Neuron Explorer

Qwen3-0.6B

"on" after "of all living things" phrases

A neuron in layer 22 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 "on" in phrases like "the hardiest animals on the planet" or "life on Earth", where a place or world comes next. Turned up, the model talked about the Earth and its surface over and over.

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

    .<|endoftext|>Even the hardy tardigrade might not stand up to climate change⏎Tardigrades are well-known for being one of the hardiest animals on the planet, but even they have their limits. New research has shown that they

  2. #2 peak

    and let your imagination run WILD!<|endoftext|>Baleen and sperm whales, known collectively as the great whales, include the largest animals in the history of life on Earth. With high metabolic demands and large populations, whales probably had a strong influence

  3. #3 peak

    , digestive problems, numbness in the extremities, and skin cancer.⏎Water is the essence of life, but not just human life. Almost every living thing on the planet must have water to survive, and that includes a whole host of bacteria

  4. #4 peak

    Mamba. …⏎- White Rhinoceros.⏎What is Africa’s deadliest animal?⏎Despite being vegetarian, hippos are considered the most dangerous terrestrial animal on the African continent. On average, they are responsible for the deaths of almost

  5. #5 peak

    and many other things dirty. There are many type of pollutions such as air pollution, water pollution and soil pollution. Air pollution is harmful to all living organisms on this earth. Air pollution can exist in many type of states such as open burning

  6. #6 peak

    ocean...⏎and hydrologic sciences⏎Hydrology is the study of the movement, distribution, and quality of water on Earth and other planets, including the hydrologic cycle, water resources and environmental watershed

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: in the context of the future, we will have the same as in the past, but with the same content in the same capacity, and the same use, and the same identity, in the same content, in the same context, in

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

    Everything You Need⏎Who was the first president to live in the White House? Learn about this

  2. #2 peak

    reality of his actions is that he only speaks in order to appear sincere as a politician, to look good rather then actually be good. ” to thine own self be true, And it must follow, as the night the day, Thou can

  3. #3 peak

    Twitter. The Nintendo Wii. Google Maps. Google Translate. Artificial hearts and livers. Face transplants. Self-cleaning windows. Invisibility cloaking. Graphene. The fuel-cell bike

  4. #4 peak

    the entire Saturn system. A Titan colony is self sustaining as long as it has power to supply these processes, and invaluable food resources

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. L20:2975 suppresses
  2. L19:2280 suppresses
  3. L21:1210 suppresses
  4. L0:2559 excites
  5. L21:3017 suppresses
  6. L21:797 suppresses
  7. L12:1693 weakly suppresses
  8. L13:1213 weakly excites
  9. L20:1722 weakly suppresses
  10. L19:2967 weakly excites
  11. L17:1165 weakly excites
  12. L2:2795 weakly excites
  13. L9:2862 weakly suppresses
  14. L9:1120 weakly excites
  15. L12:2240 weakly suppresses
  16. L15:349 weakly excites
  17. L21:1949 weakly suppresses
  18. L1:2478 weakly suppresses
  19. L21:2469 weakly suppresses
  20. L5:1318 weakly excites
  21. L21:792 weakly suppresses
  22. L11:1 weakly excites
  23. L21:508 weakly excites
  24. L16:2690 weakly excites

Feeds later layers

  1. L26:948 excites
  2. L27:995 excites
  3. L27:2905 excites
  4. L25:530 excites
  5. L24:2125 excites
  6. L26:2384 weakly excites
  7. L25:2813 weakly excites
  8. L24:2651 weakly excites
  9. L25:1282 weakly excites
  10. L25:312 weakly excites
  11. L25:2811 weakly suppresses
  12. L25:2432 weakly suppresses
  13. L24:2566 weakly excites
  14. L25:2895 weakly excites
  15. L26:1910 weakly excites
  16. L26:981 weakly excites
  17. L24:716 weakly excites
  18. L24:1842 weakly excites
  19. L24:1054 weakly excites
  20. L23:152 weakly excites
  21. L23:1455 weakly excites
  22. L27:2244 weakly excites
  23. L25:499 weakly excites
  24. L24:3011 weakly excites 'of' in national and founding phrases

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.

2 values between -8.86 and -7.933 values between -7.01 and -6.083 values between -5.16 and -4.234 values between -4.23 and -3.315 values between -3.31 and -2.3814 values between -2.38 and -1.45162 values between -1.45 and -0.533230 values between -0.53 and 0.40368 values between 0.40 and 1.3245 values between 1.32 and 2.2517 values between 2.25 and 3.1813 values between 3.18 and 4.1010 values between 4.10 and 5.037 values between 5.03 and 5.953 values between 5.95 and 6.884 values between 6.88 and 7.815 values between 7.81 and 8.732 values between 8.73 and 9.664 values between 10.58 and 11.511 values between 11.51 and 12.432 values between 12.43 and 13.361 values between 15.21 and 16.141 values between 17.99 and 18.911 values between 20.77 and 21.691 values between 27.25 and 28.17offnegativepositiveoften

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. earth
  2. Earth
  3. 地面
  4. earth
  5. ground
  6. 地球
  7. Earth
  8. -earth
  9. 地
  10. 地上

Suppresses

  1. UserManager
  2. _SCOPE
  3. AGIC
  4. 前瞻性
  5. oping
  6. ryptography
  7. ileged
  8. UserService
  9. ?>><?
  10. cox