Neuron Explorer

Qwen3-0.6B

year after a month and day

A neuron in layer 17 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 responds weakly to the year in a written date, such as the 2021 in "August 13, 2021." It also has a weak opposite side on brackets in measurements. Turning it up or down gave text that did not seem to be about dates.

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

    years. This is because the photons have to travel about twice the distance from the Earth to the Moon to<|endoftext|>Posted by Admin on August, 13, 2021⏎Every flame possesses some specific characteristics in general. They produce

  2. #2 peak

    Thompson Chalkboard Paper Kit. Fonts used: school script dashed, Rudiment and Homemade Apple.⏎Posted by Craftyminx on October 12, 2011 at

  3. #3 peak

    3.0604 / 3°3'37"W⏎OS Eastings: 328136⏎OS Northings: 291451⏎OS

  4. #4 peak

    B) hypertonic C) hypotonic D) none of these I think it's B. I'm not too sure though.⏎November 27, 2007<|endoftext|>Hopefully your child will not experience at first hand the impact

  5. #5 peak

    .codeguru.com/algorithms/factorial.shtml I don't happen to think that the code there is easily understandable though...⏎January 10th, 2004 01:32 AM⏎Tim_axe,

  6. #6 peak

    lock (During the 1834 Coorg Campaign)". www.archerfamily.org.uk. Archer Family. Archived from the original on 13 November 2013. Retrieved 13 November 2013.⏎

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: the other way that I could be used as a "solicitor" or as a "solicent" – as the "solicent" in the other way – as a "solicent" by the other 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

    |Dimensions||H: 11 1/16 in. (28.2 cm.), W: 8 7/8 in. (22.5 cm.)|⏎|Location||Alte Pinakothek

  2. #2 peak

    Business District as seen from the top of the Carlton Centre⏎|Municipality||City of Johannesburg|⏎|• Total||1.09 km2 (0.42 sq mi)|⏎|• Density||13,000

  3. #3 peak

    Weber River|⏎|• Total||659 sq mi (1,710 km2)|⏎|• Land||576 sq mi (1,490 km2)|⏎|• Water||83 sq

  4. #4 peak

    ika 316 U.S. 584 (1942) and 319 U.S. 103 (1943) raised important issues regarding free exercise of religion. In the first Jones

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. L10:36 excites
  2. L16:80 suppresses
  3. L15:20 weakly excites
  4. L14:2228 weakly excites
  5. L4:568 weakly excites
  6. L14:835 weakly excites
  7. L1:288 weakly suppresses
  8. L10:804 weakly suppresses
  9. L13:1411 weakly excites
  10. L13:2086 weakly excites
  11. L14:2182 weakly suppresses
  12. L8:91 weakly suppresses
  13. L15:32 weakly suppresses
  14. L15:41 weakly excites
  15. L8:1331 weakly excites
  16. L15:2215 weakly excites
  17. L1:2415 weakly excites
  18. L16:1607 weakly excites
  19. L11:1725 weakly excites
  20. L14:1288 weakly excites
  21. L6:1041 weakly excites
  22. L5:3042 weakly suppresses
  23. L15:653 weakly suppresses
  24. L3:2310 weakly suppresses

Feeds later layers

  1. L19:964 weakly excites
  2. L20:1998 weakly excites
  3. L18:1514 weakly suppresses
  4. L18:2826 weakly excites
  5. L22:2629 weakly excites
  6. L19:2051 weakly excites
  7. L23:370 weakly excites
  8. L19:2160 weakly excites
  9. L19:1458 weakly excites
  10. L23:1317 weakly excites
  11. L19:207 weakly suppresses
  12. L18:2951 weakly excites
  13. L18:2530 weakly suppresses
  14. L21:2338 weakly suppresses
  15. L21:1101 weakly excites
  16. L20:893 weakly excites
  17. L20:2648 weakly suppresses
  18. L27:2735 weakly excites
  19. L22:2421 weakly suppresses
  20. L19:1657 weakly excites
  21. L21:2068 weakly excites
  22. L20:2786 weakly excites
  23. L19:571 weakly excites
  24. L19:1844 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 -1.90 and -1.721 values between -1.72 and -1.541 values between -1.17 and -0.984 values between -0.98 and -0.802 values between -0.80 and -0.6233 values between -0.62 and -0.43156 values between -0.43 and -0.25590 values between -0.25 and -0.061253 values between -0.06 and 0.121043 values between 0.12 and 0.31568 values between 0.31 and 0.49197 values between 0.49 and 0.6749 values between 0.67 and 0.865 values between 0.86 and 1.041 values between 1.23 and 1.411 values between 1.41 and 1.591 values between 1.78 and 1.961 values between 2.33 and 2.511 values between 5.27 and 5.46offnegativepositiveoften

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. ycles
  2. -even
  3. even
  4. telegram
  5. 闳
  6. Even
  7. igr
  8. antino
  9. gue
  10. avings

Suppresses

  1. upon
  2. ")));⏎
  3. atoire
  4. onn
  5. adar
  6. erh
  7. 梅州
  8. oug
  9. RLF
  10. ")));