awesome-RLHF
A curated list of reinforcement learning with human feedback resources (continually updated)
What is awesome-RLHF?
A curated list of reinforcement learning with human feedback resources (continually updated)
Overview of awesome-RLHF
What it is
This is a curated list of research papers, codebases, datasets, blogs, and books on Reinforcement Learning with Human Feedback (RLHF). The repository is continuously updated to track the frontier of RLHF. RLHF uses reinforcement learning methods to optimize a language model with human feedback, enabling alignment of models trained on general text corpora with complex human values. The list covers RLHF for large language models (LLMs) and video games (e.g., Atari).
Key features
- Curated collection of RLHF research papers organized by year (2020 and before through 2026)
- Includes codebases, datasets, blogs, and books related to RLHF - Covers RLHF applications for LLMs and video games - Provides an overview and detailed explanation of RLHF concepts (e.g., Inverse Reinforcement Learning, Apprenticeship Learning, Interactive Machine Learning, Human-in-the-Loop RL)
- Continuously updated to track the latest developments in RLHF
Use cases
- Game playing: human feedback helps agents learn strategies and tactics (e.g., Go)
- Personalized recommendation systems: feedback from users helps learn individual preferences - Robotics: human feedback aids safe and efficient interaction with physical environments - Education: AI-based tutors use teacher feedback to personalize learning experiences
Getting started
See the repository README.
License and pricing
Licensed under Apache-2.0. Open source.
Based on the project's public README and repository metadata.