Neuron Explorer

Qwen3-0.6B

unclear: document ends and instruction verbs

A neuron in layer 14 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 is hard to read. It fires most on the end-of-document marker between web pages, but also on some other spots. It is strongly negative on action verbs in how-to instructions, like 'take' or 'plant'. Steering did not give a clear picture of what it does.

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

    ’ or ‘raw’ and ‘war’. Use the update function to automatically scan for overlaps, the numbers should update every time you click outside the word list.<|endoftext|>ambience: dingoes howling⏎Australian Aborigines first named these wild

  2. #2 peak

    . FormsCan be used for anything<|endoftext|>We found 1,708 reviewed resources for state of the union address⏎How can

  3. #3 peak

    ates and the data subjected to analysis of variance and the means were compared by Tukey test.⏎News published in Agência FAPESP Newsletter about the scholarship:<|endoftext|>Image: Blazing volcano⏎"This is an erupting volcano, showing the process

  4. #4 peak

    instructions and pre-fetching. Relevant Computer Science topics such as models of computation and formal grammars are addressed, and their practical value explained.⏎Share Low-Level Programming<|endoftext|>New Delhi: For possibly the first

  5. #5 peak

    utf8mb4) bytes per character can be used. So if one specifies VARCHAR(50) CHARSET utf8mb4, …[Read more]<|endoftext|>To actually reduce pounds in a healthier manner, it is perfect to burn fat gradually

  6. #6 peak

    also become unstable. In most cases, the treated tooth is then crowned. How exactly to proceed further, the doctor and patient discuss in the root canal aftercare<|endoftext|>Black locust is a member of the legume

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: A question that is a problem, a question that can be solved, the problem can be solved, a problem can be solved... a problem. The problem, as the problem... a problem. The problem can be solved. The problem,

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

    considerations for when and how to take the medication include:⏎- Monopril comes in tablet form. It is taken once a day.⏎- It is best to take it on an empty stomach at least one hour before a meal. However, it

  2. #2 peak

    provide the community with great swimming, boating and fishing opportunities.⏎The Garvey-Glenn drains 17.5 square kilometres of farmland and empties directly into Lake Huron,

  3. #3 peak

    succulents also need nutrients for their growth and development.⏎How to propagate succulents:⏎Cut a single leaf or stem from the succulent very gently and plant this cutting separately.⏎Where to place succulents?⏎Succul

  4. #4 peak

    needs a space in Linux you can use a backward slash followed by a space, to escape the space in a name. If this isn't done, Linux will treat anything after the space as an option or additional command.⏎Finally, escape characters are

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. L12:2347 suppresses
  2. L13:2636 suppresses
  3. L12:1022 excites
  4. L9:1444 excites
  5. L13:1456 excites
  6. L6:487 suppresses
  7. L6:2326 suppresses
  8. L6:2692 excites
  9. L12:1653 suppresses
  10. L10:1395 excites
  11. L10:132 suppresses
  12. L12:525 suppresses
  13. L13:1928 suppresses
  14. L13:1906 suppresses
  15. L6:1829 suppresses
  16. L13:2299 suppresses
  17. L13:792 excites
  18. L12:1066 suppresses
  19. L12:2680 suppresses
  20. L11:487 excites
  21. L6:1001 suppresses
  22. L4:118 weakly suppresses
  23. L13:645 weakly suppresses
  24. L5:1535 weakly excites

Feeds later layers

  1. L16:1934 strongly suppresses
  2. L15:1028 suppresses
  3. L15:1702 suppresses
  4. L18:2830 suppresses
  5. L16:1077 suppresses
  6. L15:1047 suppresses
  7. L17:429 excites
  8. L18:2243 suppresses
  9. L16:1551 suppresses
  10. L15:2633 suppresses
  11. L16:2180 suppresses
  12. L18:2842 suppresses
  13. L16:2157 suppresses
  14. L19:2228 suppresses
  15. L18:2971 suppresses
  16. L18:1299 suppresses
  17. L19:398 suppresses
  18. L16:2612 suppresses
  19. L18:867 suppresses
  20. L18:358 excites
  21. L16:1313 suppresses
  22. L15:27 suppresses
  23. L17:663 suppresses
  24. L15:37 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 -5.58 and -5.351 values between -4.41 and -4.171 values between -4.17 and -3.941 values between -3.47 and -3.241 values between -3.00 and -2.772 values between -2.77 and -2.533 values between -2.53 and -2.303 values between -2.30 and -2.066 values between -2.06 and -1.836 values between -1.83 and -1.599 values between -1.59 and -1.3610 values between -1.36 and -1.1221 values between -1.12 and -0.8931 values between -0.89 and -0.6654 values between -0.66 and -0.42181 values between -0.42 and -0.191586 values between -0.19 and 0.051403 values between 0.05 and 0.28462 values between 0.28 and 0.5283 values between 0.52 and 0.7530 values between 0.75 and 0.993 values between 0.99 and 1.221 values between 1.46 and 1.692 values between 1.69 and 1.931 values between 1.93 and 2.161 values between 2.40 and 2.632 values between 2.86 and 3.101 values between 3.10 and 3.332 values between 3.57 and 3.80offnegativepositiveoften

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. -state
  2. conut
  3. -Identifier
  4. ilitary
  5. .getLog
  6. national
  7. ulatory
  8. uel
  9. .lin
  10. abetic

Suppresses

  1. disposed
  2. 大片
  3. 它们
  4. 这两
  5. 这支
  6. гран
  7. 这只
  8. 这种方法
  9. esto
  10. 這些