Fine-tuning
Expanded definition
Fine-tuning updates model weights, unlike prompting or RAG which leave the base checkpoint unchanged. Parameter-efficient methods such as LoRA train adapters instead of every parameter. You still need labeled data, evals, and a serving plan for the new artifact. Fine-tuning is not a substitute for retrieval when the facts change daily, and it does not grant rights beyond the base-model license. Compare against a frozen-base plus prompt/RAG baseline before paying for a custom checkpoint.
Related terms
Explore adjacent ideas in the knowledge graph.
Fine-tuning FAQ
What is Fine-tuning?
Fine-tuning continues training a pretrained model on a smaller, task-specific dataset so it follows that task more reliably.
How is Fine-tuning used in AI systems?
Fine-tuning updates model weights, unlike prompting or RAG which leave the base checkpoint unchanged. Parameter-efficient methods such as LoRA train adapters instead of every parameter. You still need labeled data, evals, and a serving plan for the new artifact. Fine-tuning is not a substitute for retrieval when the facts change daily, and it does not grant rights beyond the base-model license.
Related
Comparisons, tools, and models that connect to this idea.