Neuron Explorer

Qwen3-0.6B

verb right before its object

A neuron in layer 23 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 a verb or preposition that is about to be followed by its object, such as 'picked', 'chose', 'showed' or 'submerges it in'. Its negative side fires on 'how' after words like 'when and'. Turning it up made the model repeat phrases like 'at least a few'.

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

    G) vs. electro-galvanized⏎Hot dip galvanization (HDG) is a process that takes cleaned and treated steel and submerges it in a molten zinc bath. The zinc metal at these temperatures bonds with the steel

  2. #2 peak

    student figured out who it was — Italian Prime Minister Silvio Berlusconi.⏎Todorov's next example illustrated how facial recognition can have serious consequences. He showed the students a photo of Darrell Edwards, a man who is currently fighting a

  3. #3 peak

    -D structure, but are as far away from morphine and codeine as possible," Manglik says.⏎The team ran 3 million possibilities through the computer and picked the <|endoftext|>Research on the relationship between microbiome and cancer has made significant progress

  4. #4 peak

    healthy like replacing full-fat version of foods such as milk, yogurt, and salad dressing with the fat-free or low-fat kind.⏎The best way to give your body the balanced nutrition it needs is by eating a variety of nutrient-packed foods every day

  5. #5 peak

    hydrogen peroxide reacts with organic material it breaks down into oxygen and water.⏎Whiten Clothes – An Alternative to Bleach⏎Add a cup of Peroxide to white clothes in your laundry to whiten them. Peroxide is great to

  6. #6 peak

    on in the same way, so I try to work in changes and transitions that make my lectures more interesting and engaging.⏎At the beginning of each lecture, I show and discuss “student questions of the day” that pertain to the previous lecture

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: on a day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day in the day

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

    , the ancestors of all people trace to Africa. That much is clear. We are all Africans. But there’s been considerable room for debate about exactly when and how many times modern humans made their way out of Africa to

  2. #2 peak

    sure though – the fascination with space exploration isn’t likely to go away any time soon.⏎“There’s a driving desire within us to find out who we are and how we fit and what the world around us is about,” said Kistler.⏎

  3. #3 peak

    future? I think my future is better."⏎Walesa spoke of his past – and of his fight. By analyzing our past, we can"understand how to build the future," he said. He then told the crowd that even though

  4. #4 peak

    about the body from the media.⏎"People don't remember how many factors are related to body images and the role models being pushed upon our children

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. L22:554 suppresses
  2. L18:1339 excites
  3. L17:264 suppresses
  4. L20:593 excites
  5. L19:272 suppresses
  6. L19:2406 weakly excites
  7. L17:2363 weakly suppresses
  8. L20:417 weakly excites
  9. L19:1501 weakly suppresses
  10. L21:1851 weakly excites
  11. L19:1980 weakly excites
  12. L7:403 weakly suppresses
  13. L16:832 weakly excites
  14. L22:353 weakly suppresses
  15. L8:508 weakly suppresses
  16. L18:2153 weakly suppresses
  17. L17:1557 weakly suppresses
  18. L9:1787 weakly suppresses
  19. L11:810 weakly excites
  20. L19:1305 weakly suppresses unclear: verbs after a noun or 'that'
  21. L18:1601 weakly suppresses
  22. L20:728 weakly suppresses
  23. L16:1860 weakly suppresses
  24. L3:2730 weakly suppresses

Feeds later layers

  1. L25:1581 weakly excites
  2. L25:1132 weakly excites
  3. L24:380 weakly excites the word 'at' in place phrases
  4. L25:2791 weakly suppresses
  5. L25:664 weakly suppresses
  6. L25:696 weakly excites
  7. L26:898 weakly excites
  8. L24:1785 weakly suppresses
  9. L26:880 weakly excites
  10. L27:1484 weakly excites
  11. L26:148 weakly excites
  12. L24:655 weakly suppresses
  13. L27:2184 weakly excites
  14. L24:1929 weakly suppresses
  15. L24:1070 weakly suppresses
  16. L26:2452 weakly excites
  17. L24:1354 weakly suppresses
  18. L25:1865 weakly excites 'in' in 'keep in mind'
  19. L27:168 weakly excites
  20. L27:213 weakly suppresses
  21. L24:722 weakly excites 'about' or 'approximately' before a number
  22. L25:1157 weakly excites
  23. L25:1836 weakly suppresses
  24. L26:474 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 -17.78 and -16.701 values between -16.70 and -15.611 values between -15.61 and -14.532 values between -13.44 and -12.351 values between -12.35 and -11.275 values between -11.27 and -10.1811 values between -10.18 and -9.104 values between -9.10 and -8.0115 values between -8.01 and -6.9322 values between -6.93 and -5.8444 values between -5.84 and -4.7658 values between -4.76 and -3.67116 values between -3.67 and -2.58206 values between -2.58 and -1.50445 values between -1.50 and -0.411722 values between -0.41 and 0.67419 values between 0.67 and 1.76243 values between 1.76 and 2.84180 values between 2.84 and 3.93133 values between 3.93 and 5.0288 values between 5.02 and 6.1050 values between 6.10 and 7.1942 values between 7.19 and 8.2730 values between 8.27 and 9.3622 values between 9.36 and 10.4415 values between 10.44 and 11.536 values between 11.53 and 12.616 values between 12.61 and 13.7011 values between 13.70 and 14.791 values between 14.79 and 15.873 values between 15.87 and 16.961 values between 18.04 and 19.132 values between 19.13 and 20.211 values between 20.21 and 21.301 values between 24.56 and 25.64offnegativepositiveoften

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. trer
  2. icens
  3. atics
  4. alyze
  5. osit
  6. off
  7. ưỡ
  8. ateurs
  9. [out
  10. oplan

Suppresses

  1. 为基础
  2. often
  3. ithub
  4. usually
  5. ushman
  6. Cp
  7. often
  8. 大全
  9. 为主
  10. 有些人