Neuron Explorer

Qwen3-0.6B

'of' in set phrases like 'of course'

A neuron in layer 15 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

Fires on the word 'of' in phrases like 'of course', 'of all' or 'of interest'. It goes quiet or negative on 'of' after a number or quantity, such as 'one of' or 'a dozen of'. Turning it up made the model's continuations wander without a clear theme.

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

    and the telescope) can see. Those objects are getting farther away all the time because the universe is expanding, and moreover the expansion is accelerating. Strangest of all, the luminous matter of the galaxies—everything we see shining in the

  2. #2 peak

    a floral shop.⏎Even more influential of Liberia<|endoftext|>Oxford University Press⏎2017 | 200

  3. #3 peak

    was spoken around 3500 B.C.E. by a group of nomads called the Aryans living in what is now Southern Russia. It is of course critical to mention the improper usage of the word "Aryan" in N

  4. #4 peak

    primary process, few stop to ask a very important question: why do we have primaries in the first place? In the history of American democracy, political primaries are of relatively recent origin, so when and why did we adopt them?⏎Before answering these

  5. #5 peak

    use some, or all, of this content as you see fit, whether that’s to compliment, vary or boost work set by your child's school or, of course, just to give some extra structure to your usual EdPlace approach.⏎How

  6. #6 peak

    seems to be unique in the Indo<|endoftext|>A POTENTIALLY deadly and virulent bug that attacks the gut may soon be treated with the most stomach-churning of remedies.⏎A study published on Thursday in The New England Journal of Medicine has found

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 day the 2010, 2010, 2 the year, 2020, 2010, 20 10, 2 1 2 2

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

    in the Federal Register, Rumsfeld decided to take things in hand. He assembled Jack Grayson, the chairman of the Price Commission, and about a dozen of our CLC staff and said that we wouldn’t be leaving until we had the

  2. #2 peak

    uranium. Heavy water – because it contains an isotope of hydrogen that has already absorbed an extra neutron and has no real affinity for absorbing another one – and one of several different gases with a low neutron affinity and a low molecular density.⏎Heavy water

  3. #3 peak

    ’sot’ to the Quileute, referring to a place where there is a riffle in the water. Mora is located a mile from the mouth of the Quillayute River at its confluence with the Dickey River or

  4. #4 peak

    mealy bugs are a sign of overwatering.⏎A moisture meter is a great tool to help you keep track of your plant’s watering routine. Using one of these is a lot less hassle than trying to remember to water your succulents

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. L1:725 suppresses "to" in "referred to"
  2. L3:315 excites 'of' after a noun phrase
  3. L0:2726 excites
  4. L6:357 weakly suppresses
  5. L4:4 weakly suppresses
  6. L2:1710 weakly suppresses
  7. L14:2417 weakly suppresses
  8. L0:40 weakly suppresses 'of' in 'sort of' hedges
  9. L5:363 weakly excites
  10. L3:294 weakly excites
  11. L14:435 weakly excites
  12. L3:1138 weakly excites
  13. L6:1566 weakly excites
  14. L11:1393 weakly excites
  15. L0:707 weakly excites
  16. L5:84 weakly suppresses
  17. L4:2080 weakly suppresses
  18. L13:2895 weakly excites
  19. L9:559 weakly excites
  20. L5:290 weakly excites
  21. L5:158 weakly suppresses
  22. L5:684 weakly excites
  23. L5:734 weakly excites
  24. L7:2916 weakly excites

Feeds later layers

  1. L17:317 suppresses 'of' after a noun phrase
  2. L22:229 suppresses "of" after a noun in "X of Y"
  3. L20:1972 suppresses 'of' in 'of course'
  4. L16:1088 suppresses 'of' after a noun phrase, as in 'of course'
  5. L16:2612 suppresses
  6. L19:715 suppresses
  7. L16:2412 suppresses
  8. L19:272 weakly excites
  9. L20:2698 weakly suppresses
  10. L26:2556 weakly suppresses
  11. L17:359 weakly suppresses
  12. L16:286 weakly excites
  13. L21:226 weakly excites
  14. L17:609 weakly excites
  15. L18:2808 weakly suppresses
  16. L16:817 weakly excites
  17. L25:286 weakly suppresses
  18. L18:277 weakly excites
  19. L16:1432 weakly excites
  20. L16:1344 weakly excites
  21. L19:2716 weakly excites
  22. L16:1662 weakly excites
  23. L17:264 weakly excites
  24. L17:253 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 -9.63 and -9.361 values between -9.09 and -8.822 values between -8.82 and -8.562 values between -8.56 and -8.292 values between -8.29 and -8.021 values between -8.02 and -7.762 values between -7.76 and -7.494 values between -7.22 and -6.963 values between -6.96 and -6.692 values between -6.69 and -6.421 values between -6.42 and -6.162 values between -6.16 and -5.893 values between -5.89 and -5.625 values between -5.62 and -5.357 values between -5.35 and -5.091 values between -5.09 and -4.821 values between -4.82 and -4.558 values between -4.55 and -4.294 values between -4.29 and -4.027 values between -4.02 and -3.755 values between -3.75 and -3.496 values between -3.49 and -3.224 values between -3.22 and -2.955 values between -2.95 and -2.693 values between -2.69 and -2.422 values between -2.42 and -2.158 values between -2.15 and -1.885 values between -1.88 and -1.622 values between -1.62 and -1.355 values between -1.35 and -1.089 values between -1.08 and -0.8218 values between -0.82 and -0.5541 values between -0.55 and -0.28363 values between -0.28 and -0.022056 values between -0.02 and 0.251094 values between 0.25 and 0.52200 values between 0.52 and 0.7923 values between 0.79 and 1.05offnegativepositiveoften

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. سطين
  2. ・
  3. ện
  4. Throws
  5. ottom
  6. ival
  7. 一面
  8. rances
  9. 之
  10. ственно

Suppresses

  1. 经开
  2. bral
  3. chá
  4. ","+
  5. =',
  6. ApiKey
  7. =admin
  8. mys
  9. ذي
  10. >[]