Neuron Explorer

Qwen3-0.6B

comma after a conjunction starting a clause

A neuron in layer 3 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 responds to a comma placed right after a joining word like "and", "or" or "yet" when a new clause or action follows, as in "and, taking the money". Its negative side is weak and not clearly meaningful. Turning it up gave text that did not clearly relate to this pattern.

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

    , or Red River; then following the course of Rio Roxo westward, to the degree of longitude 100 west . . . , then, crossing the said Red River, and running thence, by a line due north

  2. #2 peak

    always the same according to different sources and not always exact. So, how do you know what information to believe? It is all very well reading books, but, have you ever considered that you will never get a sense of shape and size from

  3. #3 peak

    exchange it, what has been exchanged as well as its substitute is sacred. It is not to be redeemed."⏎Then let these things be exchanged for money, and, taking the money in your hand

  4. #4 peak

    is perpetrated onto someone as in the story of “St George and the Dragon.” However, we can redeem the evil dragon by turning him into a protector, but, again, he would have to protect someone or something.⏎From St George to George

  5. #5 peak

    human exploration of the solar system and beyond. In that expectation, we were cruelly disappointed. Partisan politics, a collective failure of will and ambition, and, yes, US victory in the Cold War all played a part, because the US

  6. #6 peak

    amusement.⏎There was a half-decayed log lying by the roadside, and Grant sat down on it, calmly pulled a cigar from his cigar case, and, requesting one of his staff who stood nearby to furnish him a sheet of paper,

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: a number of the same number of the same number of the same number of the same number of the same number of the same number of the same number of the same number of the same number of the same number of the same number of the

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

    measured unless subject to both ethical and commercial consideration.⏎Home page url⏎Download or read it online for free here:⏎by Flora Klick

  2. #2 peak

    algae, such as common⏎species of Sargassum.⏎Hawai‘i: Northwest Hawaiian⏎Islands, O‘ahu, Maui, and Hawai‘i Island.⏎Worldwide: Eastern Atlantic,⏎Caribbean, Indian and Pacific O

  3. #3 peak

    the three descriptions.⏎A. Ask a volunteer to leave the room. When s/he returns, s/he will be asked to see if s/he can guess an<|endoftext|>

  4. #4 peak

    the raku firing process result in exciting colors, surfaces, and metallic luster effects.⏎|Presenter:||Lori Mills (Faculty)|⏎|Location:||Outside NE Corner Tower|⏎|Time:||9:30 am

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. L2:409 suppresses
  2. L1:38 suppresses
  3. L2:564 excites
  4. L2:1283 suppresses
  5. L2:375 suppresses
  6. L0:2370 excites
  7. L2:1036 weakly excites
  8. L0:11 weakly excites
  9. L2:267 weakly excites
  10. L2:173 weakly excites
  11. L1:2648 weakly excites
  12. L0:899 weakly excites
  13. L0:1563 weakly excites
  14. L1:807 weakly suppresses
  15. L1:1 weakly excites
  16. L2:45 weakly excites
  17. L1:34 weakly excites
  18. L0:1626 weakly suppresses
  19. L1:349 weakly excites
  20. L1:8 weakly excites
  21. L2:1427 weakly suppresses
  22. L2:1534 weakly excites
  23. L1:3033 weakly excites
  24. L2:1322 weakly suppresses

Feeds later layers

  1. L9:268 excites
  2. L4:260 excites
  3. L4:155 suppresses
  4. L4:43 excites
  5. L6:660 excites
  6. L10:704 excites
  7. L7:876 excites
  8. L5:107 weakly suppresses
  9. L8:1468 weakly excites
  10. L5:44 weakly suppresses
  11. L4:5 weakly suppresses
  12. L6:258 weakly suppresses
  13. L4:559 weakly excites
  14. L6:491 weakly excites
  15. L9:802 weakly excites
  16. L15:933 weakly suppresses
  17. L6:2652 weakly excites
  18. L12:1196 weakly suppresses
  19. L13:1243 weakly excites
  20. L7:589 weakly excites
  21. L7:222 weakly suppresses
  22. L7:85 weakly excites
  23. L8:73 weakly suppresses
  24. L15:1298 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.

2 values between -0.71 and -0.621 values between -0.62 and -0.543 values between -0.54 and -0.458 values between -0.45 and -0.3717 values between -0.37 and -0.2890 values between -0.28 and -0.20301 values between -0.20 and -0.11902 values between -0.11 and -0.031569 values between -0.03 and 0.06652 values between 0.06 and 0.15250 values between 0.15 and 0.2372 values between 0.23 and 0.3222 values between 0.32 and 0.4011 values between 0.40 and 0.492 values between 0.49 and 0.572 values between 0.57 and 0.661 values between 0.66 and 0.741 values between 0.83 and 0.911 values between 2.28 and 2.361 values between 2.62 and 2.70offnegativepositiveoften

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. allis
  2. izzato
  3. vides
  4. Study
  5. urry
  6. olio
  7. Curso
  8. Studies
  9. ologies
  10. Read

Suppresses

  1. 当时
  2. ำ
  3. whim
  4. 凼
  5. ]) ⏎ ⏎
  6. ảo
  7. 车上
  8. 当时的
  9. 机器
  10. 市场的