What Is RLCD? The Secret Behind Jev

(di-zhang-llm.github.io)

47 points by tnspacetime5 hours ago

4 comments

  • firejake30819 minutes ago
    &gt; The operational signal was always relative preference. The scalar merely hid it.<p>Is this another Claude-ism? &quot;X was always Y. The Z merely hid it.&quot; Or am I overcalling it?
  • daemonk52 minutes ago
    Yeah the calibration is really what makes it useful in practice for quick, small decisions. Asking a LLM to give scores to a problem will yield inconsistently scaled&#x2F;anchored results that changes at a whim.<p>The blog is pretty heavy on statistics. I&#x27;ll have to study it more when I have time. Is it essentially bootstrapping results to statistically normalize the answers?
    • tnspacetime1 minute ago
      I have not studied it properly too. Good that it has both code and note though.
  • WalterGR1 hour ago
    RLCD, not defined in the article, is Reinforcement Learning for Calibrated Decisions.
    • borgel23 minutes ago
      Ah, so not Reflective LCD [1] then.<p>[1] <a href="https:&#x2F;&#x2F;www.e3displays.com&#x2F;reflective-lcd-display-monitor&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.e3displays.com&#x2F;reflective-lcd-display-monitor&#x2F;</a>
  • jackb40401 hour ago
    [flagged]