Skip to content
/ SCORE Public

Author's implementation of SCORE in "False Correlation Reduction for Offline Reinforcement Learning"

Notifications You must be signed in to change notification settings

familyld/SCORE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SCORE

A PyTorch implementation for our paper "False Correlation Reduction for Offline Reinforcement Learning" published on IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). Our code is built off of TD3-BC and rlkit.

Link to our paper:

Prerequisites

  • PyTorch 1.4.0 with Python 3.7
  • MuJoCo 2.00 with mujoco-py 2.0.2.13
  • d4rl 1.1 or higher (with v2 datasets)
  • rlkit 0.2.1

Usage

For training SCORE on Envname (e.g. walker2d-medium-v2), run:

python main.py --env_name=Envname --version=VersionName --gpu=0 

The results are collected in ./output/Envname/SCORE(VersionName)/, where

  • debug.log records the log data,
  • params.pkl records the final model parameters,
  • progress.csv records the log data in the .csv format for analysis purpose,
  • variant.json records the hyperparameters.

Bibtex

@article{deng2023score,
  author={Deng, Zhihong and Fu, Zuyue and Wang, Lingxiao and Yang, Zhuoran and Bai, Chenjia and Zhou, Tianyi and Wang, Zhaoran and Jiang, Jing},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={False Correlation Reduction for Offline Reinforcement Learning}, 
  year={2023},
  volume={},
  number={},
  pages={1-12},
  doi={10.1109/TPAMI.2023.3328397}}

About

Author's implementation of SCORE in "False Correlation Reduction for Offline Reinforcement Learning"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages