Overview
Foundation models are the 'raw material' of modern AI. They are trained on broad data and can be fine-tuned for specific applications like chatbots, code assistants, or image generators.
Characteristics
- Scale: Billions of parameters and massive training datasets.
- Emergence: They often develop capabilities not explicitly taught during training.
- Versatility: One model can handle many different types of tasks.