fine-tuning
Further training an existing model on your own examples to change its behaviour, style or format.
In more detail
Fine-tuning shapes how a model responds rather than reliably teaching it new facts; facts that change belong in retrieval. It needs labelled data and its own evals, and a fine-tuned model has to be retrained when the base model is retired.
Related terms
- retrieval-augmented generation — Answering a question by first retrieving relevant passages from your own documents and adding them to the prompt, so the model answers from them.
- model training — The process that sets a model's weights by adjusting them, over many examples, until its predictions fit the data.
- in-context learning — A model picking up a task from instructions and examples in its prompt, with no change to its weights.
Taught in
- What An AI Engineer Does — AI Engineering: LLM Apps, RAG & Agents
- Choosing A Model — AI Engineering: LLM Apps, RAG & Agents
- Zero-Shot And Few-Shot — AI Engineering: LLM Apps, RAG & Agents
- How RAG Works And When To Use It — AI Engineering: LLM Apps, RAG & Agents
References
- OpenAI: Model optimization (fine-tuning)
- Lewis et al.: Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (2020)