Neuron Explorer

Qwen3-0.6B

zeros inside large money or count numbers

A neuron in layer 8 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

The neuron fires on a zero in the middle of a big number written with commas, like the second zero in '$350,000'. It is a narrow pattern about digits, and turning it up did not visibly change the sample text.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. 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

    of effectiveness) each year produces over $350,000 more income for his/her students compared to an average teacher. On the the

  2. #2 peak

    our organization in Chicago was called upon to assemble some detailed cost figures from butter manufacturers whose 77 plants produced almost 150,000,000 pounds of butter annually. These figures reflected accurately the big increases in the

  3. #3 peak

    institute said its efforts had provided research opportunities to 15,000 undergraduates, attracted more women and minorities and assisted nearly 50,000 science teachers and students at local schools across the country engaged in efforts to improve

  4. #4 peak

    5,000 in today’s dollars. Today, that home—even without major remodels or updates—sells for about $350,000.⏎Compare this with nearby Lakeview, NY, a predominantly African American town

  5. #5 peak

    found itself subject to a migrating swarm of nearly 20,000 honey bees. The species is close to extinction, so the crew on base

  6. #6 peak

    910.1200(g)), revised in 2012, requires that the chemical manufacturer, distributor, or importer provide Safety Data Sheets (SDSs) (formerly MSDSs or Material Safety Data Sheets) for

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: [A] This suggests that I have been thinking about lately, and [B] could be something. This might be something. I am not sure about, and [X] is a particular case of [B], and [X]

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

    Regulations, 1992: Main aim of this legislation is establishing tour operators' responsibility towards customers and providing remedies to them (Gibson, 2016).⏎1.2 Surface, sea and air transport law in

  2. #2 peak

    : Gwartney, J. D., Stroup, R. L., Sobel, R. S., & Macpherson, D. A. (2018). Economics: Private⏎and public choice (16

  3. #3 peak

    , word-of-mouth marketing, consumption patterns, information traits, credibility.⏎ANTHEUNIS, M.L., & REIJMERSDAL, E.A. (2012). Social connections and the persuasiveness of viral campaigns in

  4. #4 peak

    <|endoftext|>FROM ISSUE NUMBER 15 ~ SPRING 2013 GO TO TABLE OF CONTENTS⏎The

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. L4:0 suppresses
  2. L7:323 excites
  3. L4:1272 suppresses
  4. L2:882 excites
  5. L7:1889 excites
  6. L1:122 suppresses 'World War II' and the 'II' after 'War'
  7. L3:21 weakly excites
  8. L3:1057 weakly excites
  9. L4:179 weakly suppresses
  10. L2:1169 weakly suppresses
  11. L3:48 weakly excites
  12. L6:284 weakly excites
  13. L5:875 weakly suppresses
  14. L4:2187 weakly excites
  15. L4:2561 weakly suppresses
  16. L6:4 weakly suppresses
  17. L0:735 weakly excites
  18. L7:2588 weakly excites
  19. L4:400 weakly excites
  20. L7:1376 weakly excites
  21. L3:188 weakly excites
  22. L2:2469 weakly suppresses
  23. L1:580 weakly suppresses
  24. L4:674 weakly excites

Feeds later layers

  1. L9:898 weakly excites
  2. L12:478 weakly excites
  3. L9:307 weakly excites
  4. L13:1121 weakly suppresses
  5. L9:826 weakly suppresses
  6. L19:247 weakly excites
  7. L9:1185 weakly suppresses
  8. L10:699 weakly excites
  9. L26:2543 weakly suppresses
  10. L11:222 weakly excites
  11. L14:820 weakly suppresses
  12. L12:3061 weakly suppresses
  13. L10:1397 weakly excites
  14. L14:2397 weakly suppresses
  15. L17:1017 weakly suppresses
  16. L23:93 weakly excites
  17. L10:208 weakly suppresses
  18. L9:314 weakly suppresses
  19. L16:1121 weakly excites
  20. L12:2965 weakly suppresses
  21. L12:2442 weakly suppresses
  22. L12:761 weakly suppresses
  23. L15:1718 weakly suppresses
  24. L12:2584 weakly suppresses

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.

1 values between -1.91 and -1.751 values between -1.75 and -1.592 values between -1.59 and -1.445 values between -1.44 and -1.283 values between -1.28 and -1.1214 values between -1.12 and -0.9618 values between -0.96 and -0.8027 values between -0.80 and -0.6439 values between -0.64 and -0.48110 values between -0.48 and -0.33427 values between -0.33 and -0.171350 values between -0.17 and -0.011296 values between -0.01 and 0.15266 values between 0.15 and 0.31123 values between 0.31 and 0.4781 values between 0.47 and 0.6349 values between 0.63 and 0.7826 values between 0.78 and 0.9425 values between 0.94 and 1.1011 values between 1.10 and 1.267 values between 1.26 and 1.428 values between 1.42 and 1.584 values between 1.58 and 1.742 values between 1.74 and 1.894 values between 1.89 and 2.052 values between 2.05 and 2.212 values between 2.21 and 2.373 values between 2.53 and 2.691 values between 3.00 and 3.161 values between 4.27 and 4.43offnegativepositiveoften

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. metry
  2. ละ
  3. łów
  4. arkin
  5. ourcem
  6. overe
  7. appers
  8. ót
  9. ictionaries
  10. trackers

Suppresses

  1. 枨
  2. uda
  3. domest
  4. 优化
  5. room
  6. 发育
  7. 位
  8. 膜
  9. Moder
  10. membr