Neuron Explorer

Qwen3-0.6B

commas inside citation author names

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 the comma right after an author's surname in reference lists, like the comma in 'Cuenca, C.' or 'Easton, Matthew'. When it was turned up, the model's writing got stuck repeating a single word or phrase, such as 'sunny' or 'Mr. Obama'.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. Think it is wrong? Flag it (opens an email to the maker with the neuron already filled in).

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

    concerning a stereo photography studio that Dalmau opened in 1854.⏎Download data is not yet available.⏎How to Cite⏎Cuenca, C. (2019). The The “Animated Polystereoscope

  2. #2 peak

    .D., Illustrated Bible Dictionary, Third Edition,⏎published by Thomas Nelson, 1897. Public Domain, copy freely.⏎Bibliography InformationEaston, Matthew George. "Entry for Tamarisk". "Easton

  3. #3 peak

    above questions, in APA format. Your analysis needs to relate the relevant economic concepts learned so far and be sure to justify your arguments.⏎references: Gwartney, J. D., Stroup, R. L., Sobel, R.

  4. #4 peak

    provided by Indiana University-Purdue University Indianapolis School of Science. Note: Content may be edited for style and length.⏎Cite This Page:<|endoftext|>Briers, Mississippi facts for kids⏎Quick facts for kids⏎|Elevation||62

  5. #5 peak

    complex, global issue that threatens the sustainability and resilience of our fishing communities,<|endoftext|>Suicide and Social Change in China⏎- Cite this article as:⏎- Phillips, M.R., Liu, H. & Zhang, Y. Cult Med Psychiatry

  6. #6 peak

    emys scripta elegans) on survival rates of the European pond turtle (Emys orbicularis)⏎- Cite this article as:⏎- Cadi, A. & Joly, P. B

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 thinking of using a single object to create a single object and create another object for another object, and then access it further, and use it again, and use it again, and use it again, and create another object for the

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

    ation found in the catalog.⏎April 1, 2003⏎by Baker Academic⏎Written in English⏎|The Physical Object|⏎|Number of Pages||180|⏎Hermeneutics gives the reader not only an

  2. #2 peak

    |The Physical Object|⏎|Number of Pages||136|⏎Learn why the Common Core is important for your child.

  3. #3 peak

    violation of the generalized second law will disappear when solutions are available for black holes embedded in arbitrary backgrounds.⏎|Original language||English (US)|⏎|Number of pages||12|⏎|Journal||Classical and Quantum Gravity|⏎|State||

  4. #4 peak

    by Theron N. Ford Stevenson|⏎|The Physical Object|⏎|Pagination||vi, 157 leaves ;|⏎|Number of Pages||157|⏎An adoptee’s personal meaning of adoption will be determined

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. L14:704 excites
  2. L12:2294 suppresses
  3. L13:979 excites
  4. L12:250 suppresses
  5. L14:1744 excites
  6. L13:299 excites
  7. L14:197 suppresses
  8. L14:1012 excites
  9. L14:1911 suppresses
  10. L14:1587 weakly excites
  11. L14:1628 weakly suppresses
  12. L13:897 weakly excites
  13. L13:3013 weakly excites
  14. L3:1536 weakly suppresses
  15. L13:2952 weakly excites
  16. L10:1769 weakly excites
  17. L14:1150 weakly suppresses
  18. L13:1967 weakly excites
  19. L10:1863 weakly suppresses
  20. L11:868 weakly excites
  21. L14:641 weakly excites
  22. L11:2977 weakly suppresses
  23. L14:1323 weakly suppresses
  24. L9:100 weakly suppresses

Feeds later layers

  1. L16:1599 excites
  2. L16:2457 excites
  3. L16:1849 excites
  4. L16:1147 excites
  5. L20:2730 excites
  6. L18:1413 suppresses
  7. L16:354 excites
  8. L19:462 excites
  9. L16:1956 suppresses
  10. L21:632 excites
  11. L19:2639 excites
  12. L16:1234 weakly suppresses
  13. L16:1929 weakly excites
  14. L18:2544 weakly excites
  15. L17:1605 weakly suppresses
  16. L18:1767 weakly excites
  17. L16:1830 weakly excites
  18. L18:115 weakly excites
  19. L17:820 weakly excites
  20. L20:769 weakly excites
  21. L16:351 weakly suppresses
  22. L17:2264 weakly excites
  23. L20:333 weakly excites
  24. L19:2450 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 -3.52 and -3.322 values between -1.69 and -1.497 values between -1.29 and -1.0825 values between -1.08 and -0.8824 values between -0.88 and -0.68101 values between -0.68 and -0.47361 values between -0.47 and -0.271344 values between -0.27 and -0.071867 values between -0.07 and 0.14156 values between 0.14 and 0.3415 values between 0.34 and 0.551 values between 0.75 and 0.951 values between 0.95 and 1.161 values between 1.36 and 1.561 values between 2.17 and 2.381 values between 4.41 and 4.61offnegativepositiveoften

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. erg
  2. odox
  3. iris
  4. ме
  5. rief
  6. eward
  7. ainers
  8. incer
  9. rees
  10. ards

Suppresses

  1. \n
  2. fä
  3. /current
  4. /////
  5. voi
  6. feud
  7. 一笑
  8. ...\
  9. 雅黑
  10. фт