Over the past year, we’ve been honored by your creativity, insights, and shared passion for advancing generative AI through InstructLab. Whether you added a new “knowledge” via pull request, offered feedback, joined a community call, or helped translate documentation, you’ve shaped our project in meaningful ways. Thank you.
To better align with evolving technical needs, we’re announcing an evolution for the InstructLab community. We will be refactoring the project by separating the components out to improve its maintainability and usability, primarily as a framework SDK for model tuning.
To enhance the long-term viability and efficiency of the InstructLab project, a strategic decision has been made to relocate its foundational building blocks into separate, dedicated project repositories. This carefully considered shift is anticipated to yield substantial benefits, primarily in the areas of maintainability and independent component maturation. This independent development will foster greater agility, allowing for more focused improvements and faster iteration cycles for individual parts of the project.
We’re excited about this next chapter and believe it will lead to more robust, flexible, and powerful tools for the generative AI community. We encourage you to follow the individual component projects in their new homes and continue contributing to their growth.
SDG: https://github.com/Red-Hat-AI-Innovation-Team/sdg_hub
Training: https://github.com/Red-Hat-AI-Innovation-Team/training_hub
InstructLab uses Large-Scale Alignment for ChatBots [1] (LAB), a new alignment tuning method for LLMs that leverages synthetic data. 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)