This repository serves as a curated collection and tracker of research papers exploring the intersection of Large Language Models (LLMs) and Reinforcement Learning (RL), specifically focusing on their combined application in control tasks. It targets researchers and practitioners interested in leveraging LLMs to enhance RL agent capabilities in domains like robotics, game playing, and autonomous systems.

How It Works

The project categorizes papers based on how LLMs are integrated into RL workflows, including direct action generation, indirect guidance, data preference, environment configuration, reward function design, and state representation. This structured approach allows for a clear understanding of the diverse methodologies employed to synergize LLM's reasoning and language understanding with RL's learning and decision-making.

Quick Start & Requirements

This repository is a collection of research papers and does not have a direct installation or execution command. It serves as a reference and index for the listed papers.

Highlighted Details

• Comprehensive taxonomy of LLM-RL integration methods.
• Links to numerous research papers with summaries and overviews.
• Categorization of papers by their specific LLM-RL application (e.g., reward design, policy teaching, state representation).
• Focus on control-oriented applications like robotics and game agents.

Maintenance & Community

The repository is community-driven, with an invitation for contributions via Pull Requests to add new relevant papers.

Licensing & Compatibility

The licensing information for the repository itself is not explicitly stated in the README. The linked papers will have their own respective licenses.

Limitations & Caveats

This repository is a curated list and does not provide any executable code or frameworks. Its utility is solely as a reference for the state of research in LLM-RL integration.