Neuron Explorer

Qwen3-0.6B

line break in the middle of a sentence

A neuron in layer 4 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 weakly responds to a line break that cuts a sentence in two, the kind you see in text copied from a PDF or a narrow column. Its negative side is weak and not clearly meaningful. Turning it up gave scrambled text, so it tells us little about 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

    any of the other auberges were situated on a piazza, is not⏎certain but shall be discussed later on.⏎Unfortunately, only three of⏎Cassar's seven auberges have survived. The German aub

  2. #2 peak

    - Shin-to ('the way of the gods'), as opposed to Butsu-dō ('the way of Buddha')." (2)⏎On the left is a very⏎small detail from a pentaptych by Hokusai

  3. #3 peak

    was responsible for the defence of the new city. At the same time, the⏎hostels had to have a central location.⏎Cassar's predecessor and⏎teacher, Francesco Laparelli, had already planned to build each auberge

  4. #4 peak

    ⏎review new products and technologies, select a particular supplier,⏎get details on a specific application, or access a library of solar⏎“Much of today’s advanced⏎design solutions require more than a list of part numbers and⏎suppliers in stock

  5. #5 peak

    Sensee River after capturing⏎the villages of Cherisy and Vis-en-Artois.⏎The 3rd Division met with⏎stubborn opposition but succeeded in capturing Bois de Vert, Bois du S

  6. #6 peak

    too⏎You can have mutable values in the IO or (ST s) monads, if you need them.⏎> That is, I can't change a⏎> component of somethingNew without creating y. The point is there's⏎>

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:

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

    coordinator Juan⏎Pedro Vela Arreola of the Alianza de Ribereños y⏎Armadores, an association of fishermen and producers in Mexico.⏎"Fishermen are desperate."⏎Gilly blames the most recent El Niño in the

  2. #2 peak

    May morning to view the medows gay,⏎I met a pretty dairy-maid all in the new mown hay.⏎* 4 *⏎As I was ridding out one day I saw some pooks of hay;⏎

  3. #3 peak

    yells,⏎A Yankee they had found,⏎With curses bid me “face about,”⏎And march away to town.⏎They marched me down with many more,⏎Through mud and mire to town,⏎And there with bars and prisons

  4. #4 peak

    household not naturalized."⏎- Homesteading Land The person had to have initiated the⏎naturalization process to be eligible for free land through homesteading.⏎- Voter Registration Lists Is

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. L3:428 excites
  2. L3:468 excites
  3. L3:2 suppresses
  4. L3:65 excites
  5. L3:20 suppresses
  6. L3:1 suppresses
  7. L0:46 suppresses
  8. L3:673 suppresses
  9. L3:421 excites
  10. L3:298 suppresses
  11. L3:1598 suppresses
  12. L3:225 suppresses
  13. L3:1011 suppresses
  14. L2:154 weakly excites
  15. L3:42 weakly excites
  16. L3:740 weakly excites
  17. L3:1093 weakly suppresses
  18. L3:559 weakly excites
  19. L0:971 weakly excites
  20. L1:727 weakly suppresses
  21. L0:1967 weakly suppresses
  22. L3:141 weakly excites
  23. L3:1622 weakly suppresses
  24. L3:128 weakly excites

Feeds later layers

  1. L6:203 strongly excites
  2. L6:102 strongly suppresses
  3. L5:140 strongly excites
  4. L5:312 strongly excites
  5. L5:181 strongly suppresses
  6. L8:211 strongly excites
  7. L5:273 strongly suppresses
  8. L9:675 strongly excites
  9. L8:445 suppresses
  10. L8:239 suppresses
  11. L8:530 suppresses
  12. L5:268 suppresses
  13. L9:2053 suppresses
  14. L8:80 suppresses
  15. L5:148 suppresses
  16. L8:28 excites
  17. L9:606 suppresses
  18. L11:1018 excites
  19. L8:235 excites
  20. L9:2548 excites
  21. L11:35 excites
  22. L10:1710 excites
  23. L8:979 suppresses
  24. L8:310 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.

2 values between -1.40 and -1.331 values between -1.13 and -1.062 values between -1.06 and -0.993 values between -0.99 and -0.933 values between -0.93 and -0.865 values between -0.86 and -0.795 values between -0.79 and -0.724 values between -0.72 and -0.6511 values between -0.65 and -0.599 values between -0.59 and -0.5216 values between -0.52 and -0.4516 values between -0.45 and -0.3819 values between -0.38 and -0.3238 values between -0.32 and -0.2580 values between -0.25 and -0.18222 values between -0.18 and -0.11668 values between -0.11 and -0.051925 values between -0.05 and 0.02577 values between 0.02 and 0.09163 values between 0.09 and 0.1668 values between 0.16 and 0.2329 values between 0.23 and 0.2911 values between 0.29 and 0.366 values between 0.36 and 0.4310 values between 0.43 and 0.506 values between 0.50 and 0.562 values between 0.56 and 0.631 values between 0.63 and 0.702 values between 0.77 and 0.841 values between 0.90 and 0.972 values between 1.04 and 1.111 values between 1.24 and 1.31offnegativepositiveoften

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. resco
  2. kel
  3. utely
  4. reflexivity
  5. emás
  6. ApplicationController
  7. createState
  8. theano
  9. SCRIPTION
  10. ******************************************************************************⏎

Suppresses

  1. 质量
  2. люд
  3. uD
  4. izen
  5. 触动
  6. after
  7. wj
  8. واجب
  9. 质量问题
  10. D