Artificial Intelligence Back to glossary
Fine-tuning
Fine-tuning consists of partially retraining an existing AI model on a company-specific dataset, specializing its behavior without starting from scratch.
Why it matters
Fine-tuning is more costly and rigid than RAG or prompt engineering; it is only justified for high-volume use cases that require a very specific style or format.
Concrete example
A customer support company fine-tunes a model on thousands of historical tickets to reproduce its tone and business terminology.
Related notions
RetrainingSpecialized modelProprietary dataset
Related Kaliteq diagnostics
Assess your organization's maturity on this topic
Start the diagnostic