What is a Fine-tuning?
The process of further training a pre-trained AI model on specialised data to improve performance on specific tasks.
Pronunciation: /faɪn ˈtjuːnɪŋ/
Fine-tuning explained in plain English
Fine-tuning takes an already-trained AI model and trains it further on a smaller, task-specific dataset. This adapts the model's general capabilities to perform better on particular domains, styles, or tasks — like teaching a general practitioner to specialise in cardiology.\n\nFine-tuning is distinct from prompt engineering (guiding the model at inference time) and from training a model from scratch.
Analogy
Fine-tuning is like hiring an experienced chef and teaching them your restaurant's specific recipes. They already know how to cook; you are refining their skills for your particular kitchen.
Example
A healthcare company fine-tunes an LLM on medical literature and clinical notes so it can assist doctors with terminology and context specific to their speciality.
How is Fine-tuning used?
Organisations fine-tune models for brand-specific writing, domain-specific analysis (legal, medical), code generation in proprietary languages, and customer service with company-specific knowledge.
Common misconceptions about Fine-tuning
History
Fine-tuning has been a core technique in machine learning for decades. With LLMs, parameter-efficient methods like LoRA (2021) made fine-tuning accessible without massive compute resources.
Related terms
People also read
- LLM
A type of AI model trained on vast amounts of text to understand and generate human language.
- Prompt
The input text or instruction given to an AI model to guide its response.
- Embedding
A numerical representation of text, images, or other data that captures semantic meaning.
- Inference
The phase when a trained model is actually used — taking new input and producing a prediction or response.
- AI slop
Output from a generative AI system that favors quantity over quality.
- Artificial Intelligence
Any system that performs tasks requiring human judgment — understanding language, recognising images, making decisions, or solving problems — by finding patterns in data rather than thinking or feeling.
- average precision at k
A metric for summarizing a model's performance on a single prompt that generates ranked results, such as a numbered list of book recommendations.
- Backpropagation
The process that tells a neural network which internal settings caused an error and how to adjust them, working backwards through layers.
- BERT
A model architecture for text representation.
- Chain-of-Thought Prompting
Asking an AI to show its reasoning step by step before giving a final answer, which often improves accuracy on complex tasks.