Markus Krimmel comments

Results 18 comments of


                                            Markus Krimmel

Questions regarding mujoban environment

For anyone experiencing a similar issue: If we do ``` task._arena.mjcf_model.size.nconmax = 200 ``` it *seems* to be working. I doubt that this is the right way to do it,...

Questions regarding mujoban environment

Thanks for the response! I have tried several different versions of `dm_control`, `dm_env`, `labmaze` etc., especially those around the publication date of the paper. Currently I am using more recent...

dm_control.rl.control.PhysicsError: Physics state is invalid. Warning(s) raised: mjWARN_CONTACTFULL

Somewhat connected to https://github.com/deepmind/deepmind-research/issues/307

New Gym API compatibility updates

Hey, thanks for the PR! Are there any news on this? Could it make sense to go directly to Gym 0.26?

Conflicting Dependencies

IIRC, I just relaxed the version requirements in `requirements.txt`

Changing `Model.sense` makes (feasible) relaxation infeasible

Thanks for the quick response - I have only tested it with CBC so far - The issue persists in 1.14.0 - If I set `model.sense = mip.MAXIMIZE` before the...

Changing `Model.sense` makes (feasible) relaxation infeasible

I don't know whether the maximization problem is actually unbounded, that's what I want to find out (it might also be optimal, which occurs for some MIPLIB instances). But maximization...

Changing `Model.sense` makes (feasible) relaxation infeasible

I have tested it with Gurobi and maximization should indeed be unbounded.

Environment inverted_pendulum never returns done=True

Some info for debugging: This problem doesn't occur with `legacy_spring=True`. Looking at a rollout, the hinge joint doesn't move at all. Same story for inverted_double_pendulum.

Make Experiment objects picklable

Are there any news on this? :)