Neuron Explorer

Qwen3-0.6B

names of agencies and institutions

A neuron in layer 8 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron fires most on the middle words of official names of agencies, services and commissions, such as Selective Service and Office of. It is weak and noisy, and steering did not produce a clear change in the written text.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    or a $250,000 fine.⏎And the system has not been updated since women were allowed to serve in combat roles.⏎The Selective Service System has disbanded

  2. #2 peak

    3 x 4¼ in.⏎This item is published by PomegranateKids®, an imprint of Pomegranate Communications, and is CPSIA compliant.⏎⚠️WARNING: CHOKING HAZARD - small parts. Not

  3. #3 peak

    except for the state of Texas, women cannot be drafted.⏎Within 30 days of turning 18, all male citizens must register for the Selective Service System,

  4. #4 peak

    dramatically enhanced if these multiple electrons can be produced via a single-photon absorption event.⏎Research on carrier multiplication at Los Alamos is funded by the DOE’s Office of Basic Energy Sciences and by Los Alamos

  5. #5 peak

    section 3: Extended Library Functions, Volume 2 Oracle Solaris 11.1 Information Library|⏎- load exponent of a floating point number⏎c99 [

  6. #6 peak

    Greece’s national digital strategy 2014-2020, the Ministry of Infrastructure, Transports and Networks, the National Telecommunications and Post Commission and the Ministry of Education co-la<|endoftext|>Student or Learner⏎Brass:

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 connection to a character's background The question is about the connection to a character's background The connection to a character's background is not about using a character to understand things, but about using the character to describe the world The

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

    ⏎- Configure Spark to modify the Spark properties, environmental variables, or logging properties⏎- Monitor Spark using the web UIs, metrics, and external instrumentation⏎- Understand performance tuning considerations⏎- This

  2. #2 peak

    better-informed...⏎7th - 11th Social Studies & History⏎President Obama's State of the Union Address⏎Students watch the State of the Union Address from last year, make a PowerPoint about their expectation for this year's Address,

  3. #3 peak

    ikipedia.org/wiki/Cookie<|endoftext|>Search through our library of courses, select a category you might be interested in, use the search bar, or filter options on the left side of the page.⏎Are you looking for a unique way to teach storytelling elements

  4. #4 peak

    suggest that the first thing you should do is to just start printing stuff so that you familiarize yourself with operating the 3D printer. Watch the getting started videos, load the machine with some 3D printing filament, and go find a

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. L7:1971 suppresses
  2. L7:1120 suppresses
  3. L7:562 suppresses
  4. L7:2841 suppresses
  5. L7:3050 suppresses
  6. L7:1236 excites
  7. L5:153 suppresses
  8. L7:1657 suppresses
  9. L7:248 suppresses
  10. L7:278 excites
  11. L6:566 suppresses
  12. L5:2626 suppresses
  13. L6:719 weakly excites
  14. L7:155 weakly excites
  15. L1:1090 weakly suppresses
  16. L7:494 weakly suppresses
  17. L5:246 weakly excites
  18. L6:1120 weakly excites
  19. L5:1480 weakly suppresses
  20. L7:2899 weakly excites
  21. L1:737 weakly suppresses
  22. L3:616 weakly suppresses
  23. L7:2657 weakly excites
  24. L7:1658 weakly suppresses

Feeds later layers

  1. L10:2529 excites
  2. L9:558 suppresses
  3. L10:1880 suppresses
  4. L9:368 excites
  5. L12:1149 suppresses
  6. L9:1048 suppresses
  7. L9:696 suppresses
  8. L10:1897 excites
  9. L10:1342 excites
  10. L11:1316 suppresses
  11. L15:1434 suppresses
  12. L11:1500 suppresses
  13. L9:2412 suppresses
  14. L10:1903 suppresses
  15. L11:1797 suppresses
  16. L9:999 excites
  17. L18:1600 suppresses
  18. L9:407 excites
  19. L11:1181 suppresses
  20. L12:1768 suppresses
  21. L12:2104 excites
  22. L12:407 suppresses words before 'signs' or 'symptoms'
  23. L11:937 suppresses
  24. L12:2869 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 -1.94 and -1.861 values between -1.78 and -1.691 values between -1.28 and -1.191 values between -1.11 and -1.032 values between -1.03 and -0.942 values between -0.94 and -0.866 values between -0.86 and -0.784 values between -0.78 and -0.703 values between -0.70 and -0.6112 values between -0.61 and -0.539 values between -0.53 and -0.4522 values between -0.45 and -0.3637 values between -0.36 and -0.2869 values between -0.28 and -0.20194 values between -0.20 and -0.11545 values between -0.11 and -0.031497 values between -0.03 and 0.05714 values between 0.05 and 0.14388 values between 0.14 and 0.22218 values between 0.22 and 0.30104 values between 0.30 and 0.3942 values between 0.39 and 0.4714 values between 0.47 and 0.5511 values between 0.55 and 0.641 values between 0.64 and 0.727 values between 0.72 and 0.802 values between 0.97 and 1.051 values between 1.30 and 1.38offnegativepositiveoften

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. orne
  2. 人も
  3. 记者采访
  4. termin
  5. 人流
  6. stabbed
  7. 电子产品
  8. setDate
  9. ขาย
  10. Termin

Suppresses

  1. *__
  2. (('
  3. .*(
  4. (\'
  5. proposition
  6. ött
  7. ('
  8. ((__
  9. (!(
  10. particulière