Neuron Explorer

Qwen3-0.6B

the word 'page' in references

A neuron in layer 24 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

It fires on the word 'page' when it appears in book references and printed material, such as 'page 119' or 'Page 5'. Turning it down made the model repeat the word 'page' over and over.

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

    (by 2020) the industries can’t put on the market products containing microplastic as defined by the REACH restriction.<|endoftext|>« Previous Page⏎Local anesthesia is a method to deliver pain control medicine to⏎a specific area

  2. #2 peak

    ...⏎however you want Santa to look!).⏎- Close template window when done printing to return to this screen.⏎- Set page margins to zero if you have trouble fitting the template on one page (FILE,

  3. #3 peak

    ner's Absalom, Absalom!,⏎1938: John Dos Passos's U.S.A. trilogy,⏎19<|endoftext|>A view can become a page or a block. There is no difference in work between a page and a block

  4. #4 peak

    out of her bag of verse she almost proposed to her hero in rhyme.George Meredith: The Egotist, iii.⏎To bell the cat. (See page 119, B

  5. #5 peak

    well as the viewers’s recognition of it. Simply put, aesthetic appeals concerns the level to which a masterpiece adds to the visual experience. A things is evaluated<|endoftext|>Page 5 – The late 1800s⏎Although more English and

  6. #6 peak

    tool doesn't work. Likewise, if the shape of a protein is warped, then it fails to do its job.⏎Spetner (Not by Chance, page 69) writes,⏎The number of amino acids in contact [when two

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: p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p.

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

    to facilitate children’s science exploration through everyday kitchen substances and experiences such as mixing, pouring, heating, and cooling. Full text» [PDF, 162KB]⏎Marvelous Explorations Through Science and Stories (MESS) and

  2. #2 peak

    son charts the cultural and political forces behind the overvaluation—and masculinization—of wakefulness in the United States.⏎2013 | 240 pages | Cloth $55.00⏎American History / Women's Studies

  3. #3 peak

    95 73 pages⏎The Adult Student's Guide to Survival & Success 2008 US$ 12.00 176 pages⏎Teacher Evaluation to Enhance Professional Practice 2000 US$

  4. #4 peak

    Interpretation found in the catalog.⏎April 1, 2003⏎by Baker Academic⏎Written in English⏎|The Physical Object|⏎|Number of Pages||180|⏎Hermeneutics gives the reader not only

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. L21:1701 excites
  2. L20:1875 suppresses
  3. L22:1765 suppresses
  4. L20:967 excites
  5. L22:2574 excites
  6. L22:1233 excites
  7. L21:912 suppresses
  8. L11:2091 weakly excites
  9. L18:951 weakly excites
  10. L6:1545 weakly excites
  11. L19:1920 weakly suppresses
  12. L7:2480 weakly excites
  13. L19:2060 weakly suppresses
  14. L20:307 weakly suppresses
  15. L20:1532 weakly excites
  16. L10:2153 weakly suppresses
  17. L16:1926 weakly excites
  18. L22:2046 weakly suppresses
  19. L23:2352 weakly suppresses
  20. L21:998 weakly suppresses
  21. L4:97 weakly suppresses
  22. L3:1654 weakly suppresses
  23. L23:465 weakly excites
  24. L19:2053 weakly excites

Feeds later layers

  1. L27:1788 strongly suppresses
  2. L25:394 suppresses
  3. L26:600 suppresses
  4. L25:584 suppresses
  5. L25:2569 suppresses
  6. L26:1591 weakly suppresses
  7. L26:1360 weakly suppresses
  8. L27:736 weakly suppresses
  9. L26:2349 weakly suppresses
  10. L27:420 weakly suppresses
  11. L27:417 weakly suppresses
  12. L27:2160 weakly suppresses
  13. L27:346 weakly suppresses
  14. L25:1155 weakly suppresses
  15. L25:841 weakly suppresses
  16. L26:1720 weakly suppresses
  17. L25:453 weakly excites
  18. L26:2696 weakly excites
  19. L25:1134 weakly suppresses
  20. L27:2983 weakly suppresses
  21. L27:2533 weakly suppresses
  22. L27:418 weakly suppresses
  23. L27:2774 weakly excites
  24. L26:382 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 -8.59 and -8.171 values between -6.94 and -6.531 values between -6.11 and -5.702 values between -5.70 and -5.291 values between -5.29 and -4.881 values between -4.88 and -4.472 values between -4.47 and -4.055 values between -3.64 and -3.234 values between -3.23 and -2.822 values between -2.82 and -2.405 values between -2.40 and -1.995 values between -1.99 and -1.5813 values between -1.58 and -1.1734 values between -1.17 and -0.76192 values between -0.76 and -0.342269 values between -0.34 and 0.071149 values between 0.07 and 0.48170 values between 0.48 and 0.8924 values between 0.89 and 1.318 values between 1.31 and 1.725 values between 1.72 and 2.133 values between 2.95 and 3.371 values between 3.37 and 3.783 values between 3.78 and 4.191 values between 4.19 and 4.602 values between 4.60 and 5.011 values between 5.43 and 5.841 values between 6.25 and 6.662 values between 7.49 and 7.90offnegativepositiveoften

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. earrings
  2. elong
  3. спор
  4. aea
  5. ública
  6. squad
  7. فجر
  8. SND
  9. نس
  10. ureka

Suppresses

  1. page
  2. page
  3. 页
  4. Page
  5. pages
  6. Page
  7. _page
  8. (page
  9. →page
  10. pages