takuseno / d3rlpy

An offline deep reinforcement learning library

Home Page:https://takuseno.github.io/d3rlpy

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[BUG] The gym version of D4RL and d3rlpy doesn't seem match.

CTID282 opened this issue · comments

Hi,
First, I want to say thank you for making such a wonderful library.

I checked that you solved this problem before. #337
But I am facing the following error.

AttributeError: 'dict' object has no attribute 'env_specs'

Unlike what is shown in #337, it doesn't work in environments like Hopper or Walker2d.

Can you tell me how to fix this? Thank you!

@CTID282 Hi, thank you for the issue. #323 is the actual relevant issue here. The easiest fix is as follows:

# assuming that you're using d3rlpy==2.2.0
$ pip uninstall d4rl gym
$ d3rlpy install d4rl

d3rlpy provides some useful CLI commands. d3rlpy install d4rl will install all necessary dependencies to experiment with D4RL.

I appreciate your quick and accurate response.
I've resolved the issue.
I will now close this issue.
Thank you !