// AI
Fine-Tuning
Additional training of a base model on domain-specific data to specialise its behaviour.
In depth
In 2023 fine-tuning was the default customisation lever. By 2026, RAG plus long context has replaced most fine-tuning use cases, except for style adaptation, low-latency task-specific models, and regulated domains where a private model is required.
Related terms
LLM
Large Language Model, the transformer-based foundation model behind ChatGPT, Claude, Gemini and every 2024+ AI product.
RAG
Retrieval-Augmented Generation, grounding an LLM's response in retrieved documents to reduce hallucination.
Embedding
A dense numeric vector representation of text, image or audio, the input format for semantic search and RAG.