Neuron Explorer

Qwen3-0.6B

"in" in "which in turn" phrases

A neuron in layer 20 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 lights up on the word "in" in phrases like "which in turn" or "and in return", where one thing leads to another. It goes quiet, and slightly negative, on "on" or "the" in phrases like "on the other hand" or "on the part of". Turning it down made the model repeat "on the part of" and "on the other hand" over and over, while turning it up gave loose, unrelated 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

    . The trees that die as a result of rising water levels attract insects, which in turn feed woodpeckers, whose holes later provide homes for other wildlife.⏎

  2. #2 peak

    an interaction of nature, nurture and the decisions and choices that people make. Biological programming is your genetics. Genes provide the instructions for structuring amino acids which in turn influence the proteins within our body cells thus programming chemical basis of our biology.

  3. #3 peak

    the group changed its name to the STEMG.⏎The STEMG reports to the ASD Product Support Specification Group (PSSG) which in turn reports to the ASD Services Commission (SVC).<|endoftext|>Biology exquis

  4. #4 peak

    production will be the most affected by climate change by 2030. This would lead to food insecurity, which in turn could lead to social instability, as was already the case in Mozambique in

  5. #5 peak

    is an example of mutualism because the nectar from the flower nourishes the bee and in return, the bee pollinates the flower.⏎Activity Three: Humans and the

  6. #6 peak

    the initial state of the atmosphere is critical. Even if Doppler radar information may seem highly localized, we find that it enhances the regional atmospheric conditions, which, in turn, can significantly improve the dynamic prediction of how the monsoon depression will move

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: on the one hand, on the part of the part of the part on the part on the part of the part part on the part on the part part on the part on the part on the part on the part part on the part on

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

    his responsibilities, and the means by which his duties are to be performed. One of the chief causes of failure in teaching is the lack of well-defined purpose on the part of those entrusted with the training of pupils. A knowledge of the end to

  2. #2 peak

    . Hard work can lead to a deep sense of accomplishment and mastery, but can also be time-consuming and may not be the most effective approach. Smart work, on the other hand, can save time and increase productivity, but may lead to a

  3. #3 peak

    to perform specific functions.⏎Microcontrollers are designed to handle simple tasks, making them a popular choice in low-cost, low-power embedded systems. FPGAs, on the other hand, are designed to handle complex logic functions that require high performance and

  4. #4 peak

    to achieve and have clear steps and timelines for completion. They are usually long-term and require dedication and hard work in order to be achieved. Resolutions, on the other hand, are more short-term commitments that a person makes in order to improve

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. L19:2530 strongly excites
  2. L19:934 strongly excites
  3. L18:2007 excites
  4. L15:2777 weakly suppresses
  5. L7:2179 weakly excites
  6. L19:1644 weakly suppresses
  7. L3:1289 weakly suppresses
  8. L7:3031 weakly suppresses
  9. L6:17 weakly suppresses
  10. L19:417 weakly excites
  11. L16:2008 weakly excites
  12. L11:43 weakly suppresses
  13. L18:569 weakly excites
  14. L17:1259 weakly suppresses
  15. L19:212 weakly excites
  16. L10:17 weakly suppresses
  17. L9:8 weakly excites
  18. L19:2308 weakly excites
  19. L15:2926 weakly excites
  20. L17:2541 weakly suppresses comma before but also in not only constructions
  21. L17:2988 weakly excites
  22. L7:934 weakly suppresses
  23. L11:1253 weakly suppresses
  24. L11:72 weakly excites

Feeds later layers

  1. L27:177 strongly excites
  2. L27:721 strongly suppresses
  3. L26:951 strongly suppresses
  4. L24:1796 strongly suppresses
  5. L23:1438 suppresses
  6. L26:2330 excites
  7. L25:2282 suppresses
  8. L27:1194 suppresses
  9. L25:2569 suppresses
  10. L24:1898 excites
  11. L23:2742 suppresses
  12. L25:666 suppresses
  13. L21:1237 suppresses
  14. L24:1785 weakly suppresses
  15. L25:1892 weakly suppresses
  16. L24:1669 weakly excites
  17. L21:2136 weakly excites
  18. L23:476 weakly suppresses
  19. L24:1141 weakly suppresses
  20. L27:2013 weakly suppresses
  21. L23:2234 weakly suppresses
  22. L22:1190 weakly excites
  23. L22:1368 weakly excites
  24. L22:2554 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 -5.54 and -5.351 values between -4.36 and -4.171 values between -3.58 and -3.382 values between -3.38 and -3.192 values between -3.19 and -2.992 values between -2.60 and -2.404 values between -2.40 and -2.203 values between -2.20 and -2.013 values between -2.01 and -1.813 values between -1.81 and -1.615 values between -1.61 and -1.421 values between -1.42 and -1.225 values between -1.22 and -1.025 values between -1.02 and -0.838 values between -0.83 and -0.6330 values between -0.63 and -0.4375 values between -0.43 and -0.24537 values between -0.24 and -0.042179 values between -0.04 and 0.16526 values between 0.16 and 0.35236 values between 0.35 and 0.55101 values between 0.55 and 0.7585 values between 0.75 and 0.9435 values between 0.94 and 1.1433 values between 1.14 and 1.3412 values between 1.34 and 1.535 values between 1.53 and 1.731 values between 1.73 and 1.933 values between 1.93 and 2.124 values between 2.12 and 2.32offnegativepositiveoften

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. 逵
  2. ��
  3. daf
  4. lotte
  5. EPROM
  6. 茴
  7. congratulations
  8. 櫃
  9. ottes
  10. 熟悉的

Suppresses

  1. on
  2. On
  3. On
  4. .on
  5. →on
  6. _on
  7. →On
  8. on
  9. -on
  10. (on