Neuron Explorer

Qwen3-0.6B

'at' in idioms like 'at work'

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' when it starts a set phrase such as 'at work', 'at least', or 'at the heart of'. Turning it up made the model repeat 'work' and 'least' many times.

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

    be and negation. If you ask her, "Give me the ball that is not red," she'll give you the red ball. So negation is at the same level of abstraction, it would seem, as time.⏎How does this

  2. #2 peak

    of care (signs, interfaces with the electronic medical record, and smart phone apps).⏎Why is this decision support ineffective? Three key facets of human psychology are at least partially responsible: 1. We are bad at knowing when we need help

  3. #3 peak

    been so pervasive – though certainly degrees have varied in different times and places – that regarding it as coincidental seems farfetched; a structural process seems to be at work.⏎The only force in the known world that provides functional and enduring structure among

  4. #4 peak

    Mexico and parts of Southern California was barely worth noting compared to the ongoing, major forces that shape planet Earth. In a geologic sense, there are greater forces at work, from the coming and going of ice ages to an ongoing imbalance in the

  5. #5 peak

    from therapeutic breaches to unthinkable crimes that defy our understanding — can be rehabilitated.⏎That last point – whether empathy can be rehabilitated, or taught, is also at the heart of a fascinating melding of art and science that panelist Alice Fl

  6. #6 peak

    factor that made her quite atypical was her chosen field of art specialization in sculpture rather than the standard popular domain

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: if I could, with my current to be to be, and with my current to be to be, then... Well, obviously, if I can't, then... Well, obviously, if I can't, then... Well, obviously

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

    from conflict to cyclone relief.”⏎Senior emergency response specialist, United Nations Office for the Coordination of Humanitarian Affairs⏎Cyclones Idai and Kenneth that swept

  2. #2 peak

    Top Program to states that joined Common Core. Federal Race to the Top money is being used to fund the tests for Common Core, which are extensive

  3. #3 peak

    Nearly 7,000 people have been displaced and more than 5,000 homes destroyed, damaged or flooded,” said the UN Office for the Coordination of Humanitarian Affairs (OCHA), which also warned that the death

  4. #4 peak

    had been done. It encourages the doer to double up and seek remedy in ways unimaginable. In other words, the feeling of ‘There is light at the end of the tunnel‘ will spur one to do better.⏎3 Knowledgeable —

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. L19:2538 excites "at" ending a phrase
  2. L17:1325 excites
  3. L18:895 suppresses
  4. L4:430 suppresses
  5. L8:2038 suppresses
  6. L21:2846 excites 'at' after a job title, before an institution
  7. L19:2572 weakly excites
  8. L13:2374 weakly excites
  9. L18:2550 weakly suppresses
  10. L22:3023 weakly suppresses
  11. L14:707 weakly suppresses
  12. L0:2837 weakly excites
  13. L22:1284 weakly excites
  14. L22:2683 weakly suppresses
  15. L4:1054 weakly suppresses
  16. L4:862 weakly suppresses 'to' after 'too big' or 'right angles'
  17. L6:2979 weakly excites
  18. L6:2614 weakly suppresses
  19. L16:1530 weakly suppresses
  20. L10:2041 weakly excites
  21. L0:1417 weakly suppresses
  22. L16:1216 weakly suppresses
  23. L12:1487 weakly suppresses
  24. L21:1935 weakly excites

Feeds later layers

  1. L25:965 excites
  2. L24:2924 excites
  3. L27:1598 weakly excites
  4. L25:770 weakly excites
  5. L27:1895 weakly excites
  6. L24:2968 weakly excites
  7. L24:2870 weakly excites
  8. L24:2114 weakly excites
  9. L26:840 weakly excites
  10. L24:1879 weakly suppresses
  11. L26:2780 weakly excites
  12. L24:973 weakly excites
  13. L24:214 weakly suppresses
  14. L27:946 weakly excites
  15. L24:1429 weakly excites
  16. L24:216 weakly excites
  17. L25:1212 weakly excites
  18. L24:428 weakly excites
  19. L24:1298 weakly suppresses
  20. L25:1730 weakly excites
  21. L24:1167 weakly suppresses
  22. L24:1731 weakly suppresses
  23. L24:2690 weakly excites
  24. L26:2705 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 -28.20 and -25.411 values between -25.41 and -22.621 values between -19.83 and -17.042 values between -14.25 and -11.466 values between -11.46 and -8.6714 values between -8.67 and -5.8816 values between -5.88 and -3.081234 values between -3.08 and -0.292555 values between -0.29 and 2.5036 values between 2.50 and 5.2913 values between 5.29 and 8.0811 values between 8.08 and 10.873 values between 10.87 and 13.662 values between 13.66 and 16.453 values between 19.24 and 22.041 values between 30.41 and 33.201 values between 41.57 and 44.361 values between 44.36 and 47.151 values between 49.95 and 52.742 values between 69.48 and 72.271 values between 77.86 and 80.653 values between 80.65 and 83.44offnegativepositiveoften

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. rest
  2. least
  3. Least
  4. once
  5. (rest
  6. 工作
  7. least
  8. odds
  9. work
  10. night

Suppresses

  1. ohon
  2. 决
  3. 迅速
  4. boxed
  5. explode
  6. activex
  7. umpy
  8. jected
  9. alyze
  10. umping