From the course: RAG Fine-Tuning: Advanced Techniques for Accuracy and Model Performance
Join today to access over 24,800 courses taught by industry experts.
Fine-tuning and inference
From the course: RAG Fine-Tuning: Advanced Techniques for Accuracy and Model Performance
Fine-tuning and inference
- [Instructor] Now, let's explore the fine-tuning process. We start with our carefully prepared training data, including questions and document sets, which are structured into clear reasoning steps. Next, we use supervised learning to train the model. During training, the model learns to identify and prioritize relevant information from Oracle documents while ignoring the distractor documents, and it's trained to generate chain-of-thought style answers, learning to reason step by step, and cite relevant source documents. The model gradually adapts to understand domain-specific patterns. So throughout the process, the model learns to generate detailed and step-by-step answers. RAFT has four crucial training objectives. Like a skilled researcher, the model learns to spot the exact information needed to answer the question. This is crucial. The model learns not to be misled by any sort of irrelevant information, and by learning from distractor documents, it manages to hone that…
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.