Fine-tuning is further training of an existing model on a smaller, task-specific dataset so it adopts a particular style, format or skill. It changes how a model behaves without the cost of training a new one from scratch.
RAG gives a language model relevant documents at the moment it answers, while fine-tuning retrains the model's weights on custom examples. RAG suits changing facts that need citing; fine-tuning suits fixed styles, formats or specialised skills.
A small language model is a language model with relatively few parameters, built to run cheaply and quickly on a single server, device or private environment.
Training data is the set of examples a machine learning model learns from. Its quality, coverage and labelling largely determine how well the model performs.
Deep learning is a type of machine learning that uses neural networks with many layers to learn complex patterns. It excels with large amounts of unstructured data such as images, audio and text.
A large language model (LLM) is a neural network trained on very large amounts of text to predict and generate language. It powers chat assistants, summarisers, coding tools and AI agents.
RLHF is a training method that uses people's preferences between model outputs to shape a model's behaviour. It is mainly used to make language models more helpful, honest and safe.
A clear proven process, that delivers
You've seen our ideas - now go deeper.
Discover more insights, guides, and engineering practice.