<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Rescore on INFINI Pizza</title><link>https://pizza.rs/tags/rescore/</link><description>Recent content in Rescore on INFINI Pizza</description><generator>Hugo</generator><language>en</language><atom:link href="https://pizza.rs/tags/rescore/index.xml" rel="self" type="application/rss+xml"/><item><title>Rescore</title><link>https://pizza.rs/docs/references/search/request/rescore/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://pizza.rs/docs/references/search/request/rescore/</guid><description>&lt;h1 id="rescore">
 Rescore
 &lt;a class="anchor" href="#rescore">#&lt;/a>
&lt;/h1>
&lt;p>&lt;code>rescore&lt;/code> re-ranks the top window of hits with a second query — an
expensive precision pass (phrase, vector similarity, a narrow filter
match) applied only where it can change the final page, instead of to
every matching document.&lt;/p>
&lt;div class="highlight">&lt;pre tabindex="0" class="chroma">&lt;code class="language-sh" data-lang="sh">&lt;span class="line">&lt;span class="cl">POST /my-collection/_search
&lt;/span>&lt;/span>&lt;/code>&lt;/pre>&lt;/div>&lt;h2 id="how-it-works">
 How it works
 &lt;a class="anchor" href="#how-it-works">#&lt;/a>
&lt;/h2>
&lt;p>Rescore is engine-native. When a request carries a &lt;code>rescore&lt;/code> block, the
collector widens its gather to &lt;code>max(from + size, window)&lt;/code> candidates, the
first-pass query ranks them, and the store layer then applies the rescore
to the collected set (&lt;code>apply_rescore_to_result&lt;/code> in
&lt;code>lib/engine/src/store/layered/generic_reader.rs&lt;/code>, blending in
&lt;code>lib/engine/src/search/scorer/mod.rs&lt;/code>):&lt;/p></description></item></channel></rss>