Neuron Explorer

Qwen3-0.6B

decimal points inside numbers

A neuron in layer 26 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 dot in numbers like 3.0, 2.5, 1.25 or 10.4, and on version or citation numbers. When it was turned up, the model fell into repeating strings of code-like symbols, which tells us little about the neuron itself.

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

    List of Users.⏎Sometimes, it’s also handy to define a list of multiple Entities and Structures.⏎Until 184.108.40.2

  2. #2 peak

    for forest health, natural resources and silviculture images operated by The Bugwood Network at the University of Georgia and the USDA Forest Service., CC BY 3.

  3. #3 peak

    - Amount of protein: Experts recommend adults over age 65 consume .45 to .68 grams of protein per pound of body weight daily (1.0-1.5 gm/kg/d), depending on

  4. #4 peak

    the FN FAL, the HK G3 and the AK-47.⏎A magazine-fed, select-fire rifle firing the .308 Winchester/7.62x51mm NATO was, at the time, seen as the

  5. #5 peak

    on this website is licensed under a Creative Commons Attribution 2.5 Australia Licence together with any terms, conditions and exclusions as set out in

  6. #6 peak

    all the texts on this journal belong to the respective authors without restrictions.⏎This journal is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License (full legal code).⏎See also our Open Access Policy.

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 know that, well, two of the top top! top! top! top! top! top! top! top! top! top! top! top! top! top! top! top! top! top! top!

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

    return to a shielded base.⏎To continue reading this article, subscribe to receive access to all of newscientist.com, including 20 years of archive content.<|endoftext|>Fly Created To Study How A Normal Cell Turns Cancerous⏎The study

  2. #2 peak

    will review what you’ve submitted and determine whether to revise the article.Join Britannica's Publishing Partner Program and our community of experts to gain a global audience for your work!⏎One of the youngest of the world’s major cities, Johannesburg was founded in

  3. #3 peak

    explaining how they use data to make predictions and offering a comprehensive understanding of this fascinating topic. Get ready to explore the world of decision trees and discover how they make predictions that are accurate, reliable, and effective.⏎II. What are Decision Trees?⏎

  4. #4 peak

    and considering dental treatments might aid in detecting and preventing oral diseases before they become more serious. Supreme Dental CT helps you to keep a beautiful smile and improve your overall health by making these practices a habit.<|endoftext|>Teen Attitude: What is normal?

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. L20:44 strongly excites
  2. L18:2322 strongly suppresses end-of-text token after a finished passage
  3. L17:1088 strongly excites
  4. L17:658 strongly suppresses
  5. L16:235 strongly suppresses
  6. L25:808 strongly suppresses
  7. L15:774 strongly excites
  8. L18:926 excites
  9. L17:1523 suppresses
  10. L17:68 suppresses
  11. L17:1029 excites
  12. L25:118 excites
  13. L16:2510 excites
  14. L15:2675 excites
  15. L16:3064 excites
  16. L15:101 excites
  17. L17:2723 excites
  18. L17:3036 excites end of document boundary
  19. L18:791 suppresses
  20. L15:917 suppresses
  21. L17:2848 suppresses
  22. L14:667 suppresses
  23. L24:36 excites
  24. L17:450 suppresses

Feeds later layers

  1. L27:725 strongly excites
  2. L27:200 suppresses
  3. L27:1078 weakly excites
  4. L27:136 weakly excites
  5. L27:114 weakly suppresses
  6. L27:502 weakly excites
  7. L27:531 weakly excites
  8. L27:220 weakly suppresses end-of-text token between documents
  9. L27:1899 weakly excites
  10. L27:1530 weakly excites
  11. L27:170 weakly suppresses unclear: spaces in mixed-language text
  12. L27:207 weakly suppresses
  13. L27:2795 weakly excites
  14. L27:697 weakly excites
  15. L27:690 weakly suppresses
  16. L27:1554 weakly suppresses
  17. L27:1255 weakly suppresses
  18. L27:1300 weakly excites
  19. L27:313 weakly suppresses
  20. L27:1909 weakly excites
  21. L27:2964 weakly excites
  22. L27:1399 weakly excites
  23. L27:126 weakly suppresses
  24. L27:1089 weakly excites

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 -17.06 and -16.341 values between -10.52 and -9.801 values between -9.80 and -9.071 values between -8.34 and -7.621 values between -5.43 and -4.712 values between -4.71 and -3.981 values between -3.98 and -3.252 values between -3.25 and -2.538 values between -2.53 and -1.8010 values between -1.80 and -1.0785 values between -1.07 and -0.352161 values between -0.35 and 0.381477 values between 0.38 and 1.1168 values between 1.11 and 1.8312 values between 1.83 and 2.5615 values between 2.56 and 3.2922 values between 3.29 and 4.0116 values between 4.01 and 4.7411 values between 4.74 and 5.478 values between 5.47 and 6.191 values between 6.19 and 6.921 values between 6.92 and 7.651 values between 7.65 and 8.371 values between 9.10 and 9.831 values between 11.28 and 12.01offnegativepositiveoften

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. <|endoftext|>
  2. 返回搜狐
  3. […]⏎⏎
  4. ↩
  5. ')}}</
  6. ⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎⏎
  7. ⏎ ⏎ ⏎ ⏎
  8. }}</
  9. ']}}</
  10. [...]⏎⏎

Suppresses

  1. _',
  2. odont
  3. IVITY
  4. Ń
  5. opsis
  6. üstü
  7. полу
  8. .minecraft
  9. adata
  10. ulos