Neuron Explorer

Qwen3-0.6B

"of" after "regardless"

A neuron in layer 0 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 "of" in the phrase "regardless of". Turning it up did not produce any clear change in what the model wrote about.

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

    grow female parts, even if it’s genes are male. Similarly, testosterone levels during fetal development will determine what gender the person will identify with later in life regardless of their genetic sex. During early childhood, testosterone drives the rambunctious behavior

  2. #2 peak

    further claims that a long-life may not be the best ideal to strive for; because people are scared of death, they want to live as long as possible regardless of whether or not they will be able to fill that extra time with something profound.

  3. #3 peak

    Pioneer researchers say this winter's weather could have some impact on pest populations, but many pests are just lying in wait.⏎"We see pest problems every year, regardless of conditions," reports Paula Davis, senior manager for insect and disease traits at DuPont

  4. #4 peak

    reduction in the energies. This energy effect operates regardless of the interpretation of red-shifts. It must be applied as a matter of course

  5. #5 peak

    , year-round in 2010 made 77 cents for every dollar men earned across the country. This comparison includes all male and female workers regardless of occupation

  6. #6 peak

    ability that overwhelms the circulatory system of their prey, which leads to death. Then the anaconda swallows it whole! They will hunt any animal regardless of size, and if the opportunity strikes, this slimy giant won’t hesitate in

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 problem of a single person in a room. I have been thinking about how to approach it. I think it's possible to have a single person in a room, but I can't help to make a decision. I have

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

    drawing a picture of.⏎"God," he said.⏎"Why Johnny," she said, "you can't do thatbecause nobody knows what God looks like."⏎"They will when I'm finished," he assured

  2. #2 peak

    average-this-probably-will-work value, or is there more to it than that? And if it is just an average, what knowledge is necessary<|endoftext|>from The Century Dictionary and Cyclopedia⏎- n. A grain of rice.⏎-

  3. #3 peak

    -ocean ridges. Now, the obtained cores showed: No piece of the drilled ocean floor was older than 200 million years, and therefore<|endoftext|>this is an origami ruler. you can make one out of scrap paper, so

  4. #4 peak

    … Read more<|endoftext|>problem 1: Define the term IT. describe different IT trends and their advantages to

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

None: this is the first layer.

Feeds later layers

  1. L1:344 suppresses
  2. L1:119 suppresses
  3. L1:204 excites
  4. L15:1474 excites
  5. L1:907 excites
  6. L3:558 suppresses
  7. L14:620 weakly suppresses
  8. L14:2545 weakly excites
  9. L14:327 weakly suppresses
  10. L1:1019 weakly suppresses
  11. L9:2376 weakly suppresses
  12. L22:1404 weakly excites
  13. L1:189 weakly suppresses
  14. L3:2748 weakly excites
  15. L19:2274 weakly excites
  16. L21:366 weakly suppresses
  17. L3:468 weakly excites
  18. L18:2502 weakly excites
  19. L13:2163 weakly excites
  20. L16:1382 weakly excites comma after an introductory phrase
  21. L27:1244 weakly excites
  22. L11:2175 weakly excites
  23. L4:2156 weakly excites
  24. L15:2597 weakly excites end of a web document

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 -0.74 and -0.721 values between -0.55 and -0.531 values between -0.42 and -0.391 values between -0.39 and -0.361 values between -0.36 and -0.341 values between -0.34 and -0.311 values between -0.28 and -0.263 values between -0.23 and -0.208 values between -0.20 and -0.178 values between -0.17 and -0.1521 values between -0.15 and -0.1250 values between -0.12 and -0.09146 values between -0.09 and -0.07281 values between -0.07 and -0.04588 values between -0.04 and -0.011165 values between -0.01 and 0.02646 values between 0.02 and 0.04431 values between 0.04 and 0.07265 values between 0.07 and 0.10147 values between 0.10 and 0.1268 values between 0.12 and 0.1536 values between 0.15 and 0.1819 values between 0.18 and 0.216 values between 0.21 and 0.236 values between 0.23 and 0.262 values between 0.26 and 0.293 values between 0.29 and 0.312 values between 0.31 and 0.34offnegativepositiveoften

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. :');⏎
  2. 直升
  3. ).</
  4. aving
  5. ⾊
  6. ""); ⏎
  7. ])):⏎
  8. 淅
  9. ');</
  10. '/');⏎

Suppresses

  1. atar
  2. arian
  3. ology
  4. eker
  5. 亿美元
  6. alignment
  7. guarantees
  8. гаранти
  9. тон
  10. .Identifier