An open-source Large Language Model (LLM) is a publicly available AI model that has been trained on vast amounts of data to understand and generate text, images, videos, audio, etc. These models are freely accessible, allowing developers and researchers to use and modify them, and to contribute to their development.
In contrast to Open-weights Models, which only make their weights available, open-source LLMs provide the full model architecture, training data, and codebase, enabling others to replicate, improve, and build upon the model.
Examples of open-source LLMs include MPT-30B by Databricks (Mosaic ML), the Falcon series from the Technology Innovation Institute (1.3B, 7.5B, 40B, and 180B), and GPT-J (6B) by EleutherAI.
Recent posts