Fine-Tuning
Definition
Fine-tuning is the process of further training a pre-trained AI model on a smaller, task-specific dataset to improve its performance on that particular task. It allows organizations to customize general-purpose models for domain-specific applications without training from scratch. Techniques like LoRA and QLoRA have made fine-tuning accessible on consumer hardware.
How It Works
The pre-trained model's weights are updated using gradient descent on the new dataset, typically with a lower learning rate to preserve existing knowledge. Parameter-efficient methods like LoRA freeze most weights and train small adapter matrices, reducing memory requirements by 90% or more. The fine-tuned model retains its general capabilities while gaining specialized knowledge in the target domain.