fine-tuning

Fine-tuning is the process of taking a model that has already been trained and training it a bit further on a narrower, more specific set of examples. The result is a version specialised for a particular task, tone or domain, without the enormous cost of building a model from scratch.

It can be remarkably effective. In one case, a hedge fund fine-tuned a free open-weight model on its own experts’ judgment and beat every big-name closed AI on finance tasks, at a fraction of the cost. The lesson generalises: for specialised work, a smaller model shaped by your own data can outperform a larger, general-purpose one.

Fine-tuning is a key reason open-weight models are so valuable to businesses: you can adapt them to your needs while keeping both the model and your data under your control.