Fine-tuning

See also How do you fine-tune a model using peft for a more practical point of view

Fine-tuning is the process of training an already-trained model to make it better as at a specific task. (Train a neural network)

For example, if you have a general purpose text generation model, you can fine-tune it to make it better at translation by giving it multi-lingual pieces of texts that the model will try to complete.

See QLoRA and the paper LoRA on an algorithm to make fine-stuning (and training) faster and for a more mathematical outlook.