Neuron Explorer

Qwen3-0.6B

the digit 3 inside numbers

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 digit 3 when it appears inside a number, such as the 3 in a Psalm reference, a percentage, a word count, or a year like 2015. Turning it up made the model's continuations drift into unrelated topics like 3D objects and weather, so the effect on the text is not easy to read.

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

    6:1-13; Psalms 1:1-6; Psalms 2:1-12; Psalms 3:1-8; Psalms 4:1-8; Ps

  2. #2 peak

    of 1009 specimens were reported as positive for influenza viruses, 685 (67.9%) were typed as influenza A, 324

  3. #3 peak

    be assessed by means of a combination of one 1000 word assessed essay [15%], one 2000 word assessed essay [35%] and one two-hour unseen written examination paper in which the student will

  4. #4 peak

    83 cases with known vaccination status, 87% were unvaccinated, 8% were vaccinated with one dose of measles-containing vaccine, 3% were vaccinated with two or more doses, and 2% were vaccinated with

  5. #5 peak

    /her own and does not want to move to Curry Island with his/her...⏎This section contains 570 words|⏎(approx. 2 pages at 300 words per page)<|endoftext|>Of all the technology featured at CES this week

  6. #6 peak

    status silver objects.⏎Elements of the hoard will be on display for the first time at King’s Museum, University of Aberdeen from 20 January to 31 May 2015

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 is the first non-1010 to 101010 to the 1010...?...? How do I proceed? How do I write the terms?? A... for the first,

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

    6 lay persons. Participants were presented with a questionnaire using either the term intellectual disability or mental retardation and asked to report on stereotypes, prejudice, and discrimination in response to a vignette. Moderate levels of stigma were found. An indirect effect was

  2. #2 peak

    . My specific interests are related to species’ responses to environmental change, and alpine plants provide an incredible system to understand changes in range margins and community composition in response to global change. In the European Alps, plant species are shifting their range to

  3. #3 peak

    anatomy of a spacesuit⏎A spacesuit is made up of practically a half dozen completely different elements and may have as much as 16 layers, in response to NASA.⏎NASA revealed a floor prototype of the brand new exploration extraveh

  4. #4 peak

    logical links in the /dev directory. The devfsadm command also maintains the path_to_inst instance database.⏎Updates to the /dev and /devices directories in response to dynamic reconfiguration events or file system accesses are handled by devfsadmd

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. L4:347 strongly excites
  2. L16:134 strongly excites
  3. L17:994 strongly excites
  4. L0:1108 strongly suppresses
  5. L16:671 strongly excites
  6. L2:1774 strongly excites
  7. L2:1866 suppresses
  8. L16:112 excites
  9. L16:260 excites
  10. L10:887 suppresses the B in vitamin B names
  11. L16:374 excites
  12. L17:134 excites
  13. L10:715 excites
  14. L7:2095 excites
  15. L17:253 suppresses
  16. L10:1473 suppresses
  17. L17:594 excites
  18. L16:215 suppresses
  19. L15:129 suppresses
  20. L16:56 excites
  21. L15:82 suppresses
  22. L16:827 suppresses
  23. L15:96 suppresses
  24. L5:1984 excites

Feeds later layers

  1. L20:1308 strongly excites the digit 3 inside years
  2. L19:196 strongly suppresses
  3. L19:647 strongly excites
  4. L21:1189 excites
  5. L22:1550 suppresses
  6. L21:252 suppresses
  7. L21:815 suppresses
  8. L21:325 suppresses
  9. L20:411 suppresses
  10. L20:133 suppresses
  11. L20:247 suppresses
  12. L19:1021 suppresses
  13. L20:183 suppresses
  14. L19:322 suppresses
  15. L19:1615 suppresses
  16. L20:1310 suppresses
  17. L21:2715 suppresses
  18. L20:211 suppresses
  19. L25:2807 excites
  20. L22:715 excites
  21. L19:593 suppresses
  22. L19:879 suppresses
  23. L19:2848 suppresses
  24. L19:2826 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.

2 values between -3.34 and -2.3751 values between -2.37 and -1.40622 values between -1.40 and -0.443190 values between -0.44 and 0.5315 values between 0.53 and 1.507 values between 1.50 and 2.461 values between 2.46 and 3.434 values between 3.43 and 4.392 values between 9.23 and 10.191 values between 11.16 and 12.121 values between 15.99 and 16.951 values between 17.92 and 18.892 values between 22.75 and 23.721 values between 23.72 and 24.681 values between 24.68 and 25.651 values between 26.62 and 27.581 values between 28.55 and 29.521 values between 31.45 and 32.412 values between 32.41 and 33.382 values between 34.35 and 35.31offnegativepositiveoften

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. rd
  2. -quarters
  3. rd
  4. arters
  5. →rd
  6. مائة
  7. cheers
  8. RD
  9. lectric
  10. rbrace

Suppresses

  1. Warp
  2. GetUser
  3. HttpResponse
  4. GetString
  5. ọi
  6. rag
  7. 률
  8. SError
  9. CAS
  10. ivant