Neuron Explorer

Qwen3-0.6B

unclear: word fragments like 'limpet' and 'lim'

A neuron in layer 2 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 is unclear. It fires on pieces of words, most often the 'limpet' sea creature and word starts like 'lim' and 'kelp', and it is weak. Turning it up or down did not produce an obvious change.

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

    better and more robust vehicles—such as bicycles, boats, and race cars—in the not-so-distant future.⏎"It's about translating design principles found in limpets to form structures that are strong, yet light," says study leader Asa Barber

  2. #2 peak

    contain—which makes them weaker than smaller, less flawed, structures. (Also see "Watery Gecko Grip Could Lead to Stickier Tape.")⏎Not so for limpets.⏎"Limpets are clever because they use mineral fibers below a particular size,

  3. #3 peak

    instance, racing bikes and race cars use light, strong composite materials to cope with stress, and a limpet-tooth-inspired design could help engineers build a stronger chassis that doesn't sacrifice speed

  4. #4 peak

    avitreal injection of pharmacologic agent.⏎Courtesy of Stephen J. Kim, MD.⏎Injections can be accomplished safely 3–4 mm posterior to the limbus. Commonly used methods for anesthesia administration before intravitreal injections include pledgets

  5. #5 peak

    , which cruised in the Baltic made no attack on German territory, and the immunity of such places as Kiel and Danzig was, no doubt, due to the formidable batteries which guarded the

  6. #6 peak

    flowers on hand for one vase, use two, or even three. Do not crowd flowers.⏎-<|endoftext|>Pacific ecosystem: An ochre sea star makes a kelp forest home in Monterey Bay, California.⏎Image courtesy of Kip F. Evans

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 lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately

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

    agua is the highest peak in both the Western and Southern Hemispheres.⏎McKinley, North America⏎At the Denali National Park in the US state of

  2. #2 peak

    graduating in 1952. One year later he married Louise Jones, an educator and activist in her own right. The couple has two daughters.⏎McKinney served as pastor of Olney Street Baptist Church in Providence, Rhode Island between

  3. #3 peak

    .⏎Our research is ongoing and aims to best understand these elusive creatures in order to best manage them and the essential ecological processes they provide.⏎Ms Jeannine McManus is the research and field manager. She is currently doing her PhD with the

  4. #4 peak

    J.B. Kaufman, who wrote the exhibit’s catalog, called the film “the culmination” of Disney’s previous work on short, animated films featuring Mickey Mouse, among other characters. “It established a gold standard for the craft of making

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. L0:1301 weakly suppresses
  2. L0:1224 weakly suppresses
  3. L0:178 weakly suppresses
  4. L0:1320 weakly suppresses
  5. L1:1029 weakly excites
  6. L0:1075 weakly excites
  7. L1:1464 weakly excites
  8. L1:189 weakly suppresses
  9. L0:6 weakly excites
  10. L0:2069 weakly excites
  11. L0:1 weakly suppresses
  12. L1:1674 weakly excites
  13. L0:1646 weakly excites
  14. L1:1959 weakly suppresses
  15. L1:331 weakly suppresses
  16. L1:194 weakly suppresses
  17. L1:2922 weakly excites
  18. L1:146 weakly excites
  19. L0:2002 weakly excites
  20. L1:1460 weakly excites
  21. L0:611 weakly suppresses
  22. L1:2111 weakly excites
  23. L1:535 weakly suppresses
  24. L0:799 weakly suppresses

Feeds later layers

  1. L11:1982 weakly suppresses
  2. L10:2704 weakly suppresses
  3. L5:2646 weakly excites
  4. L7:2342 weakly suppresses
  5. L6:2611 weakly suppresses
  6. L11:2567 weakly excites
  7. L19:1265 weakly suppresses
  8. L23:2708 weakly suppresses
  9. L17:2068 weakly suppresses
  10. L5:1880 weakly excites
  11. L12:212 weakly excites
  12. L16:2822 weakly suppresses
  13. L17:361 weakly suppresses
  14. L5:2188 weakly suppresses
  15. L11:1764 weakly suppresses
  16. L5:2521 weakly excites
  17. L15:197 weakly excites
  18. L3:2642 weakly suppresses
  19. L4:576 weakly suppresses
  20. L15:2654 weakly suppresses
  21. L10:2067 weakly suppresses
  22. L24:962 weakly suppresses
  23. L13:2831 weakly suppresses
  24. L7:2399 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 -0.89 and -0.851 values between -0.80 and -0.761 values between -0.63 and -0.592 values between -0.47 and -0.422 values between -0.42 and -0.383 values between -0.38 and -0.344 values between -0.34 and -0.3017 values between -0.30 and -0.2637 values between -0.26 and -0.2160 values between -0.21 and -0.17133 values between -0.17 and -0.13241 values between -0.13 and -0.09557 values between -0.09 and -0.041067 values between -0.04 and -0.001076 values between -0.00 and 0.04370 values between 0.04 and 0.08172 values between 0.08 and 0.1383 values between 0.13 and 0.1737 values between 0.17 and 0.2118 values between 0.21 and 0.2512 values between 0.25 and 0.298 values between 0.29 and 0.342 values between 0.34 and 0.382 values between 0.38 and 0.421 values between 0.50 and 0.551 values between 0.76 and 0.80offnegativepositiveoften

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. ırl
  2. tures
  3. inis
  4. atoria
  5. ǐ
  6. icap
  7. 社会资本
  8. inea
  9. Fatal
  10. Neither

Suppresses

  1. 才知道
  2. 활
  3. silence
  4. annunci
  5. 笑了
  6. ám
  7. dess
  8. allowed
  9. 蟆
  10. eração