Fine-tuning
Further training a pre-trained model on specific data so its weights change to suit a task or style. It is one of several ways a system can learn, alongside memory, retrieval and workflow changes.
Fine-tuning takes a model that has already been trained on broad data and continues training it on a smaller, specific dataset. The model's weights are updated, so the behavior is changed in the model itself: a particular tone, format, domain vocabulary or task.
It is only one of the ways a system can "learn". Today the idea splits into several different mechanisms. Weights can be updated through training or fine-tuning; external memory can accumulate; an agent can revise a plan after observing a result; and a workflow can change routing or tool choice without the underlying model learning anything at all. Calling all of these "learning" blurs what actually changed.
Choosing a mechanism
For new knowledge, retrieval (RAG) is often simpler and easier to update than retraining an LLM. Fine-tuning suits stable behavior that prompts struggle to produce. In every case, re-evaluate afterwards, because a changed model has a new runtime address.
This distinction echoes Cahit Arf's point that learning changes the category of machine: see intibak kabiliyeti.
Read more in Cahit Arf's 1959 question about thinking machines.
Related terms
LLM
Large language model: an AI model trained on very large amounts of text to predict and generate language, which can be used to answer, summarise, write, reason about and act on text.
RAG
Retrieval-augmented generation: a pattern where a model is given relevant documents or data retrieved at question time, so its answers rest on current, specific sources rather than only its training.
Intibak kabiliyeti
Turkish for the ability to adapt. In Cahit Arf's 1959 lecture on thinking machines, it is the threshold that matters: not how many problems a machine solves, but whether it can adapt to new ones.