It learns reusable representations from large and varied training data rather than being built for one narrow prediction target. Fine-tuning, retrieval, prompting, or task-specific heads can specialize its general capabilities.
A foundation model is a broadly trained model that can be adapted or prompted for many downstream tasks.
It learns reusable representations from large and varied training data rather than being built for one narrow prediction target. Fine-tuning, retrieval, prompting, or task-specific heads can specialize its general capabilities.