Neuron Explorer

Qwen3-0.6B

"of" inside titles and names

A neuron in layer 18 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 word "of" when it sits inside a title or a fixed name, like "The Diary of Anne Frank" or "Music of the Spheres". When it was turned up, the model's writing drifted into rambling, repeated phrases.

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

    the report. "The estimate is transparent and conservative," says particle physicist Georg Weiglein at the University of Durham, UK. "I ...<|endoftext|>The Diary of Anne Frank Essay written by l

  2. #2 peak

    in American History: Selected Documents Celebrating the Manuscript Division’s First 100 Years features drafts of Langston Hughes’ poem, “The Ballad of Booker T.” In this 1941 poem,<|endoftext|>Whenever the st

  3. #3 peak

    in her lifetime, a towering legacy left behind. Rest in peace, brave comrade,” she penned.⏎Former Chair of the<|endoftext|>The Kalima or the Declaration of Faith is two short sentences and is the essence of Islam. It is given below

  4. #4 peak

    5:13-16 (The Prayer of Faith)⏎Have the kids find James 5:13-16 in

  5. #5 peak

    equal to the sum of the squares of the sides containing the right angle." But Pythagoras had an all-inclusive vision, which also included the "music of the spheres" (the sound that no doubt had to be produced by the planets

  6. #6 peak

    the 13th USENIX Security Symposium, 2004.⏎Garfinkel, S. and Shelat, A., "Remembrance of Data Passed: A Study of Disk Sanitization Practices," IEEE Security and Privacy

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: for example, the famous mathematician, 1942, 1975, 1985, 1995, 1995, 2001, 201

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

    the speeds of two satellites placed in stable circular orbits of radius R and 2R respectively?⏎C) 1 / 2⏎D

  2. #2 peak

    's own Apple II.⏎Jobs' vision of "good design" meant not only building aesthetically pleasing machines, but devices that had a utility of function like user-friendly GUIs akin to those seen on early Macs.⏎Near

  3. #3 peak

    Like⏎their giant counterparts, Humboldt squid are enigmatic. No one has⏎seen them mate or lay eggs. No one has watched them develop from⏎egg to adult. No one knows how many exist.⏎These tentacled titans

  4. #4 peak

    of the Journal of the American Academy of Religion eschews this empirical approach. Twelve scholars of Hinduism three of them practicing Hindus attempt to answer this question by methodical analysis. None can be said to successfully respond to the question, but their

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. L15:1888 suppresses
  2. L16:1594 suppresses
  3. L11:2859 excites
  4. L11:1168 suppresses
  5. L14:935 suppresses
  6. L16:652 excites
  7. L17:30 excites
  8. L10:1644 suppresses
  9. L17:317 excites 'of' after a noun phrase
  10. L10:365 excites
  11. L5:493 suppresses
  12. L1:788 weakly excites
  13. L12:2554 weakly excites
  14. L10:155 weakly suppresses
  15. L17:147 weakly excites
  16. L17:1721 weakly excites 'by' introducing a method or means
  17. L4:23 weakly excites
  18. L16:1088 weakly excites 'of' after a noun phrase, as in 'of course'
  19. L17:188 weakly suppresses 'a' before a noun phrase
  20. L9:227 weakly suppresses
  21. L16:1286 weakly suppresses
  22. L6:16 weakly excites
  23. L6:352 weakly suppresses
  24. L16:413 weakly excites

Feeds later layers

  1. L19:1180 strongly excites
  2. L19:1023 suppresses
  3. L19:1427 suppresses
  4. L25:284 excites
  5. L21:1400 excites
  6. L25:2973 excites
  7. L19:2522 excites
  8. L19:2115 excites
  9. L19:626 excites
  10. L19:2593 excites
  11. L24:359 excites
  12. L27:250 excites
  13. L20:1047 weakly excites
  14. L24:443 weakly excites
  15. L24:360 weakly suppresses
  16. L23:79 weakly excites
  17. L23:258 weakly excites
  18. L25:2331 weakly excites
  19. L25:403 weakly excites
  20. L19:535 weakly excites
  21. L22:533 weakly excites
  22. L19:663 weakly suppresses
  23. L27:2096 weakly excites
  24. L23:1394 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 -7.65 and -7.251 values between -6.08 and -5.692 values between -5.29 and -4.903 values between -4.51 and -4.121 values between -4.12 and -3.732 values between -3.73 and -3.346 values between -3.34 and -2.946 values between -2.94 and -2.5510 values between -2.55 and -2.1613 values between -2.16 and -1.7714 values between -1.77 and -1.3821 values between -1.38 and -0.9958 values between -0.99 and -0.59446 values between -0.59 and -0.202045 values between -0.20 and 0.191023 values between 0.19 and 0.58202 values between 0.58 and 0.9730 values between 0.97 and 1.374 values between 1.37 and 1.765 values between 1.76 and 2.154 values between 2.15 and 2.541 values between 2.54 and 2.932 values between 2.93 and 3.321 values between 3.32 and 3.721 values between 3.72 and 4.112 values between 4.11 and 4.501 values between 5.67 and 6.072 values between 6.07 and 6.461 values between 7.63 and 8.02offnegativepositiveoften

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. dives
  2. itled
  3. erties
  4. ponents
  5. states
  6. .getError
  7. rates
  8. stances
  9. [val
  10. очных

Suppresses

  1. 芒
  2. 戶
  3. GORITHM
  4. 军
  5. OCUS
  6. 冈
  7. 솜
  8. 逊
  9. 户
  10. Axe