Neuron Explorer

Qwen3-0.6B

"set" before out, off or up

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 "set" when it is followed by a direction word, as in "set out", "set off" or "set up". Turning it up made the model repeat "off" and "about", which matches the words it helps the model guess next.

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

    hand. He immediately visited the hospital alongside mechanical engineer Alessandro Romaioli to see the valve themselves, upon learning about the situation. Afterward, the pair set about creating a prototype of the valve to test on the patients. Speaking to BBC

  2. #2 peak

    9. Can you give Pirate Pete directions to find Percy Parrot?⏎Great work, Shipmates! Now Pirate Pete has gathered his things, he is ready to set sail.⏎Your teachers are sooo proud of you, keep up the good work

  3. #3 peak

    tied to the idea that farmers who sowed seeds around the full moon had a better harvest.⏎So, over the years, scientists–a naturally curious bunch–set out to see if there is any true link. And what did they find?⏎

  4. #4 peak

    -Galmier (Loire).⏎Seven years later, he found a partnership with Bessac and the painter Joséphus Veyrat. Then, he set up his own business in Grigny, around ten kilometres from Lyon. Such

  5. #5 peak

    home.” –LHBGC, p. 37⏎When Bailey joined the Horticulture faculty at Cornell University in 1888, he immediately set to work expanding that program and also establishing his family in a comfortable home on Cornell

  6. #6 peak

    to the film. Rather than Australia becoming a frozen tundra, “[we] would be fried,” according to Kusenko.⏎And then, when our heroes set off a stellar bomb successfully in the heart of the sun, it would take much

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: what is the print echo print echo on a print echo on a print echo print echo print echo print print print print print print print print print print print print print print print print print print print print print print print print print print print print print

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

    head of the State and all executive acts are done in his name. However, the Governor is bound by the aid and advise of the elected State Government under article 163 of the Constitution. It cannot be said that the Governor, who

  2. #2 peak

    of two molecules with complementary rather than identical structures in juxtaposition. We see here a direct connection to molecular packing in molecular crystals. Aleksandr Isaakovich Kitaigorodsky (left) published a brief paper in 1945

  3. #3 peak

    do just that. After discussion with another friend, Wes Prichard, Prichard suggested to Wennberg<|endoftext|>Home > Preview⏎The flashcards below were created by user⏎on FreezingBlue Flashcards.⏎What factors could affect feet and

  4. #4 peak

    the volcano has fumaroles that emit gas in the crater on the main summit of Kibo.⏎Mount Elbrus, Europe’s highest peak stands at 5,642

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. L20:1511 excites
  2. L21:2174 weakly suppresses
  3. L19:2831 weakly excites
  4. L11:2971 weakly excites
  5. L21:2022 weakly suppresses
  6. L8:112 weakly suppresses
  7. L2:379 weakly suppresses "the" right after "without"
  8. L5:2642 weakly excites
  9. L19:2126 weakly suppresses
  10. L10:706 weakly suppresses
  11. L8:2100 weakly suppresses
  12. L1:2440 weakly suppresses
  13. L20:1793 weakly suppresses
  14. L22:530 weakly suppresses
  15. L11:1195 weakly suppresses
  16. L17:1403 weakly excites
  17. L2:2183 weakly excites
  18. L18:1844 weakly suppresses
  19. L20:579 weakly suppresses
  20. L15:1850 weakly excites
  21. L16:1613 weakly suppresses
  22. L5:1772 weakly suppresses
  23. L20:2681 weakly suppresses
  24. L8:2811 weakly excites

Feeds later layers

  1. L27:1150 suppresses
  2. L27:1484 excites
  3. L25:2283 suppresses
  4. L27:864 weakly suppresses
  5. L24:2727 weakly excites
  6. L26:1297 weakly excites
  7. L24:1173 weakly excites
  8. L27:1058 weakly excites
  9. L25:2543 weakly excites
  10. L25:1726 weakly excites
  11. L27:850 weakly excites
  12. L26:822 weakly excites
  13. L24:708 weakly excites
  14. L24:1151 weakly excites
  15. L26:2182 weakly excites
  16. L26:1369 weakly excites 'Pr' starting capitalised names
  17. L24:2956 weakly excites
  18. L25:947 weakly suppresses
  19. L24:1669 weakly excites
  20. L24:1009 weakly excites
  21. L24:1577 weakly excites
  22. L25:1718 weakly excites
  23. L26:2005 weakly excites
  24. L24:2863 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 -11.78 and -10.901 values between -9.14 and -8.261 values between -8.26 and -7.381 values between -7.38 and -6.506 values between -6.50 and -5.624 values between -5.62 and -4.742 values between -4.74 and -3.863 values between -3.86 and -2.987 values between -2.98 and -2.1023 values between -2.10 and -1.22388 values between -1.22 and -0.343200 values between -0.34 and 0.55189 values between 0.55 and 1.4332 values between 1.43 and 2.3116 values between 2.31 and 3.198 values between 3.19 and 4.079 values between 4.07 and 4.952 values between 4.95 and 5.833 values between 5.83 and 6.713 values between 6.71 and 7.591 values between 7.59 and 8.471 values between 9.35 and 10.233 values between 10.23 and 11.111 values between 12.87 and 13.751 values between 16.39 and 17.271 values between 17.27 and 18.161 values between 22.56 and 23.44offnegativepositiveoften

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. off
  2. Off
  3. -off
  4. (off
  5. off
  6. about
  7. Off
  8. About
  9. offs
  10. ers

Suppresses

  1. print
  2. print
  3. console
  4. →print
  5. println
  6. println
  7. Console
  8. console
  9. prints
  10. (print