Neuron Explorer

Qwen3-0.6B

you and your in how-to addresses

A neuron in layer 4 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on 'you' or 'your' when a piece of writing speaks directly to the reader, as in instructions or calls to action. It is strongly negative on the 's in 'Here's', and turning it up did not produce an obvious change in the model's replies.

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

    Add your thoughts below. You can sign in with Google, Twitter, Facebook, Yahoo!

  2. #2 peak

    of the mark. You can do this by using a percentage calculator. You can use it to find the value of a number. Once you have the value, you can write it down in terms of the percent. It’s easy to do this

  3. #3 peak

    allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.<|endoftext|>Permanova’s knowledge centre for industrial lasers⏎

  4. #4 peak

    Pilot, your job as the Flight Test Instrumentation Engineer is to create gauges, cameras,

  5. #5 peak

    . © 2006 Elsevier Ltd. All rights reserved.<|endoftext|>Are you tired of being sick and tired of being sick and tired? Then, get in the driver’s seat on managing your health by adopting a new lifestyle. Join

  6. #6 peak

    stream. Then, I came face to face with flying creatures, fighting to get near me. I even endured webmakers spinning my hair into a shiny maze.⏎Where did I go? Into the woods, of course. Why? I wanted to

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: 1. 20000000000000000000000000000000000000000000

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

    combination which merged was the C<|endoftext|>How lego does it today.⏎I found little information about Autodesk Maya being the 3D modeling software being used.⏎Here`s a blurb about their current design process :⏎Primary concept and development work takes place

  2. #2 peak

    can also be beneficial for children. Here’s how it works: With your child, choose a few categories for saving and budget

  3. #3 peak

    and history, including the Holocaust, are presented in Eastern Europe and the way African American (including slave) history and heritage are presented in the Charleston area.⏎Here's a brief essay I wrote, for the web site of the Drayton Hall

  4. #4 peak

    this spectacle, drawing on evidence gathered by NASA’s Galaxy Evolution Explorer and the Pan-STARRS1 telescope located in Hawaii. Here’s how NASA describes what you’re seeing in the clip above:⏎Some of the stellar

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. L2:765 excites
  2. L3:524 weakly suppresses
  3. L3:1135 weakly excites
  4. L2:126 weakly excites
  5. L3:828 weakly suppresses
  6. L3:1306 weakly suppresses
  7. L2:36 weakly excites
  8. L1:2470 weakly suppresses
  9. L3:464 weakly excites
  10. L0:1677 weakly excites
  11. L2:75 weakly suppresses
  12. L0:2726 weakly suppresses
  13. L0:2976 weakly suppresses
  14. L1:1580 weakly excites
  15. L1:1014 weakly excites
  16. L3:487 weakly suppresses
  17. L3:342 weakly excites
  18. L3:774 weakly suppresses
  19. L0:3009 weakly suppresses
  20. L3:889 weakly suppresses
  21. L2:70 weakly excites
  22. L3:320 weakly excites
  23. L0:400 weakly suppresses
  24. L2:632 weakly excites

Feeds later layers

  1. L5:1620 suppresses
  2. L8:2382 weakly suppresses wording that leads into 'below' or 'find it'
  3. L9:2586 weakly suppresses
  4. L15:2701 weakly suppresses
  5. L10:590 weakly excites
  6. L21:2484 weakly excites
  7. L7:2507 weakly excites
  8. L22:1039 weakly excites
  9. L19:1475 weakly suppresses
  10. L9:2092 weakly suppresses
  11. L24:1948 weakly suppresses
  12. L12:2209 weakly suppresses
  13. L6:821 weakly suppresses
  14. L22:752 weakly excites
  15. L5:2927 weakly excites
  16. L13:1635 weakly excites
  17. L11:1693 weakly suppresses
  18. L20:1192 weakly suppresses
  19. L21:2872 weakly excites
  20. L17:2993 weakly suppresses
  21. L20:49 weakly suppresses
  22. L6:2167 weakly excites
  23. L20:1321 weakly suppresses
  24. L13:2326 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 -0.83 and -0.781 values between -0.73 and -0.681 values between -0.68 and -0.631 values between -0.63 and -0.586 values between -0.58 and -0.533 values between -0.53 and -0.484 values between -0.48 and -0.435 values between -0.43 and -0.3912 values between -0.39 and -0.3424 values between -0.34 and -0.2973 values between -0.29 and -0.24136 values between -0.24 and -0.19252 values between -0.19 and -0.14390 values between -0.14 and -0.09591 values between -0.09 and -0.04818 values between -0.04 and 0.01639 values between 0.01 and 0.06408 values between 0.06 and 0.11248 values between 0.11 and 0.16126 values between 0.16 and 0.2181 values between 0.21 and 0.2639 values between 0.26 and 0.3025 values between 0.30 and 0.355 values between 0.35 and 0.403 values between 0.40 and 0.454 values between 0.45 and 0.502 values between 0.50 and 0.551 values between 0.55 and 0.602 values between 0.60 and 0.652 values between 0.65 and 0.701 values between 0.70 and 0.752 values between 0.80 and 0.851 values between 0.85 and 0.901 values between 1.09 and 1.14offnegativepositiveoften

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. ilage
  2. IonicModule
  3. itative
  4. continued
  5. Nile
  6. uelle
  7. uguay
  8. ilinear
  9. ivariate
  10. ijo

Suppresses

  1. 给您
  2. çek
  3. ayan
  4. aan
  5. 教你
  6. 告诉你
  7. 给你
  8. list
  9. 把你
  10. 帮你