Neuron Explorer

Qwen3-0.6B

journal citation numbers and punctuation

A neuron in layer 11 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 responds to the bits of journal references that come after a title, such as the year, month, volume and issue numbers and their brackets. When turned up, the writing did not clearly change in a recognisable way, though it kept producing repetitive text.

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

    too word<|endoftext|>Sore Throat Complications Not Increased with Fewer Antibiotics⏎Am Fam Physician. 2006 May 1;73(9):1619-1634.⏎Clinical Question: Does

  2. #2 peak

    or maintaining the 180 degrees rotational symmetry of the two flagella.⏎Rights and Permissions⏎Citation: J Cell Biol. 1983 Sep;97(3):9

  3. #3 peak

    an-RM⏎Am Ind Hyg Assoc J 1980 Oct; 41(10):758-765⏎

  4. #4 peak

    people.⏎|Printer friendly Cite/link Email Feedback|⏎|Title Annotation:||British Empire|⏎|Article Type:||Brief Article|⏎|Date:||Jan 24, 20

  5. #5 peak

    you only have<|endoftext|>Poor communication in just one area can affect the entire health care experience.⏎Fam Pract Manag. 2013;20(5):40⏎Author disclosure: no relevant financial affiliations disclosed.⏎As

  6. #6 peak

    ⏎DOMINGUES, Beatriz Helena. The guarani jesuitic missions in the context of Enlightenment. História [online]. 2006, vol.25, n.1, pp. 44-6

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: all the personal, academic, and professional issues, and the like. Yes, I know. I have to feel about everything. I can't. I have to feel about everything. I have to feel about everything. I can't.

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

    different concepts to you effectively. You can submit your school, college or university level homework or assignment to us and we will make sure that you get the answers related to classification

  2. #2 peak

    operates to trip the circuit breaker.⏎Email Based, Online Homework Assignment Help in Classification Of Relays⏎Transtutors is the best place to get answers to all your doubts regarding classification of relays with examples. Transtutors has a vast panel

  3. #3 peak

    home and on the road to reach the goal. You can chat with other challenge-takers at dailypress.com/gogreenforum and exchange ideas, frustrations or inspiration.⏎Intentionally, the series will not veer into the political debate

  4. #4 peak

    tip to offer your fellow DBAs and developers? The best tips submitted will receive a cool prize. Submit your tip today!⏎- Ask your technical Oracle and SQL questions -- or help out your peers by answering<|endoftext|>On the Lap of the Mighty Sag

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. L8:46 weakly excites
  2. L9:3030 weakly suppresses
  3. L7:1863 weakly excites
  4. L1:1166 weakly suppresses
  5. L4:2135 weakly suppresses
  6. L9:817 weakly excites
  7. L10:2885 weakly suppresses
  8. L9:1541 weakly excites
  9. L0:2491 weakly suppresses
  10. L1:1869 weakly excites
  11. L9:2790 weakly excites
  12. L3:2387 weakly excites
  13. L9:1667 weakly suppresses "before" after a do-this-first advice clause
  14. L4:2681 weakly excites the word 'flat' as adjective
  15. L1:2764 weakly suppresses
  16. L4:869 weakly suppresses
  17. L3:1088 weakly suppresses
  18. L8:1168 weakly excites
  19. L6:2180 weakly suppresses
  20. L2:1657 weakly excites
  21. L7:745 weakly suppresses
  22. L7:1745 weakly suppresses
  23. L9:1521 weakly suppresses
  24. L1:3014 weakly suppresses

Feeds later layers

  1. L17:42 weakly suppresses
  2. L13:2291 weakly excites
  3. L13:2051 weakly excites
  4. L20:2246 weakly suppresses
  5. L16:1267 weakly suppresses
  6. L14:2143 weakly excites
  7. L23:2603 weakly excites
  8. L24:942 weakly excites
  9. L18:2032 weakly suppresses
  10. L13:1992 weakly excites
  11. L18:3012 weakly suppresses
  12. L16:2933 weakly suppresses
  13. L12:2414 weakly suppresses
  14. L13:451 weakly excites
  15. L12:1611 weakly excites
  16. L17:835 weakly suppresses
  17. L19:2164 weakly excites
  18. L12:550 weakly excites
  19. L15:2163 weakly excites
  20. L13:1203 weakly excites
  21. L16:2606 weakly suppresses
  22. L13:1465 weakly excites
  23. L12:1113 weakly suppresses
  24. L12:2013 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 -3.92 and -3.751 values between -3.06 and -2.891 values between -2.55 and -2.371 values between -2.03 and -1.861 values between -1.86 and -1.692 values between -1.52 and -1.352 values between -1.35 and -1.186 values between -1.00 and -0.836 values between -0.83 and -0.6622 values between -0.66 and -0.49110 values between -0.49 and -0.32438 values between -0.32 and -0.151821 values between -0.15 and 0.021216 values between 0.02 and 0.20234 values between 0.20 and 0.3736 values between 0.37 and 0.542 values between 0.54 and 0.712 values between 0.88 and 1.051 values between 1.05 and 1.221 values between 1.22 and 1.392 values between 1.57 and 1.742 values between 2.76 and 2.94offnegativepositiveoften

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. arded
  2. URY
  3. 槁
  4. shortcode
  5. 六年
  6. finder
  7. forgiving
  8. aced
  9. uida
  10. vaccinated

Suppresses

  1. دد
  2. اخ
  3. Anything
  4. !"
  5. ê
  6. Something
  7. .eclipse
  8. anything
  9. phem
  10. nhu