Neuron Explorer

Qwen3-0.6B

"at" before end, bottom or start

A neuron in layer 23 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 word "at" in phrases about a position in a text or place, like "placed at the end", "at the bottom" or "at the beginning". Turning it up made the model write "at the end of the day" over and over.

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

    surname or family name.⏎Is surname your last name?⏎In the English-speaking world, a surname is commonly referred to as a last name because it is usually placed at the end of a person’s full name, after any given names.⏎What exactly

  2. #2 peak

    about it! Check out the brainteasers below to see the power of the brain in action, and answer the questions at the bottom of the page.⏎Your brain can see things that aren’t there.

  3. #3 peak

    . If you want the list double-spaced, highlight all the text and set the paragraph to be double-spaced. Do not simply hit the Enter button twice at the end of each line, it will become an editing nightmare and when the sorting

  4. #4 peak

    . Students continue solving problems and finding answers as they move through the Fact Land. For some extra fun, students collect special tokens in certain boxes and tally them up at the bottom. Once students have made their way through the “land,” they count

  5. #5 peak

    .”)⏎He likes to eat popcorn when he goes to see a movie.⏎To play in the snow is one of life’s greatest pleasures.⏎(The infinitive appears at the beginning of the sentences. This sounds very formal.)⏎He went downstairs to get

  6. #6 peak

    postures) and movements designed to increase strength and flexibility. Breathing is actively encouraged throughout the class and a relaxation exercise (savasana) is practised at the beginning and/or end of class. You will leave class feeling energised and

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: what if we can create a system that can just immediately and immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately immediately

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

    any paper they like. Here’s the easiest way I know.⏎- Choose any piece of paper.⏎- Fold the paper in half.⏎- Starting and ending at the fold, make the first stroke of the number 2.⏎- Cut along the

  2. #2 peak

    than just black blobs that mimic the shapes of things casting them. At the center, known as the umbra, the shadow is very dark indeed. But at the edges is a lighter, softer-edged area called the penumbra. When

  3. #3 peak

    focus at all points in a scene. We see the center of a scene with ideal light sensitivity and perfect focus and then focus and brightness will fall off increasingly toward the periphery. Therefore,

  4. #4 peak

    eater or steam through conduction. When the air circulates around the radiator using convention, the room is heated.⏎The distance from the center of a circle to the circumference. One-half of the diameter of a circle.⏎A tool used for checking

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. L19:1989 suppresses
  2. L22:1467 excites
  3. L21:2680 excites
  4. L22:717 excites
  5. L21:893 suppresses
  6. L20:2201 excites 'the' in location phrases like 'at the end'
  7. L21:2835 excites
  8. L21:2851 suppresses
  9. L20:2458 suppresses
  10. L20:2571 suppresses
  11. L20:1604 suppresses
  12. L20:2257 suppresses
  13. L19:1967 excites
  14. L21:970 excites
  15. L21:2498 excites
  16. L18:2946 excites
  17. L22:1205 excites
  18. L22:1470 excites
  19. L19:1920 suppresses
  20. L19:2448 suppresses
  21. L16:623 excites
  22. L18:2149 excites
  23. L22:1450 suppresses
  24. L22:2371 weakly suppresses

Feeds later layers

  1. L27:599 strongly excites
  2. L26:835 strongly excites
  3. L24:1252 strongly excites
  4. L25:501 suppresses
  5. L24:856 excites
  6. L26:1210 excites
  7. L27:1119 excites
  8. L26:840 suppresses
  9. L26:2537 suppresses
  10. L26:1800 weakly excites
  11. L24:2468 weakly excites
  12. L24:2353 weakly excites
  13. L27:3040 weakly suppresses
  14. L27:2816 weakly excites
  15. L24:1282 weakly excites
  16. L25:2274 weakly suppresses
  17. L26:507 weakly excites
  18. L24:380 weakly suppresses the word 'at' in place phrases
  19. L24:2573 weakly excites
  20. L26:971 weakly excites
  21. L25:1268 weakly excites
  22. L25:724 weakly excites
  23. L27:2999 weakly excites
  24. L27:2323 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 -15.84 and -13.783 values between -7.60 and -5.542 values between -5.54 and -3.4868 values between -3.48 and -1.423678 values between -1.42 and 0.64130 values between 0.64 and 2.709 values between 2.70 and 4.769 values between 4.76 and 6.822 values between 6.82 and 8.881 values between 8.88 and 10.942 values between 10.94 and 13.001 values between 19.18 and 21.241 values between 23.30 and 25.361 values between 64.50 and 66.56offnegativepositiveoften

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. at
  2. At
  3. At
  4. _at
  5. at
  6. →at
  7. -at
  8. .at
  9. (at
  10. "At

Suppresses

  1. hone
  2. antry
  3. ddl
  4. udson
  5. ypress
  6. 埠
  7. umont
  8. isle
  9. 呼
  10. orda