search

coax-dev / coax

Modular framework for Reinforcement Learning in python
https://coax.readthedocs.io
MIT License
168 stars 17 forks source link

Frozen Lake example has an invalid gym signature. #31

Closed dbleyl closed 2 years ago

dbleyl commented 2 years ago

Describe the bug

The example for Frozen Lake in the main branch of the docs isn't fully updated for the new version of gym's signature.

ValueError Traceback (most recent call last) in 77 78 a = pi.mode(s) ---> 79 s, r, done, info = env.step(a) 80 81 env.render()

ValueError: too many values to unpack (expected 4)

Expected behavior

Executing the notebook should not result in a ValueError.

To Reproduce

Colab notebook to repro the bug:

- https://colab.research.google.com/...

Runtime used for this colab notebook: ... (e.g. CPU/GPU/TPU)

Any.

Additional context

Simple fix, happy to contribute a pull request.

dbleyl commented 2 years ago

Pull request sent.

KristianHolsheimer commented 2 years ago

Thanks for reporting this!

This was fixed in #28, but we didn't bump the pypi version. I'll do that now.

KristianHolsheimer commented 2 years ago

Apologies, I didn't see that the evaluation episode loop hadn't been updated yet. Will check out #32 now.