Neuron Explorer

Qwen3-0.6B

empty table cell separator bars

A neuron in layer 9 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

This neuron fires on the vertical bar symbols that start or separate empty cells in wiki-style tables, the spot where a new table cell is about to begin. Turning it up mostly pushed the text into lists of numbers, which fits the table style, but turning it down did not give a clear theme.

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

    to remove this template message)|⏎The Central Business District as seen from the top of the Carlton Centre⏎|Municipality||City of Johannesburg|⏎|• Total||1.09 km2 (0.42 sq mi)|⏎|

  2. #2 peak

    ||Announced||May 4, 1990|⏎|-||Restored||August 21, 1991|⏎|-||Total||64,589 km2 (124th)⏎2

  3. #3 peak

    , the following chart applies.⏎Fall 2001 Jewish Holiday Calendar⏎|Biblical Name||Modern Holiday||Hebrew Date||English Date|⏎|Feast of Trumpets (Lev. 23:23–

  4. #4 peak

    200 m||200000 mm|⏎|Numbers||Mm in m||M in mm|⏎Some another meter to mm

  5. #5 peak

    2:15|⏎|Last Modified:||11 May 2011 04:

  6. #6 peak

    ivut||678 - 687|⏎|Julkaisu||JOURNAL OF LANGUAGE TEACHING AND RESEARCH|⏎

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 know that the concept of a "Math is Beautiful" at the moment in the website's location, and that's what I was talking about. Let me know if you need more information. The question is related to the mathematical and

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

    , and Interactions⏎|Herb||Why People May Use This Herb||Examples of Side Effects||<|endoftext|>In the statements, It rains; it is cold

  2. #2 peak

    Succeeded by||Daniel François Malan|⏎3 September 1919 – 30 June 1924⏎|Preceded by||Louis Botha|⏎|Succeeded by||James Barry Munnik Hertz

  3. #3 peak

    reports, he continued "wielding his machete until the foe had<|endoftext|>|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|⏎Well-to-well tracer tests contribute

  4. #4 peak

    Plant Operator⏎Tasks | Tools & Technology | Knowledge | Skills | Abilities | Work Activities | Detailed Work Activities | Work Context | Job Zone | Education | Interests | Work Styles | Work Values | Wages & Employment | Job Openings⏎-

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. L7:335 excites
  2. L6:1691 suppresses
  3. L8:1384 suppresses
  4. L6:1056 weakly suppresses
  5. L5:809 weakly suppresses
  6. L6:1841 weakly suppresses
  7. L8:1394 weakly excites
  8. L6:2156 weakly excites
  9. L1:871 weakly excites
  10. L7:62 weakly excites
  11. L6:234 weakly suppresses
  12. L6:2650 weakly suppresses
  13. L5:1311 weakly excites
  14. L8:18 weakly suppresses
  15. L5:222 weakly suppresses
  16. L6:1081 weakly excites
  17. L7:2073 weakly suppresses
  18. L5:258 weakly excites
  19. L8:1023 weakly excites
  20. L5:1906 weakly excites
  21. L2:2952 weakly excites
  22. L7:77 weakly suppresses
  23. L4:947 weakly excites
  24. L2:493 weakly suppresses

Feeds later layers

  1. L11:1496 excites
  2. L12:1387 excites
  3. L18:1621 excites
  4. L11:2508 excites
  5. L12:2336 excites
  6. L13:2294 weakly excites
  7. L11:280 weakly excites
  8. L10:2274 weakly excites
  9. L15:1864 weakly excites
  10. L11:2434 weakly excites
  11. L24:401 weakly suppresses 'it' referring back to a noun
  12. L17:1547 weakly suppresses
  13. L17:2611 weakly excites
  14. L16:3014 weakly excites
  15. L16:1550 weakly excites
  16. L11:2383 weakly excites
  17. L22:729 weakly suppresses
  18. L15:2545 weakly excites
  19. L13:2370 weakly suppresses
  20. L13:2024 weakly suppresses
  21. L19:476 weakly excites
  22. L19:1367 weakly excites
  23. L24:411 weakly excites
  24. L19:2595 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.

2 values between -0.62 and -0.554 values between -0.55 and -0.495 values between -0.49 and -0.4214 values between -0.42 and -0.3641 values between -0.36 and -0.2982 values between -0.29 and -0.23158 values between -0.23 and -0.16369 values between -0.16 and -0.10827 values between -0.10 and -0.031407 values between -0.03 and 0.04546 values between 0.04 and 0.10268 values between 0.10 and 0.17100 values between 0.17 and 0.2345 values between 0.23 and 0.3021 values between 0.30 and 0.366 values between 0.36 and 0.433 values between 0.43 and 0.503 values between 0.50 and 0.561 values between 0.76 and 0.821 values between 0.82 and 0.891 values between 0.89 and 0.951 values between 0.95 and 1.021 values between 1.09 and 1.151 values between 1.87 and 1.941 values between 1.94 and 2.00offnegativepositiveoften

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. ystate
  2. 的动作
  3. 家中
  4. engers
  5. 标志
  6. же
  7. usterity
  8. illage
  9. หาย
  10. rei

Suppresses

  1. /cop
  2. LOC
  3. locale
  4. _memcpy
  5. .Win
  6. fits
  7. 是一款
  8. López
  9. LPARAM
  10. PHY