Neuron Explorer

Qwen3-0.6B

web page boilerplate headings

A neuron in layer 6 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron fires most strongly on the headings and navigation text that website templates add, such as 'Top Suggestions' on worksheet pages and 'Results 1 of 7 matches'. Its other reactions are weaker and less clear. Turning it up or down didn't produce an obvious theme in the written 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

    :⏎Top Suggestions Vocabulary Worksheet Mix :⏎Vocabulary Worksheet Mix Whether you re planning a summer lesson or

  2. #2 peak

    22, at 888-835-5322 (TTY) or at email@example.com.<|endoftext|>Table of Contents :⏎Top Suggestions Senior Kindergarten Writting Worksheet :⏎Senior Kindergarten Writting Worksheet In

  3. #3 peak

    (6-8) Show all Grade Level⏎Resource Type: Activities⏎Results 1 - 7 of 7 matches⏎From Grid to Home⏎Lisa Gardiner, National Center for Atmospheric Research; Marie

  4. #4 peak

    University of Cambridge. All rights reserved.⏎'Trick or Treat' printed from http://nrich.maths.org/⏎Emily and Tristan from St James' School⏎answered the whole problem as follows:⏎I think the solution

  5. #5 peak

    and then its right child.⏎T / \ I S / \ D E<|endoftext|>Original URL: http://www.theregister.co.uk/2006/08/21/mars_geysers/⏎Martian pole freck

  6. #6 peak

    whole current is then transferred to the trip coil which operates to trip the circuit breaker.⏎Email Based, Online Homework Assignment Help in Classification Of Relays⏎Transtutors is the best place to get answers to all your doubts regarding classification of relays

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: So, the problem is to figure out the solution, but here's the problem. Now, let's say that we have a time machine and a number of things: - The key things are: - The problem is

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

    is very familiar with the role of standards. Well-known standards, such as ISO 17799 for IT security and the ISO 9000 series for management, establish guidelines and general principles that codify industry best practices.

  2. #2 peak

    will see substantive progress in Cancun towards an internationally binding agreement.⏎This hope was echoed by a group of 30 young climate champions

  3. #3 peak

    and configuration of the garden complex for the entire territory of Flanders.⏎Citizen science in support⏎A related STEREO project (type 'Dissemination and Support') named GARVAL has just been

  4. #4 peak

    .⏎The next step is to determine the identifier for the drive you want to make an image of. In the terminal window, type in the command⏎sudo fdisk -l⏎Our 1GB drive is /<|endoftext|>During the period from

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. L5:268 excites
  2. L5:256 excites
  3. L4:67 excites
  4. L5:181 excites
  5. L5:273 suppresses
  6. L4:138 excites
  7. L5:312 excites
  8. L5:148 suppresses
  9. L4:143 excites
  10. L3:79 suppresses
  11. L3:268 excites
  12. L4:32 weakly excites
  13. L5:140 weakly excites
  14. L4:437 weakly excites
  15. L5:1952 weakly suppresses
  16. L4:233 weakly suppresses line break in the middle of a sentence
  17. L2:700 weakly suppresses
  18. L3:1073 weakly suppresses
  19. L2:605 weakly suppresses
  20. L0:894 weakly excites
  21. L1:728 weakly excites
  22. L0:651 weakly suppresses
  23. L2:774 weakly excites
  24. L0:1561 weakly suppresses

Feeds later layers

  1. L8:211 suppresses
  2. L8:239 excites
  3. L9:675 suppresses
  4. L8:445 excites
  5. L8:449 suppresses
  6. L8:80 excites
  7. L15:387 weakly suppresses
  8. L7:2540 weakly suppresses
  9. L18:998 weakly excites
  10. L10:230 weakly suppresses
  11. L8:235 weakly suppresses
  12. L10:2315 weakly excites line break inside wrapped text
  13. L7:1240 weakly suppresses
  14. L13:1213 weakly suppresses
  15. L8:308 weakly suppresses
  16. L8:310 weakly suppresses
  17. L16:2145 weakly suppresses
  18. L8:667 weakly suppresses
  19. L9:2053 weakly excites
  20. L15:326 weakly excites
  21. L9:808 weakly excites
  22. L14:908 weakly suppresses
  23. L13:583 weakly excites
  24. L8:180 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 -0.76 and -0.721 values between -0.53 and -0.491 values between -0.49 and -0.451 values between -0.45 and -0.413 values between -0.38 and -0.348 values between -0.34 and -0.3010 values between -0.30 and -0.2652 values between -0.26 and -0.2280 values between -0.22 and -0.18136 values between -0.18 and -0.14227 values between -0.14 and -0.10343 values between -0.10 and -0.07451 values between -0.07 and -0.03625 values between -0.03 and 0.01527 values between 0.01 and 0.05517 values between 0.05 and 0.09335 values between 0.09 and 0.13247 values between 0.13 and 0.17152 values between 0.17 and 0.2197 values between 0.21 and 0.2538 values between 0.25 and 0.2834 values between 0.28 and 0.327 values between 0.32 and 0.368 values between 0.36 and 0.403 values between 0.40 and 0.441 values between 0.44 and 0.481 values between 0.52 and 0.561 values between 0.71 and 0.751 values between 0.75 and 0.79offnegativepositiveoften

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. Ge
  2. weights
  3. Ge
  4. itt
  5. Se
  6. _weights
  7. ид
  8. (event
  9. 树
  10. dates

Suppresses

  1. ABCDEFGHIJKLMNOP
  2. opies
  3. iative
  4. ril
  5. royalties
  6. ACITY
  7. icism
  8. sublic
  9. Pinterest
  10. CDDL