Neuron Explorer

Qwen3-0.6B

'the' or 'of' after a role or title

A neuron in layer 25 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 the small word 'the' (or 'of') right after a phrase that sets up a job, title or role, like 'is the Messenger', 'known as the' or 'Minister of'. Turning it up made the model's text drift into unrelated topics and repeat itself, so the effect on the writing was messy.

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

    them as false gods.⏎The second part of this declaration is that we believe that the Holy Prophet Muhammad, may peace and blessings of Allah be upon him, is the Messenger of Allah. Since he has come for the final, universal religion of Islam

  2. #2 peak

    .D. 697-707, and was one of the great poets of the early days of Japan; he is known as the rival of Akahito Yamabe (see next verse), and after death was

  3. #3 peak

    did reach Istanbul in 1918, Russia had made a separate peace with Germany and declared itself the enemy of all bourgeois states, France and Britain prominent among them.⏎What made you

  4. #4 peak

    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 called him as

  5. #5 peak

    came to India in 1917 after acquiring a high degree of doctorate in economics from Britain.⏎According to his agreement, he worked as the Minister of Finance in the princely state of Baroda for some time. But he did

  6. #6 peak

    greater now than it was before that Revolution which sought to bring about Islamist rule across the Arab world with the Shia Ayatollah Khomeini attempting to act as the leader of worldwide Islam. Khomeini hoped that Shi'ites in Iraq would

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: what if the "streamery" (ery, streamery) streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery streamery stream

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

    , that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.⏎Locke didn’t stop there, however

  2. #2 peak

    , that all Men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness." Government's place in our lives is to provide us the basic God given

  3. #3 peak

    the interplay of human emotions, thoughts, and actions by taking mythological metaphors. Kamayani has personalities like Manu, Ida and Śraddhā who are found in the Vedas. The great deluge described in

  4. #4 peak

    could have been avoided⏎Jefferson tried to explain as best he can the philosophical principles of the war and of the DOI⏎life, liberty, and the pursuit of happiness⏎what Jefferson said that every human being had

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. L20:1391 excites
  2. L7:247 suppresses
  3. L24:8 weakly suppresses
  4. L22:589 weakly excites
  5. L10:315 weakly excites
  6. L18:967 weakly suppresses
  7. L2:562 weakly suppresses
  8. L19:1228 weakly excites
  9. L24:573 weakly excites
  10. L10:2738 weakly suppresses
  11. L10:1042 weakly suppresses
  12. L19:827 weakly suppresses
  13. L20:1089 weakly suppresses
  14. L7:1314 weakly suppresses
  15. L4:1380 weakly excites
  16. L21:241 weakly excites
  17. L24:75 weakly suppresses
  18. L20:1159 weakly excites
  19. L19:175 weakly excites
  20. L20:677 weakly excites
  21. L15:1274 weakly excites
  22. L16:1378 weakly suppresses
  23. L10:2370 weakly suppresses
  24. L10:934 weakly suppresses

Feeds later layers

  1. L27:1972 weakly suppresses
  2. L26:2334 weakly suppresses the 'de' syllable starting words
  3. L26:1717 weakly excites
  4. L27:2522 weakly excites
  5. L27:2446 weakly excites
  6. L26:2499 weakly suppresses
  7. L26:1655 weakly excites
  8. L26:2460 weakly excites
  9. L26:2257 weakly excites
  10. L26:901 weakly excites
  11. L26:1719 weakly excites
  12. L26:2836 weakly excites
  13. L26:760 weakly excites
  14. L27:1731 weakly excites
  15. L26:2187 weakly excites
  16. L26:1376 weakly suppresses
  17. L27:984 weakly suppresses
  18. L26:1763 weakly excites
  19. L26:2230 weakly excites
  20. L27:2751 weakly suppresses
  21. L27:3018 weakly suppresses
  22. L27:1959 weakly excites
  23. L27:2299 weakly suppresses
  24. L27:1384 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.

1 values between -22.44 and -21.381 values between -17.14 and -16.071 values between -12.89 and -11.832 values between -11.83 and -10.775 values between -9.71 and -8.652 values between -8.65 and -7.594 values between -7.59 and -6.539 values between -6.53 and -5.4725 values between -5.47 and -4.4121 values between -4.41 and -3.3570 values between -3.35 and -2.29148 values between -2.29 and -1.23896 values between -1.23 and -0.172211 values between -0.17 and 0.90264 values between 0.90 and 1.9688 values between 1.96 and 3.0250 values between 3.02 and 4.0833 values between 4.08 and 5.1424 values between 5.14 and 6.2015 values between 6.20 and 7.269 values between 7.26 and 8.3211 values between 8.32 and 9.384 values between 9.38 and 10.443 values between 10.44 and 11.502 values between 11.50 and 12.561 values between 12.56 and 13.623 values between 13.62 and 14.681 values between 15.74 and 16.802 values between 17.86 and 18.922 values between 18.92 and 19.98offnegativepositiveoften

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. Sql
  2. сти
  3. DbType
  4. upper
  5. DataBase
  6. sql
  7. lower
  8. UserDao
  9. şı
  10. sql

Suppresses

  1. ery
  2. stream
  3. [string
  4. 一行
  5. mitter
  6. iral
  7. -stream
  8. Stream
  9. String
  10. iet