Fine-tuning

Additional training process on a pre-trained model, using a specific dataset to specialize it for a particular task. For example, you can fine-tune Llama 7B with legal documents to create a model expert in law. Popular techniques include LoRA and QLoRA.