Fine-tuning
TrainingFurther training a pre-trained AI model on a smaller, task-specific dataset to specialize its behavior.
Full Explanation
Pre-trained foundation models are general-purpose. Fine-tuning adapts them for specific use cases — a customer support chatbot, a medical coding tool, or a legal document analyzer. It's far cheaper than training from scratch. Techniques include supervised fine-tuning (SFT) on examples of desired behavior, and RLHF to align with human preferences.
OpenAI fine-tunes GPT-4 on customer service conversations to create specialized support chatbots.
Related Terms
Reinforcement Learning from Human Feedback — a training technique used to align AI models with human preferences and values.
A large AI model trained on broad data at scale that can be adapted for many different downstream tasks.
A type of AI model trained on vast amounts of text data that can generate, summarize, translate, and reason about language.