Neuron Explorer

Qwen3-0.6B

number right after 'top' or 'bottom'

A neuron in layer 11 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featured, so it has steering outputs.

What this neuron does

Fires on the number that follows the word 'top' or 'bottom' in a ranking, such as 'top 10' or 'bottom 15 percent'. When turned up, the model's writing moved toward rankings and 'highest' lists.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. Think it is wrong? Flag it (opens a prefilled issue on GitHub).

Fires most on

The text windows where this neuron's activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

  1. #1 peak

    low or no earnings, perhaps 40 percent of college-educated workers make less than the top 25 percent of high school graduates.⏎This alternative reading of the figure helps to

  2. #2 peak

    08 NBA season, there were a total of 245,811 points scored by 450 players (data here). The top 20% of scorers (led by #1 Kobe Bryant with 2

  3. #3 peak

    5 years, the main causes of death in the world have remained fairly consistent.⏎The top 10 leading causes of death in the world:⏎- Heart disease⏎- Cancer

  4. #4 peak

    ondall State School 's Year 3 Literacy Test results have increased significantly since the ELF program began. The percentage of students identified as being in the bottom 15 percent of the state has decreased from 21.6 percent to

  5. #5 peak

    than any other continent.⏎What continent has the largest animals?⏎List of largest land carnivorans⏎|Rank||Common name||Native range by continent|⏎|1||Polar bear||North America, Eurasia|⏎|2||Brown bear

  6. #6 peak

    . The figure also shows, for example, that something approaching half of college graduates (the figure doesn’t allow for too much precision) make less than the top 25 percent of high school graduates. And somewhere approaching half of high school graduates

Steering

Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.

turned down a lot
Prompt

Here is something I have been thinking about lately: "I don't mean to say, but I see. I mean that I mean." So far, I have been thinking about, and I have been using. I can see. I mean it is clear, so it is clear.

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

  1. #1 peak

    site, additional surface material was collected and one 1 X 2 meter test pit was excavated to determine the site's approximate extent, cultural affinity, and stratigraphic position. From cultural material recovered on the surface and in the test excavation

  2. #2 peak

    (6-8) Show all Grade Level⏎Resource Type: Activities⏎Results 1 - 7 of 7 matches⏎From Grid to Home⏎Lisa

  3. #3 peak

    marriage behind everyone’s back, everything seems to be testing them; including a fight that broke out and ended in Tybalt’s murder and Romeo being<|endoftext|>Show Contents Previous Page Next Page⏎Chapter 3 - The Apache Module Architecture and API⏎

  4. #4 peak

    |⏎Widely distributed along the base of the Southern Alps lies a series of unfossiliferous sedimentary beds, consisting for the most part of well-stratified gravels, sands, and clays, with occasional lignite, whose

Connections

Show on map

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Fed by earlier layers

  1. L10:2995 excites
  2. L10:1207 suppresses
  3. L5:1902 excites
  4. L6:2088 suppresses
  5. L6:1934 suppresses
  6. L6:2402 weakly excites
  7. L7:2566 weakly suppresses
  8. L7:2337 weakly suppresses
  9. L10:2133 weakly excites
  10. L6:2045 weakly suppresses
  11. L9:5 weakly excites
  12. L9:647 weakly excites
  13. L6:12 weakly excites
  14. L2:1895 weakly excites
  15. L10:0 weakly excites
  16. L8:2872 weakly excites
  17. L9:3040 weakly suppresses
  18. L8:1523 weakly excites
  19. L9:2569 weakly excites
  20. L3:1832 weakly excites
  21. L7:1776 weakly suppresses
  22. L8:1800 weakly excites
  23. L10:2363 weakly suppresses
  24. L9:1232 weakly suppresses

Feeds later layers

  1. L17:2737 excites
  2. L15:2789 excites
  3. L27:2575 excites
  4. L14:2092 excites
  5. L18:2149 excites
  6. L22:969 excites
  7. L22:1605 excites
  8. L19:1263 excites
  9. L26:1525 excites
  10. L22:2560 excites
  11. L21:1581 excites
  12. L13:513 excites
  13. L12:2921 excites
  14. L19:2558 weakly excites
  15. L21:542 weakly excites
  16. L19:2312 weakly excites
  17. L20:1150 weakly excites
  18. L23:2127 weakly excites
  19. L13:571 weakly excites
  20. L17:1168 weakly suppresses
  21. L13:2110 weakly excites
  22. L18:2465 weakly excites
  23. L18:2934 weakly excites
  24. L17:2688 weakly excites

Activation histogram

How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.

4 values between -0.91 and -0.7052 values between -0.70 and -0.49356 values between -0.49 and -0.281419 values between -0.28 and -0.081876 values between -0.08 and 0.13175 values between 0.13 and 0.3418 values between 0.34 and 0.554 values between 0.55 and 0.761 values between 0.76 and 0.971 values between 0.97 and 1.181 values between 4.31 and 4.521 values between 7.23 and 7.44offnegativepositiveoften

Logit lens

The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layers instead.

Promotes

  1. 排序
  2. ranking
  3. highest
  4. highest
  5. 最快的
  6. 排名第
  7. 排名
  8. 最高的
  9. largest
  10. 排名第一

Suppresses

  1. arel
  2. aras
  3. asions
  4. .Done
  5. aton
  6. utions
  7. اث
  8. ardless
  9. nung
  10. วร