138 lines
7.1 KiB
HTML
138 lines
7.1 KiB
HTML
<!--
|
|
This is free software for the public good of a permacomputer hosted at
|
|
permacomputer.com, an always-on computer by the people, for the people.
|
|
One which is durable, easy to repair, & distributed like tap water
|
|
for machine learning intelligence.
|
|
|
|
The permacomputer is community-owned infrastructure optimized around
|
|
four values:
|
|
|
|
TRUTH First principles, math & science, open source code freely distributed
|
|
FREEDOM Voluntary partnerships, freedom from tyranny & corporate control
|
|
HARMONY Minimal waste, self-renewing systems with diverse thriving connections
|
|
LOVE Be yourself without hurting others, cooperation through natural law
|
|
|
|
This software contributes to that vision by making machine learning
|
|
accessible to everyone through a free, open, embeddable chat interface.
|
|
Code is seeds to sprout on any abandoned technology.
|
|
-->
|
|
<!DOCTYPE html>
|
|
<html lang="en">
|
|
|
|
<head>
|
|
<meta charset="UTF-8">
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|
<meta name="theme-color" content="#43a047">
|
|
<meta name="color-scheme" content="light dark">
|
|
<title>Reverse Retrieval Augmented Generation | uncloseai.com</title>
|
|
<meta name="description" content="Whitepaper: Reverse RAG, a client-side context injection technique that makes small language models punch above their weight by extracting live page content and injecting it directly into the conversation.">
|
|
|
|
<!-- PicoCSS -->
|
|
<link rel="stylesheet" href="/css/pico.classless.min.css">
|
|
<!-- ChunkFive Font -->
|
|
<link rel="stylesheet" href="/css/chunkfive/stylesheet.css" type="text/css" charset="utf-8" />
|
|
<!-- Sidebar Theme -->
|
|
<link rel="stylesheet" href="/css/sidebar-theme.css">
|
|
|
|
<!-- Theme Switcher Script -->
|
|
<script>
|
|
function switchTheme(theme) {
|
|
if (theme === "auto") {
|
|
document.documentElement.removeAttribute('data-theme');
|
|
} else {
|
|
document.documentElement.setAttribute('data-theme', theme);
|
|
}
|
|
}
|
|
|
|
// Disable custom styling for uncloseai.js - use PicoCSS instead
|
|
window.UNCLOSEAI_CUSTOM_STYLING = false;
|
|
</script>
|
|
|
|
<script src="https://uncloseai.com/uncloseai.js" type="module"></script>
|
|
</head>
|
|
|
|
<body>
|
|
<!-- Mobile menu toggle -->
|
|
<button class="sidebar-toggle" onclick="document.querySelector('.sidebar').classList.toggle('open')">
|
|
☰
|
|
</button>
|
|
|
|
<!-- Left sidebar - Main Navigation -->
|
|
<aside class="sidebar">
|
|
<div class="table-of-contents">
|
|
<nav>
|
|
<ul>
|
|
<li><a href="/">Home</a></li>
|
|
<li><a href="/c-examples.html">C Examples</a></li>
|
|
<li><a href="/csharp-examples.html">C# Examples</a></li>
|
|
<li><a href="/dart-examples.html">Dart Examples</a></li>
|
|
<li><a href="/elixir-examples.html">Elixir Examples</a></li>
|
|
<li><a href="/go-examples.html">Go Examples</a></li>
|
|
<li><a href="/java-examples.html">Java Examples</a></li>
|
|
<li><a href="/kotlin-examples.html">Kotlin Examples</a></li>
|
|
<li><a href="/nodejs-examples.html">Node.js Examples</a></li>
|
|
<li><a href="/php-examples.html">PHP Examples</a></li>
|
|
<li><a href="/python-examples.html">Python Examples</a></li>
|
|
<li><a href="/ruby-examples.html">Ruby Examples</a></li>
|
|
<li><a href="/rust-examples.html">Rust Examples</a></li>
|
|
<li><a href="/swift-examples.html">Swift Examples</a></li>
|
|
<li><a href="/uncloseai-js.html">uncloseai.js Docs</a></li>
|
|
<li><a href="/uncloseai-js-styleguide.html">Styleguide</a></li>
|
|
<li><a href="/cli.html">uncloseai-cli</a></li>
|
|
<li><a href="/browser-toys.html">Browser Toys</a></li>
|
|
<li><a href="/inference.html">Inference Setup</a></li>
|
|
<li><a href="/text-to-speech.html">Text-to-Speech</a></li>
|
|
<li><a href="/crawler.html">Our Crawler</a></li>
|
|
<li><a href="/reverse-retrieval-augmented-generations-rag.html" class="active">Reverse RAG</a></li>
|
|
<li><a href="/languages" target="_blank">All Languages</a></li>
|
|
<li><a href="https://shop.unturf.com/p/8486f492-a93e-11f0-b477-02dfe05770ee/uncloseai-machine-learning-reference-guide-to-inference-clients" target="_blank">Book</a></li>
|
|
</ul>
|
|
</nav>
|
|
</div>
|
|
</aside>
|
|
|
|
<!-- Main content -->
|
|
<main>
|
|
<header>
|
|
<hgroup>
|
|
<a href="https://uncloseai.com"><h1 class="unturf" style="font-family: 'ChunkFiveRegular';">uncloseai.</h1></a>
|
|
<p>Reverse Retrieval Augmented Generation</p>
|
|
</hgroup>
|
|
<nav>
|
|
<ul>
|
|
<li><a href="#" onclick="switchTheme('auto')">Auto</a></li>
|
|
<li><a href="#" onclick="switchTheme('light')">Light</a></li>
|
|
<li><a href="#" onclick="switchTheme('dark')">Dark</a></li>
|
|
</ul>
|
|
</nav>
|
|
</header>
|
|
|
|
<h2>Client-Side Context Injection for Small Language Models</h2>
|
|
|
|
<p><em>How live DOM extraction makes 8B models punch above their weight class.</em></p>
|
|
|
|
<p><em>russell@unturf, cthegray, TimeHexOn, foxhop</em></p>
|
|
|
|
<h3>Abstract</h3>
|
|
|
|
<p>Traditional Retrieval Augmented Generation (RAG) requires a server-side pipeline: chunk documents, embed them into vectors, store them in a database, and retrieve by similarity search at query time. This architecture demands infrastructure, indexing latency, and maintenance of embedding models and vector stores.</p>
|
|
|
|
<p><strong>Reverse Retrieval Augmented Generation (Reverse RAG)</strong> inverts this entirely. Instead of the server fetching documents to augment the prompt, the client extracts live content from the page the user currently views and injects it directly into the conversation context. The data comes to the model. No vector database. No embeddings. No indexing pipeline. No server-side retrieval.</p>
|
|
|
|
<p><a href="/uncloseai-js.html">uncloseai.js</a> implements this technique as an AGPL-3.0-only algorithm. <code>uncloseai.js</code> serves as the entrypoint of a modular application that adds a machine learning chat interface to any webpage. By feeding the model the full, fresh content of whatever page the user visits, small 8B-parameter models produce answers that rival much larger models on page-specific questions.</p>
|
|
|
|
<p><a href="/reverse-rag.pdf" target="_blank" role="button">Read the full whitepaper (PDF)</a></p>
|
|
|
|
<h3>Citation</h3>
|
|
<pre><code>russell@unturf, cthegray, TimeHexOn, foxhop. "Reverse Retrieval Augmented
|
|
Generation: Client-Side Context Injection for Small Language Models."
|
|
uncloseai.com, 2026. https://uncloseai.com/reverse-retrieval-augmented-generations-rag.html</code></pre>
|
|
|
|
<footer>
|
|
<small>© uncloseai. 2026</small>
|
|
<br>
|
|
<small>Stylesheets by <a href="https://picocss.com" target="_blank">PicoCSS</a></small>
|
|
</footer>
|
|
</main>
|
|
</body>
|
|
</html>
|