Grid2op/l2rpn-baselines

About the "DoubleDuelingDQN" baseline

Closed this issue · 2 comments

ydlu commented

Hi, when I try yo train the "DoubleDuelingDQN" baseline using the example provided in the README，I encounter the following error：（in the DoubleDuelingDQN.py file）
AttributeError: 'Environment_rte_case14_realistic' object has no attribute 'reward_helper'

I can't find the method “reward_helper” anywhere.The only method related to it is "_reward_helper".
Could you help me explain this? Thank you very much!

BDonnot commented

Hello,

This is due to the refactorization of grid2op and a better separation of the public / private part.

If you use the grid2op version of the neurips competition:

pip install -U grid2op==1.2.2

Then the problem is solved.

It might also be solved on the "master" branch of this repository (i don't remember if i merged the PR that solves the issue)

ydlu commented

Thanks for your help.I can run it successfully now.

  • 👍1