PKU-Alignment/align-anything
A modular framework for aligning multi-modal large language models with human feedback using RLHF, DPO, and PPO techniques.
Not currently ranked — collecting fresh signals.
star history
Align-Anything is an open-source framework for training all-modality large language models (any-to-any models) to align with human intentions and values. It provides modular implementations of alignment algorithms including SFT, DPO, PPO, and O1-like training approaches. The framework supports fine-tuning diverse multi-modal models across image, video, audio, and text modalities, offering a unified CLI for experimenting with different alignment methods.
Frequently asked
- What is PKU-Alignment/align-anything?
- A modular framework for aligning multi-modal large language models with human feedback using RLHF, DPO, and PPO techniques.
- Is align-anything open source?
- Yes — PKU-Alignment/align-anything is open source, released under the Apache-2.0 license.
- What language is align-anything written in?
- PKU-Alignment/align-anything is primarily written in Python.
- How popular is align-anything?
- PKU-Alignment/align-anything has 4.7k stars on GitHub.
- Where can I find align-anything?
- PKU-Alignment/align-anything is on GitHub at https://github.com/PKU-Alignment/align-anything.