Neuron Explorer

Qwen3-0.6B

chinese characters inside mixed text

A neuron in layer 19 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron responds to Chinese characters, especially inside text that mixes Chinese with English or other languages, such as a bracketed Chinese term or a blank in a classic poem. It is most active on the character itself. Turning it up gave English text with no clear Chinese link, so the steering did not reveal much.

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

    systems, it also means that pollutants that originate indoors are getting trapped inside.⏎“People normally<|endoftext|>Spring Dawn by Meng Haoran 春眠不觉晓 处处闻啼茑 夜来风雨声 花落知多少⏎

  2. #2 peak

    scholar, politician and philosopher who lived in the Spring and Autumn period. The main body of the Confucian canon comprises the Four Books and the Five Classics (四書五經), texts which have been traditionally attributed to Confucius himself,

  3. #3 peak

    be an accessible topic for middle school students.19<|endoftext|>毋忘在莒(Pinyin: Wu Wang Zai Ju) is a famous Chinese idiom (成語, Pinyin: Cheng Yu). It is sort of like the Latin id

  4. #4 peak

    udoki 肥前國風土記 (first part of the 8th century) 10. The reading (furigana 振り假名) of those texts, although, probably does

  5. #5 peak

    : place 闻 : hear 啼 : cry 茑 : bird⏎Meaning for Line 3 夜 : night 来 : come 风 :

  6. #6 peak

    the learner<|endoftext|>n.交響樂 [an elaborate instrumental composition in three or more movements, similar in form to a sonata but written for an orchestra (管弦樂隊) and usually of far grander proportions and more varied elements.

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: the use of 360度点子云图图谱图谱子集集子集子集子集子集子集子集子集子集子集子集子集子集子集子集子

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

    法、分析法来进行研究,探讨泰国学生在学习汉语助动词中的困难及偏误情况,分析他们产生偏误的原因。

  2. #2 peak

    to be an accessible topic for middle school students.19<|endoftext|>毋忘在莒(Pinyin: Wu Wang Zai Ju) is a famous Chinese

  3. #3 peak

    wiki: In the past, syndromes were often named after the physician or scientist who identified and described the condition in an initial publication⏎ex. 唐氏

  4. #4 peak

    being explored and the most valuable part of the ship (the stern castle, where the captain’s cabin would have concealed the most precious cargos of all) has yet to be located.⏎Priceless Knowledge: Devon’s Bronze Age wreck⏎Salcombe

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. L18:811 excites
  2. L16:1293 suppresses
  3. L17:1590 suppresses
  4. L16:38 suppresses
  5. L17:257 suppresses
  6. L18:501 excites
  7. L9:146 excites
  8. L18:125 suppresses
  9. L8:763 excites
  10. L16:286 suppresses
  11. L18:1169 suppresses
  12. L8:423 suppresses
  13. L16:1432 excites
  14. L17:259 excites
  15. L3:305 excites
  16. L18:569 excites
  17. L11:1872 suppresses
  18. L16:2908 excites
  19. L14:374 suppresses
  20. L17:446 excites
  21. L2:1321 suppresses
  22. L7:631 excites
  23. L5:667 suppresses
  24. L10:576 suppresses

Feeds later layers

  1. L20:1476 suppresses
  2. L21:2623 suppresses
  3. L20:2613 suppresses
  4. L20:2426 suppresses
  5. L20:486 suppresses
  6. L20:272 suppresses
  7. L22:530 weakly suppresses
  8. L27:507 weakly excites
  9. L26:1539 weakly suppresses
  10. L27:535 weakly excites
  11. L20:824 weakly suppresses unclear: opposite pairs and citation digits
  12. L20:438 weakly excites
  13. L20:404 weakly suppresses
  14. L23:238 weakly excites
  15. L27:443 weakly excites
  16. L20:110 weakly suppresses
  17. L27:973 weakly excites
  18. L27:157 weakly excites
  19. L27:367 weakly excites
  20. L24:400 weakly excites
  21. L25:194 weakly suppresses
  22. L20:111 weakly suppresses
  23. L22:685 weakly suppresses
  24. L20:2553 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.

10 values between -0.71 and -0.4778 values between -0.47 and -0.221962 values between -0.22 and 0.021664 values between 0.02 and 0.26165 values between 0.26 and 0.5126 values between 0.51 and 0.751 values between 0.75 and 0.991 values between 0.99 and 1.241 values between 8.77 and 9.02offnegativepositiveoften

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. theirs
  2. _Syntax
  3. !_
  4. (Have
  5. пром
  6. зи
  7. Tehran
  8. 辘
  9. Garc
  10. ymes

Suppresses

  1. 事
  2. Pref
  3. 信
  4. ler
  5. <|endoftext|>
  6. 干
  7. 用
  8. 하
  9. 악
  10. bing