Neuron Explorer

Qwen3-0.6B

"in" before a time span

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 "in" when it introduces how long something takes, like "in a few seconds" or "in under twenty-four hours". When it was turned down, the model got stuck repeating "in", and when turned up it wrote long strings of numbers or code-like 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

    the maneuvering. All Carpenter did was destroy the group’s paralysis with a vision of where to aim first. They figured out on their own how to get there in the same way they landed the jet five years before.⏎So, there are lessons

  2. #2 peak

    one of the popular development methodologies that allows development teams to create a working product with high-quality code in a predictable time frame. Below we will describe Agile cycles and also indicate the types

  3. #3 peak

    shells mark the scene of a few hours illegal fishing, where a whole population of this globally threatened species can be wiped out in a couple of hours.⏎These threats to the species have meant that the freshwater pearl

  4. #4 peak

    machine Colossus at Bletchley Park, thanks to the efforts of British Intelligence, meant the messages scrambled by the enigma machine could eventually be decoded in under twenty-four hours, leading the allies to an early victory.⏎Laurence Fisher

  5. #5 peak

    , and traditional educational pathways need to be rethought and redeveloped.⏎Some experts have criticized university curricula for lacking agility. Many programs haven't been updated in years even though the technology refresh cycle is happening at a faster pace. Continuous or

  6. #6 peak

    requires the user to have a compass and a general idea of the satellite’s location. Slowly turning the terminal will soon indicate signal capture, which can be done in less than a minute for an experienced user with a good signal.⏎Some limitations include

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: in the in the in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in

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

    ways. Arranging eight different tiems in a line can be done in 8×7×6×5×4×3×2×

  2. #2 peak

    the history of our planet has EVER created!⏎We can deduce this because counting such arrangements is very simple. Arranging three different items in a line can be done in 3×2×1

  3. #3 peak

    , how long does it take to learn Hebrew fluently? A conservative estimate might be something like 2 years, though a really motivated and talented student might get there as soon as, say, 18 months.⏎Can I learn Hebrew for

  4. #4 peak

    learn and Implement⏎PHP is easier to learn than its competitors such as like ASP, CGI, and others. It is so easy, that a beginner can learn php write a program within few

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. L21:2347 excites
  2. L22:2338 suppresses
  3. L19:1917 excites
  4. L21:1027 excites
  5. L22:825 suppresses 'for' before a span of time
  6. L22:2932 excites
  7. L20:696 excites
  8. L20:1335 weakly suppresses
  9. L20:1404 weakly excites
  10. L22:439 weakly excites
  11. L21:175 weakly suppresses
  12. L19:2887 weakly suppresses
  13. L21:3026 weakly excites
  14. L19:1764 weakly excites
  15. L18:2846 weakly suppresses
  16. L12:734 weakly excites
  17. L17:2760 weakly excites
  18. L19:1118 weakly excites
  19. L20:2041 weakly excites
  20. L8:1482 weakly excites
  21. L18:1750 weakly suppresses the blank slot before a year
  22. L2:108 weakly excites
  23. L18:291 weakly suppresses
  24. L11:1604 weakly excites

Feeds later layers

  1. L24:893 strongly suppresses
  2. L26:207 strongly excites
  3. L27:405 strongly suppresses
  4. L26:825 suppresses
  5. L25:2608 suppresses
  6. L27:2429 suppresses
  7. L25:2282 excites
  8. L25:1790 suppresses the digit 1 starting a statistic
  9. L25:2902 excites
  10. L24:116 weakly excites
  11. L24:1503 weakly excites
  12. L26:1401 weakly suppresses
  13. L25:1369 weakly excites
  14. L26:584 weakly suppresses
  15. L27:918 weakly suppresses
  16. L25:2440 weakly suppresses
  17. L26:667 weakly excites
  18. L27:973 weakly suppresses
  19. L25:421 weakly excites
  20. L25:2569 weakly excites
  21. L24:587 weakly excites
  22. L24:2028 weakly excites
  23. L24:2950 weakly excites
  24. L25:2756 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 -11.17 and -10.162 values between -6.10 and -5.083 values between -5.08 and -4.071 values between -4.07 and -3.058 values between -3.05 and -2.0428 values between -2.04 and -1.021674 values between -1.02 and -0.012148 values between -0.01 and 1.0132 values between 1.01 and 2.021 values between 2.02 and 3.043 values between 3.04 and 4.052 values between 4.05 and 5.071 values between 5.07 and 6.081 values between 6.08 and 7.101 values between 10.14 and 11.161 values between 23.33 and 24.351 values between 28.41 and 29.42offnegativepositiveoften

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. ervative
  2. ato
  3. igators
  4. hec
  5. asons
  6. rogate
  7. toHaveLength
  8. lace
  9. ató
  10. requently

Suppresses

  1. in
  2. In
  3. In
  4. 在
  5. >In
  6. _in
  7. dalam
  8. 'in
  9. →in
  10. -in