Neuron Explorer

Qwen3-0.6B

'which' or 'whom' after 'of'

A neuron in layer 8 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on the word "which" or "whom" (and "them") when it follows a "some of" or "most of" phrase, like "a large chunk of which". Turning it up or down gave unrelated text, so it's not clear what the output does.

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

    , tar removal and methanation of the syngas.⏎Despite recycling efforts, the UK produces millions of tons of rubbish each year, a large chunk of which goes to methane-releasing landfill. Gasification and methanation are well understood

  2. #2 peak

    Dwarfs: The Creation of a Classic,” the exhibit celebrates Disney’s vision and the artistry of his staff—over 1,200 of whom labored for almost four years to make the film.⏎Diane Disney Miller,

  3. #3 peak

    the tau, and their respective neutrinos.⏎All known leptons have a negative or neutral charge. There are 6 types of lepton, 3 of which are negative, the other 3 neutral.<|endoftext|>From: Nicholas Holton,

  4. #4 peak

    of God (Anderson) saw potential in his work, and began mass-marketing them. He created several versions over the years, the last and most popular of which was this one created in 1941.⏎Christ at Heart's Door

  5. #5 peak

    as lithium carbonate and lithium hydroxide.⏎Lithium reserves worldwide are estimated at 39.5 million tons, about 5.5 million of which are in the United States, including at the Salton Sea, according to the

  6. #6 peak

    on the U.S. Team in the 1936 Olympic Games held in Berlin, Germany. They won 14 medals — eight of them gold: a quarter of the 56 medals won by the entire U.S

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: the one that is about the heart. And I don't know what it is. Let me think. So, I'm thinking, I'm thinking, I'm thinking, I'm thinking, I'm thinking, I'm thinking, I

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

    . This one, called iBART, was a thing of beauty. Free, too. It was written by two former high-school buddies—Ian Leighton, a sophomore at UC Berkeley, and David Hodge, a sophomore at the University

  2. #2 peak

    they drain faster than clay soils. If you live in an area with sandy soils, water two or three times a week, 20-30 gallons total during the first two years after planting, followed by half as much for the

  3. #3 peak

    , when her blood work came back everything was reported as normal. Understandably, she was both frustrated and confused about what was going on in her body. Sound familiar?⏎If you’re convinced your thyroid isn’t functioning properly but all your lab work

  4. #4 peak

    say you pose a multiple-choice question for which there is no single correct answer but for which there are perhaps more justifiable answers and less justifiable answers. Something like “Which of the following motivations best explains so-and-so’s actions in such-and

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. L7:975 excites
  2. L7:323 excites
  3. L7:1253 suppresses
  4. L2:2319 weakly excites
  5. L2:520 weakly excites
  6. L7:620 weakly excites
  7. L5:1619 weakly suppresses
  8. L4:307 weakly suppresses
  9. L7:298 weakly suppresses
  10. L7:931 weakly suppresses
  11. L7:94 weakly excites
  12. L3:156 weakly excites
  13. L7:2264 weakly excites
  14. L7:1714 weakly excites
  15. L7:808 weakly excites
  16. L7:1089 weakly excites
  17. L4:632 weakly excites
  18. L7:1170 weakly suppresses
  19. L3:191 weakly suppresses
  20. L6:581 weakly excites
  21. L7:1886 weakly excites
  22. L7:1264 weakly excites
  23. L6:2493 weakly excites
  24. L7:567 weakly excites

Feeds later layers

  1. L9:454 suppresses
  2. L9:2181 weakly suppresses
  3. L22:505 weakly excites
  4. L11:2219 weakly suppresses
  5. L15:503 weakly suppresses
  6. L9:854 weakly suppresses
  7. L20:812 weakly suppresses
  8. L10:2324 weakly suppresses pronouns referring back to a noun
  9. L12:2896 weakly excites
  10. L15:2548 weakly excites
  11. L11:434 weakly suppresses
  12. L10:976 weakly suppresses
  13. L10:2259 weakly excites
  14. L17:1439 weakly excites
  15. L13:2316 weakly excites
  16. L13:2031 weakly suppresses
  17. L10:2579 weakly excites
  18. L16:2489 weakly excites
  19. L13:1126 weakly suppresses
  20. L9:562 weakly excites
  21. L13:2540 weakly suppresses
  22. L13:2365 weakly excites
  23. L11:2863 weakly suppresses
  24. L17:400 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 -2.60 and -2.481 values between -1.67 and -1.551 values between -1.44 and -1.324 values between -1.21 and -1.094 values between -1.09 and -0.972 values between -0.86 and -0.744 values between -0.74 and -0.638 values between -0.63 and -0.5121 values between -0.51 and -0.3964 values between -0.39 and -0.28261 values between -0.28 and -0.16844 values between -0.16 and -0.042010 values between -0.04 and 0.07468 values between 0.07 and 0.19119 values between 0.19 and 0.3035 values between 0.30 and 0.4222 values between 0.42 and 0.5412 values between 0.54 and 0.659 values between 0.65 and 0.777 values between 0.77 and 0.884 values between 0.88 and 1.002 values between 1.00 and 1.122 values between 1.12 and 1.231 values between 1.23 and 1.352 values between 1.93 and 2.05offnegativepositiveoften

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. "_"
  2. ��
  3. "|"
  4. ":""
  5. riority
  6. са
  7. Idx
  8. WXYZ
  9. 塍
  10. ętr

Suppresses

  1. reate
  2. 也不例外
  3. ves
  4. ewe
  5. isten
  6. 一路
  7. finally
  8. 积极配合
  9. 汲
  10. ::_('