Neuron Explorer

Qwen3-0.6B

'of' after quantity words

A neuron in layer 2 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 about an amount, such as 'a large percentage of', 'the majority of' or 'a fraction of'. Turning it up did not make the model's text clearly reflect this pattern.

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

    the nutrition labels on the products they buy, and a large percentage of them don’t read the labels at all. The question is why?⏎Researchers at

  2. #2 peak

    vision nor mission statements include the words “aeronautics” or “space”. When the Space Shuttle program ended, a substantial fraction of the general public (along with many news reporters) surmised that NASA was

  3. #3 peak

    process for creating it.⏎Assessing data health⏎To understand the health of your organization’s data, start by asking questions, such as:⏎- What percentage of data is considered complete?⏎- What percentage of data contains inaccurate, incomplete, or

  4. #4 peak

    of play-based pedagogies. In addition, a large majority of teaching staff is female.⏎Similar to what research has found to be true in countries

  5. #5 peak

    incentive mechanism for OEMs to produce zero-emission and low-emission vehicles, involving a relaxation in the specific CO2 emissions of an OEM provided that its share of such vehicles exceeds 15% of

  6. #6 peak

    alcohol consumption, and poor diet contribute heavily to extrinsic skin aging. But, in most individuals, it is over-exposure to sunlight that accounts for the majority of pre-mature skin aging – up to 90%!⏎Over exposure to

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: "I am a little bit. I have a little bit. I have a little bit. I have a little bit." Let's see, I have a little bit. I have a little bit. I have a little bit. I

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

    ) законодательная власть⏎4) Associate Justice d) беспартийная система⏎5) a right of voting e) гражданин государ

  2. #2 peak

    between exposure to hazardous chemicals and preterm birth in northern Puerto Rico. Particular attention is given to chlorinated volatile organic compounds and phthalates, although biom

  3. #3 peak

    thritis:⏎Collagen (type II, bovine) is used to induce arthritis in animal studies. These studies are preformed to evaluate anti-inflammatory treatments. Particular mice (a strain called DBA/1) which are particularily sensitive

  4. #4 peak

    I react negatively to dilation drops, that won’t stop me from having my eyes dilated each visit. I want to prevent any damage to my eyes from cataracts or

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. L0:474 weakly excites
  2. L1:1975 weakly suppresses
  3. L0:1657 weakly suppresses
  4. L0:1020 weakly excites
  5. L0:1318 weakly suppresses
  6. L1:2467 weakly suppresses
  7. L0:1633 weakly excites
  8. L1:2151 weakly excites
  9. L1:62 weakly excites
  10. L1:1739 weakly suppresses
  11. L1:1032 weakly suppresses
  12. L0:2134 weakly suppresses
  13. L1:2550 weakly suppresses
  14. L0:2962 weakly excites
  15. L1:18 weakly excites
  16. L1:3045 weakly excites
  17. L0:644 weakly suppresses
  18. L1:1289 weakly suppresses
  19. L1:2225 weakly suppresses
  20. L0:1630 weakly excites
  21. L0:2147 weakly suppresses
  22. L0:897 weakly suppresses
  23. L0:2440 weakly suppresses
  24. L0:2778 weakly excites

Feeds later layers

  1. L14:2630 weakly excites
  2. L5:832 weakly suppresses
  3. L22:909 weakly suppresses
  4. L3:2742 weakly suppresses
  5. L4:1995 weakly excites
  6. L12:249 weakly excites
  7. L14:2067 weakly excites
  8. L4:1361 weakly excites
  9. L14:597 weakly excites unclear: document ends and list breaks
  10. L14:26 weakly excites
  11. L19:550 weakly excites
  12. L3:709 weakly excites mid-word pieces in anatomy and place names
  13. L7:1065 weakly suppresses
  14. L7:2940 weakly suppresses
  15. L16:1727 weakly suppresses
  16. L7:1662 weakly suppresses
  17. L21:1174 weakly suppresses
  18. L6:1051 weakly excites
  19. L7:2918 weakly suppresses
  20. L15:1193 weakly suppresses
  21. L6:638 weakly excites
  22. L3:1778 weakly suppresses
  23. L20:862 weakly excites
  24. L10:2365 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.

1 values between -1.18 and -0.982 values between -0.98 and -0.773 values between -0.77 and -0.569 values between -0.56 and -0.36168 values between -0.36 and -0.152645 values between -0.15 and 0.051030 values between 0.05 and 0.2631 values between 0.26 and 0.477 values between 0.47 and 0.674 values between 0.67 and 0.881 values between 0.88 and 1.081 values between 1.08 and 1.292 values between 1.29 and 1.502 values between 1.50 and 1.701 values between 1.91 and 2.111 values between 6.85 and 7.06offnegativepositiveoften

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. ье
  2. isers
  3. doInBackground
  4. 用户提供
  5. setError
  6. endTime
  7. .getHost
  8. ьте
  9. Cosby
  10. iores

Suppresses

  1. bu
  2. alus
  3. 炮
  4. token
  5. fu
  6. rence
  7. token
  8. icht
  9. sp
  10. ve