Abstraction for Offline Goal-Conditioned Reinforcement Learning

This folder contains a lightweight training entrypoint (main.py). All agent implementations are found under ogb_impls/.

Hyperparameters

Full experiment configurations are in ogb_impls/sweep.sh.

Getting Started

ARL requires Python 3.9+. To install the full dependencies, simply run pip install -r requirements.txt

Acknowledgment

These implementations are based on and adapted from OGBench environments and agent implementations: https://github.com/seohongpark/ogbench/tree/master.