Neuron Explorer

Qwen3-0.6B

email address endings in examples

A neuron in layer 9 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 end of fake email addresses and web addresses, like the part right after the at sign in a placeholder email. When it was turned up, the model's writing mostly became repetitive or drifted into odd loops rather than showing a clear theme.

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

    how your school can get involved with Bush Builders contact firstname.lastname@example.org⏎Brothers Island tuatara⏎For just over six years, Wellington

  2. #2 peak

    firstname.lastname@example.org.<|endoftext|>The Musical Representation⏎How human musical experience emerges from the audition of

  3. #3 peak

    the publication for benchmarking.⏎To explore opportunities for collaboration with subject-matter experts on your assessment and evaluation efforts, contact PRA Well-Being at firstname.lastname@example.org.⏎This resource was first shared in 2019.<|endoftext|>Social

  4. #4 peak

    the Department of Construction, Chaoyang University of Technology, Taichung 41349, Taiwan (e-mail: firstname.lastname@example.org).⏎Cite:Ming-Ju Lee, Ming-Gin Lee, Y

  5. #5 peak

    wide variety of languages, catering for beginning students to background speakers.⏎Our Faculty Members⏎Mr John Norberry: email@example.com⏎Mrs Laura Wade: firstname.lastname@example.org⏎Quite often as Language teachers, we are asked why students should have to

  6. #6 peak

    8|⏎Article from a library⏎|Web site with no author|⏎The examples above are reproduced from https://owl.english.purdue.edu/owl/resource/747/22/⏎

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: The original problem: Let's say I am. I. J. K. Question asked by I. J. K. **Question** **Answer** **Answer** **Question** **Answer** **Answer** **Question

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

    op icthealth.nl. Link: https://ictandhealth.com/ucla-looks-into-rewiring-brain-in-order-to-help-with-depressions/news⏎UCLA looks into rewiring brain in order to help with depressions⏎

  2. #2 peak

    provider:|| Postal: Katajanokanlaituri 6B, 00160 Helsinki|⏎Web page: http://www.wider.unu.edu/⏎More information through EDIRC⏎Please report citation or reference errors

  3. #3 peak

    2018;15(2):57-61 DOI:http://dx.doi.org/10.4314/aas.v15i2.5⏎© 2018 Author. This

  4. #4 peak

    .org for more information. Updated June 25, 2015. (taken from http://genealogy.about.com/od/surname_meaning/p/boyle.htm ) BOYLE Last Name Meaning & Origin: A variant of

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. L5:79 suppresses
  2. L3:254 suppresses
  3. L1:70 excites
  4. L2:25 suppresses
  5. L2:464 suppresses
  6. L3:21 excites
  7. L0:36 excites
  8. L3:171 excites
  9. L3:871 suppresses
  10. L4:58 excites
  11. L0:45 excites
  12. L0:29 suppresses
  13. L2:9 suppresses sentence-final periods vs dots in code
  14. L2:1775 excites
  15. L8:143 suppresses
  16. L1:1079 suppresses
  17. L5:1304 weakly excites
  18. L8:687 weakly excites
  19. L2:1170 weakly suppresses
  20. L3:195 weakly suppresses
  21. L7:5 weakly excites
  22. L7:1302 weakly suppresses
  23. L7:1913 weakly excites
  24. L5:227 weakly suppresses

Feeds later layers

  1. L15:722 strongly suppresses
  2. L10:134 strongly excites
  3. L10:342 suppresses
  4. L14:817 excites
  5. L15:1057 excites
  6. L12:1740 suppresses
  7. L15:2259 suppresses
  8. L10:856 suppresses
  9. L12:142 excites
  10. L10:984 excites
  11. L10:909 suppresses
  12. L15:2425 excites
  13. L15:241 suppresses
  14. L10:250 weakly suppresses
  15. L17:1775 weakly suppresses
  16. L17:603 weakly excites inside organisation and product names
  17. L14:345 weakly excites
  18. L15:625 weakly excites
  19. L10:2083 weakly suppresses
  20. L14:254 weakly suppresses
  21. L10:1642 weakly suppresses
  22. L16:2296 weakly suppresses
  23. L20:494 weakly excites
  24. L10:2225 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 -1.19 and -1.142 values between -0.98 and -0.931 values between -0.71 and -0.662 values between -0.61 and -0.552 values between -0.55 and -0.504 values between -0.50 and -0.455 values between -0.45 and -0.391 values between -0.39 and -0.346 values between -0.34 and -0.296 values between -0.29 and -0.2321 values between -0.23 and -0.1859 values between -0.18 and -0.13117 values between -0.13 and -0.07241 values between -0.07 and -0.02461 values between -0.02 and 0.03527 values between 0.03 and 0.09550 values between 0.09 and 0.14467 values between 0.14 and 0.19406 values between 0.19 and 0.25326 values between 0.25 and 0.30242 values between 0.30 and 0.35169 values between 0.35 and 0.41115 values between 0.41 and 0.4664 values between 0.46 and 0.5147 values between 0.51 and 0.5730 values between 0.57 and 0.6215 values between 0.62 and 0.6710 values between 0.67 and 0.738 values between 0.73 and 0.782 values between 0.78 and 0.831 values between 0.89 and 0.94offnegativepositiveoften

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. 荆
  2. ipping
  3. Truman
  4. tom
  5. %"><
  6. ']}'
  7. ores
  8. UIT
  9. cent
  10. raphics

Suppresses

  1. ereal
  2. aved
  3. those
  4. iese
  5. uelve
  6. 那些
  7. 犇
  8. alyze
  9. épo
  10. وقع