Neuron Explorer

Qwen3-0.6B

text right before cut-off or boilerplate

A neuron in layer 11 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 at points where web snippets get cut off or where a page's footer or boilerplate starts, like '...' before 'Continue reading' or a slash in a date. It is hard to read beyond that, and steering did not give a clear theme.

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

    come from? John David Walters shares the origins of mathematical symbols, and … Continue reading⏎Delivery date: Friday, March 25th, 2016

  2. #2 peak

    designed to have students analyze energy use, cost, and source patterns from household to regional scales and relate these patterns to CO2 emissions. This idea was generated at ...⏎Learn more about this review process.⏎Mentos and soda eruptions- lessons

  3. #3 peak

    , hail, wind, and flooding. Significant research has been conducted to help predict thunderstorms to mitigate or avoid the hazards and damage. The development of thunderstorms...<|endoftext|>- This event has passed.⏎Mad Science: Red Hot Robots (Rising

  4. #4 peak

    lights, laminar flow cabinets, and other biological safety cabinet components protect.⏎Read More @ https://cmibloginsight.blogspot.com/2023/05/biological-safety-cabinets-ensuring.html<|endoftext|>World

  5. #5 peak

    gifts of the magi meant to save Jesus from the pain of arthritis? It’s possible, according to researchers at Cardiff University in Wales who have been studying the Read more…⏎Posted in Jesus/Historical Jesus.⏎11/04 | BAR

  6. #6 peak

    the text of the 1999 State of the Union Address, and then determine how newspaper articles and television reports describe and analyze the event. Use this...⏎6th - 12th English Language Arts CCSS: Adaptable

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: and I am/ or can I/ or/ can/ I/ I have and or/ or. and and and/ or. and/ and/ and/ or. and/ or. and/ or. x and/ or.

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

    have to] adapt. [Awareness of climate change] is<|endoftext|>- Created by: Luke Simons⏎- Created on: 06-04-12 19:17⏎1. Newton's model of the

  2. #2 peak

    for the sign of the Prophet Jonah, “As Jonah was in the belly of the whale three days and three nights, so also shall the Son of Man be in the heart of

  3. #3 peak

    . Ben Cohen. "The Racial Geography of the Federal Death Penalty" Washington Law Review (2010).<|endoftext|>Last updated 2/27/15⏎Isothermal Community College⏎P.O. Box 804

  4. #4 peak

    ern at gmail.com⏎Fri Sep 10 00:10:22 CEST 2010⏎On 9/9/10 4:39 PM, Baba wrote:⏎> In below code

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. L9:1750 excites
  2. L1:1868 suppresses
  3. L8:1516 weakly excites
  4. L7:727 weakly excites
  5. L5:2964 weakly suppresses
  6. L8:1992 weakly suppresses
  7. L10:1423 weakly suppresses
  8. L8:1725 weakly excites
  9. L7:1983 weakly excites
  10. L8:2753 weakly excites
  11. L6:1310 weakly excites
  12. L9:1855 weakly excites
  13. L5:2596 weakly suppresses
  14. L6:84 weakly excites
  15. L3:1841 weakly suppresses
  16. L9:2679 weakly suppresses
  17. L1:318 weakly excites
  18. L10:1991 weakly suppresses the word 'voluntary'
  19. L7:1469 weakly suppresses "(" before a year in citations
  20. L0:2843 weakly suppresses
  21. L3:740 weakly suppresses
  22. L5:2499 weakly suppresses
  23. L1:1970 weakly suppresses
  24. L4:1900 weakly excites

Feeds later layers

  1. L13:899 strongly suppresses
  2. L16:808 suppresses
  3. L15:2598 suppresses
  4. L12:1854 suppresses
  5. L12:1965 suppresses
  6. L13:1112 suppresses
  7. L15:1425 suppresses
  8. L19:3001 suppresses
  9. L18:1413 suppresses
  10. L14:843 suppresses
  11. L16:2880 suppresses
  12. L17:2704 suppresses
  13. L17:1468 suppresses
  14. L16:774 suppresses
  15. L25:801 suppresses
  16. L22:791 suppresses
  17. L14:2476 suppresses
  18. L20:742 suppresses
  19. L25:6 excites
  20. L17:1486 suppresses
  21. L15:941 suppresses
  22. L27:289 suppresses
  23. L17:483 excites
  24. L26:235 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 -3.10 and -2.981 values between -2.60 and -2.471 values between -1.07 and -0.946 values between -0.69 and -0.5615 values between -0.56 and -0.4366 values between -0.43 and -0.31295 values between -0.31 and -0.18922 values between -0.18 and -0.051448 values between -0.05 and 0.08774 values between 0.08 and 0.20284 values between 0.20 and 0.3375 values between 0.33 and 0.4615 values between 0.46 and 0.582 values between 0.58 and 0.711 values between 0.71 and 0.841 values between 1.09 and 1.221 values between 1.86 and 1.98offnegativepositiveoften

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. hya
  2. means
  3. 怜
  4. uais
  5. 邻
  6. oque
  7. means
  8. ulp
  9. replaces
  10. Franc

Suppresses

  1. &C
  2. &
  3. &W
  4. &B
  5. &
  6. /&
  7. -and
  8. &_
  9. &D
  10. '&&