Neuron Explorer

Qwen3-0.6B

'to' after 'unable'

A neuron in layer 1 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 strongly on the word 'to' right after 'unable', as in 'unable to recall' or 'unable to venture out'. It is very consistent on that phrase. Turning it up or down did not produce an obvious change in the model's writing.

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

    black-throated gray warbler is exclusively western in distribution, from our southern border to British Columbia. Though I have seen it many times, I am unable to recall any especially salient characteristics possessed by the species. Like others of the family

  2. #2 peak

    romantic relationship with Mayor Guillén de Guzmán who bore to him an illegitimate daughter Beatrice.⏎Due to Violant's young age, she was unable to get pregnant for several years. Alfonso came to believe that his wife was

  3. #3 peak

    elasticity. As the temperature falls, the rubber continues to lose flexibility and its ability to resist abrasion, impact, and cutting. Eventually, the belt is unable to trough and pass around pulleys. The covers and the rubber between the plies

  4. #4 peak

    it comes to protecting the body from infection. When germs get onto our skin or into a cut or other break in the skin’s surface, they are unable to invade deeper into body tissues due to the acid mantle (pH) of our

  5. #5 peak

    are lacking and can lead to psychological wounding associated with shame and loneliness.⏎Growing awareness of this trait has generated much inquiry, yet psychologists have so far been unable to agree on a single defining attribute. If you identify with the description, there are

  6. #6 peak

    óa last resort in case of a grizzly charge. The bears were obviously aware of my presence, but upwind, and in the dark, they were unable to locate my exact position. After over a minute of intense sniffing and scanning,

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 have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately

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

    th century? That’s easier. The Zeppelin airship. The first modern vacuum cleaner. The first aeroplane. The vacuum tube diode. Bakelite.⏎And what about the first decade of the 21st century? The

  2. #2 peak

    impulsively and irrationally. Salekin uses a laptop computer and LCD player to explain brain plasticity and how they can

  3. #3 peak

    News | Troubled youth in the custody of the Alabama Department of Youth Services have been given a second chance at success thanks to the work of Dr. Randy Salekin, a professor in the Department of Psychology

  4. #4 peak

    troubled youth at the facility than typical methods, like empathy training or strict discipline. The treatments were tested at UA’s Disruptive Behavior Clinic in advance of Salekin’s partnership with DYS.⏎The teens and pre-teens Salekin and his

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. L0:2989 suppresses
  2. L0:178 suppresses
  3. L0:1301 suppresses
  4. L0:2870 weakly suppresses
  5. L0:1224 weakly suppresses
  6. L0:1343 weakly suppresses
  7. L0:644 weakly suppresses
  8. L0:2743 weakly suppresses
  9. L0:2585 weakly suppresses
  10. L0:2563 weakly excites
  11. L0:2111 weakly suppresses
  12. L0:1256 weakly suppresses
  13. L0:1743 weakly excites
  14. L0:2078 weakly excites
  15. L0:1713 weakly excites
  16. L0:963 weakly suppresses
  17. L0:2911 weakly suppresses
  18. L0:1715 weakly excites
  19. L0:1483 weakly excites
  20. L0:1766 weakly suppresses
  21. L0:589 weakly excites
  22. L0:718 weakly suppresses
  23. L0:1581 weakly excites
  24. L0:791 weakly excites

Feeds later layers

  1. L16:2754 excites
  2. L15:786 excites
  3. L15:1104 excites
  4. L17:443 excites
  5. L14:1859 excites
  6. L17:1917 excites
  7. L16:1828 excites
  8. L4:9 excites
  9. L13:383 excites
  10. L21:1349 excites
  11. L20:2810 excites
  12. L15:519 excites
  13. L4:1512 excites
  14. L9:3000 excites
  15. L9:54 excites
  16. L19:1301 excites
  17. L17:2492 excites
  18. L16:1692 excites
  19. L8:140 excites
  20. L16:1221 excites
  21. L16:752 excites
  22. L21:1970 excites
  23. L16:697 excites
  24. L23:2518 excites sentence-initial 'Not' starting a clause

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 -3.69 and -3.451 values between -1.23 and -0.982 values between -0.98 and -0.7417 values between -0.74 and -0.49226 values between -0.49 and -0.242365 values between -0.24 and 0.001226 values between 0.00 and 0.2530 values between 0.25 and 0.5012 values between 0.50 and 0.747 values between 0.74 and 0.993 values between 0.99 and 1.234 values between 1.23 and 1.483 values between 1.48 and 1.732 values between 1.73 and 1.971 values between 1.97 and 2.223 values between 2.71 and 2.961 values between 2.96 and 3.211 values between 3.94 and 4.191 values between 4.44 and 4.681 values between 5.18 and 5.421 values between 5.91 and 6.16offnegativepositiveoften

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. 打扰
  2. trầm
  3. hề
  4. 征求意见
  5. intage
  6. riers
  7. phony
  8. artz
  9. ocrat
  10. ocurrency

Suppresses

  1. asts
  2. arDown
  3. nets
  4. апр
  5. казалось
  6. izations
  7. несен
  8. nm
  9. Sap
  10. iedades