The term foundation model was popularized by Stanford CRFM in 2021 to describe large models pretrained on broad data and then adapted to many downstream tasks. GPT-4, Claude Sonnet, Gemini, and Llama 3 are all foundation models under this definition. Pre-training is done self-supervised, then Fine-tuning or RLHF specializes the model for particular behaviors. Almost every layer of today's AI product stack — chatbots, agents, RAG systems — is built on top of one or more foundation models.
MEVZU N°124ISTANBULYEAR I — VOL. III
Glossary · Intermediate · 2021
Foundation Model
A large-scale AI model pretrained on broad data that can be adapted to many downstream tasks.
- EN — English term
- Foundation Model
- TR — Turkish term
- Temel Model (Foundation Model)