adding '--disable-gpu' flag on windows [here](https://github.com/pyppeteer/pyppeteer/blob/5daef2ae92458c5ebd75a38ffcdc43cfd1de0705/pyppeteer/launcher.py#L392) is no longer necessary, according to this [PR](https://github.com/puppeteer/puppeteer/pull/4523) on puppeter.
Currently, the sensor and controller update periods are set at engine level, while it should be specific to every robots. It would be natural to move the controller as an...
It is possible to speedup simulation by pre-compiling computationally intensive loops. For C++ core, the main bottlenecks are related to the contact solver, namely the decomposition of the mass matrix...
- `panda3d-sync` and `meshcat` backends should run simultaneously to avoid having to rely on dirty hack based on `pyppeteer`. It makes sense since `panda3d-sync` is always available and is computationally...