Neuron Explorer

Qwen3-0.6B

opening parenthesis before an acronym

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 fires on the opening bracket that comes right after a full name, just before its short form, such as "International Monitoring System (IMS)". Turning it up did not give a clear theme in the writing.

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

    quite a scare with the local population!⏎In a press release from the Near Earth Object (NEO) program, the explosion was detected by many International Monitoring System (IMS) infrasound stations,

  2. #2 peak

    of Music, University of Chicago⏎—Anthony Cascardi, University of California, Berkeley⏎—Jesse J. Prinz, John J. Rogers Professor of Philosophy, University of North Carolina, and author of Furnishing the Mind: Concepts

  3. #3 peak

    at the Idaho Reading Indicator Homepage⏎The Idaho Language Proficiency Assessment (IELA) is a federally mandated assessment for all students served in a Limited English Proficiency (LEP) program. The IELA is administered annually every spring and calculates

  4. #4 peak

    not. Get to the end of the story. The Next Steps.⏎How Do I Set Up A Website Using Iis?⏎You can access the Internet Information Services (IIS) Manager by selecting Control Panel > Administrative Tools. Select Add Website from

  5. #5 peak

    closure through alternative burial rituals.⏎With a team of four counselors, psychologist Noé Kasali — who heads Bethesda's counseling program<|endoftext|>Human Endogenous Retroviruses (HERVs)⏎

  6. #6 peak

    an OS provides, e.g., multiple processes or threads of execution, signals, IPC and so on.⏎IPC - In computer science, 'Inter Process Communication' (IPC) refers specifically to the mechanisms an operating system provides to allow processes it manages

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: I have been thinking about lately: I have been thinking about lately: I have been thinking about lately, and I have been thinking about lately, and I have been thinking about lately, and I have been thinking about lately, and I have

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

    • Teacher Resources • Frequently Asked Questions • Glossary • References • Links to Other Mussel Sites⏎Privacy • FOIA • FirstGov • Contact⏎Department of the Interior • U.S. Fish & Wildlife Service • U.S. Geological Survey⏎

  2. #2 peak

    1978)⏎- National Medal of Science (1986); the APA’s Award for Outstanding Lifetime Contributions to

  3. #3 peak

    ’Av is the culmination, of course, of two national tragedies: the loss of the first Beis HaMikdash as well as the Second Beis HaMikdash.⏎To understand the historical background that existed in the years before the

  4. #4 peak

    name, roughly follows the outlines of Dick’s book with a number of changes. For example, instead of a book within a book, we have a film within a film–an underground video portraying the world in which the Allies win. Subsequent

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. L7:6 strongly suppresses
  2. L7:26 suppresses
  3. L5:5 suppresses digit after a label like paper or type
  4. L6:1752 excites
  5. L8:517 suppresses
  6. L8:232 suppresses
  7. L8:25 suppresses
  8. L6:101 suppresses
  9. L6:516 excites
  10. L4:4 suppresses
  11. L8:1346 excites
  12. L8:706 suppresses
  13. L3:940 excites
  14. L8:1163 suppresses
  15. L6:863 suppresses
  16. L6:158 excites
  17. L7:8 excites
  18. L6:43 excites
  19. L6:10 excites
  20. L6:1801 excites
  21. L6:2086 excites
  22. L6:184 excites
  23. L6:655 excites
  24. L8:296 excites 'of' inside fixed noun phrases

Feeds later layers

  1. L10:67 excites
  2. L10:40 suppresses
  3. L11:23 suppresses
  4. L10:121 excites
  5. L13:308 suppresses
  6. L14:143 excites 'the' inside journal and institution names
  7. L15:867 suppresses
  8. L10:152 excites
  9. L15:11 suppresses
  10. L13:364 suppresses
  11. L13:293 suppresses
  12. L13:27 suppresses
  13. L15:376 suppresses
  14. L16:1855 excites
  15. L11:300 excites
  16. L25:102 suppresses
  17. L21:94 suppresses
  18. L19:216 suppresses
  19. L16:312 suppresses
  20. L13:112 suppresses
  21. L11:55 suppresses
  22. L13:175 suppresses
  23. L15:182 suppresses
  24. L19:337 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 -4.54 and -4.321 values between -3.02 and -2.813 values between -2.81 and -2.593 values between -2.59 and -2.373 values between -2.37 and -2.165 values between -2.16 and -1.949 values between -1.94 and -1.7315 values between -1.73 and -1.5127 values between -1.51 and -1.2929 values between -1.29 and -1.0849 values between -1.08 and -0.86108 values between -0.86 and -0.65471 values between -0.65 and -0.43824 values between -0.43 and -0.21687 values between -0.21 and 0.00648 values between 0.00 and 0.22359 values between 0.22 and 0.44218 values between 0.44 and 0.65164 values between 0.65 and 0.8797 values between 0.87 and 1.0864 values between 1.08 and 1.3036 values between 1.30 and 1.5233 values between 1.52 and 1.7320 values between 1.73 and 1.956 values between 1.95 and 2.169 values between 2.16 and 2.387 values between 2.38 and 2.605 values between 2.60 and 2.813 values between 2.81 and 3.032 values between 3.03 and 3.251 values between 3.25 and 3.461 values between 3.89 and 4.11offnegativepositiveoften

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. oton
  2. atu
  3. UT
  4. ña
  5. enco
  6. ivery
  7. nom
  8. obe
  9. mina
  10. firearms

Suppresses

  1. 和其他
  2. -warning
  3. Guar
  4. -template
  5. -how
  6. _requires
  7. ious
  8. 及其他
  9. -he
  10. =sub