Yema

Hi, using `pip install git+https://github.com/benelot/pybullet-gym` or` python setup.py install` did not put anything in the `assets` for me. After testing, I found that after modifying the [#36](https://github.com/benelot/pybullet-gym/blob/master/setup.py#L36) of setup.py to...

Hi, thanks for your great work for saving and restoring state in pybullet-gym. It helped me a lot. But through the following test, I found that there is a problem...

Hi, I have the same issue on WSL2 Ubuntu 20.04 with nvim 0.8.1. Through checking the theme config of windows terminal, I found that the color matching of telescope in...

> > Here is a complete self-contained snippet that, for me, consistently errors with `mujoco.FatalError: Offscreen framebuffer is not complete, error 0x8cdd` > > ```python > > from multiprocessing import...

Add an option on `generate` recipe (e.g. `chat_format`), to determine whether to wrap the prompt as chat templates? Just like below: ```python # # You are a friendly chatbot who...

Sure, I would be happy to!

A minimalist script: ```python scorer = AutoScoringJudge() # AutoScoring solution = "$\\therefore$ k = \\boxed{-6}.$" prediction = "$\\therefore$ k = \\boxed{-6}. \\\\\n k = \\boxed{-6}.$" if "\\boxed{" in solution and...

It seems that I underestimated GPT-4o as a powerful model. The repeated output of \boxed is more common on weaker, smaller models. However, the significant performance discrepancy compared to the...

> I have the similar issue, and fix it through delete the inject envionment variables produce by `torchrun`. > > The scenario I use is to instance an vllm model...

hi, may I ask if there is any follow-up for this? I found that turning on the eager mode can run normally, but it will increase inference time