Quick Links: Documentation | FAQ | Hugging Face | Calendar | Slack | YouTube | X
InstructLab is an open source, accessible, and model-agnostic AI project that facilitates contributions to existing large language models (LLMs). Our community's mission is to enable anyone to shape the future of generative AI.
- Install InstructLab locally
- Contribute to the project
- Connect with the community to share feedback and ideas, ask questions, and troubleshoot issues.
Many projects are rapidly embracing and extending permissively-licensed AI models, but they face three main challenges:
- Direct contributions to LLMs are not possible. They show up as forks, which are expensive for model creators to maintain, and force consumers to choose a “best-fit” model that isn’t easily extensible.
- The barrier to entry is high. One’s ability to contribute ideas is limited by their AI/ML expertise. One has to learn how to fork, train, and refine models to see their idea move forward.
- There is no direct community governance or best practice around review, curation, and distribution of forked models.
InstructLab solves these problems by using Large-Scale Alignment for ChatBots [1] (LAB), a new alignment tuning method for LLMs that leverages synthetic data.
InstructLab's model-agnostic technology gives model upstreams the ability to regularly create builds of their open-source-licensed models. This is achieved by composing new skills and knowledge into the model, as opposed to rebuilding and retraining it.
Take a look at LAB-enhanced models on the InstructLab Hugging Face page.
To learn more about InstructLab’s origins, visit the About Taxonomy page.
[1] Shivchander Sudalairaj*, Abhishek Bhandwaldar*, Aldo Pareja*, Kai Xu, David D. Cox, Akash Srivastava*. "LAB: Large-Scale Alignment for ChatBots", arXiv preprint arXiv: 2403.01081, 2024. (* denotes equal contributions)
