search

vwxyzjn / cleanrl

High-quality single file implementation of Deep Reinforcement Learning algorithms with research-friendly features (PPO, DQN, C51, DDPG, TD3, SAC, PPG)
http://docs.cleanrl.dev
Other
5.26k stars 602 forks source link

Problem with multi-agent atari #280

Closed Matkicail closed 1 year ago

Matkicail commented 1 year ago

Problem Description

Trying to run the multi-agent atari games and getting a specific error. " AssertionError: info dtype is <class 'list'> while supported dtype is dict. This may be due to usage of other wrappers in the wrong order. "

Checklist

I did not install via poetry since it was giving me this specific error: File "/usr/lib/python3/dist-packages/poetry/repositories/pypi_repository.py", line 11, in from cachecontrol import CacheControl ModuleNotFoundError: No module named 'cachecontrol'

Followed the outline in this doc for the multi-agent running commands : https://docs.cleanrl.dev/rl-algorithms/ppo/#ppo_pettingzoo_ma_ataripy

Current Behavior

When trying to run a manually installed version it gives me the above assertion error. When trying to run the poetry installation it gives me the error in the checklist.

Expected Behavior

Should be running a multi-agent atari game

Steps to Reproduce

  1. Created conda environment in 3.7.1 Python or 3.9 python and attempt to install using poetry
vwxyzjn commented 1 year ago

Thanks for reporting. The error is due to the dependency version (you probably installed gym 0.25.1 but we pinned at 0.23.1). We use poetry to lock versions to specifically avoid this class of issue, and the CI passes

Hmm not sure what happened with poetry installation. Based on a google search it seems to be related to linux distribution. You might have to play with it a bit.

Since this is a poetry installation issue I suggest raising a ticket in the poetry repo if you are unable to resolve this issue. I am going to close this issue now.

Matkicail commented 1 year ago

Okay thank you for the assistance.