Also known as: foundation model (FM) Β· Foundation Models Β· FM
A very large pre-trained model trained on an enormous and diverse training set. A foundation model can do both of the following: - Respond well to a wide range of requests. - Serve as a base model for additional fine-tuning or other customization. In other words, a foundation model is already very capable in a general sense but can be further customized to become even more useful for a specific task.
Large, powerful models that are trained on vast amounts of data, which often spans multiple modalities like text, images, video, and audio. These models use statistical modeling to predict likely responses to prompts and to generate new content. They learn patterns from their training data, such as language patterns for text generation and diffusion techniques for image generation.
A large pre-trained model (text, image, multimodal) adaptable to many tasks (e.g., via prompting, fine-tuning or RAG).
A side-by-side comparison of General-Purpose AI and Foundation Model. Understand how one describes broad adaptability and the other describes a large pretrained base model supporting downstream tasks.
A side-by-side comparison of Large Language Model and Foundation Model. Understand why an LLM is language-centered, while a foundation model is a broad pretrained base model that can support many downstream tasks.