Neuron Explorer

Qwen3-0.6B

'World' in organisation names

A neuron in layer 25 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 word 'World' in names like World Health Organization, World Factbook and World Heritage, and also on a few 'are' and 'was' words after a definition. Turning it up made the text talk about things to consider and the world in general, while turning it down gave loose, rambling text.

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

    locate.<|endoftext|>Science Fair Project Encyclopedia⏎The World Factbook⏎The World Factbook is an annual publication by the Central Intelligence Agency

  2. #2 peak

    PM2.5, mainly from coal.⏎As it stands now, only about two percent of China’s nearly 1.4 billion people breathe air that meets World Health Organization (WHO) guidelines, the IEA said.⏎That proportion would rise

  3. #3 peak

    in Poland in 1473 and as a young man became fascinated by math, medicine and science, particularly astronomy. His studies led him to challenge what was then a widely accepted belief — that the Earth stood still in outer space while the

  4. #4 peak

    of brucellosis was considered confirmed if results of serologic tests, such as ELISA or complement fixation test, were positive, in accordance with the World Health Organization case definition (<|endoftext|>The

  5. #5 peak

    While earlier versions of the laws dealt with cattle and livestock, today Texas cruelty laws protect a variety of animals from inhumane treatment. These laws define which actions are “cruel”, explain what punishment or penalties can be imposed against those whom

  6. #6 peak

    or neither?⏎Parallel lines are the lines that never intersect and perpendicular lines are the lines that intersect at 90 degrees.⏎Answer: Lines with the same slope are parallel and if the slope of one line is the negative reciprocal of the second line

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 outting out out of the best of my best outing outing outing outing outing outing outing outing outing outing outing outing outing outing outing outing outing out

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 the monument. Where it is not available, the satellite view is shown instead.⏎Latitude: 52.5162 / 52°30'58"N⏎Longitude: -3.0604 /

  2. #2 peak

    UNESCO World Heritage Committee has put the Site on its List for the following reasons:⏎The realization of an architectural and sacred art’s work into a natural landscape for didactic and religious aims, finds its⏎highest expression in northern Italy Sacred Mounts

  3. #3 peak

    individually, and asked the confederate to leave for a few minutes. After the confederate had left, the experimenter debrief<|endoftext|>Passchendaele at Home⏎Passchendaele at Home was a nationwide initiative

  4. #4 peak

    understanding of his culture and the injustices Navajos faced. Bruchac sprinkles historical facts throughout the book in a way that feels fascinating, rather than didactic. I learned so much from Code Talker. My high school boyfriend was

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. L22:2721 strongly excites
  2. L20:2988 excites
  3. L15:1531 excites
  4. L18:2451 suppresses
  5. L19:2286 suppresses
  6. L11:1484 weakly suppresses
  7. L9:66 weakly excites
  8. L0:1609 weakly suppresses
  9. L21:2482 weakly suppresses
  10. L16:2661 weakly suppresses
  11. L20:2406 weakly suppresses
  12. L10:2934 weakly excites
  13. L22:2071 weakly excites
  14. L8:1219 weakly suppresses
  15. L19:1612 weakly excites
  16. L19:386 weakly suppresses
  17. L19:2748 weakly suppresses
  18. L7:1265 weakly suppresses
  19. L17:2645 weakly suppresses
  20. L15:599 weakly excites
  21. L20:1872 weakly suppresses
  22. L19:1884 weakly suppresses
  23. L19:2495 weakly suppresses
  24. L20:2121 weakly suppresses

Feeds later layers

  1. L27:1849 strongly excites
  2. L27:834 weakly excites
  3. L27:879 weakly excites
  4. L27:526 weakly excites
  5. L27:310 weakly suppresses
  6. L27:2319 weakly suppresses
  7. L26:2023 weakly suppresses
  8. L26:612 weakly suppresses end-of-text boundary tokens
  9. L27:2291 weakly suppresses
  10. L26:2540 weakly excites
  11. L27:431 weakly excites
  12. L27:1282 weakly excites
  13. L26:2369 weakly suppresses
  14. L27:2677 weakly suppresses
  15. L27:2829 weakly excites
  16. L26:1688 weakly excites
  17. L26:1349 weakly excites
  18. L27:1049 weakly suppresses
  19. L27:855 weakly suppresses
  20. L26:1877 weakly suppresses
  21. L27:997 weakly suppresses
  22. L26:2383 weakly excites
  23. L27:1399 weakly suppresses
  24. L26:1585 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 -16.33 and -15.271 values between -13.15 and -12.091 values between -11.03 and -9.972 values between -9.97 and -8.911 values between -8.91 and -7.853 values between -7.85 and -6.793 values between -6.79 and -5.734 values between -5.73 and -4.6711 values between -4.67 and -3.6115 values between -3.61 and -2.5515 values between -2.55 and -1.49419 values between -1.49 and -0.432954 values between -0.43 and 0.63230 values between 0.63 and 1.6997 values between 1.69 and 2.7552 values between 2.75 and 3.8126 values between 3.81 and 4.8710 values between 4.87 and 5.9314 values between 5.93 and 6.9913 values between 6.99 and 8.057 values between 8.05 and 9.117 values between 9.11 and 10.174 values between 10.17 and 11.233 values between 11.23 and 12.291 values between 12.29 and 13.354 values between 13.35 and 14.411 values between 14.41 and 15.474 values between 17.58 and 18.642 values between 18.64 and 19.701 values between 19.70 and 20.761 values between 21.82 and 22.881 values between 25.00 and 26.06offnegativepositiveoften

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. consider
  2. considered
  3. 考虑
  4. Consider
  5. consideration
  6. considers
  7. Consider
  8. consider
  9. consid
  10. considerations

Suppresses

  1. pel
  2. ault
  3. 确定
  4. UTES
  5. 定
  6. emat
  7. 现有的
  8. "path
  9. ать
  10. amework