Neuron Explorer

Qwen3-0.6B

'of' after a category noun

A neuron in layer 6 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 fires on the word 'of' when it follows a word for a kind or count of things, like 'types of', 'species of' or 'classes of'. Its weakest side shows up in phrases like 'all of', and turning it up did not produce any clear change we could tie to this habit.

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

    high altitudes are less than ideal for reptiles. These animals are ectothermic, meaning they gather body heat from their surroundings. There are only three species of reptiles in the park, with a fourth suspected to be present.⏎The common

  2. #2 peak

    .⏎Since 2019 hundreds of trees and shrubs have been planted by the group close to the Bucketts. Koalas eat many different species of eucalypts

  3. #3 peak

    .5km long, which is the same as 1,517 tyrannosaurus rexes laid end to end⏎- 10 species of bat are regularly seen in Bridgend County Borough<|endoftext|>Measurement, as a method to

  4. #4 peak

    08:41:37⏎The prospect of untreatable gonorrhoea has provoked alarm around the world, and there are no new classes of antibiotics in development.⏎In this week's Scrubbing Up column, Peter Green

  5. #5 peak

    ans with 10 limbs.⏎- Crab’s shell is a skeleton on the outside of its body.⏎- There are about 5,000 species of crabs.⏎- Crabs have claws on their front two legs.⏎- Cr

  6. #6 peak

    detector, the quicker you will be alerted to smoke or fire.⏎There are currently two different types of technologies used to detect smoke and fire. Photoelectric smoke detectors rely on an electric

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 first two, the first two, the first two, the first two, the first two, the first two. (This) the the the the the the the the the the the the the the the the the the the the 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

    through all of it.⏎So what’s the key? Goals have to be actionable. That is

  2. #2 peak

    Buy second hand where possible and upcycle⏎- 9. Hang your clothes to dry rather than using a tumble dryer⏎- 10. Use all of your food leftovers⏎1. Reduce your meat intake⏎You don’t have to

  3. #3 peak

    keep on your desktop. And all the files get the “.mind” extension added to their filenames. Thus, the infection will not block you from accessing all of your files. However, it can do a lot of other things that can be

  4. #4 peak

    you’re internet dating a person from a further country, you will need to understand the customs that they are from so you can learn how to communicate properly with all of them. This can help you to enjoy your romance and avoid

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. L4:1763 suppresses
  2. L3:31 excites
  3. L3:6 suppresses
  4. L3:763 suppresses
  5. L4:4 weakly suppresses
  6. L4:731 weakly excites
  7. L5:198 weakly excites unclear: weak "and" in lists
  8. L5:472 weakly excites
  9. L1:1796 weakly suppresses
  10. L0:784 weakly excites 'of' after 'think' or 'out'
  11. L3:624 weakly excites
  12. L3:1440 weakly excites
  13. L5:2838 weakly excites
  14. L2:1439 weakly suppresses
  15. L1:59 weakly suppresses 'the' after a verb or preposition
  16. L3:722 weakly suppresses
  17. L1:2567 weakly excites
  18. L5:38 weakly suppresses
  19. L2:2226 weakly suppresses
  20. L1:1430 weakly excites
  21. L4:1517 weakly excites
  22. L0:916 weakly suppresses
  23. L5:1372 weakly excites
  24. L5:1747 weakly suppresses

Feeds later layers

  1. L7:207 strongly suppresses
  2. L8:252 strongly suppresses 'of' after 'each' or 'one' or 'all'
  3. L14:280 strongly suppresses
  4. L16:1320 strongly suppresses
  5. L10:167 excites
  6. L8:636 excites
  7. L14:313 suppresses
  8. L17:985 excites
  9. L19:663 excites
  10. L13:975 excites
  11. L16:896 excites
  12. L17:1056 suppresses
  13. L19:1129 excites
  14. L19:1013 suppresses
  15. L19:1023 excites
  16. L21:274 suppresses
  17. L18:217 suppresses
  18. L15:557 suppresses
  19. L25:544 suppresses
  20. L21:310 suppresses
  21. L25:16 suppresses
  22. L7:173 excites
  23. L18:376 suppresses
  24. L18:602 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 -5.18 and -4.931 values between -3.96 and -3.721 values between -3.72 and -3.481 values between -3.24 and -2.991 values between -2.51 and -2.271 values between -1.30 and -1.054 values between -1.05 and -0.816 values between -0.81 and -0.5744 values between -0.57 and -0.32727 values between -0.32 and -0.082721 values between -0.08 and 0.16357 values between 0.16 and 0.4021 values between 0.40 and 0.659 values between 0.65 and 0.895 values between 0.89 and 1.133 values between 1.37 and 1.621 values between 1.62 and 1.861 values between 1.86 and 2.101 values between 2.34 and 2.591 values between 2.59 and 2.831 values between 4.29 and 4.53offnegativepositiveoften

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. inion
  2. 成品
  3. result
  4. chte
  5. olarity
  6. biology
  7. RAM
  8. carp
  9. résultat
  10. clamation

Suppresses

  1. cade
  2. odesk
  3. 的一些
  4. 的一面
  5. dì
  6. 的时间
  7. pread
  8. 的缘
  9. 一些
  10. greSQL