LLM Zoo is a curated repository tracking open- and closed-source Large Language Models (LLMs) released after ChatGPT. It serves as a comprehensive catalog for researchers and developers interested in the rapidly evolving LLM landscape, providing key details for each model.

How It Works

The project functions as a structured database, meticulously collecting and presenting information on various LLMs. Each entry includes release time, model size, supported languages, domain, training data specifics, and direct links to resources like GitHub, HuggingFace, papers, demos, and official blogs. This systematic approach allows for easy comparison and discovery of new models.

Quick Start & Requirements

This project is a data repository, not a runnable application. No installation or specific requirements are needed to view the information.

Highlighted Details

• Comprehensive table of over 20 open-source LLMs (e.g., LLaMA, Alpaca, Vicuna, ChatGLM) with detailed metadata.
• Includes a section for prominent closed-source LLMs (e.g., ChatGPT, Claude, GPT-4, Bard).
• Provides links to model resources, papers, and demos for further exploration.
• Regularly updated to reflect the latest LLM releases.

Maintenance & Community

The repository is actively maintained, with a stated commitment to regular updates. Contributions are welcomed via issues and pull requests. The project is associated with DAMO-NLP-SG.

Licensing & Compatibility

The repository itself is likely under a permissive license given its nature as a data collection. However, the LLMs referenced within the repository are subject to their own individual licenses, which may vary significantly and could include restrictions on commercial use.

Limitations & Caveats

The repository focuses on cataloging LLMs and does not provide direct access to or functionality for running these models. The "TODO List" indicates planned additions like systematic training data reviews and performance evaluations, suggesting current incompleteness in these areas.