base models

Pre-trained AI models on massive, diverse datasets: they serve as the versatile, general-purpose foundation for specialized applications.

Base models are the foundational core of modern AI: a single, large-scale model pre-trained on vast, diverse data. This paradigm shift (coined by Stanford HAI in 2021) allows for rapid deployment. Instead of building a bespoke model for every task, developers fine-tune a base model like OpenAI's GPT-4 or Google's BERT. This transfer learning approach drastically cuts development time and computational cost for downstream applications: a critical efficiency gain in the industry.

https://hai.stanford.edu/research/foundation-models