Neuron Explorer

Qwen3-0.6B

commas inside lists before 'etc'

A neuron in layer 26 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 a comma that sits in the middle of a list, such as 'curb cuts, ramps, restrooms, etc.' It is at its strongest right where the next item is about to come or the list is about to end. When it was pushed up, the model got stuck repeating words like 'either', and when pushed down it repeated 'etc'.

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

    XML-based data on the server, and to transform it into plain-vanilla HTML, XHTML, or other formats (PDF, plain text, rich text files, and so forth) for delivery to Web clients. This helps sidestep the

  2. #2 peak

    3. Belonging/Social—family, affection, friends, relationships, etc.⏎4. Esteem—achievement, recognition, independence, responsibility, reputation, respect, etc.⏎5. Self-Actualization—realizing potential, self

  3. #3 peak

    .⏎b. (of two or more attributes of an object) unable to belong to the object simultaneously; inconsistent.⏎5. (of positions, functions, ranks, etc.) unable to be held simultaneously by one person.⏎6. Med.of or

  4. #4 peak

    00-342-3736.⏎Ask such questions as whether the service location is physically accessible (curb cuts, ramps, restrooms, etc.) as well as whether there is a Telecommunication Device for the Deaf

  5. #5 peak

    problems, finding the roots of nonlinear equations, functions written in Octave's own language, or by using dynamically loaded modules written in C, C++, Fortran, or other languages.⏎Donations to support the software can be made at https://

  6. #6 peak

    conservation of electric charge. There are also many approximate conservation laws, which apply to such quantities as mass, parity, lepton number, baryon number, strangeness, hypercharge, etc. These quantities are conserved in certain classes

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: etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc etc等等等等等等等等等等等等等等等等等等等等等等等

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

    wind 雨 : rain 声 : sound⏎Meaning for Line 4 花 : flower 落 : fall 知 : know 多 : how many 少 : missing⏎Background Story A

  2. #2 peak

    al., 2008), it’s obviously extremely successful, although it should be noted that the virulence of each Wolbachia strain is different (Min & Benzer, 1997), and each species is has a

  3. #3 peak

    at DARPA.⏎The first connection came 4 years later in October 1969, between UCLA (Los Angeles) & the Stanford Research Institute in Menlo Park, California (.. which would become part of Silicon Valley, just south

  4. #4 peak

    Halton Chesters, Northumberland)⏎The Romano-British name for the Rudchester fort is recorded in two of the major classical geographical works, the Notitia Dignitatum of the late-fourth/early-fifth centuries and

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. L24:1087 strongly suppresses
  2. L21:2512 strongly excites
  3. L25:162 strongly excites
  4. L18:2938 strongly excites
  5. L19:1265 strongly suppresses
  6. L23:784 suppresses
  7. L23:489 suppresses
  8. L22:2580 suppresses
  9. L21:2248 suppresses
  10. L25:454 excites
  11. L22:311 excites
  12. L24:1528 suppresses
  13. L22:2309 suppresses
  14. L25:2691 suppresses
  15. L20:1731 suppresses
  16. L18:2722 suppresses
  17. L17:2209 excites
  18. L25:409 suppresses
  19. L22:201 weakly excites
  20. L16:1139 weakly suppresses 'and' in lists of places or names
  21. L22:1259 weakly suppresses
  22. L21:2160 weakly excites
  23. L20:3010 weakly suppresses
  24. L16:2649 weakly suppresses

Feeds later layers

  1. L27:1233 suppresses
  2. L27:1593 weakly suppresses
  3. L27:922 weakly suppresses
  4. L27:2774 weakly suppresses
  5. L27:759 weakly excites
  6. L27:385 weakly suppresses
  7. L27:1677 weakly suppresses
  8. L27:1458 weakly suppresses
  9. L27:942 weakly excites
  10. L27:694 weakly suppresses
  11. L27:2111 weakly excites
  12. L27:808 weakly excites
  13. L27:1388 weakly excites
  14. L27:1364 weakly suppresses
  15. L27:979 weakly suppresses
  16. L27:1317 weakly suppresses
  17. L27:2017 weakly suppresses
  18. L27:168 weakly excites
  19. L27:1556 weakly excites
  20. L27:1061 weakly suppresses
  21. L27:788 weakly excites
  22. L27:1357 weakly suppresses
  23. L27:268 weakly suppresses
  24. L27:446 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.

3 values between -4.37 and -3.662 values between -3.66 and -2.951 values between -2.95 and -2.243 values between -2.24 and -1.5312 values between -1.53 and -0.82530 values between -0.82 and -0.103013 values between -0.10 and 0.61305 values between 0.61 and 1.3213 values between 1.32 and 2.034 values between 2.03 and 2.742 values between 2.74 and 3.454 values between 3.45 and 4.162 values between 4.16 and 4.883 values between 4.88 and 5.592 values between 6.30 and 7.013 values between 7.01 and 7.721 values between 7.72 and 8.431 values between 9.14 and 9.851 values between 10.57 and 11.281 values between 11.28 and 11.991 values between 21.94 and 22.661 values between 23.37 and 24.08offnegativepositiveoften

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. arsed
  2. onna
  3. ually
  4. either
  5. either
  6. EITHER
  7. gresql
  8. |#
  9. >|
  10. >{

Suppresses

  1. 等
  2. etc
  3. etc
  4. 等等
  5. 等内容
  6. 等方面
  7. 等领域
  8. 等地
  9. 等问题
  10. 等因素