Neuron Explorer

Qwen3-0.6B

blank before a size number

A neuron in layer 21 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 fires on the empty spot right before a measurement, such as the gap in "grows to [blank] feet tall" or "about [blank] inches long". Turning it up gave text about sizes and heights.

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

    Then you would notice that he was slightly different, based on the brow ridge and facial features.⏎It is worth remembering that the average height of Homo erectus was 6 ft., they

  2. #2 peak

    inea pigs are considered large weighing approximately 700 and 1200 grams or 1.5 to 2.5 pounds and measuring around 20 to

  3. #3 peak

    Maturity: 45–60 days⏎Native: Widespread, probably Asia⏎Hardiness: Frost-tolerant annual⏎Plant Dimensions: 16"–22" tall⏎Variety Information: Large, dark

  4. #4 peak

    Glow," a carnelian cherry dogwood (Cornus mas) cultivar, tolerates a range of moist soils, including well-drained clay. Standing 15 to 25 feet high, with a 12- to

  5. #5 peak

    ), formerly classed with lemurs on account of its woolly aspect, monkey-like dexterity, and strong prehensile tail. Its body is about 18 inches long, and its tail 12 inches; and it is

  6. #6 peak

    , the ears are rounded and small, not reaching the muzzle, and it has small feet with a keeled calcar. Body measurements are: total length, 87 to 103 mm; tail, 37 to

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 a 48000$ on the first year of my $48000$ $38$ $200$ $200$ $200$ $38$ $2

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

    everything they<|endoftext|>Variety of electromagnetic radiation with frequencies between a few and a few hundred quadrillion oscillations per second, corresponding to wave-lengths between a few hundred and a few billionths of metres. Known in everyday life as that part of

  2. #2 peak

    regular microLEDs.⏎In addition to the advantages at the chip (LED) level, the nano-membranes (which are thin, at 100nm) can be broken using mechanical force and so the microLEDs can be

  3. #3 peak

    the HiRISE camera onboard Mars Reconnaissance Orbiter -- the first image ever of a spacecraft descending to the surface of another planet. Taken from 750

  4. #4 peak

    EM waves with frequencies below and above that of humans a better definition must be made. We therefore define the visual wavelength range as wavelengths of EM between 300nm and 800nm.

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. L10:2845 suppresses
  2. L18:1546 suppresses
  3. L19:2129 excites
  4. L19:2176 suppresses
  5. L6:1030 suppresses
  6. L16:1154 suppresses
  7. L10:1911 excites
  8. L10:294 suppresses
  9. L20:1411 excites
  10. L15:861 excites
  11. L6:3064 suppresses
  12. L3:1251 weakly excites
  13. L20:876 weakly suppresses
  14. L18:336 weakly excites
  15. L10:1496 weakly excites
  16. L19:1158 weakly suppresses
  17. L20:1510 weakly excites
  18. L20:2838 weakly suppresses
  19. L5:1247 weakly suppresses
  20. L16:1882 weakly suppresses
  21. L20:1128 weakly excites
  22. L9:2048 weakly suppresses
  23. L12:2002 weakly excites
  24. L20:640 weakly excites

Feeds later layers

  1. L24:601 excites
  2. L26:1508 excites
  3. L25:918 excites
  4. L23:1742 excites
  5. L22:557 excites
  6. L27:1404 excites
  7. L24:2403 excites
  8. L27:2106 excites
  9. L23:2862 excites
  10. L23:2296 excites
  11. L24:2592 excites
  12. L22:608 excites
  13. L23:2206 excites
  14. L26:1469 weakly excites
  15. L22:2783 weakly excites
  16. L22:2854 weakly excites
  17. L22:2474 weakly excites
  18. L24:1809 weakly excites
  19. L24:3016 weakly suppresses
  20. L24:1238 weakly excites blank space before a number
  21. L22:1020 weakly excites
  22. L27:1830 weakly excites
  23. L22:1027 weakly excites
  24. L24:1097 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 -10.83 and -10.131 values between -8.74 and -8.051 values between -8.05 and -7.351 values between -7.35 and -6.663 values between -6.66 and -5.962 values between -5.96 and -5.272 values between -4.57 and -3.886 values between -3.88 and -3.185 values between -3.18 and -2.4811 values between -2.48 and -1.7961 values between -1.79 and -1.09904 values between -1.09 and -0.402555 values between -0.40 and 0.30248 values between 0.30 and 0.9947 values between 0.99 and 1.6925 values between 1.69 and 2.3812 values between 2.38 and 3.084 values between 3.08 and 3.772 values between 4.47 and 5.165 values between 5.16 and 5.864 values between 5.86 and 6.562 values between 6.56 and 7.251 values between 7.25 and 7.951 values between 7.95 and 8.641 values between 10.03 and 10.731 values between 12.81 and 13.511 values between 15.59 and 16.291 values between 16.29 and 16.98offnegativepositiveoften

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. size
  2. larger
  3. smaller
  4. -size
  5. taller
  6. サイズ
  7. size
  8. sizes
  9. 大小
  10. -sized

Suppresses

  1. eways
  2. acebook
  3. hesion
  4. ę
  5. pery
  6. adow
  7. redi
  8. 去了
  9. xDE
  10. adox