Optimizing foundation models
Course Overview
Two ways to make a smart helper smarter
The story: You hire a clever general assistant. There are two ways to make them useful in your business: hand them your company's filing cabinet to look things up, or send them on a course so they absorb your field for good. This course follows two companies, one trying each approach.
In AI/AWS terms: This course looks at two ways to improve a foundation model, RAG (the filing cabinet) and fine-tuning (the course), through two business cases. Afterwards you should be able to:
- Name the AWS services that store embeddings in vector databases
- Explain the role of agents in multi-step tasks
- Evaluate FM performance and decide whether it meets business objectives
- Describe methods for fine-tuning an FM and how to prepare the data
For the exam: RAG looks things up at answer time. Fine-tuning changes the model through extra training.