Neuron Explorer

Qwen3-0.6B

'of' after a noun phrase

A neuron in layer 10 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 'of' when it comes right after a noun, as in 'the time of Moses' or 'the 34th day of 2013'. The same word in some other 'of' phrases pushes it strongly negative, so it is not a clean signal. Turning it up produced rambling text that did not clearly relate to the word 'of'.

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

    hearts or arteries preserved enough to study. Of those, nine had evidence of blockage from atherosclerosis. “This disease has been around since before the time of Moses,” said Randall Thompson

  2. #2 peak

    of Pakistan borders Afghanistan, India, Iran. Gross domestic product (GDP) is of about 315000 millions of dollars dollars.⏎K2 is

  3. #3 peak

    .noaanews.noaa.gov⏎– Erich Luening⏎Print this page<|endoftext|>Today is Sunday, Feb. 3, the 34th day of 2013. There are 331 days left in the

  4. #4 peak

    s) have not been enumerated in any of the Census since 1951 till day”. The submission was in response to a plea by the State Government of Maharashtra seeking a direction to the Census Department to collect information on Backward Class of

  5. #5 peak

    After Einstein’s death, his brain was preserved for future study. Scientists were naturally curious to see how the brain of this genius compared with the brain of a person of ordinary intelligence. Would there be an abundance of neurons (grey matter) or some

  6. #6 peak

    creating "cricket reactors"—enclosed cricket breeding contraptions fed by recycled organic wastes like lawn clippings and kitchen scraps.⏎His apartment contains one of a few working prototypes. (His wife almost kicked him out when a few escaped,

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 mind and the brain. The brain. The brain. The brain as a subject. The brain, the brain as a subject, the brain. The brain. The brain as a subject. The brain. The brain. The brain.

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

    ) discusses the long term use of fossil fuels, further exploring the notion that certain reserves of oil, gas and coal should not be extracted and used due to concerns about rising

  2. #2 peak

    failure, the toxic components of lily of the valley affect your dog's heart. These components are called cardiac

  3. #3 peak

    lives. If women united internationally and with the wider struggles of working<|endoftext|>|Part of a series on|⏎|Types of creationism|⏎|

  4. #4 peak

    situated at the valley of the mountain. These statues represent the style of Phnom Da. Among the five

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. L4:4 suppresses
  2. L9:1428 excites
  3. L8:1003 suppresses "the" or "of" before a noun phrase
  4. L4:1686 excites
  5. L5:363 excites
  6. L3:315 excites 'of' after a noun phrase
  7. L1:679 suppresses
  8. L2:895 weakly suppresses
  9. L6:357 weakly suppresses
  10. L9:2264 weakly excites 'of' after superlatives and 'the X of'
  11. L7:1152 weakly suppresses
  12. L4:315 weakly suppresses
  13. L8:668 weakly suppresses
  14. L5:384 weakly suppresses
  15. L3:518 weakly excites
  16. L2:2786 weakly excites
  17. L9:1446 weakly suppresses
  18. L5:1614 weakly suppresses
  19. L5:1404 weakly suppresses
  20. L1:2033 weakly suppresses
  21. L9:1942 weakly suppresses
  22. L8:688 weakly excites
  23. L6:915 weakly suppresses
  24. L4:561 weakly excites

Feeds later layers

  1. L22:229 suppresses "of" after a noun in "X of Y"
  2. L13:132 suppresses "of" after a noun
  3. L12:2757 suppresses "of" after a measure like inches or teaspoon
  4. L13:1062 suppresses
  5. L12:2022 suppresses 'of' after type or class nouns
  6. L15:801 suppresses 'of' inside organisation names
  7. L12:850 excites
  8. L15:232 excites
  9. L17:1835 suppresses
  10. L17:385 suppresses
  11. L14:2075 excites
  12. L16:805 suppresses
  13. L17:1507 suppresses
  14. L17:317 suppresses 'of' after a noun phrase
  15. L21:399 suppresses
  16. L14:298 suppresses
  17. L15:1554 weakly suppresses
  18. L20:283 weakly suppresses
  19. L19:207 weakly suppresses
  20. L17:2302 weakly suppresses 'in' after a noun
  21. L13:1876 weakly suppresses 'in' or 'on' before a place or date
  22. L15:395 weakly suppresses
  23. L17:1417 weakly suppresses
  24. L14:1519 weakly suppresses

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 -10.15 and -9.861 values between -9.86 and -9.581 values between -9.01 and -8.721 values between -8.72 and -8.441 values between -8.44 and -8.161 values between -7.59 and -7.301 values between -7.30 and -7.022 values between -7.02 and -6.731 values between -6.73 and -6.455 values between -6.45 and -6.165 values between -6.16 and -5.884 values between -5.88 and -5.591 values between -5.59 and -5.315 values between -5.31 and -5.024 values between -5.02 and -4.747 values between -4.74 and -4.455 values between -4.45 and -4.173 values between -4.17 and -3.8810 values between -3.88 and -3.607 values between -3.60 and -3.318 values between -3.31 and -3.039 values between -3.03 and -2.7411 values between -2.74 and -2.4614 values between -2.46 and -2.1726 values between -2.17 and -1.8933 values between -1.89 and -1.6128 values between -1.61 and -1.3229 values between -1.32 and -1.0451 values between -1.04 and -0.7590 values between -0.75 and -0.47117 values between -0.47 and -0.18698 values between -0.18 and 0.101966 values between 0.10 and 0.39737 values between 0.39 and 0.6722 values between 0.67 and 0.962 values between 0.96 and 1.24offnegativepositiveoften

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. imer
  2. aux
  3. ifique
  4. 均有
  5. avan
  6. 断
  7. imientos
  8. 总有
  9. 总是
  10. 均衡

Suppresses

  1. sorts
  2. building
  3. CLUD
  4. ark
  5. .tie
  6. wards
  7. circuits
  8. closed
  9. @@⏎
  10. civil