---
title: "LLM Neuroanatomy: How I Topped the AI Leaderboard Without Changing a Single Weight"
url: https://daily.dev/posts/llm-neuroanatomy-how-i-topped-the-ai-leaderboard-without-changing-a-single-weight-ra1ocifbe
source_url: https://dnhkng.github.io/posts/rys/
type: article
source: "Hacker News"
published: 2026-03-10T17:03:02.452Z
updated: 2026-03-10T17:03:36.427Z
tags: ["llm"]
reading_time: 28
upvotes: 2
comments: 0
language: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# LLM Neuroanatomy: How I Topped the AI Leaderboard Without Changing a Single Weight

**[Hacker News](https://daily.dev/sources/hn)** · 28 min read · 2 upvotes · 0 comments

## Summary

A researcher topped the HuggingFace Open LLM Leaderboard in mid-2024 without training or modifying any weights. The method, called RYS (Repeat Yourself), works by identifying a specific block of middle transformer layers in a 72B model and duplicating them. Starting from two observations — that LLMs can reason in Base64 and that the Goliath-120B frankenmerge worked despite violating distribution assumptions — the author hypothesized that transformers have a functional anatomy: early layers encode input, late layers decode output, and middle layers contain discrete reasoning circuits. A 'brain scanner' was built by sweeping all possible layer duplication configurations and scoring them with two orthogonal probes: hard math guessing and emotional quotient benchmarks. The optimal config for Qwen2-72B was duplicating 7 specific middle layers, yielding +17.72% on MuSR and +8.16% on MATH with no weight changes. Heatmaps revealed that only circuit-sized blocks improve performance — single-layer duplication fails — suggesting middle layers form indivisible multi-step reasoning pipelines. Descendants of this model still dominate the leaderboard as of early 2026.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://dnhkng.github.io/posts/rys/>

## Similar posts on daily.dev

- [LLM Neuroanatomy II: Modern LLM Hacking and hints of a Universal Language?](https://daily.dev/posts/llm-neuroanatomy-ii-modern-llm-hacking-and-hints-of-a-universal-language--sedj667om) · Hacker News · 1 upvotes · 0 comments
- [GitHub - alainnothere/llm-circuit-finder: I replicated Ng's RYS method and found that duplicating 3 specific layers in Qwen2.5-32B boosts reasoning by 17% and duplicating layers 12-14 in Devstral-24B ](https://daily.dev/posts/github---alainnothere-llm-circuit-finder-i-replicated-ng-s-rys-method-and-found-that-duplicating-3--fawqmnh1t) · Hacker News · 1 upvotes · 0 comments

---

Tags: [#llm](https://daily.dev/tags/llm)

[View this post on daily.dev](https://daily.dev/posts/llm-neuroanatomy-how-i-topped-the-ai-leaderboard-without-changing-a-single-weight-ra1ocifbe)