Neuron Explorer

Qwen3-0.6B

closing bracket after an acronym

A neuron in layer 3 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

It fires on the closing bracket that ends a short name or abbreviation written in brackets, such as the ')' in '(WWF)' or '(WHO)'. Turning it up did not bring out this pattern in the sample outputs.

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

    .<|endoftext|>A random variable X has the cumulative distribution function⏎Calculate the variance of X.⏎The answer specifies that the density function is⏎I got the f(x) = x-1 part, and I got how to calculate the variance after you

  2. #2 peak

    is based⏎Two NGOs -- the World Wide Fund for Nature (WWF) and a consortium led by Greenpeace -- have put forward blueprints that could be

  3. #3 peak

    Treaties and Documents⏎1949 Conventions and Additional Protocols, and their Commentaries⏎Historical Treaties and Documents⏎Convention (III) relative to the Treatment of Prisoners of War. Geneva, 12 August

  4. #4 peak

    ) 1999 RQ36.”⏎The asteroid was discovered in 1999 by the Lincoln Near Earth Asteroid Research (LINEAR) survey at MIT’s Lincoln Laboratory. LINEAR is part of NASA’s Near Earth

  5. #5 peak

    problems and other post-COVID complications.⏎The new method developed by KAUST — known as Deep-Lung Parenchyma-Enhancing (DLPE) — overlays

  6. #6 peak

    truth trials, but there’s a lot left to do for women,” said Julissa Mantilla Falcón, Interamerican Commission for Human Rights (CIDH) Commisioner, told The Herald in an interview. “Even after the trials

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: the effect of a previous decision to be in the same time I have been thinking about lately... the effect of a previous decision to be in the same time I have been thinking about lately... the effect of a previous decision to be in the

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

    1989 (Aptian), G. crenulata (Matsumoto, 1938; Albian–lower Cenomanian) and G. maedai sp. nov. (middle to ?

  2. #2 peak

    drug in the world, but did its advent really change society in Britain and even the way we eat?⏎In 1965 something lurking under the meniscus of British society punctured the surface.⏎A man named

  3. #3 peak

    an optical show in the city. By analyzing this case, I will attempt to illustrate the visual novelty, the level of acceptance and then great popularity of the stereoscope compared to other entertainment in this period. Finally, I will present new data concerning

  4. #4 peak

    of the shock of jumps and landings and helps to distribute joint fluid evenly to lubricate and nourish the knee.⏎In the United States, a torn meniscus is the most common reason for knee surgery.⏎The symptoms of a torn men

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. L1:341 suppresses
  2. L0:2146 weakly excites
  3. L2:359 weakly suppresses
  4. L1:86 weakly suppresses
  5. L0:870 weakly excites
  6. L0:1424 weakly excites
  7. L1:2535 weakly suppresses
  8. L2:70 weakly excites
  9. L1:1506 weakly suppresses
  10. L1:1384 weakly excites
  11. L1:1674 weakly excites
  12. L1:367 weakly suppresses
  13. L2:75 weakly suppresses
  14. L1:360 weakly excites
  15. L2:1320 weakly suppresses
  16. L0:70 weakly excites
  17. L2:455 weakly suppresses
  18. L2:1657 weakly excites
  19. L1:1029 weakly excites
  20. L0:2179 weakly suppresses
  21. L1:124 weakly suppresses
  22. L2:780 weakly excites
  23. L1:899 weakly suppresses 'which' after 'find out' or 'choose'
  24. L1:151 weakly excites

Feeds later layers

  1. L8:577 excites
  2. L11:293 excites
  3. L7:6 suppresses
  4. L15:316 suppresses
  5. L13:2620 weakly excites
  6. L11:55 weakly excites
  7. L14:756 weakly excites
  8. L4:1013 weakly suppresses
  9. L5:38 weakly suppresses
  10. L15:616 weakly excites closing parenthesis ending an aside
  11. L21:1024 weakly suppresses
  12. L10:860 weakly excites
  13. L13:685 weakly excites
  14. L8:1789 weakly excites
  15. L6:37 weakly suppresses
  16. L8:455 weakly excites
  17. L10:679 weakly excites
  18. L17:2516 weakly suppresses
  19. L17:382 weakly excites closing bracket after a fact in parentheses
  20. L14:2097 weakly excites
  21. L4:1394 weakly suppresses
  22. L12:187 weakly suppresses
  23. L9:29 weakly suppresses opening parenthesis before an acronym
  24. L14:2816 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 -1.55 and -1.371 values between -1.37 and -1.201 values between -1.20 and -1.022 values between -0.85 and -0.671 values between -0.67 and -0.5014 values between -0.50 and -0.32293 values between -0.32 and -0.152129 values between -0.15 and 0.031315 values between 0.03 and 0.21117 values between 0.21 and 0.387 values between 0.38 and 0.564 values between 0.56 and 0.731 values between 0.73 and 0.911 values between 0.91 and 1.083 values between 1.08 and 1.262 values between 1.26 and 1.431 values between 1.43 and 1.613 values between 1.78 and 1.963 values between 1.96 and 2.131 values between 2.13 and 2.311 values between 2.31 and 2.482 values between 2.48 and 2.661 values between 2.83 and 3.011 values between 3.18 and 3.361 values between 4.24 and 4.412 values between 5.29 and 5.46offnegativepositiveoften

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. reta
  2. aye
  3. sx
  4. eczy
  5. =wx
  6. gx
  7. ẫ
  8. maries
  9. ayne
  10. Raqqa

Suppresses

  1. ories
  2. toll
  3. Pipe
  4. 产出
  5. тро
  6. Product
  7. Install
  8. �
  9. Installer
  10. hog