Neuron Explorer

Qwen3-0.6B

unclear: year digits and prepositions

A neuron in layer 16 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron is mixed. It fires on some digits in years, such as the "1" in "1870s", and on words like "Suggested" in lists of resources. It is quieter on prepositions like "for" and "to" after a noun. Turning it up or down did not give a clear theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    addition to our editors' articles, see section below for printable documents, Ballotechnics books and related discussion.⏎Suggested Pdf Resources⏎Suggested News Resources⏎Suggested Web Resources⏎Great care has been taken to prepare the information on this page

  2. #2 peak

    2004, the increase varied widely among states. Residents of New York gained 4.3 years, the most of any state, during the 13-year period studied and can expect to live 79.2 years.

  3. #3 peak

    (During the 1834 Coorg Campaign)". www.archerfamily.org.uk. Archer Family. Archived from the original on 13 November 2013. Retrieved 13 November 2013.⏎-

  4. #4 peak

    2018, Schuldt et al. 2020). Under these circumstances, tree´s survival primarily depends on the extent to which tree function is impaired by drought (i.e. resistance, Lloret

  5. #5 peak

    the latest changes. If you use the database to generate your MLA citation, please make sure to compare the citation against the core elements, changes, and examples listed above.<|endoftext|>Progressivism is a political term that emerged in the late 19

  6. #6 peak

    three cases are the most fun to prove: a(5) = 4, a(8) ≤ 6, and a(30) ≤ 24. Try it!⏎By the way, K. Knop and L

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: how to use a combination of the methods of the K-12 curriculum and the principles of the Socratic method and the philosophy of education, specifically with the help of the MATLAB 2023a, which is a tool for

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

    tell our children to remain out of these<|endoftext|>Though I've long been intrigued by the idea of design thinking, it was the recent launch of a "Design Thinking for Educators" workshop here at Edutopia that compelled

  2. #2 peak

    in interesting developments in consumer markets.⏎So what is the relevance for business schools? They do not create big data. They do not have 1

  3. #3 peak

    's ownership rights and society's desire to use his works. Although legal scholars continue to debate whether fair use is a right or a privilege, the distinction is irrelevant to teachers and students.

  4. #4 peak

    Governance⏎- Civil Society⏎Sunday, September 19, 2021⏎MONTEVIDEO, Mar 10 2010 (IPS) - “Nature is wise, and if we take the time to observe

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. L15:965 excites
  2. L15:327 excites
  3. L15:2476 suppresses
  4. L12:1449 suppresses
  5. L15:2742 excites
  6. L15:3065 suppresses
  7. L8:824 suppresses
  8. L13:1075 suppresses
  9. L15:2825 suppresses
  10. L13:1817 excites
  11. L14:2071 suppresses
  12. L15:1088 weakly excites
  13. L15:1727 weakly excites
  14. L12:98 weakly suppresses
  15. L14:627 weakly suppresses
  16. L14:2488 weakly excites
  17. L15:2979 weakly excites
  18. L13:1466 weakly excites
  19. L14:872 weakly excites
  20. L10:51 weakly excites
  21. L13:3041 weakly suppresses
  22. L5:1196 weakly excites
  23. L13:2037 weakly excites
  24. L13:1326 weakly excites

Feeds later layers

  1. L17:2635 suppresses
  2. L18:2466 suppresses
  3. L19:2605 suppresses
  4. L18:1288 suppresses
  5. L18:826 suppresses
  6. L20:45 excites
  7. L17:2153 suppresses
  8. L17:2965 suppresses
  9. L24:367 suppresses
  10. L19:483 weakly suppresses
  11. L24:2396 weakly excites
  12. L25:609 weakly excites
  13. L17:876 weakly excites
  14. L19:192 weakly suppresses
  15. L18:307 weakly excites
  16. L20:312 weakly excites
  17. L17:2327 weakly suppresses
  18. L17:1084 weakly suppresses
  19. L23:348 weakly suppresses
  20. L21:12 weakly suppresses
  21. L17:1846 weakly suppresses
  22. L18:1536 weakly suppresses end of web page disclaimer
  23. L19:225 weakly excites
  24. L19:2035 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 -9.13 and -8.831 values between -3.75 and -3.451 values between -3.45 and -3.154 values between -3.15 and -2.865 values between -2.86 and -2.561 values between -2.56 and -2.2617 values between -2.26 and -1.9620 values between -1.96 and -1.6630 values between -1.66 and -1.3674 values between -1.36 and -1.07125 values between -1.07 and -0.77224 values between -0.77 and -0.47475 values between -0.47 and -0.171324 values between -0.17 and 0.13878 values between 0.13 and 0.43388 values between 0.43 and 0.73187 values between 0.73 and 1.0384 values between 1.03 and 1.3238 values between 1.32 and 1.6219 values between 1.62 and 1.928 values between 1.92 and 2.221 values between 2.22 and 2.523 values between 2.52 and 2.82offnegativepositiveoften

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. ibal
  2. ogle
  3. ocation
  4. скор
  5. eways
  6. zeich
  7. ecal
  8. icies
  9. icular
  10. ogy

Suppresses

  1. Tutor
  2. gL
  3. _bot
  4. g
  5. 崆
  6. transformer
  7. .asm
  8. ainen
  9. tutor
  10. uem