Neuron Explorer

Qwen3-0.6B

'to' after 'for someone' (purpose infinitive)

A neuron in layer 4 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 'to' when it comes after 'for' plus a person or thing, as in 'for people to take shelter'. It switches off, strongly, on the word 'tips'. Turning it up did not lead to a clear change in topic.

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

    to European excellence in imaging for life sciences researchers,” says interim-director Professor John Eriksson from Finland. “All of the facilities are excellent and open for anyone to use, with nominal fees. We provide training, data management services and access to

  2. #2 peak

    historic world from your own imagination⏎A thick jungle scene puzzle with dinosaurs hidden for others to find⏎Well, I could go on and on. That's just a sample

  3. #3 peak

    into the school gymnasium.⏎With this is mind I wanted to ensure that the data that was being sourced within the session would be instantly displayed for the students to analyse. To make this possible I setup a

  4. #4 peak

    experienced by His flesh, and even the humiliation he encountered when the soldiers beat him and stripped off his clothes, and hung Him on the cross for the world to see. Even Through rejection, deprivation, persecution, and many other dealings with sinful

  5. #5 peak

    see how the superpowers of Egypt and Hatti fought each other to a standstill, leaving, for a short period, the land open for God’s people to take over.⏎Special thanks to Dr. Robert L Diss Jr. for his materials

  6. #6 peak

    can withstand the dynamic shock of his touching the ground at speeds over 100 miles an hour, and a brain that can process information fast enough for him to react to his surroundings at high speed.⏎DC Comics dispenses with such explanations with

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: when you have a question for someone who is trying to get a better answer for the question. So, when you're thinking about a question to get a better answer to the question, this is for someone who is thinking about a question to

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

    printable page, download the free Adobe Acrobat Reader.)⏎Click for the answers to this week's puzzle.⏎Click for Sudoku Puzzles from previous weeks.⏎Click for tips for teaching students to do Sudoku puzzles.⏎Copyright © 2006 Education

  2. #2 peak

    gum disease, and tooth loss. Here are some tips on how to take care of teeth.⏎1. Brush your teeth twice daily:⏎

  3. #3 peak

    getting organized. It offers tips for setting short- and long-term goals, choosing activities and fitting them into your

  4. #4 peak

    answers to tough Oracle questions from Brian Peasland are available.⏎- The Best Oracle Web Links: tips, tutorials, scripts, and more.⏎- Have an Oracle or SQL tip to

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. L3:2666 suppresses
  2. L3:3020 weakly suppresses
  3. L3:1791 weakly suppresses
  4. L0:2068 weakly excites
  5. L3:1873 weakly suppresses
  6. L3:2092 weakly suppresses
  7. L3:115 weakly suppresses
  8. L3:2361 weakly excites
  9. L3:3000 weakly excites
  10. L1:1109 weakly suppresses
  11. L3:2263 weakly excites the verb 'guide' or 'navigate'
  12. L0:122 weakly suppresses
  13. L1:2805 weakly suppresses
  14. L3:1210 weakly suppresses
  15. L1:2099 weakly suppresses
  16. L3:3040 weakly suppresses
  17. L2:1235 weakly excites
  18. L2:3027 weakly suppresses
  19. L3:1608 weakly suppresses
  20. L3:2517 weakly excites
  21. L2:1710 weakly suppresses
  22. L3:1641 weakly suppresses
  23. L3:2194 weakly suppresses
  24. L0:2042 weakly suppresses

Feeds later layers

  1. L22:2213 weakly suppresses
  2. L6:2512 weakly suppresses
  3. L6:2388 weakly suppresses
  4. L8:2050 weakly suppresses
  5. L5:1837 weakly suppresses
  6. L13:1438 weakly suppresses
  7. L6:431 weakly suppresses
  8. L6:442 weakly suppresses
  9. L12:1983 weakly excites
  10. L7:2498 weakly excites
  11. L11:1706 weakly suppresses
  12. L5:1604 weakly suppresses
  13. L10:3019 weakly suppresses
  14. L5:2174 weakly suppresses
  15. L27:2531 weakly suppresses
  16. L20:3063 weakly excites
  17. L9:2811 weakly suppresses
  18. L11:653 weakly suppresses
  19. L16:917 weakly suppresses 'of' inside organization names
  20. L7:2351 weakly suppresses
  21. L5:2538 weakly suppresses
  22. L8:2362 weakly excites
  23. L12:1358 weakly suppresses
  24. L6:1367 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 -4.24 and -4.102 values between -2.72 and -2.581 values between -2.02 and -1.893 values between -1.61 and -1.472 values between -1.47 and -1.331 values between -1.33 and -1.191 values between -1.19 and -1.051 values between -1.05 and -0.923 values between -0.92 and -0.782 values between -0.78 and -0.642 values between -0.64 and -0.5018 values between -0.50 and -0.36126 values between -0.36 and -0.22675 values between -0.22 and -0.091903 values between -0.09 and 0.05961 values between 0.05 and 0.19182 values between 0.19 and 0.3314 values between 0.33 and 0.475 values between 0.47 and 0.611 values between 0.61 and 0.751 values between 0.75 and 0.882 values between 1.02 and 1.161 values between 1.16 and 1.30offnegativepositiveoften

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. trand
  2. allis
  3. ignment
  4. iationException
  5. igr
  6. -state
  7. urger
  8. alled
  9. state
  10. agnet

Suppresses

  1. 要点
  2. Osborne
  3. _oct
  4. cps
  5. cps
  6. cciones
  7. dyn
  8. 窍
  9. ally
  10. Tips