Neuron Explorer

Qwen3-0.6B

commas in citations and name lists

A neuron in layer 9 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 commas that sit inside formal references and lists, such as in a citation's volume number, a journal date, or a name followed by a title. When it was turned up, the output became repetitive rather than showing a clear topic.

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

    erry, India. You may make a digital copy or printout of this text for your personal, non-commercial use under the condition that you copy this document without modifications and in its

  2. #2 peak

    and Chia-Liang Chiang, "Purification Study of Pervious Concrete Pavbement," International Journal of Engineering and Technology vol. 5, no. 5, pp. 532-535,

  3. #3 peak

    une-system⏎- Kennedy, David O. “B Vitamins and the Brain: Mechanisms, Dose and Efficacy—A Review.” Nutrients, vol. 8, no. 2, Jan. 201

  4. #4 peak

    aim of this legislation is establishing tour operators' responsibility towards customers and providing remedies to them (Gibson, 2016).⏎1.2 Surface, sea and air transport law in relation to the carriage of passengers⏎Below explained are

  5. #5 peak

    anxieties are indicators of healthy curiosity and strong moral fiber, while others are a source of severe stress. Knowing which is which can help you to navigate your personal, professional, and intellectual life more effectively.⏎Charlie Kurth, assistant professor in philosophy

  6. #6 peak

    per kilo of bodyweight than do younger people. But that’s because their lean body mass is less.⏎In a book called Biomarkers, authors William Evans, Ph.D. and

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 number of U.S. dollars in circulation in the United States Mint Museum Collection since March 15,000,001,001,001,001 in 1918-

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 Code of Practice Concerning the Employment of Young Persons in Licensed Premises which a number of organisations have signed up to. IBEC, the employers’ organisation, recommends that pupils in examination classes should not

  2. #2 peak

    hearing aids when they are not in use.⏎Keep replacement batteries and small aids away from children and pets.⏎Certain conditions, including age, illness, and genetics, may contribute to hearing loss. Over several generations, modern life has added a host

  3. #3 peak

    methodology and data sources used for the calculation of the European steel recycling rate have been once again reviewed independently by the energy and waste specialist consultancy, Eunomia, who confirm the credibility of APEAL’s methodology and data.⏎www.ape

  4. #4 peak

    Future Bioenergy Workforce:⏎- K-12 Education: The McCall Outdoor Science School (MOSS), and Facing the Future, a non-profit organization, develop education curricula on bioenergy and biofuels, which is tested

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. L4:3 strongly excites commas after citation fields
  2. L5:16 excites
  3. L3:71 suppresses
  4. L2:66 excites
  5. L5:1469 suppresses
  6. L6:798 excites
  7. L1:86 excites
  8. L6:620 suppresses "and" in lists of three or more
  9. L0:1596 excites
  10. L0:146 suppresses
  11. L6:427 weakly suppresses
  12. L1:1464 weakly excites
  13. L1:2683 weakly excites
  14. L6:1811 weakly excites
  15. L8:70 weakly excites 'and' between stacked superlatives
  16. L7:1223 weakly suppresses "and" joining two decades or times
  17. L6:912 weakly suppresses
  18. L0:1001 weakly excites
  19. L4:19 weakly suppresses
  20. L4:363 weakly suppresses
  21. L4:25 weakly excites
  22. L3:421 weakly suppresses
  23. L6:2945 weakly suppresses
  24. L5:102 weakly excites

Feeds later layers

  1. L14:35 strongly excites
  2. L18:366 strongly excites comma inside numbers and dates
  3. L10:335 strongly excites commas in catalog, date, and list metadata
  4. L15:588 strongly excites comma after an introductory phrase
  5. L19:726 excites comma inside large numbers
  6. L23:184 excites
  7. L17:118 excites comma after a clause ending
  8. L16:240 excites comma after 'sometimes' or 'the truth'
  9. L19:383 excites commas after a clause or list item
  10. L16:846 excites comma opening an aside or explanation
  11. L17:2285 excites
  12. L11:154 excites commas after citation numbers
  13. L22:144 excites
  14. L20:1298 excites
  15. L16:241 suppresses
  16. L16:1082 excites comma linking two clauses after 'and'
  17. L19:1506 excites
  18. L17:116 suppresses
  19. L17:585 excites
  20. L15:727 excites
  21. L18:1270 excites
  22. L20:476 excites commas that lead into a continuation
  23. L17:2001 excites
  24. L19:1826 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 -2.31 and -2.001 values between -1.70 and -1.39169 values between -0.78 and -0.482279 values between -0.48 and -0.171223 values between -0.17 and 0.1369 values between 0.13 and 0.4435 values between 0.44 and 0.7430 values between 0.74 and 1.0522 values between 1.05 and 1.3512 values between 1.35 and 1.667 values between 1.66 and 1.9610 values between 1.96 and 2.277 values between 2.27 and 2.5711 values between 2.57 and 2.884 values between 2.88 and 3.184 values between 3.18 and 3.494 values between 3.49 and 3.796 values between 3.79 and 4.104 values between 4.10 and 4.401 values between 4.40 and 4.711 values between 4.71 and 5.011 values between 5.01 and 5.323 values between 5.32 and 5.621 values between 6.23 and 6.541 values between 7.15 and 7.451 values between 7.45 and 7.761 values between 9.59 and 9.89offnegativepositiveoften

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. 以及其他
  2. Hezbollah
  3. 以及
  4. および
  5. która
  6. ahkan
  7. +⏎⏎
  8. "⏎⏎
  9. кер
  10. 和支持

Suppresses

  1. on
  2. ,
  3. ira
  4. 后再
  5. rex
  6. Ix
  7. 在这
  8. I
  9. ось
  10. 对他们