Neuron Explorer

Qwen3-0.6B

end of document token

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 invisible marker that separates one web page from the next in the training text. When it was turned up, the model's output did not show a clear change in theme. It also goes quiet or negative on numbers inside lists or counting sequences.

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

    countless texts in the canon of English literature. The subject occupies a prominent position in a diverse array of genres and forms, among which is that of the modern...<|endoftext|>The Hundred Years' War began in the reign of Edward

  2. #2 peak

    .⏎You can also purchase the Almanac Monthly at the iTunes Store !<|endoftext|>Learning in dissolving boundaries⏎Talking about changes inherent in the move to what the

  3. #3 peak

    Virginia.⏎To summarize, then, the Civil War Center is committed to the following:⏎WHY KENNESAW STATE UNIVERSITY IS AN⏎<|endoftext|>[Last Updated: April 2nd, 2015]⏎Con

  4. #4 peak

    the globe.⏎The definitions of ‘buddy’ and ‘bro’ in the OED have recently been revised. We explore their history and increase in popularity.<|endoftext|>Native Plants for the Mid-Atlantic⏎On this page⏎- What Are Native

  5. #5 peak

    . Please see the bottom of the page for more information or visit our Terms and Conditions.⏎Thanks for your feedback.⏎37 of 42 found this helpful<|endoftext|>Television was hardly a reliable broadcast medium

  6. #6 peak

    of all bourgeois states, France and Britain prominent among them.⏎What made you want to look up "Constantinople Agreement"? Please share what surprised you most...<|endoftext|>Your healthcare provider will ask about your medical and family history and your overall health.

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 have a very simple but actually very important and simple and important question that is actually very simple but in which I can also take a point of view that I can also have an important and important question that I have to ask of the other

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

    center activities!⏎- Engaging morning work!⏎- Homework variety!⏎- Individual pages for facts 1-12 (-1, -2, -3, etc. up to -12)⏎- Mixed facts practice pages⏎-

  2. #2 peak

    index i and decrementing the value of i by 1.⏎If you run this program, it will print the below output:⏎5 4 3 2 1⏎We need to use the range function to iterate over the array using

  3. #3 peak

    5, 27, 28, 31, 32, 33, 34, 35, 36, 37, 39, 49, 51

  4. #4 peak

    a matrix⏎Find nth maximum⏎Matlab Basics II - Create a vector with a repeated entry⏎Create sequnce 1 4 9 16 25.........⏎Basics: Divide integers to get integer outputs in all

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. L13:2412 weakly excites
  2. L7:2830 weakly suppresses
  3. L13:15 weakly suppresses
  4. L13:1281 weakly suppresses
  5. L13:1757 weakly suppresses
  6. L13:1139 weakly suppresses
  7. L5:1128 weakly excites
  8. L13:996 weakly excites
  9. L14:1627 weakly suppresses
  10. L14:692 weakly excites word right before a section number
  11. L14:2807 weakly excites
  12. L13:2773 weakly excites
  13. L9:1351 weakly suppresses
  14. L2:333 weakly excites
  15. L14:1 weakly excites
  16. L14:1952 weakly excites
  17. L12:1713 weakly suppresses
  18. L1:1112 weakly suppresses
  19. L8:1484 weakly excites
  20. L14:8 weakly suppresses
  21. L3:1922 weakly excites
  22. L11:1764 weakly suppresses
  23. L10:2410 weakly excites
  24. L1:2173 weakly excites

Feeds later layers

  1. L17:386 weakly excites
  2. L17:2509 weakly excites
  3. L17:1992 weakly excites
  4. L19:1887 weakly excites
  5. L22:110 weakly excites
  6. L16:1210 weakly excites
  7. L22:2702 weakly suppresses
  8. L17:1092 weakly excites
  9. L22:263 weakly excites
  10. L18:1476 weakly suppresses
  11. L24:226 weakly excites
  12. L19:2524 weakly suppresses
  13. L16:2699 weakly suppresses
  14. L17:436 weakly excites
  15. L25:1634 weakly suppresses
  16. L20:2151 weakly suppresses
  17. L19:2500 weakly suppresses
  18. L16:2446 weakly suppresses
  19. L20:2002 weakly suppresses
  20. L21:38 weakly excites
  21. L25:470 weakly suppresses
  22. L22:2614 weakly suppresses
  23. L24:741 weakly suppresses
  24. L22:1264 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 -2.89 and -2.631 values between -2.11 and -1.853 values between -1.85 and -1.592 values between -1.33 and -1.089 values between -0.82 and -0.56173 values between -0.56 and -0.301095 values between -0.30 and -0.041926 values between -0.04 and 0.22640 values between 0.22 and 0.4850 values between 0.48 and 0.741 values between 1.00 and 1.261 values between 1.51 and 1.771 values between 2.29 and 2.552 values between 4.10 and 4.361 values between 4.36 and 4.621 values between 4.88 and 5.141 values between 7.21 and 7.47offnegativepositiveoften

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. istribution
  2. ourage
  3. 슬
  4. edata
  5. ugin
  6. tings
  7. lename
  8. endency
  9. ạn
  10. ictions

Suppresses

  1. 正
  2. BO
  3. 待
  4. ROS
  5. either
  6. 即时
  7. 此刻
  8. head
  9. moment
  10. =