Neuron Explorer

Qwen3-0.6B

'yet' and 'haven't' after have

A neuron in layer 4 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 fires on words like 'yet', 'haven't' and 'have now' that show something has or has not happened so far. The examples where it is strongly negative are mostly the name Ramses, but that is a weaker pattern. Turning it up did not give a clear theme in the replies.

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

    of existing ties.⏎In addition, volunteering can be a risk-free method to try out a new profession. If you're thinking about a career in nursing but have yet to gain much experience, volunteering

  2. #2 peak

    . We can multiply by 5 and divide by 5.⏎- 1/5 times 5 is just going to be 1. So we haven't changed the value of the expression.⏎- But when we do it this way,

  3. #3 peak

    commemorate the Jewish holiday, Yom Ha’ Shoah, will begin at 5 p.m. in the Hatfield

  4. #4 peak

    school-business partnership.⏎- Design & Thinking Official Trailer (02:28)⏎Design & Thinking is a documentary released this year about design thinking. I haven't seen it yet, but I'm loving the approach -- the cacophony of

  5. #5 peak

    shield against certain kinds of flu virus.⏎Some individuals suggest drinking Echinacea tea to shorten the length of cold and respiratory illness symptoms. However, this result has yet to be tried by research.⏎As we know garlic

  6. #6 peak

    Yom Ha' Shoah is not just a Jewish event — it touches everyone,” she says.⏎“Yom Ha' Shoah is a wake-up call that genocide is happening now and we have

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 power of a king in the capital: the god of a god of the goddess of a goddess. The king is in the city of a. The god of a god of a. The goddess of a. The queen of a.

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 being constructed to continue the cult of the deceased Pharaoh and to honor Amon, was also built in reverent memory of Seti's father, Ramses I, who died before constructing a temple of his own. It was not completed

  2. #2 peak

    1190–87 bce), founder of the 20th dynasty (1190–1075 bce), Ramses found Egypt upon his accession only recently recovered from the unsettled political conditions that had

  3. #3 peak

    , instead of Apap, the serpent of darkness, the great antagonist of the beneficent Osiris. The hatred of the Egyptians for the very name of Set was carried so far that it was chiseled out of the monuments; the

  4. #4 peak

    and Mount Carmel. The attempted invasion ended Egyptian pretensions to a Syro-Palestinian hegemony.⏎Two more years of peace ensued, but in Ramses’ 11th year a new coalition of Libyan tribes infiltrated the western

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. L3:464 excites
  2. L3:1933 excites
  3. L1:2444 weakly suppresses
  4. L2:70 weakly excites
  5. L3:421 weakly suppresses
  6. L3:2369 weakly suppresses
  7. L3:1496 weakly excites period in abbreviations and initials
  8. L3:50 weakly excites
  9. L2:75 weakly suppresses
  10. L3:512 weakly suppresses
  11. L1:743 weakly suppresses
  12. L3:691 weakly excites
  13. L3:2351 weakly suppresses
  14. L3:953 weakly suppresses
  15. L3:769 weakly suppresses
  16. L0:1416 weakly suppresses
  17. L1:180 weakly suppresses
  18. L3:1260 weakly suppresses
  19. L3:2440 weakly suppresses
  20. L3:2501 weakly excites
  21. L3:1633 weakly excites
  22. L3:2473 weakly suppresses
  23. L2:1321 weakly suppresses
  24. L2:3042 weakly suppresses

Feeds later layers

  1. L5:660 excites
  2. L10:817 suppresses
  3. L20:1632 suppresses
  4. L11:1926 suppresses
  5. L5:2738 suppresses
  6. L5:214 weakly suppresses
  7. L5:2623 weakly suppresses
  8. L22:2657 weakly suppresses
  9. L6:2001 weakly suppresses
  10. L9:2449 weakly excites
  11. L17:1931 weakly suppresses
  12. L13:1295 weakly suppresses
  13. L18:911 weakly excites comma after 'and' or 'and then' clause
  14. L5:17 weakly suppresses
  15. L10:1150 weakly suppresses
  16. L9:629 weakly excites
  17. L16:2656 weakly excites
  18. L7:1144 weakly excites
  19. L14:2875 weakly suppresses
  20. L19:2948 weakly excites
  21. L12:2697 weakly excites
  22. L21:144 weakly excites
  23. L21:1281 weakly suppresses
  24. L11:1714 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 -2.84 and -2.741 values between -2.14 and -2.041 values between -1.94 and -1.841 values between -1.74 and -1.641 values between -1.13 and -1.031 values between -1.03 and -0.931 values between -0.93 and -0.835 values between -0.73 and -0.632 values between -0.63 and -0.5311 values between -0.53 and -0.4336 values between -0.43 and -0.33148 values between -0.33 and -0.23397 values between -0.23 and -0.13946 values between -0.13 and -0.031322 values between -0.03 and 0.07681 values between 0.07 and 0.17231 values between 0.17 and 0.2784 values between 0.27 and 0.3727 values between 0.37 and 0.477 values between 0.47 and 0.571 values between 0.67 and 0.771 values between 0.87 and 0.982 values between 1.08 and 1.18offnegativepositiveoften

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. ivant
  2. <Class
  3. categor
  4. المادة
  5. MX
  6. here
  7. here
  8. 惯
  9. \TestCase
  10. (class

Suppresses

  1. 谁能
  2. 一族
  3. leneck
  4. 所能
  5. .cloudflare
  6. щения
  7. ucken
  8. 报仇
  9. çıktı
  10. ились