Neuron Explorer

Qwen3-0.6B

vibr- word pieces (vibration)

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 reacts to words about vibrating or shaking, such as 'vibratory', 'vibrate', 'vibrato' and 'shake'. Turning it up did not make the model write about vibration, so the link is only visible in the text it reads.

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

    research program was conducted at the St. Anthony Falls Hydraulic Laboratory to investigate impulse pressures caused by the collapse of transient cavitation bubbles. The impulse pressures associated with vibratory as well as hydrodynamic cavitation were measured. A stationary specimen subjected to vibr

  2. #2 peak

    I learned, had arrived at a music venue near my apartment. But why? What was happening there? Then I saw pictures of other friends enjoying fancy milkshakes at a trendy restaurant. Suddenly, my simple domestic pleasures paled in comparison with

  3. #3 peak

    :⏎Blueberry plants host so many insects that only some of them can be mentioned. The chief pollinators are bees. Inside the flower, bees often vibrate their wings vigorously, shaking loose the pollen. Look for flowers punctured at the

  4. #4 peak

    able multivibrators in digital logic probes, DC-DC converters, tachometers, analog

  5. #5 peak

    peggios like a guitar. Several later models of the Omnichord added MIDI compatibility, a greater selection of sounds for the Sonic Strings, vibrato, and chord memory, called Chord Computer.⏎The OM27 was the

  6. #6 peak

    carton.⏎- Demonstrate the sound matching activity to your child. Shake, listen, match. You can say, “May I show you how we

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 it, and I have been thinking about it. I have been thinking about it. I have been thinking about it. I have been thinking about it. I have been thinking about it. I have been thinking about

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

    ).⏎What nutrients are depleted by alcohol?⏎Vitamins B-1, B-3 and B-6, or thiamine, niacin and pyridoxine, are directly or indirectly involved in alcohol metabolism, and they are among

  2. #2 peak

    balancing the potential benefits and risks. Talk to your doctor about the right medicines for you.⏎A common and safe remedy is to take supplemental vitamin B6 (pyridoxine), to a maximum of 100 mg/day. The American

  3. #3 peak

    marker. Battleship U.S.S. Ohio (approx<|endoftext|>“Snow White,” the beloved, first

  4. #4 peak

    became the first Gavi-eligible country to introduce the inactivated polio vaccine (IPV). The introduction as a component of the routine immunization schedule in Nepal is part

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. L1:2590 suppresses
  2. L0:1301 suppresses
  3. L0:178 weakly suppresses
  4. L0:1224 weakly suppresses
  5. L1:951 weakly excites
  6. L1:92 weakly excites
  7. L1:1973 weakly suppresses
  8. L1:1357 weakly excites
  9. L1:2470 weakly suppresses
  10. L1:2553 weakly excites
  11. L1:1099 weakly excites
  12. L0:2639 weakly excites unclear: weak hits on 'holiday' and 'playing', negative on 'Ada-fruit'
  13. L0:1290 weakly suppresses
  14. L0:320 weakly suppresses the digit 0 inside numbers
  15. L0:2334 weakly excites
  16. L1:1603 weakly excites
  17. L1:2906 weakly suppresses
  18. L1:311 weakly suppresses
  19. L0:2249 weakly excites medical history in health text
  20. L0:2865 weakly excites
  21. L0:2248 weakly excites
  22. L0:3063 weakly excites
  23. L1:1857 weakly suppresses
  24. L0:763 weakly suppresses

Feeds later layers

  1. L4:2932 excites
  2. L7:1506 weakly excites
  3. L9:2784 weakly excites unclear: deity names strongly negative
  4. L20:2928 weakly excites
  5. L10:476 weakly excites
  6. L6:929 weakly excites
  7. L9:2116 weakly suppresses
  8. L6:870 weakly excites
  9. L6:869 weakly suppresses
  10. L12:2332 weakly excites
  11. L11:2412 weakly excites
  12. L4:2976 weakly excites
  13. L14:2042 weakly suppresses comma after 'For X,' introducing a viewpoint
  14. L3:2575 weakly excites
  15. L25:1183 weakly suppresses
  16. L22:2801 weakly excites
  17. L8:3053 weakly excites
  18. L19:1663 weakly suppresses
  19. L9:1356 weakly suppresses
  20. L3:2881 weakly suppresses
  21. L3:2225 weakly excites
  22. L3:2754 weakly suppresses
  23. L5:1242 weakly suppresses
  24. L12:503 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.

1 values between -1.84 and -1.711 values between -1.47 and -1.352 values between -1.35 and -1.221 values between -1.10 and -0.981 values between -0.98 and -0.853 values between -0.61 and -0.491 values between -0.49 and -0.3627 values between -0.36 and -0.24222 values between -0.24 and -0.121792 values between -0.12 and 0.011650 values between 0.01 and 0.13186 values between 0.13 and 0.2515 values between 0.25 and 0.371 values between 0.37 and 0.502 values between 0.50 and 0.621 values between 0.62 and 0.741 values between 1.48 and 1.601 values between 2.95 and 3.07offnegativepositiveoften

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. ensively
  2. urch
  3. 一下
  4. reaching
  5. 奄
  6. cntl
  7. uja
  8. rama
  9. \$
  10. aram

Suppresses

  1. ced
  2. lder
  3. isters
  4. gué
  5. oste
  6. ister
  7. 利器
  8. ettle
  9. ück
  10. inter