Martin Schuck
Martin Schuck
Yes, I already suspected that any environment that uses Mujoco and allows for solver warm starts might have this issue. By the way, I am not sure about the performance...
Should I update my PR to fix all of them?
I pushed a new fix (https://github.com/Farama-Foundation/Gymnasium-Robotics/pull/208/commits/5e14ea14d1ac46c60aaed7a00edc4f997510e705) that includes both the `mujoco` and `mujoco_py` envs. It passes all tests on my side including the new reproducibility tests, feel free to have...
I could try, but this would leave `RigidTransform` broken. In fact, we would not even have a `RigidTransform` class anymore. If this is supposed to be a branch purely for...
Sure, I can do that if you are fine with the rather confusing intermediate state of that PR. Are you certain this is the way forward?
Made a copy with otherwise 0 changes in #22975
> I don't see anything typed in the Cython code so the resulting compiled code won't make too much of a difference in terms of performance. The very first function...
The new commits follow the steps you outlined in your last message. When reviewing, be sure to review the change between individual commits, otherwise git will merge them into a...
Seems like `xp.take` causes an issue in one 32bit environment. I'd like to avoid manually casting to i32 if possible. Any chance there is something already in place to handle...
> thanks @amacati ! I had a look through the Python RigidTransform class and this is looking pretty good! Would you be able to make some benchmarks again for the...