Neuron Explorer

Qwen3-0.6B

citation and list breaks, mixed

A neuron in layer 10 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 is weak and messy. It sometimes fires at the end of a citation or just before a text break, and it does not show a clear single pattern. Turning it up gave off-topic rambling.

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

    laser. This technique could be used to construct ultra-small photonic crystals, and could pave the way for all-optical circuits (M Mützel et al 2002 Phys. Rev. Lett. 88 0

  2. #2 peak

    utor(s):||Arndt, T., Boesman, P., Olarte, L., Salaman, P. & Olaciregui , C.|⏎|Facilitator/Compiler(s):||Benstead, P

  3. #3 peak

    your hands with soap and water for at least 20 seconds or use hand sanitiser with a minimum of 60% alcohol base.⏎9. Wash<|endoftext|>Date Updated: 12/17/2020⏎St

  4. #4 peak

    that has been aged is often mixed or "blended" with other coffee that is not aged.⏎"Roasting" is one of the important stages in making<|endoftext|>For those of us who

  5. #5 peak

    fire. This led to a dramatic change in wind direction and speed, sending the fire racing to the south three times faster than it had previously been moving. And<|endoftext|>In Danny Boyle’s 2007 outer space thriller Sunshine, a group

  6. #6 peak

    predict a price decline.⏎- Prediction Markets⏎Shorting Bitcoin can also be done through prediction markets, where you place bets on future events. The prediction markets in<|endoftext|>I

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: What if there is a single line in the line of a single tree. If I didn't know the distance between the two points, I can't connect the two points. What is the length of the distance? 1111

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

    itis. Seborrheic dermatitis is a kind of dermatitis which is caused by physical stress by actions such as travelling and neurological conditions such as Parkinsonís disease. Conditions that affect blood flow in the legs and cause varic<|endoftext|>Res

  2. #2 peak

    groups. Sections 1 through 7 tell the story of agriculture and farmworkers in our area by decade over the past thirty years.⏎In each of the Centerís three decades, certain ethnic communities have been prominent in the diverse farmworker population.

  3. #3 peak

    of principal information and facts for historical examination. But just what is a map? This is a deceptively simple query, before you are motivated to provide an respond to — you may find it significantly more difficult than you feel. But we experience

  4. #4 peak

    the entire crisis on p-values.⏎I think to develop a better understanding of why there is a “crisis”, we need to step back and look across differend fields of study. There is one key question you should be asking yourself: Is

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. L6:660 strongly suppresses
  2. L4:1617 excites
  3. L9:909 suppresses
  4. L7:184 excites
  5. L6:71 excites
  6. L6:179 excites
  7. L4:1446 excites
  8. L6:75 excites
  9. L6:689 excites
  10. L4:872 excites
  11. L4:414 excites
  12. L5:1142 suppresses
  13. L3:627 excites
  14. L6:192 suppresses
  15. L9:268 suppresses
  16. L4:98 excites
  17. L4:1361 suppresses
  18. L9:110 suppresses
  19. L4:1293 excites
  20. L3:1069 suppresses
  21. L3:446 suppresses
  22. L3:389 excites
  23. L8:209 excites
  24. L5:339 suppresses

Feeds later layers

  1. L18:326 strongly suppresses
  2. L16:1004 suppresses
  3. L14:1808 suppresses
  4. L18:1186 suppresses
  5. L17:2343 suppresses
  6. L16:822 suppresses
  7. L14:2231 suppresses
  8. L15:2087 suppresses
  9. L18:467 suppresses
  10. L15:952 suppresses
  11. L17:848 weakly suppresses
  12. L17:2160 weakly suppresses
  13. L17:1314 weakly suppresses
  14. L18:2038 weakly suppresses
  15. L21:714 weakly suppresses
  16. L19:93 weakly suppresses
  17. L16:1731 weakly suppresses
  18. L16:718 weakly suppresses
  19. L15:1856 weakly suppresses
  20. L13:1243 weakly suppresses
  21. L16:1821 weakly suppresses
  22. L14:118 weakly suppresses
  23. L15:335 weakly excites
  24. L19:315 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.

2 values between -2.43 and -2.311 values between -2.19 and -2.071 values between -2.07 and -1.961 values between -1.84 and -1.724 values between -1.60 and -1.485 values between -1.36 and -1.247 values between -1.24 and -1.127 values between -1.12 and -1.0016 values between -1.00 and -0.8823 values between -0.88 and -0.7644 values between -0.76 and -0.6454 values between -0.64 and -0.5299 values between -0.52 and -0.41202 values between -0.41 and -0.29294 values between -0.29 and -0.17514 values between -0.17 and -0.05836 values between -0.05 and 0.07758 values between 0.07 and 0.19565 values between 0.19 and 0.31323 values between 0.31 and 0.43111 values between 0.43 and 0.5529 values between 0.55 and 0.675 values between 0.67 and 0.794 values between 0.79 and 0.911 values between 1.15 and 1.271 values between 1.27 and 1.381 values between 2.22 and 2.34offnegativepositiveoften

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. uish
  2. IonicModule
  3. ]: ⏎
  4. [], ⏎
  5. "", ⏎
  6. '', ⏎
  7. 燮
  8. (), ⏎
  9. INESS
  10. ": ⏎

Suppresses

  1. ide
  2. fase
  3. oppers
  4. expected
  5. realise
  6. GH
  7. inn
  8. eff
  9. Gr
  10. 可在