Neuron Explorer

Qwen3-0.6B

unclear: weak, scattered activations

A neuron in layer 2 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 barely fires, and the places it fires on look unrelated: a word inside a hyphenated compound, 'Schedule' in drug and legal texts, and 'Action' in an organisation name. Nothing clear ties these together, and turning it up only made the model repeat words like 'company'.

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

    and eventually communism crumbled.⏎Walesa called for that persistence from his listeners. The message was only slightly different from his speech four years earlier at a College-sponsored event attended by, among others, local mayors and elected officials.⏎Things have

  2. #2 peak

    It grows from egg to adult in three stages, with its appearance changing during the molting process for each stage, noted Amy Korman, an extension educator based in Northampton County.⏎“Unless you are paying daily attention to changes in this insect

  3. #3 peak

    cell phone.⏎Jen Hensely, who works for the company installing them, explained: “People are on their phones all the time.<|endoftext|>National Conservative Political Action Committee⏎The National Conservative Political Action Committee (NCPAC; pronounced "nick

  4. #4 peak

    sympathetic dystrophy (RSD), fibromyalgia, and chronic headaches to determine whether ketamine infusions would effectively reduce their pain. Ketamine, a Schedule III drug, has previously been used successfully to treat chronic pain in adults. Schedule

  5. #5 peak

    , wetland plants and trees that grew near the ancient town.<|endoftext|>For decades, the rightwing gadfly Ed Blum, his group Students for Fair Admissions, and his shadowy rightwing funders have been trying to end affirmative action

  6. #6 peak

    ”⏎When the Public Health (Tobacco Control) Regulations Jamaica 2013 were being drafted, the issue of e-cigarettes was noted. The First Schedule referred to the banning of smoking in public places and carried specific reference to electronic cigarettes

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 can use a special kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind of kind

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

    РАБОТА ПО АНГЛИЙСКОМУ ЯЗЫКУ⏎1. Соотнесите значения слов и выражений, данных в левой колонке, с их английскими эквив

  2. #2 peak

    the three descriptions.⏎A. Ask a volunteer to leave the room. When s/he returns, s/he will be asked to see if s/he can guess an<|endoftext|>Allergies and

  3. #3 peak

    or is not clear. Often, they will propose revised sentences, a different subject organization, reordering of paragraphs—all to help the author effectively develop and convey his/her subject. They also challenge the author to make his/her research interesting and relevant.

  4. #4 peak

    six grade-appropriate math problems: one each of addition, subtraction, multiplication, division, fractions, and percents. I stressed "grade-appropriate" because single digit addition, for example,

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. L1:1057 suppresses
  2. L0:2942 weakly excites
  3. L1:222 weakly excites
  4. L0:1224 weakly suppresses
  5. L0:178 weakly suppresses
  6. L1:1797 weakly excites
  7. L1:1849 weakly suppresses
  8. L1:3041 weakly excites
  9. L0:1083 weakly suppresses
  10. L1:583 weakly suppresses
  11. L1:2354 weakly excites
  12. L1:1229 weakly excites
  13. L0:1061 weakly excites
  14. L1:1099 weakly excites
  15. L0:1634 weakly suppresses
  16. L1:3009 weakly suppresses
  17. L1:1464 weakly excites
  18. L1:1522 weakly suppresses
  19. L0:1013 weakly suppresses
  20. L0:2374 weakly excites
  21. L1:2463 weakly excites
  22. L0:2766 weakly excites
  23. L0:1301 weakly suppresses
  24. L1:536 weakly suppresses

Feeds later layers

  1. L3:1525 excites
  2. L9:446 suppresses
  3. L7:887 excites
  4. L17:664 excites
  5. L10:2432 weakly excites
  6. L9:1462 weakly excites
  7. L6:97 weakly excites "as well" after "as"
  8. L9:2913 weakly excites
  9. L4:267 weakly excites
  10. L12:2509 weakly suppresses
  11. L20:1597 weakly excites
  12. L16:903 weakly excites
  13. L8:1398 weakly excites
  14. L8:277 weakly excites
  15. L3:1262 weakly excites
  16. L8:464 weakly excites
  17. L6:464 weakly excites
  18. L10:1042 weakly excites
  19. L9:232 weakly suppresses
  20. L12:2050 weakly excites
  21. L10:664 weakly excites
  22. L3:1351 weakly excites
  23. L7:281 weakly excites
  24. L21:752 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.

3 values between -0.39 and -0.363 values between -0.36 and -0.334 values between -0.33 and -0.3111 values between -0.31 and -0.2813 values between -0.28 and -0.2522 values between -0.25 and -0.2258 values between -0.22 and -0.2069 values between -0.20 and -0.17126 values between -0.17 and -0.14193 values between -0.14 and -0.11252 values between -0.11 and -0.09351 values between -0.09 and -0.06430 values between -0.06 and -0.03564 values between -0.03 and -0.00601 values between -0.00 and 0.02375 values between 0.02 and 0.05240 values between 0.05 and 0.08194 values between 0.08 and 0.11142 values between 0.11 and 0.1390 values between 0.13 and 0.1656 values between 0.16 and 0.1947 values between 0.19 and 0.2225 values between 0.22 and 0.2413 values between 0.24 and 0.277 values between 0.27 and 0.305 values between 0.30 and 0.337 values between 0.33 and 0.353 values between 0.35 and 0.381 values between 0.38 and 0.411 values between 0.44 and 0.462 values between 0.68 and 0.71offnegativepositiveoften

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. pawn
  2. anto
  3. avig
  4. lene
  5. rance
  6. """⏎⏎⏎
  7. velte
  8. enty
  9. ocos
  10. 柰

Suppresses

  1. 定点
  2. amd
  3. Ơ
  4. 不上
  5. ETCH
  6. одном
  7. Churchill
  8. zano
  9. mó
  10. Champ