Neuron Explorer

Qwen3-0.6B

end-of-document boundary token

A neuron in layer 16 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 invisible marker the training data uses to separate one web page from the next, which appears in the middle of a sentence or list where a new document starts. Turning it up made the model repeat newlines and dots, so it seems tied to where a text stops and another begins.

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

    994- Drexel University. All rights reserved.⏎The Math Forum is a research and educational enterprise of the Drexel University School of Education.<|endoftext|>Maybe you’ve tried intermittent fasting (IF) to shed a few (pand

  2. #2 peak

    bonsai tree is the most common reason for its decline.⏎Mary Lougee has been writing for over 10 years. She holds a Bachelor's Degree with<|endoftext|>- freely available⏎Sustainability 2013, 5(7

  3. #3 peak

    VS is a promising alternative strategy to produce the functional and thermostable reverse transcriptase.⏎All Science Journal Classification (ASJC) codes⏎- Insect Science<|endoftext|>An account of the religion of

  4. #4 peak

    the China site (in Chinese or English) for best site performance. Other MathWorks country sites are not optimized for visits from your location.⏎Contact your local office<|endoftext|>It’s PrAACtically Martin Luther King Jr Day⏎Looking for activities and

  5. #5 peak

    the China site (in Chinese or English) for best site performance. Other MathWorks country sites are not optimized for visits from your location.⏎Contact your local office<|endoftext|>The point in the Moon's orbit where it lies diametrically⏎opposite

  6. #6 peak

    S||51.00W To 54.00W|⏎|rs_class_santarem.tif||163.4 MB||<|endoftext|>English-to-Spanish medical translation will be more important than ever in the United

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: (for example, for example, in the same or or or or or or or or or or (for example, for example) and for the same or or or (or or or 1-1

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

    of allergens and irritants, but it can eventually lead to structural damage in your home if left untreated. A dehumidifier or air conditioner can help.⏎Answer: True. Radon—a colorless, tasteless and odorless gas

  2. #2 peak

    This preview has intentionally blurred sections. Sign up to view the full version.View Full Document⏎Unformatted text preview: itself, and if ( ) ( ) lim lim x

  3. #3 peak

    To access and follow an EdPlace learning plan, please download the attachment at the end of the article.⏎How can I use them to support my child?⏎There are several ways in which

  4. #4 peak

    shows pages 1–2. Sign up to view the full content.⏎This preview has intentionally blurred sections. Sign up to view the full version.View Full Document⏎Unformatted

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. L15:81 strongly suppresses line breaks inside running text
  2. L14:56 suppresses document ends and list line breaks
  3. L10:84 suppresses
  4. L13:10 excites
  5. L10:259 excites
  6. L8:32 excites
  7. L9:25 excites
  8. L1:1 excites
  9. L15:123 suppresses
  10. L14:1067 suppresses
  11. L15:82 suppresses
  12. L12:5 excites
  13. L14:51 excites colon that introduces a statement
  14. L9:7 excites
  15. L15:44 suppresses
  16. L10:306 suppresses
  17. L1:1279 suppresses
  18. L9:2729 suppresses
  19. L9:605 excites
  20. L6:216 excites
  21. L15:385 excites
  22. L9:12 weakly suppresses
  23. L0:37 weakly excites
  24. L15:1620 weakly excites period ending a sentence in essays

Feeds later layers

  1. L17:1176 strongly suppresses
  2. L17:1953 strongly excites
  3. L17:1568 strongly excites
  4. L19:1479 strongly excites
  5. L18:1950 strongly suppresses opening parenthesis after a name or title
  6. L18:2202 strongly excites
  7. L19:441 excites
  8. L19:624 excites
  9. L18:2922 excites
  10. L20:1081 suppresses
  11. L20:1085 excites
  12. L21:130 excites
  13. L22:506 weakly excites
  14. L26:2463 weakly suppresses
  15. L17:537 weakly excites
  16. L20:467 weakly excites
  17. L17:2317 weakly excites
  18. L17:1024 weakly excites
  19. L25:1274 weakly suppresses
  20. L17:1616 weakly excites
  21. L27:1065 weakly suppresses
  22. L25:1740 weakly suppresses
  23. L19:1007 weakly excites
  24. L18:705 weakly excites

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 -7.85 and -7.571 values between -7.02 and -6.752 values between -6.47 and -6.204 values between -6.20 and -5.921 values between -5.65 and -5.375 values between -5.37 and -5.102 values between -5.10 and -4.825 values between -4.82 and -4.553 values between -4.55 and -4.285 values between -4.28 and -4.006 values between -4.00 and -3.735 values between -3.73 and -3.452 values between -3.45 and -3.181 values between -3.18 and -2.905 values between -2.90 and -2.631 values between -2.63 and -2.352 values between -2.35 and -2.084 values between -2.08 and -1.806 values between -1.80 and -1.5310 values between -1.53 and -1.2511 values between -1.25 and -0.9827 values between -0.98 and -0.70120 values between -0.70 and -0.431091 values between -0.43 and -0.152330 values between -0.15 and 0.12242 values between 0.12 and 0.404 values between 0.40 and 0.672 values between 0.67 and 0.951 values between 0.95 and 1.221 values between 1.22 and 1.501 values between 1.77 and 2.053 values between 2.05 and 2.321 values between 2.60 and 2.873 values between 2.87 and 3.15offnegativepositiveoften

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. 们都
  2. less
  3. 们的
  4. 四方
  5. icks
  6. Whenever
  7. 人都
  8. bölüm
  9. Интер
  10. partes

Suppresses

  1. enstein
  2. nung
  3. omen
  4. .qq
  5. yne
  6. 亿吨
  7. �
  8. ughter
  9. etailed
  10. Hussein