Neuron Explorer

Qwen3-0.6B

unclear: word pieces around turbines and wind shear

A neuron in layer 3 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 mix of unrelated word pieces, most often 'turbines' in talk about engines and 'shear' in weather reports. It also dips on 'was' or 'is' in 'where was' questions. Turning it up made the model's writing wander without a clear theme.

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

    cables, as in vertical-lift drawbridges, power shovels, and cranes. Before the extensive use of electric motors, steam engines or water turbines often supplied the power for factory machinery. One engine or turbine might run a whole

  2. #2 peak

    September 13 local wind shear was having its effect on the storm, and Hurricane Humberto’s winds had begun

  3. #3 peak

    4 hours. They predicted that very strong wind shear, dry air and a track over marginally warm sea surface temperatures could result in

  4. #4 peak

    One might even say that Americans are a hero-worshiping people.⏎How can the American people love both equality and inequality simultaneously? How

  5. #5 peak

    mud wallows or taking a nap in a big, fluffy hay bed. No, those aren’t warthogs you are looking at—these are one of the most critically endangered pigs in

  6. #6 peak

    a lot of clay. I always dug the hole at least 25 percent deeper than the root ball and refilled that portion with watered-down good soil before placing the tree in the hole. The rest of the hole would

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: A big company, the way to get a good life, a life for a resident. Let me try to the weather. Okay, that's a life for a resident. Let me try to the weather. So, this is the way

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

    was this wreck, where was it located, why was it important? When did it happen?⏎2. What

  2. #2 peak

    expelled the only red state in the national organizaton" -- but<|endoftext|>Raising Our Children to Be Gatekeepers⏎And the LORD said unto Cain, Where is Abel thy brother? And he said, I know not: Am I my brother

  3. #3 peak

    2. Every day the boss enters the office⏎at half past nine⏎3. My friend will ring me up at eight o’clock⏎4. Where was Kate going when you met her yesterday?⏎does your brother work?⏎6. What

  4. #4 peak

    or deep? Explain the type of plate boundary it was near. Why are earthquakes common to this area?⏎Where was the deepest earthquake during the week? On what type of boundary was this?⏎Where

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. L1:1159 weakly excites
  2. L1:2536 weakly suppresses
  3. L2:1657 weakly excites
  4. L1:2654 weakly excites
  5. L2:229 weakly excites
  6. L0:710 weakly excites
  7. L2:115 weakly excites
  8. L2:75 weakly suppresses
  9. L0:433 weakly excites
  10. L0:260 weakly suppresses
  11. L2:2480 weakly suppresses
  12. L0:1036 weakly excites
  13. L1:535 weakly suppresses
  14. L1:643 weakly suppresses
  15. L2:1584 weakly excites
  16. L1:2683 weakly excites
  17. L1:2723 weakly excites mid-word syllables in capitalised names
  18. L0:1398 weakly suppresses
  19. L1:2340 weakly suppresses
  20. L0:1617 weakly suppresses
  21. L0:1568 weakly suppresses
  22. L2:2471 weakly suppresses
  23. L0:2019 weakly suppresses
  24. L0:1989 weakly excites

Feeds later layers

  1. L9:2421 weakly excites
  2. L8:505 weakly excites
  3. L15:2175 weakly suppresses
  4. L19:1418 weakly suppresses
  5. L4:1594 weakly suppresses
  6. L18:1510 weakly excites
  7. L26:168 weakly suppresses
  8. L12:1491 weakly excites
  9. L13:2407 weakly excites
  10. L22:1641 weakly suppresses
  11. L13:1603 weakly excites
  12. L8:2436 weakly suppresses
  13. L26:1050 weakly suppresses
  14. L7:939 weakly suppresses
  15. L25:1079 weakly suppresses
  16. L14:367 weakly excites
  17. L6:2911 weakly suppresses
  18. L17:1759 weakly excites
  19. L10:104 weakly excites
  20. L8:2285 weakly suppresses
  21. L4:844 weakly excites
  22. L18:1734 weakly excites
  23. L8:2041 weakly suppresses
  24. L23:3008 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 -1.95 and -1.851 values between -1.33 and -1.221 values between -1.01 and -0.9110 values between -0.49 and -0.3966 values between -0.39 and -0.28244 values between -0.28 and -0.18835 values between -0.18 and -0.071715 values between -0.07 and 0.03789 values between 0.03 and 0.14196 values between 0.14 and 0.2431 values between 0.24 and 0.347 values between 0.34 and 0.453 values between 0.45 and 0.553 values between 0.55 and 0.663 values between 0.76 and 0.871 values between 1.60 and 1.701 values between 2.01 and 2.121 values between 2.12 and 2.22offnegativepositiveoften

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. per
  2. hard
  3. pers
  4. еб
  5. 短线
  6. conde
  7. ере
  8. SSION
  9. use
  10. итель

Suppresses

  1. anthology
  2. onnement
  3. scratched
  4. 百花
  5. _Collections
  6. deductible
  7. TAIL
  8. .hu
  9. ozilla
  10. *T