Neuron Explorer

Qwen3-0.6B

friction, surfaces and photoelectric physics

A neuron in layer 9 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 physics and materials words about surfaces and electrons, such as 'photoelectric', 'work function', 'friction', 'lubric' (lubrication) and 'rake' in cutting tools. It is near zero or slightly negative on food and drug text. Turning it up or down did not clearly change the topic of the model's replies.

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

    -Edmond Becquerel, the father of the guy who discovered radioactivity. This was followed in 1887 by the discovery of the photoelectric effect by Heinrich Hertz, an ethnically half-Jewish, half-G

  2. #2 peak

    and come up with a solution that proves to be surprisingly effective. They have chosen the material Gadolinium, which has a special property: the so-called work function is very low. An electron can easily cross, out of the material an into

  3. #3 peak

    of fractured rocks is mainly composed by two mechanisms: one is the anti-shear due to the interlocking of microasperities, the other is the anti-friction owing to the relative shear of macro waves on the fractured rock surfaces. The study

  4. #4 peak

    shapes and holes during milling and can cut a variety of materials in several directions. The end mill is generally described as a positive shape or negative shape according to the rake angle of the tool bit.⏎In most cases, a positive shape with less cutting

  5. #5 peak

    (light)⏎Graph that shows the amount of light absorbed at a number of wavelengths⏎Four parts of a spectrophotometer⏎Device that isolates a photoelectric tube⏎Standard curve limitations⏎Standard curve is specific to the pigment and its buffer

  6. #6 peak

    Less Lubrication Of Vagina⏎A drop in the estrogen levels in the body during menopause may lead to the drying of the vaginal walls and hence less lubrication during sexual intercourse. During this time, you will also lose interest in sexual

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: If you want to create a story in the future, you would like to know how your characters would be like. However, what is the best place to start? The best place to start is... well, the future. So, if

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

    broth for a hearty and comforting soup that’s perfect for chilly days.⏎- Broccoli and quinoa stir-fry: Saute broccoli florets with cooked quinoa, garlic, ginger, and soy sauce for a protein-packed and satisfying meal.⏎

  2. #2 peak

    hea, and E. coli or salmonella infection.⏎How to use⏎Use Ampicillin as directed by your doctor.⏎Take Ampicillin by mouth at least one-half hour before or 2 hours after eating.⏎Take Ampicillin with

  3. #3 peak

    GI, such as white bread and white rice, cause a rapid increase in blood sugar levels. On the other side, foods with low GI, such as whole grains and sweet potato, are more slowly digested, causing a lower and gentler

  4. #4 peak

    of pet food that is designed to meet dogs’ nutritional needs. It is usually made from ingredients such as meat, grains, vegetables, and fruits. Dog food may also contain vitamins, minerals, and

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. L8:2335 weakly suppresses
  2. L7:3000 weakly suppresses
  3. L4:2519 weakly excites
  4. L7:2008 weakly suppresses
  5. L6:18 weakly excites
  6. L8:2925 weakly suppresses
  7. L5:919 weakly suppresses
  8. L1:951 weakly excites
  9. L6:1820 weakly excites
  10. L4:2736 weakly suppresses
  11. L2:1865 weakly suppresses
  12. L6:2538 weakly suppresses
  13. L8:180 weakly suppresses
  14. L1:1158 weakly excites
  15. L8:828 weakly suppresses
  16. L6:2952 weakly excites
  17. L7:1585 weakly suppresses
  18. L2:1854 weakly suppresses
  19. L8:1552 weakly suppresses
  20. L7:2289 weakly suppresses
  21. L6:85 weakly excites
  22. L5:245 weakly excites
  23. L6:2506 weakly suppresses
  24. L8:2706 weakly suppresses

Feeds later layers

  1. L10:361 suppresses
  2. L10:1818 weakly excites
  3. L10:1709 weakly excites
  4. L13:146 weakly excites
  5. L10:1595 weakly excites
  6. L12:263 weakly excites
  7. L24:2992 weakly suppresses
  8. L13:177 weakly suppresses
  9. L15:50 weakly excites
  10. L22:674 weakly excites
  11. L10:2214 weakly excites
  12. L11:2898 weakly suppresses
  13. L18:220 weakly excites
  14. L11:149 weakly excites
  15. L21:1483 weakly excites
  16. L14:391 weakly suppresses
  17. L20:1750 weakly excites
  18. L10:3068 weakly excites
  19. L25:2381 weakly excites
  20. L13:488 weakly excites
  21. L13:2669 weakly excites
  22. L19:2269 weakly suppresses
  23. L23:1627 weakly suppresses
  24. L18:387 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 -3.39 and -3.213 values between -2.31 and -2.131 values between -1.59 and -1.412 values between -1.41 and -1.233 values between -1.23 and -1.051 values between -1.05 and -0.873 values between -0.87 and -0.693 values between -0.69 and -0.5146 values between -0.51 and -0.34445 values between -0.34 and -0.161843 values between -0.16 and 0.031312 values between 0.03 and 0.21209 values between 0.21 and 0.3923 values between 0.39 and 0.571 values between 0.57 and 0.753 values between 0.75 and 0.935 values between 0.93 and 1.101 values between 1.10 and 1.281 values between 1.46 and 1.641 values between 1.64 and 1.821 values between 3.62 and 3.80offnegativepositiveoften

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. forcement
  2. 驴
  3. pin
  4. hesion
  5. _flags
  6. akh
  7. 丝
  8. flag
  9. 附
  10. Pin

Suppresses

  1. -analytics
  2. ROADCAST
  3. inality
  4. /query
  5. tragedy
  6. ARCH
  7. yans
  8. sitcom
  9. levision
  10. -analysis