https://arshadhs.github.io/tags/rag/Recent content in RAG on Arshad SiddiquiHugoen-ushttps://arshadhs.github.io/docs/ai/genai/rag/Mon, 01 Jan 0001 00:00:00 +0000https://arshadhs.github.io/docs/ai/genai/rag/<h1 id="retrieval-augmented-generation-rag"> Retrieval-Augmented Generation (RAG) <a class="anchor" href="#retrieval-augmented-generation-rag">#</a> </h1> <p><strong>Retrieval-Augmented Generation (RAG)</strong> is a system design pattern that improves an LLM’s answers by:</p> <ol> <li><strong>Retrieving</strong> relevant information from an external knowledge source, and then</li> <li><strong>Augmenting</strong> the LLM prompt with that retrieved context before generating the final response.</li> </ol> <p>RAG helps an LLM <strong>look things up first</strong>, then <strong>answer using evidence</strong>.</p> <hr> <h2 id="why-rag-is-useful"> Why RAG is Useful <a class="anchor" href="#why-rag-is-useful">#</a> </h2> <p>RAG is commonly used when:</p> <ul> <li>Your knowledge is in <strong>private documents</strong> (PDFs, policies, internal wiki)</li> <li>You need <strong>up-to-date information</strong> (things not in the model’s training data)</li> <li>You want fewer <strong>hallucinations</strong> by grounding answers in retrieved sources</li> <li>You want <strong>traceability</strong> (show “where the answer came from”)</li> </ul> <blockquote class="book-hint info"> <p>RAG does not change the model weights.<br> It changes what the model <em>sees</em> at inference time by adding retrieved context.</p>