Neuron Explorer

Qwen3-0.6B

comparative or contrasting qualifiers

A neuron in layer 13 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on words that set up a comparison or contrast, such as 'larger' individuals, 'slowly' versus sugars, 'superior' military force, or 'separate' memory. Turning it up or down did not produce a clear change in the writing.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    ), respectively. Results indicated that body size, standardized by sex, was the most important covariate determining survival of adult D. couperi, suggesting lower survival for larger individuals within each sex. We are uncertain of the mechanisms underlying this result, but

  2. #2 peak

    largely through farm-yard manure, supplemented with need-based fertiliser applications. The most significant aspect of SRI is that the fields are not kept submerged under water all the time, as is usual in rice farming, but are allowed to remain just

  3. #3 peak

    his son to go off on his own. In reality he lies about his trust for his son by sending a spy to

  4. #4 peak

    they are digested more slowly than simple sugars in white bread, for example. The slower digestion means that all

  5. #5 peak

    Whenever they are mentioned, people first think of how they can add them to a indulgent dessert, and nobody thinks of the Macadamias as a healthy food and a rich source of nutrients. Mostly the macadamias are replaced with cashews

  6. #6 peak

    athom—in no small part because of what George Washington and his soldiers achieved against all the odds that nature and a vastly superior military force could pit against them. … Because of their accomplishments, the waning

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: "I have been thinking about the way I have done it, but I have done it so much, I can't do anything, but I have done it so much, I have done it so much, I... I am... 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

    Bamyan, Afghanistan, where all that remains of the giant Buddhas carved into the cliffs are the indentations that once housed them. The Taliban blew these figures up with dynamite in 2001, viewing them as idols from a

  2. #2 peak

    indirect aggression triggers it. The "aggressor" talks behind a person's back<|endoftext|>You are using an outdated and unsupported browser.⏎This browser no longer meets the security requirements of the PCI Security Standards Council. Outdated browsers do not support the requirements

  3. #3 peak

    Inchworm Book Release⏎It wasn’t too long ago that arts education in schools was thought to be a luxury, and arts classes were cut from the curriculum to make room for more time to prepare for standardized tests. Currently

  4. #4 peak

    hey I tried right? So instead of deleting my short article I thought I would share it with my readers. I cannot use it for school purposes since I have to only use material found in our text. As a medieval scholar my final paper will

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. L12:294 suppresses
  2. L12:877 suppresses
  3. L12:1049 excites
  4. L11:1697 suppresses
  5. L9:118 excites
  6. L12:193 excites
  7. L11:1537 excites
  8. L11:146 suppresses
  9. L11:1549 suppresses
  10. L12:1626 suppresses
  11. L11:11 excites
  12. L11:748 suppresses
  13. L12:302 weakly excites
  14. L9:710 weakly excites
  15. L12:945 weakly suppresses
  16. L12:480 weakly excites
  17. L12:2924 weakly suppresses
  18. L12:2011 weakly suppresses
  19. L12:1960 weakly suppresses
  20. L12:428 weakly suppresses
  21. L12:2610 weakly excites
  22. L12:841 weakly suppresses
  23. L12:1363 weakly excites
  24. L11:1592 weakly suppresses

Feeds later layers

  1. L15:1241 suppresses
  2. L15:1669 suppresses
  3. L14:2208 suppresses
  4. L14:2465 suppresses
  5. L15:1157 suppresses
  6. L17:1472 excites
  7. L14:131 excites
  8. L14:599 suppresses
  9. L22:1937 suppresses
  10. L15:58 excites
  11. L16:2492 suppresses
  12. L15:1357 excites
  13. L18:3031 suppresses
  14. L16:1729 suppresses
  15. L15:1563 weakly suppresses
  16. L14:1712 weakly suppresses
  17. L16:295 weakly excites
  18. L16:2576 weakly suppresses
  19. L14:478 weakly excites
  20. L15:1590 weakly suppresses
  21. L14:2858 weakly suppresses
  22. L16:2212 weakly suppresses
  23. L14:2604 weakly suppresses
  24. L14:1483 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 -2.38 and -2.262 values between -2.26 and -2.131 values between -2.13 and -2.001 values between -2.00 and -1.871 values between -1.87 and -1.741 values between -1.74 and -1.623 values between -1.62 and -1.495 values between -1.49 and -1.366 values between -1.36 and -1.232 values between -1.23 and -1.105 values between -1.10 and -0.9815 values between -0.98 and -0.8515 values between -0.85 and -0.7222 values between -0.72 and -0.5943 values between -0.59 and -0.4664 values between -0.46 and -0.34129 values between -0.34 and -0.21381 values between -0.21 and -0.081689 values between -0.08 and 0.051017 values between 0.05 and 0.18326 values between 0.18 and 0.3075 values between 0.30 and 0.4335 values between 0.43 and 0.5620 values between 0.56 and 0.6913 values between 0.69 and 0.829 values between 0.82 and 0.945 values between 0.94 and 1.073 values between 1.07 and 1.204 values between 1.20 and 1.332 values between 1.33 and 1.461 values between 1.46 and 1.585 values between 1.58 and 1.714 values between 1.71 and 1.841 values between 1.84 and 1.971 values between 2.22 and 2.351 values between 2.61 and 2.73offnegativepositiveoften

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. 两类
  2. beit
  3. овый
  4. 一款
  5. mani
  6. pike
  7. voices
  8. connect
  9. 年产值
  10. ovic

Suppresses

  1. ,{"
  2. 以免
  3. 不允许
  4. 因为它
  5. 否定
  6. \Configuration
  7. freeing
  8. calor
  9. 辞职
  10. 原因