Fine-tuning
Fine-tuning is the process of adapting a trained model for another task, usually a more focused one. Additional training on new data is applied to parameters that have already been trained, across the whole network or some of its layers.
On this page
Source content checked . English translation published 2026-10-27.
In business automation
Fine-tuning changes model behavior using training examples. It may suit a style or recurring task, but an updated price list or procedure usually needs to remain in a knowledge source retrieved at runtime.
Examples
These examples illustrate a possible workflow. Adapt the connections and rules to your own systems.
Business process example
An organization collects corrected examples for document classification and keeps some separate for evaluation. It compares a fine-tuned model with a solution using instructions alone before replacing the component in the workflow.
Relevant tools
- Cohere
- Language models and adaptation capabilities depend on the service and model; check for support and data suited to the task.
What to check
- Do not train on a collection that also appears in evaluation, and do not use fine-tuning as a substitute for retrieving current information.