Neuron Explorer

Qwen3-0.6B

'the' in historical names

A neuron in layer 23 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 word "the" that sits inside a famous name, like "Ivan the Terrible" or "William the Conqueror". It points toward words for "her" and away from "he" or "him". When turned up, the text shifted toward female pronouns and when turned down it kept repeating "he".

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

    that is, agree to a Russian sphere of influence in Eastern Europe . . .⏎The second line is of an economic nature. It originated in trade contacts between Ivan the Terrible and Elizabeth I, the Queen

  2. #2 peak

    . In particular, polygamy is the practice of a man taking more than one wife, while polyandry is the practice of a woman taking more than one husband. Polygamy is a common model of marriage in some parts of the

  3. #3 peak

    also installed sticky mats on the floor<|endoftext|>Achilles was the greatest warrior to have ever lived. He was the Golden Warrior of the Bronze Age. Even Alexander the Great admired and emulated Achilles throughout his life, sleeping with a copy of the

  4. #4 peak

    which since 1981 has been strongly linking Germany, for example, first with the USSR and later with Putin’s Russia.⏎From the times of Ivan the Terrible till this day, Russia has been trying to present itself to the Western

  5. #5 peak

    , the day of St. Stephen the Martyr, everyone goes out and visits friends and relatives, weather permitting. One

  6. #6 peak

    saw an American F-86 being attacked by a Russian MiG-15 and went in pursuit.⏎Captain Semyon Alekseyevich Fedorets, commanding the 913th IAP (Istrebitel

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: he is a 55-year-old, he is a he, he is a he, he is a he... he is a he... he is a he... he is a he... he is a he... he is a he

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

    .⏎When Hamlet is engaged in a discussion with Gertrude, they are sparring with words. She is talking about one thing, he another. She tells him he has offended his father very much (meaning Claudius, who is not

  2. #2 peak

    in the past. "I stop there yesterday," though ungrammatical, gets the message across.⏎"He is here." The agreement between the third person subject "he" and the third person of the verb "is" is basically redundant.

  3. #3 peak

    a controversial figure even within the communist world. People who were his friends and those who were opposed to him held differing opinions on him and had varying interpretations and explanations of his motives for participating in the revolutions against capitalism and imperialism all over the world.⏎

  4. #4 peak

    . After his great struggle, we got freedom on 15th August 1947. He played a vital role in our independence, so we called him as the father of the nation. All our Indians with great love and affection

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. L22:1623 strongly excites
  2. L19:666 excites
  3. L17:704 excites
  4. L16:748 suppresses
  5. L16:377 excites
  6. L17:428 excites
  7. L18:438 suppresses
  8. L20:2754 suppresses
  9. L15:342 weakly suppresses
  10. L18:1039 weakly suppresses
  11. L20:2033 weakly excites
  12. L21:2560 weakly suppresses
  13. L22:520 weakly excites
  14. L6:1492 weakly suppresses
  15. L22:1981 weakly suppresses
  16. L14:1022 weakly suppresses
  17. L19:1375 weakly suppresses
  18. L5:2607 weakly excites
  19. L3:2041 weakly suppresses
  20. L12:1523 weakly excites
  21. L11:1647 weakly excites
  22. L1:911 weakly excites
  23. L17:739 weakly suppresses
  24. L21:2539 weakly excites

Feeds later layers

  1. L27:578 strongly suppresses
  2. L26:869 strongly suppresses word before him or he
  3. L24:477 suppresses
  4. L24:2385 suppresses
  5. L27:813 excites
  6. L27:1244 suppresses
  7. L26:1143 weakly excites
  8. L24:1120 weakly excites
  9. L24:1050 weakly excites
  10. L24:1615 weakly excites
  11. L25:597 weakly suppresses
  12. L26:1127 weakly suppresses
  13. L27:857 weakly suppresses
  14. L27:1190 weakly suppresses
  15. L25:310 weakly suppresses
  16. L27:1865 weakly suppresses
  17. L27:563 weakly suppresses
  18. L25:2043 weakly suppresses
  19. L26:1061 weakly excites
  20. L27:1026 weakly suppresses
  21. L26:1038 weakly suppresses
  22. L25:527 weakly suppresses
  23. L26:2320 weakly suppresses
  24. L26:2643 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 -20.89 and -20.001 values between -17.34 and -16.462 values between -14.68 and -13.802 values between -11.14 and -10.254 values between -10.25 and -9.363 values between -9.36 and -8.483 values between -7.59 and -6.706 values between -6.70 and -5.8210 values between -5.82 and -4.938 values between -4.93 and -4.0412 values between -4.04 and -3.1624 values between -3.16 and -2.2764 values between -2.27 and -1.38197 values between -1.38 and -0.502773 values between -0.50 and 0.39693 values between 0.39 and 1.2848 values between 1.28 and 2.1623 values between 2.16 and 3.0515 values between 3.05 and 3.948 values between 3.94 and 4.823 values between 4.82 and 5.711 values between 5.71 and 6.601 values between 6.60 and 7.481 values between 7.48 and 8.371 values between 10.15 and 11.031 values between 12.81 and 13.693 values between 13.69 and 14.58offnegativepositiveoften

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. her
  2. 她的
  3. 她
  4. 她在
  5. 让她
  6. 但她
  7. 她们
  8. 跟她
  9. 她说
  10. Her

Suppresses

  1. him
  2. 他
  3. he
  4. 他会
  5. 他对
  6. 他也
  7. 他又
  8. 他就
  9. his
  10. 所以他