Neuron Explorer

Qwen3-0.6B

comma after an introductory phrase

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

This neuron fires on commas that come straight after a short opening phrase, such as 'Highly visual, each topic' or 'As a community organiser, Alliah'. It is only weakly negative on the word 'but' in contrasts. Turning it up produced repetitive, looping text, and turning it down produced rambling sentences, so the steering adds little to explain it.

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 much ornamentation. A masterpiece of Juan Arellano, the bridge was completed in 1922. Paolo Alcazaren

  2. #2 peak

    the solutions to such problems. If you have an idea for an applied, data-centric note of no more than 3,000 words, please

  3. #3 peak

    -leading at a Glance series. This easy-to-read, accessible guide brings together all the key principles of caring for patients with medical and surgical needs. Highly visual, each topic is covered in one two-page spread, making it easy to quickly read

  4. #4 peak

    and Forest Ecosystems⏎Appendix 3: Estimates of Monetary Values of Ecosystem Services⏎"A landmark study on one of the most pressing problems facing society, balancing economic growth and ecological protection to achieve a sustainable future."⏎- Simon Levin,

  5. #5 peak

    see the cervix and the inside of the uterus.⏎- Laparoscopy. This minor procedure uses a laparoscope, a viewing instrument with a light, to see your uterus and pelvis. The laparoscopy is inserted through

  6. #6 peak

    .S. Fish and Wildlife Service concluded the sicklefin redhorse doesn’t need to be placed on the Endangered Species list. Though long recognized by the Cherokee, this fish was discovered by science in the early 1990s.

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, the, the, the, the, the, the, the, the, the, the, the, the, the, the, the, the,their,their,their,their,their,their,their,

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

    form or of you get them in your food, that you will be healthy and go to the doctor less.⏎In countries without universal healthcare (or ones with it but with co-pays) this means that you will end up spending a lot less

  2. #2 peak

    new st<|endoftext|>Can a small camera possibly have a large lens in the front?⏎There are really "large" lenses. This search is not the biggest one, but the look of it is great.⏎But you are referring to a large diameter compared

  3. #3 peak

    9:17⏎1. Newton's model of the universe made four assumptions which were that the universe is...?⏎- Infinitely large, Infinitely old, Static and created by god⏎- Infinitely large,

  4. #4 peak

    moment of Austen’s birth or there about, John Adams left his farm in Massachusetts for the Continental Congress in Philadelphia. Doesn’t sound particularly interesting, I know but consider this.⏎John Adams left his home in mid-December 177

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. L14:35 strongly suppresses
  2. L9:77 strongly excites commas in citations and name lists
  3. L11:154 suppresses commas after citation numbers
  4. L4:3 excites commas after citation fields
  5. L10:335 excites commas in catalog, date, and list metadata
  6. L5:16 excites
  7. L3:71 suppresses
  8. L0:146 suppresses
  9. L10:667 weakly suppresses commas inside lists of names
  10. L0:1419 weakly suppresses
  11. L13:977 weakly suppresses
  12. L6:798 weakly excites
  13. L6:620 weakly suppresses "and" in lists of three or more
  14. L10:144 weakly suppresses
  15. L2:66 weakly excites
  16. L1:86 weakly excites
  17. L5:1197 weakly suppresses
  18. L0:2921 weakly suppresses
  19. L5:5 weakly suppresses digit after a label like paper or type
  20. L7:2041 weakly suppresses 'or' offering an alternative
  21. L0:827 weakly excites
  22. L7:56 weakly excites
  23. L12:1698 weakly excites
  24. L6:502 weakly suppresses

Feeds later layers

  1. L17:118 excites comma after a clause ending
  2. L19:383 excites commas after a clause or list item
  3. L16:846 excites comma opening an aside or explanation
  4. L18:2863 excites
  5. L19:726 excites comma inside large numbers
  6. L23:184 excites
  7. L16:94 suppresses
  8. L18:366 excites comma inside numbers and dates
  9. L17:2001 excites
  10. L18:1394 excites
  11. L16:2577 excites
  12. L17:2677 suppresses
  13. L16:122 suppresses
  14. L17:585 excites
  15. L17:1841 excites
  16. L16:120 suppresses
  17. L16:17 weakly suppresses
  18. L16:91 weakly suppresses
  19. L16:558 weakly suppresses
  20. L16:1659 weakly excites
  21. L16:105 weakly suppresses
  22. L17:2250 weakly excites
  23. L16:0 weakly suppresses
  24. L16:66 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 -2.69 and -2.382 values between -2.38 and -2.076 values between -2.07 and -1.759 values between -1.75 and -1.4421 values between -1.44 and -1.1255 values between -1.12 and -0.81132 values between -0.81 and -0.49438 values between -0.49 and -0.182176 values between -0.18 and 0.14850 values between 0.14 and 0.4572 values between 0.45 and 0.778 values between 0.77 and 1.089 values between 1.08 and 1.393 values between 1.39 and 1.718 values between 1.71 and 2.0217 values between 2.02 and 2.3414 values between 2.34 and 2.657 values between 2.65 and 2.976 values between 2.97 and 3.289 values between 3.28 and 3.603 values between 3.60 and 3.918 values between 3.91 and 4.228 values between 4.22 and 4.5410 values between 4.54 and 4.854 values between 4.85 and 5.179 values between 5.17 and 5.481 values between 5.48 and 5.806 values between 5.80 and 6.113 values between 6.11 and 6.425 values between 6.42 and 6.741 values between 7.05 and 7.371 values between 7.68 and 8.001 values between 8.00 and 8.313 values between 8.63 and 8.942 values between 9.57 and 9.88offnegativepositiveoften

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. idelberg
  2. который
  3. さんは
  4. которое
  5. lette
  6. 特点是
  7. зе
  8. VIA
  9. くる
  10. rah

Suppresses

  1. ,
  2. 了自己的
  3. ”—
  4. 了
  5. inta
  6. exels
  7. elsen
  8. 在我
  9. 了些
  10. addCriterion