Neuron Explorer

Qwen3-0.6B

'one or more' before a count noun

A neuron in layer 21 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 lights up on the word 'or' in the phrase 'one or more', and on similar phrases like 'more than one'. When it was turned up, the model started repeating 'than' and 'multiple', so it seems tied to the idea of quantity.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    me that you specify where each of your posts falls amongst the categories listed below.⏎The same post can count for more than one category, but if you keep re-using the same handful of posts

  2. #2 peak

    good solution. For instance, here is a piece of a docstring that I wrote yesterday:⏎Return a dict (which I often call "response") with one or more of theThe

  3. #3 peak

    myself 50 Bitcoins to my address A" (plus transaction fees from the block).⏎- When you want to spend Bitcoins, you have to point to one or more transactions that show where you got them from, and say how many Bitcoins you

  4. #4 peak

    the shoulders of sows kept in intensive production systems, in particular in farrowing crates.⏎- Lameness: When lame, a pig is unable to use one or more limbs in a normal manner. Lame pigs often experience pain, and may

  5. #5 peak

    become mandatory when they have been incorporated into a business contract or incorporated into regulations.⏎What is a code ?⏎A code is a standard that has been adopted by one or more governmental bodies and has the force of law.⏎Why are standards effective ?⏎Stand

  6. #6 peak

    site matches the name of the publisher.⏎Omitting the city of publication. Include the the city of publication only if the version of the source differs when published in a different country. (Example: British editions of books

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 two-silloy (3,2,1,2,4,6,8,12,6,4,2,4,2,4,6,12,6,4,2,4,

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

    a longtime favorite for sleep-deprived people around the world. Arguably, the most famous and relevant of these compounds is 1,3,7-trimethylxanthine, better known as caffeine.⏎It’s pretty common knowledge that

  2. #2 peak

    grasslands, and dry deserts.⏎Pads, flowers, and fruits⏎Flavonoids, octadecadienoic acid, and hexadecanoic acid⏎Most species of prickly pears have similar effective and safe

  3. #3 peak

    ROMATIC COMPOUNDS ARE THOSE ORGANIC COMPOUNDS WHICH HAVE BENZENE RING (CYCLOHEXA-1,3,5-TRIENE). AROMATIC COMPOUNDS ARE ALWAYS

  4. #4 peak

    components and patterns on both sides of the boards. A double layer PCB is used on devices such as vending machines, phone systems, car dashboards, etc.⏎- Multi-Layer PCB – These types of PCB have more than three layers of

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. L20:1724 strongly suppresses
  2. L16:2111 strongly excites
  3. L17:2309 suppresses
  4. L18:1732 excites
  5. L18:2927 suppresses
  6. L11:1291 suppresses
  7. L14:2783 suppresses the word 'same' in 'the same'
  8. L13:1507 suppresses
  9. L20:2636 excites
  10. L13:1090 suppresses
  11. L14:2018 excites
  12. L17:2556 suppresses
  13. L16:1640 suppresses
  14. L17:1786 excites
  15. L13:2199 excites
  16. L14:1990 excites
  17. L10:2300 suppresses
  18. L20:1788 suppresses
  19. L12:2116 suppresses
  20. L17:1727 excites
  21. L19:2929 suppresses
  22. L17:3065 excites
  23. L17:904 excites
  24. L17:1789 suppresses

Feeds later layers

  1. L26:1784 strongly excites
  2. L23:888 strongly excites
  3. L25:1183 strongly excites
  4. L27:810 excites
  5. L26:1692 excites
  6. L22:825 excites 'for' before a span of time
  7. L23:751 excites
  8. L25:1756 excites
  9. L24:1576 excites
  10. L23:538 excites
  11. L24:2350 excites
  12. L27:712 weakly excites
  13. L23:231 weakly excites
  14. L27:948 weakly excites
  15. L23:408 weakly excites
  16. L23:1543 weakly excites 'up' in 'up to' limits
  17. L22:816 weakly suppresses
  18. L26:1724 weakly excites
  19. L24:1690 weakly excites
  20. L24:680 weakly excites
  21. L25:472 weakly excites
  22. L22:1016 weakly suppresses
  23. L25:2409 weakly excites
  24. L26:1089 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 -14.55 and -13.853 values between -4.74 and -4.041 values between -4.04 and -3.342 values between -2.64 and -1.9410 values between -1.94 and -1.24309 values between -1.24 and -0.542451 values between -0.54 and 0.16981 values between 0.16 and 0.8696 values between 0.86 and 1.5622 values between 1.56 and 2.267 values between 2.26 and 2.967 values between 2.96 and 3.665 values between 3.66 and 4.363 values between 4.36 and 5.063 values between 5.76 and 6.462 values between 6.46 and 7.161 values between 7.16 and 7.863 values between 7.86 and 8.561 values between 12.76 and 13.46offnegativepositiveoften

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. multiple
  2. multiple
  3. Multiple
  4. Multiple
  5. 多个
  6. _multiple
  7. 多
  8. MULT
  9. 多家
  10. mú

Suppresses

  1. tram
  2. war
  3. 嗡
  4. ierge
  5. endencies
  6. onz
  7. 三
  8. 六
  9. 四
  10. ---------- ⏎