Fine-tuning
/faɪn ˈtjuːnɪŋ/
The process of adapting a pre-trained model to a specific task by continuing training on task-specific data.
Fine-tuning leverages transfer learning by starting with pre-trained weights and updating them with smaller learning rates on new data. More efficient than training from scratch.
Examples
- Adapting BERT for sentiment analysis
- Fine-tuning GPT for code generation
- Customizing vision models for medical imaging