Neuron Explorer

Qwen3-0.6B

"as if" / "as though" after seem or look

A neuron in layer 22 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 lights up on the word "as" when it comes right after words like "seems", "looks" or "feels" and starts a comparison such as "as if" or "as though". When it was turned up, the model just kept writing the word "if" over and over.

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

    chairman Mahmoud Abbas is accusing Israel of “genocide,” and the PA’s newspaper is calling the war “Israel’s Holocaust.”⏎It seems as if every time Israel defends itself, somebody points an accusing finger and yells “

  2. #2 peak

    Frankenstein’s monster, consisting of two identical M1 Max chips bolted together using a silicon interface that serves as a bridge. This clever design makes it seem as if the conjoined chips are in fact just

  3. #3 peak

    y⏎May 27, 2012⏎When the last American astronauts blasted off from the Moon in 1972, it seemed as if they were leaving behind monuments that would stand for all time. On a life

  4. #4 peak

    (the form is at the bottom of the page).⏎Esther – Day 3⏎Xerxes wanted to show off his beautiful wife! But it seems as though he did not give her too much thought until he had drunk too much!

  5. #5 peak

    his book:⏎The two entities that were supposed to write the regulations, the Pay Board and the Price Commission, wrangled and dithered. When it looked as though they were going to miss a crucial deadline

  6. #6 peak

    because there is seldom just one. In the beginning, actinic keratoses are frequently so small that they are recognized by touch rather than sight. It feels as if you were running a finger over sandpaper. There are many times the number

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: how to make a way for the way to be seen, and how to make the way way to be seen in such a way that the way is seen, and the way is made in such a way that the way is seen. 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

    Mind and Life Institute dialogue. We highlight some of the passages Adele Diamond presented to the Dalai Lama in Dharamsala — including texts from Rabbi Heschel, Bashevis Singer, Rachel Naomi Remen, and Henri Nouwen

  2. #2 peak

    the cortical area just posterior to the auditory cortex (Heschl's gyrus) within the Sylvian fissure. It is a triangular

  3. #3 peak

    .⏎Plants were perfect and completely green and had a good amount<|endoftext|>A remarkable new species of tyrannosaur has been unearthed in Grand Staircase-Escalante National Monument (GSENM), southern Utah. The huge carnivore inhabited

  4. #4 peak

    . This class shows how such demographic dynamics influenced early jazz development and how the urban environment in promoted gender diversification and the social acceptance of jazz. Report on Hersch due. Read Lawrence Gushee, "A Preliminary Chron

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. L20:3062 strongly suppresses
  2. L20:2859 suppresses
  3. L21:1695 suppresses
  4. L19:1815 suppresses
  5. L21:1592 suppresses
  6. L19:1332 excites
  7. L19:2478 suppresses
  8. L19:2769 excites
  9. L21:1064 suppresses
  10. L21:1705 excites
  11. L18:2163 excites
  12. L20:2572 excites
  13. L18:2901 suppresses
  14. L20:726 excites
  15. L20:2600 excites
  16. L18:2503 suppresses
  17. L19:1783 weakly excites
  18. L21:2847 weakly excites
  19. L19:3062 weakly excites
  20. L20:2967 weakly suppresses
  21. L1:524 weakly suppresses
  22. L8:852 weakly excites comma inside large numbers
  23. L18:2350 weakly excites
  24. L19:2916 weakly suppresses

Feeds later layers

  1. L25:768 strongly excites
  2. L27:511 strongly excites
  3. L23:449 strongly excites
  4. L26:509 strongly excites
  5. L24:1945 excites
  6. L27:1084 excites
  7. L24:100 weakly suppresses
  8. L27:3024 weakly excites
  9. L23:1475 weakly excites
  10. L23:1813 weakly suppresses
  11. L25:1387 weakly excites end-of-document token and code keywords
  12. L27:2056 weakly suppresses the 'w' in shipwreck
  13. L24:2085 weakly suppresses
  14. L23:2514 weakly suppresses
  15. L27:987 weakly suppresses
  16. L27:2696 weakly suppresses
  17. L26:265 weakly excites
  18. L24:1002 weakly suppresses
  19. L25:1238 weakly suppresses
  20. L25:2741 weakly suppresses
  21. L26:2640 weakly suppresses
  22. L27:29 weakly excites
  23. L27:2325 weakly suppresses
  24. L25:2776 weakly suppresses

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 -10.01 and -9.032 values between -4.13 and -3.152 values between -3.15 and -2.1878 values between -2.18 and -1.202042 values between -1.20 and -0.221703 values between -0.22 and 0.7641 values between 0.76 and 1.7411 values between 1.74 and 2.723 values between 2.72 and 3.707 values between 3.70 and 4.683 values between 4.68 and 5.662 values between 5.66 and 6.643 values between 6.64 and 7.621 values between 7.62 and 8.601 values between 8.60 and 9.572 values between 9.57 and 10.551 values between 10.55 and 11.531 values between 11.53 and 12.511 values between 17.41 and 18.391 values between 19.37 and 20.341 values between 24.26 and 25.241 values between 28.18 and 29.16offnegativepositiveoften

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. if
  2. 如果
  3. if
  4. 若
  5. If
  6. If
  7. →if
  8. если
  9. nếu
  10. 如果有

Suppresses

  1. DCALL
  2. 背上
  3. 회의
  4. 那么简单
  5. 并对
  6. agnar
  7. шин
  8. icc
  9. 意义上
  10. ared