Training & Fine-tuning

LoRA (Low-Rank Adaptation)

An efficient fine-tuning technique that modifies only a small part of a model's weights.

Definition

Fine-tuning large models by updating all their parameters is extremely expensive. LoRA is a technique that fine-tunes models much more efficiently by only modifying a small set of additional parameters, while keeping the bulk of the original model frozen. The adapted model performs comparably to full fine-tuning at a fraction of the cost and computing time. LoRA is now the standard method for customising large models for specific business applications.

Why this matters for your business

LoRA makes it commercially viable for mid-sized businesses to create custom AI models fine-tuned on their specific data, without the cost of full model training.

Related Terms

Fine-tuning

Further training a pre-trained model on specific data to specialise it for a task.

QLoRA

A memory-efficient version of LoRA that uses quantisation to reduce computing requirements.

Pre-trained Model

An AI model that has already been trained on large datasets and is ready to use or customise.

Heard enough terminology — ready to talk outcomes?

We translate AI concepts into measurable business results. No upfront fees — you pay only when independently verified results are delivered.

← Back to glossary

Disclaimer

This definition is provided for educational and informational purposes only. It represents a general explanation of a technical concept and does not constitute professional, technical, or investment advice. Artificial intelligence is a rapidly evolving field; terminology, techniques, and capabilities change frequently. Coaley Peak Ltd makes no warranty as to the accuracy, completeness, or currency of the information provided. Nothing on this page should be relied upon as the sole basis for commercial, technical, legal, or investment decisions without independent professional advice.

Document reference: ISO_webpage_knowledge-base_glossary_v1

Last modified: 29 March 2026

Knowledge Base·Training & Fine-tuning·LoRA (Low-Rank Adaptation)