<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Vision-Language-Model on The Kiseki Log</title>
    <link>/tags/vision-language-model/</link>
    <description>Recent content in Vision-Language-Model on The Kiseki Log</description>
    <generator>Hugo -- 0.147.7</generator>
    <language>en</language>
    <copyright>2023-2026 Elliot Shichao Song CC BY-SA 4.0</copyright>
    <lastBuildDate>Mon, 16 Nov 2026 00:07:03 +0800</lastBuildDate>
    <atom:link href="/tags/vision-language-model/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>CaptionExam: A Challenging QA Benchmark for Captioners</title>
      <link>/posts/261029-captionexam/</link>
      <pubDate>Thu, 29 Oct 2026 00:07:13 +0800</pubDate>
      <guid>/posts/261029-captionexam/</guid>
      <description>On CaptionExam, even GPT-6 Astra only just passes at describing images, with a score of 61.2%.</description>
    </item>
  </channel>
</rss>
