Neuron Explorer

Qwen3-0.6B

the word 'figure' in 'figure out'

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 lights up on the word 'figure' (or 'figured' and 'figuring') when it is followed by 'out', as in 'figure out what's causing it'. When it was turned up, the model got stuck repeating the word 'out' 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

    take steps to clear up the haze after you figure out what’s causing it.⏎What Is Brain Fog?⏎Brain fog is not a

  2. #2 peak

    correct answers. Victor doesn’t know any answer, but is allowed to take the same test several times. Can Victor work out a strategy that guarantees that he can figure out all the answers after the 29th attempt? after the 2

  3. #3 peak

    it and how it threatens me? "- These questions sound in the office of a therapist and are driven into the search box of the browser.Let's try to figure it out.⏎What kind of substance: urea⏎Chemically it is a

  4. #4 peak

    person will make different motions, and whenever s/he changes the motion, everyone will follow. When the person who left the room returns, s/he will try to figure out who the leader is.⏎A

  5. #5 peak

    How to grow a reader⏎Tips for turning a fledgling reader into a voracious one.⏎By Peg Tyre⏎Once your child has figured out how to decode words and can actually read in a sustained way, then a

  6. #6 peak

    our genes surprised us,’ says study coordinator Tuuli Lappalainen, previously at UNIGE and now at Stanford University. ‘It is important that we figure out the general laws of how the human genome works, rather than just delving

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 can be a great hero to be a good hero, and be a good hero to be a good hero. Each time I can be a good hero, I can be a good hero to be a good hero to be a good hero

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

    would come to an end."⏎"The French empire was one of the custodians of the⏎Roman Catholic faith, and, like the Italian empire, supported

  2. #2 peak

    physics has been made at University of Konstanz. The team of scientists headed by...⏎Yersiniae cause severe intestinal infections. Studies using Yersinia pseudotuberculosis as a model organism aim to elucidate the infection mechanisms of these

  3. #3 peak

    28)⏎Design & Thinking is a documentary released this year about design thinking. I haven't seen it yet, but I'm loving the approach -- the cacophony of voices

  4. #4 peak

    isation.⏎The fundamental challenge for a young person now is how does one live with oneself, and how does one live with others in a world filled with a cacophony of voices, with

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. L20:533 excites
  2. L20:1550 excites 'find' in 'to find out'
  3. L17:241 excites
  4. L4:1366 weakly excites
  5. L16:1672 weakly excites
  6. L18:2542 weakly suppresses
  7. L22:288 weakly suppresses
  8. L8:262 weakly suppresses
  9. L20:78 weakly suppresses
  10. L19:107 weakly excites
  11. L20:256 weakly suppresses
  12. L0:785 weakly excites
  13. L22:2955 weakly excites
  14. L19:1287 weakly suppresses
  15. L20:2709 weakly excites
  16. L7:1918 weakly suppresses
  17. L1:1152 weakly excites
  18. L3:2434 weakly suppresses
  19. L17:577 weakly suppresses
  20. L14:311 weakly excites
  21. L4:2692 weakly excites
  22. L2:2821 weakly suppresses
  23. L21:443 weakly suppresses
  24. L21:305 weakly suppresses

Feeds later layers

  1. L27:850 excites
  2. L26:1474 excites
  3. L27:1965 weakly excites
  4. L27:1301 weakly excites
  5. L24:2967 weakly excites
  6. L25:1827 weakly suppresses
  7. L24:2884 weakly suppresses
  8. L25:2124 weakly excites
  9. L24:2899 weakly suppresses
  10. L26:1577 weakly suppresses
  11. L25:1453 weakly suppresses
  12. L24:872 weakly excites
  13. L25:2021 weakly excites
  14. L25:2030 weakly suppresses
  15. L27:2486 weakly excites
  16. L24:2567 weakly suppresses
  17. L24:2822 weakly suppresses
  18. L24:2236 weakly excites
  19. L25:333 weakly suppresses
  20. L27:2705 weakly suppresses
  21. L25:1541 weakly suppresses
  22. L24:2495 weakly excites
  23. L24:1513 weakly suppresses
  24. L25:2650 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 -18.34 and -17.001 values between -11.65 and -10.312 values between -8.97 and -7.633 values between -7.63 and -6.296 values between -6.29 and -4.954 values between -4.95 and -3.6114 values between -3.61 and -2.2756 values between -2.27 and -0.933455 values between -0.93 and 0.41296 values between 0.41 and 1.7523 values between 1.75 and 3.0914 values between 3.09 and 4.439 values between 4.43 and 5.779 values between 5.77 and 7.112 values between 7.11 and 8.453 values between 8.45 and 9.792 values between 9.79 and 11.134 values between 11.13 and 12.471 values between 13.81 and 15.151 values between 20.51 and 21.851 values between 28.55 and 29.891 values between 33.91 and 35.25offnegativepositiveoften

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. out
  2. -out
  3. out
  4. 出
  5. (out
  6. outs
  7. →out
  8. Out
  9. _out
  10. Out

Suppresses

  1. nist
  2. ills
  3. olio
  4. su
  5. ioni
  6. creds
  7. cem
  8. tał
  9. inated
  10. ione