Bo Liu

We are still working on it, it is on the road map of [envpool 0.7.0 release](https://github.com/sail-sg/envpool/milestone/3), hopefully :)

Basically class `task` serve as hooks in `MujocoEnv` with following methods: Class `task` - [x] `initialize_episode_mjcf` - [x] `after_compile` - [x] `initialize_episode` - [x] `before_step` - [x] `before_substep` - [x]...

Help needed due to user (developer) requests - pytorch/rl#591 - https://github.com/sail-sg/envpool/issues/189#issuecomment-1230955344 - https://github.com/sail-sg/envpool/issues/189#issuecomment-1277624434

> @zou3519 Do you think it might be possible to use `tree_map` to update parameters? Instead of doing, > > ``` > optim.zero_grad() > loss.backward() > optim.step() > ``` >...

> I'm working on testing some models using `functorch` along with `torch-mlir` and IREE. I don't see an analog of jax's `tree_map`. Is this something it makes sense for `functorch`...

> Following [#25 (comment)](https://github.com/MineDojo/Voyager/issues/25#issuecomment-1567237266), make sure the game mode is set to survival, difficulty to peaceful. This is contrary to what the documentation states if I'm reading it correctly I...

> 请问DeepSeek-VL最多支持多少张图片输入？ 测试过6-7张没出现问题。

> @XuehaiPan thank you for your reply, In my view [torchopt.stop_gradient](https://torchopt.readthedocs.io/en/latest/api/api.html#torchopt.stop_gradient) only detach the link for input tensor, but grad link between inner loop is also connected, like optimizer update...