Neuron Explorer

Qwen3-0.6B

unclear: brackets and parentheses in text

A neuron in layer 11 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

The neuron is hard to pin down. It fires most on the word right after an empty pair of square brackets, which often marks a spot where text was removed, and it goes strongly negative on opening parentheses in citations. When turned up, the model produced strings of parentheses and numbers, so it seems tied to bracket and punctuation patterns.

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

    to great violence, with swelling, soreness to the touch, and sometimes redness of the skin…. This is a very formidable and extremely painful disease, [and] generally continues for many weeks…. The consequences of this disorder are often painfully

  2. #2 peak

    , stupidity and anger,” said Mangku Puseh Juga, one of 36 priests on hand to sanctify the temple.⏎“The main goal is (to) balance the macro world of nature and the micro world within each human.”⏎

  3. #3 peak

    [and] having a forest products industry that begins to move toward oaks. If these come true, that's going to change the nature of the industry and [it] will [have to] adapt. [Awareness of climate change]

  4. #4 peak

    . with a commission visiting the battlefield on September 28, 1926.:'27 North Carolina appropriated $50,000[when?] to purchase and landscape the site and to commission Gutzon Borgl

  5. #5 peak

    said that the image of Marx has been imprinted in his heart since he was a child, adding that “Marx molded by a Chinese artist contains emotion and [warmth].”<|endoftext|>Citadel of Herat⏎This citadel, which

  6. #6 peak

    which is really useful for us, because if you have the animals behave in the same way all the time, there’s no room for flexibility in the face of [climate] change.”⏎Even with a colony containing hundreds of

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: if I were to put this into a system, I can't think of any way to think about it. Can you think of a way to put it in a system that could be something like this. For example, this is similar to

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

    . Biblical Theology "Biblical theology is the study of divine revelation as File Size: KB.⏎Get this from a library! Principles of biblical interpretation: (sacred hermeneutics). [Louis Berkhof] --

  2. #2 peak

    6.⏎Silver, Larry. “Nature and Nature’s God: Landscape and Cosmos of Albrecht Altdorfer.” Art Bulletin 81.2 (June 1999): 194–214.⏎

  3. #3 peak

    Historical, Doctrinal, and Practical Theology (3rd ed.) (edited by Philip Schaff) Vol. II at pages 684–

  4. #4 peak

    ETA⏎No references found.⏎Citations of this work BETA⏎No citations found.⏎Similar books and articles⏎Piero Antuono & Jan Beyer (19

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. L10:66 suppresses opening parenthesis after a citation or term
  2. L4:64 excites
  3. L10:2562 excites
  4. L5:1499 excites opening bracket in math and definitions
  5. L7:2152 suppresses
  6. L10:893 excites
  7. L3:263 suppresses
  8. L5:2795 weakly excites
  9. L8:354 weakly excites opening parenthesis before a citation year
  10. L6:144 weakly excites
  11. L1:155 weakly suppresses
  12. L10:1883 weakly excites closing parenthesis after a number or unit
  13. L10:2376 weakly excites
  14. L1:919 weakly excites
  15. L3:2199 weakly suppresses
  16. L0:1951 weakly suppresses
  17. L0:2963 weakly excites
  18. L4:45 weakly suppresses
  19. L1:102 weakly excites
  20. L5:75 weakly excites
  21. L5:1890 weakly suppresses
  22. L7:668 weakly suppresses
  23. L3:57 weakly excites
  24. L8:18 weakly suppresses

Feeds later layers

  1. L12:2933 strongly excites
  2. L12:1287 strongly excites
  3. L16:1967 strongly excites
  4. L19:2339 excites
  5. L15:608 excites
  6. L18:980 excites opening parenthesis after a noun phrase
  7. L12:1502 excites
  8. L14:1516 excites
  9. L13:1698 excites
  10. L18:2928 excites
  11. L26:431 excites
  12. L16:1123 weakly excites
  13. L17:1509 weakly excites
  14. L16:1513 weakly excites opening parenthesis before a numbered list
  15. L13:2571 weakly excites commas before a continuation clause
  16. L18:682 weakly excites
  17. L20:1451 weakly excites
  18. L16:869 weakly excites
  19. L22:2877 weakly excites
  20. L14:1318 weakly excites "both" and "either" before alternatives
  21. L20:2173 weakly excites
  22. L27:109 weakly excites
  23. L14:3046 weakly excites
  24. L19:2354 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 -0.97 and -0.891 values between -0.82 and -0.751 values between -0.75 and -0.671 values between -0.67 and -0.602 values between -0.60 and -0.5217 values between -0.52 and -0.4539 values between -0.45 and -0.3891 values between -0.38 and -0.30232 values between -0.30 and -0.23477 values between -0.23 and -0.15944 values between -0.15 and -0.081558 values between -0.08 and -0.01458 values between -0.01 and 0.0744 values between 0.07 and 0.1423 values between 0.14 and 0.224 values between 0.22 and 0.292 values between 0.36 and 0.442 values between 0.44 and 0.511 values between 0.51 and 0.591 values between 0.59 and 0.662 values between 0.66 and 0.731 values between 0.73 and 0.811 values between 1.03 and 1.102 values between 1.10 and 1.181 values between 1.25 and 1.321 values between 1.69 and 1.771 values between 1.91 and 1.99offnegativepositiveoften

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. 서
  2. abbrev
  3. h
  4. outer
  5. ogg
  6. 所说的
  7. собственно
  8. out
  9. A
  10. /downloads

Suppresses

  1. ega
  2. .PO
  3. WOOD
  4. MX
  5. dex
  6. .dx
  7. Dx
  8. .lambda
  9. пуск
  10. LEX