Neuron Explorer

Qwen3-0.6B

end-of-document boundary token

A neuron in layer 19 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 marker that separates one web page from the next, the point where an unrelated new text begins. When it was turned up, the model tended to start writing hash symbols (#) in a row.

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

    considered a simple problem, one that a fifth grader may find in a math book. What about the problems we face in adulthood? (read more here)<|endoftext|>Задание от гостя:Размещено 13.

  2. #2 peak

    . The result should be about 40. 000 graves.⏎You do not have the required permissions to view the files attached to this post.<|endoftext|>This overview outlines the tasks performed by field biologists, which include discovering what’s

  3. #3 peak

    shown that planets orbit at least 5 to 10 percent of all stars. But how many of these planets are Earth-size, and possibly Earth-like?<|endoftext|>& Tornado Alley⏎Jan Mikusinski developed an operational calculus which is relevant

  4. #4 peak

    be aggregated.⏎Use the app above to select a dataset with whole numbers. The app will output the dataset's stem-and-leaf plot once Submit is clicked.<|endoftext|>Americans with Disabilities Act (ADA)⏎The Americans with Disabilities Act (ADA) provides

  5. #5 peak

    to the development of camera lens.⏎Social media websites such as Twitter and facebook have provided digital photography enthusiasts a method to share their photography with family and friends. However<|endoftext|>Bread and circuses⏎"Bread and circuses" (or bread and

  6. #6 peak

    Telecommunication Device for the Deaf (TDD) and parking for people with disabilities. You may also ask if the practitioner makes house calls.⏎Your file contains<|endoftext|>The United States just won the gold medal in curling for the first time in

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 are the different possible consequences of the different different possible different different different different different different different different different of the following things: Maybe a different structure of the form of the possible consequences of the different possible different different of the different different different of

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

    accorded full status as electors in August 1920 by the Nineteenth Amendment to the federal Constitution, the...⏎Users without a subscription are not able to see the full⏎to access all content.<|endoftext|>Great Slaty

  2. #2 peak

    Vassilyevich (1928-) Russian pilot cosmonaut. Flew on Soyuz 7, Soyuz 16. More...⏎Grechko Grechko, Georgi Mikhailovich (1931

  3. #3 peak

    -835-5322 (TTY) or at email@example.com.<|endoftext|>Table of Contents :⏎Top Suggestions Senior Kindergarten Writting Worksheet :⏎Senior Kindergarten Writting Worksheet In this worksheet your students will practice their handwriting

  4. #4 peak

    number words as they follow directions on this cute worksheet. Other freebies are available to download at: www.creativeclassroomconnections.com.⏎Freebies , FREE, Pre-K and Kindergarten Fine Motor Skills.⏎Days of the Week Worksheets⏎Apple

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. L18:379 suppresses
  2. L17:833 suppresses
  3. L18:2746 suppresses
  4. L18:2722 suppresses
  5. L15:2229 excites
  6. L18:3036 excites
  7. L18:2905 excites
  8. L17:1490 suppresses
  9. L11:249 weakly excites
  10. L16:1554 weakly suppresses
  11. L13:1056 weakly excites end of document boundary tokens
  12. L18:2873 weakly suppresses
  13. L14:67 weakly suppresses
  14. L16:211 weakly suppresses
  15. L17:95 weakly suppresses
  16. L18:1028 weakly suppresses
  17. L18:1640 weakly suppresses
  18. L4:19 weakly suppresses
  19. L18:1192 weakly suppresses
  20. L6:16 weakly excites
  21. L2:2305 weakly suppresses
  22. L18:712 weakly excites
  23. L1:1692 weakly excites
  24. L14:1181 weakly suppresses

Feeds later layers

  1. L26:1040 strongly excites
  2. L23:677 strongly excites
  3. L25:907 strongly excites
  4. L20:356 excites
  5. L27:952 excites
  6. L22:284 excites
  7. L20:320 excites
  8. L22:351 excites
  9. L24:121 excites
  10. L22:161 excites
  11. L23:263 excites
  12. L26:855 excites
  13. L26:449 suppresses
  14. L21:1164 excites
  15. L27:432 weakly excites
  16. L22:731 weakly suppresses
  17. L25:692 weakly excites
  18. L22:1337 weakly suppresses
  19. L25:2352 weakly excites
  20. L23:100 weakly suppresses comma or full stop ending a clause
  21. L25:161 weakly excites
  22. L26:648 weakly suppresses
  23. L24:339 weakly excites
  24. L26:235 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 -1.80 and -1.584 values between -1.58 and -1.3615 values between -1.36 and -1.1438 values between -1.14 and -0.9273 values between -0.92 and -0.70238 values between -0.70 and -0.48667 values between -0.48 and -0.262144 values between -0.26 and -0.04708 values between -0.04 and 0.1811 values between 0.18 and 0.401 values between 0.62 and 0.841 values between 1.72 and 1.941 values between 1.94 and 2.161 values between 4.36 and 4.582 values between 4.58 and 4.801 values between 5.46 and 5.681 values between 6.12 and 6.341 values between 6.78 and 7.00offnegativepositiveoften

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. #
  2. //
  3. #
  4. //
  5. <!--
  6. #
  7. /#
  8. #get
  9. ());//
  10. #create

Suppresses

  1. ')):⏎
  2. 的重要
  3. ioned
  4. "){⏎
  5. '):⏎
  6. [${
  7. ,strong
  8. ilogue
  9. 的不同
  10. 的人来说