Neuron Explorer

Qwen3-0.6B

hyphen joining two words

A neuron in layer 2 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 lights up on a dash that joins two words into one compound, like 'drama-in-education' or 'match-the-phrase'. Turning it up made the model's text drift without any 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

    efforts of diverse Polish groups -Catholics, socialists, independents and others- to save Jewish lives despite great personal risks.⏎The London-based Polish Government-in-

  2. #2 peak

    Pathways: Children, Pedagogy and Play in the Age of Digital Technology⏎Keywords:pedagogy, children, play, day camp, drama-in-education⏎Children today are immersed in a digital environment that is marginalizing and,

  3. #3 peak

    to a glass of water, representing "tengo sed," or "I am thirsty."⏎The app also challenges its students to answer increasingly difficult typing, match-the-phrase-to-the-photo, listening, and conversational tests.⏎"With CatAcad

  4. #4 peak

    Extract Compressed Drupal⏎Extract the Drupal zip file in the /var/www/html directory using following command:⏎# cd /var/www/html⏎# unzip drupal-7.33.zip⏎It will create a folder as /var/www/html

  5. #5 peak

    chemical,petroleum,textile,and foodindustry sources.Reference mststeel/304-stainless-steel-properties/ results for this questionFeedback304 Stainless Steel Plate Midwest Steel

  6. #6 peak

    to show how the levels translate to real learning environments.⏎Through the long-term athlete development model, the Player Development Curriculum addresses the topic of proper practice/training-to-competition ratios. USA Basketball has defined competition as the act of competing against another team

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: it. In China, the market is our customers, are the main. So the problem. So the problem. So the problem. So the problem. So the problem. So the problem. So the problem. So the problem. So

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

    Labor, The Economics Daily, Demographics of Gulf War-era II veterans at https://www.bls.gov/opub/ted/2010/ted_20100805.htm (visited November 27,

  2. #2 peak

    , houses of worship, and religious schools. The full text of RLUIPA is available at http://www.usdoj.gov/crt/housing/housing_rluipa.htm. Below is a summary of the law’s key provisions, with

  3. #3 peak

    Wedlock,” Christina Gibson-Davis. Demography, July 2, 2014. DOI: 10.1007/s13524-014-0308-7<|endoftext|>

  4. #4 peak

    31(1), 131-132.⏎Available at: https://scholarworks.uni.edu/pias/vol31/iss1/25<|endoftext|>New images from NASA’s Mars Reconnaissance Orbiter show finger

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. L0:1301 suppresses
  2. L0:178 suppresses
  3. L0:1224 suppresses
  4. L1:2683 weakly excites
  5. L0:0 weakly excites
  6. L1:1096 weakly suppresses
  7. L1:2927 weakly suppresses
  8. L1:2568 weakly excites
  9. L1:576 weakly excites
  10. L1:146 weakly excites
  11. L1:824 weakly suppresses
  12. L0:1518 weakly excites
  13. L1:1699 weakly excites
  14. L1:2099 weakly excites
  15. L1:331 weakly suppresses
  16. L1:761 weakly suppresses
  17. L1:1355 weakly excites
  18. L0:46 weakly excites
  19. L1:1591 weakly suppresses
  20. L1:1492 weakly excites
  21. L1:2244 weakly suppresses
  22. L1:2940 weakly suppresses
  23. L1:1387 weakly suppresses
  24. L1:2801 weakly excites

Feeds later layers

  1. L11:102 excites
  2. L10:134 excites
  3. L4:58 excites
  4. L26:186 excites
  5. L9:11 weakly excites
  6. L3:195 weakly suppresses
  7. L4:0 weakly excites
  8. L17:927 weakly excites
  9. L3:48 weakly excites
  10. L17:364 weakly excites hyphen joining two words
  11. L25:1710 weakly suppresses
  12. L24:2894 weakly suppresses
  13. L16:975 weakly excites
  14. L6:1262 weakly excites
  15. L6:590 weakly excites
  16. L27:691 weakly suppresses
  17. L4:46 weakly excites
  18. L23:1418 weakly suppresses
  19. L6:1466 weakly excites
  20. L3:254 weakly excites
  21. L9:31 weakly excites
  22. L8:687 weakly excites
  23. L4:6 weakly excites
  24. L23:2325 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 -0.71 and -0.671 values between -0.67 and -0.631 values between -0.54 and -0.501 values between -0.50 and -0.465 values between -0.46 and -0.424 values between -0.42 and -0.386 values between -0.38 and -0.347 values between -0.34 and -0.306 values between -0.30 and -0.2616 values between -0.26 and -0.2231 values between -0.22 and -0.18100 values between -0.18 and -0.14178 values between -0.14 and -0.10391 values between -0.10 and -0.06867 values between -0.06 and -0.011111 values between -0.01 and 0.03685 values between 0.03 and 0.07311 values between 0.07 and 0.11123 values between 0.11 and 0.1547 values between 0.15 and 0.198 values between 0.19 and 0.235 values between 0.23 and 0.271 values between 0.27 and 0.311 values between 0.39 and 0.431 values between 0.88 and 0.92offnegativepositiveoften

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. ">@
  2. init
  3. ropa
  4. "><!--
  5. 操场
  6. 惯例
  7. ">⏎⏎⏎
  8. 师傅
  9. 誓言
  10. 无辜

Suppresses

  1. bose
  2. shire
  3. tele
  4. atori
  5. diffuse
  6. 雳
  7. borne
  8. ImageSharp
  9. multid
  10. ington