Neuron Explorer

Qwen3-0.6B

parentheses defining names or acronyms

A neuron in layer 12 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 the opening bracket that comes right after a full name, just before its short form, such as 'American Psychological Association (APA)'. It is strongly switched off on the word 'the' inside legal-style definitions like '(the Act)'. Turning it down made the model produce many repeated '(the NYT)' style definitions.

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

    such as in packet mode. The information transmission may or may not start immediately as requested by sender, the additional delay being caused by medium

  2. #2 peak

    il in any form as it contains compounds that can induce uterine contractions.2⏎Use

  3. #3 peak

    , and parody. Teachers may even make multiple copies for classroom use.⏎The legal standard for deciding what constitutes fair use of a copyrighted work involves examining and balancing the effects of four different relevant factors:⏎Retrieved and used with permission from ASCD http

  4. #4 peak

    constitutional anti-discrimination clause is set out in Article 23(2) that reads “no person shall be treated in a discriminatory manner by any person acting by virtue of any written law or in the performance of the functions of any public office

  5. #5 peak

    now at least 877 times the background extinction rate over geological time.⏎BioScience, published monthly, is the journal of the American Institute of Biological Sciences (AIBS; www.aibs.org). BioScience is a forum for integrating

  6. #6 peak

    one another,<|endoftext|>Down with Zero Tolerance⏎It might be time to ease up a bit when it comes to disruptive students and detention. The American Psychological Association (APA) recently reviewed 1

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 'sponsors' or the 'sponsors' (the 'sponsors' or the 'sponsors' or 'sponsors' or 'sponsors' or 'sponsors' or 'sponsors' or 'sponsors

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

    .⏎Under the Yukon Human Rights Act (the Act), harassment is defined as to:⏎“engage in a course of vexat

  2. #2 peak

    given by the trial judge was improper. The process will now be remitted back to the planning stage that was previous to the Peel Watershed Planning Commission’s (the “Commission”) Final Recommended Report (“Final Recommended Plan”), instead of being remitted

  3. #3 peak

    ak Dun v. Yukon, 2015 YKCA 18 (the “appeal decision”), the Yukon Court of Appeal unanimously allowed the Yukon government

  4. #4 peak

    workplace sexual harassment.⏎Harassment occurs in many forms, and is always wrong. However, legal protections have limits, and the Yukon Human Rights Act (the Act) only offers protection against harassment that occurs based on certain protected characteristics (called

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. L10:1934 excites
  2. L3:85 excites
  3. L11:3013 weakly excites
  4. L6:16 weakly excites
  5. L11:1624 weakly suppresses
  6. L6:1112 weakly suppresses
  7. L1:1433 weakly excites
  8. L5:1550 weakly suppresses
  9. L8:85 weakly suppresses
  10. L10:3035 weakly suppresses
  11. L7:2478 weakly suppresses
  12. L0:1388 weakly excites
  13. L5:2656 weakly excites
  14. L6:76 weakly excites
  15. L10:1632 weakly suppresses
  16. L11:2810 weakly suppresses
  17. L11:671 weakly excites
  18. L4:2353 weakly excites
  19. L1:2005 weakly suppresses
  20. L1:2908 weakly excites
  21. L10:367 weakly excites
  22. L3:1050 weakly suppresses
  23. L9:2493 weakly excites
  24. L1:2496 weakly suppresses

Feeds later layers

  1. L13:2248 suppresses
  2. L25:360 suppresses
  3. L23:224 suppresses
  4. L18:1501 suppresses
  5. L13:1360 suppresses
  6. L16:2306 suppresses
  7. L17:738 suppresses
  8. L17:95 suppresses
  9. L16:1226 weakly suppresses
  10. L15:2658 weakly suppresses
  11. L20:1923 weakly excites
  12. L26:13 weakly excites
  13. L20:995 weakly excites
  14. L14:1576 weakly suppresses
  15. L21:782 weakly suppresses
  16. L20:521 weakly suppresses
  17. L14:2455 weakly excites
  18. L23:845 weakly excites
  19. L16:978 weakly suppresses
  20. L26:2854 weakly excites
  21. L26:234 weakly suppresses
  22. L21:1503 weakly suppresses
  23. L17:2715 weakly suppresses
  24. L21:416 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 -1.15 and -1.101 values between -1.04 and -0.982 values between -0.98 and -0.931 values between -0.82 and -0.761 values between -0.65 and -0.601 values between -0.54 and -0.481 values between -0.48 and -0.435 values between -0.43 and -0.3711 values between -0.37 and -0.3230 values between -0.32 and -0.2658 values between -0.26 and -0.21124 values between -0.21 and -0.15168 values between -0.15 and -0.10332 values between -0.10 and -0.04564 values between -0.04 and 0.02547 values between 0.02 and 0.07522 values between 0.07 and 0.13411 values between 0.13 and 0.18347 values between 0.18 and 0.24263 values between 0.24 and 0.29174 values between 0.29 and 0.35117 values between 0.35 and 0.4188 values between 0.41 and 0.4656 values between 0.46 and 0.5234 values between 0.52 and 0.5715 values between 0.57 and 0.6312 values between 0.63 and 0.685 values between 0.68 and 0.749 values between 0.74 and 0.795 values between 0.79 and 0.851 values between 0.85 and 0.911 values between 0.96 and 1.021 values between 1.02 and 1.07offnegativepositiveoften

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. ropol
  2. -Level
  3. ighet
  4. ailability
  5. ipur
  6. isoft
  7. cular
  8. 点钟
  9. يلا
  10. ylum

Suppresses

  1. ("
  2. (
  3. (“
  4. ("<
  5. ("
  6. ("__
  7. (
  8. ف
  9. "](
  10. ("*