Neuron Explorer

Qwen3-0.6B

blank slot before a date

A neuron in layer 20 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 day or date, such as "on [blank] 24 April" or "on [blank] 26 September 1844". Turned up, the model's text changed but did not clearly show a date theme.

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

    , Manchester and finishing at the Manchester Museum on Oxford Road.⏎The Manchester Museum will also be hosting a special 'A History of the World Big Saturday' on Saturday 24 April.<|endoftext|>Last month, President Biden issued an executive order directing government

  2. #2 peak

    the most⏎controversial clergyman in the history of the Victorian Presbyterian Church,⏎Strong was born at Dailly, Ayrshire, Scotland on 26 September 1844 the son⏎of Rev. David Strong

  3. #3 peak

    ending the First and Second World wars. A statue of him was erected to commemorate him in London's Parliament Square.⏎Early life and education⏎He was born on 24 May 1870, at the family farm, Boven

  4. #4 peak

    unaccounted for.⏎The mission was extremely hazardous for the members of the Simcol detachments, particularly following Hitler's infamous ‘Commando Order’ issued on 18 October 1942, ordering the annihilation of all men

  5. #5 peak

    ) which⏎received the assent of the President on the 15th April 1962 and was published in the⏎Official Gazette on the 2nd May 1962. The object was equitable distribution of land

  6. #6 peak

    for the boys and they’re loving the 21 Century learning. We hope to see you at our World Travel Expo between 2-3pm on Wednesday 8 June.⏎by Lee White, Junior Boys Year

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: how to place, as a natural, in the future, the necessary and necessary and necessary and needed, and then to place something, as a natural, in the next, as a natural, and so on, as a natural, in

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

    to come.⏎Announcement: We are moving to a new web site. There may be some disruption of service from 3/23/13-3/24/13<|endoftext|>…In 197

  2. #2 peak

    extremes in temperatures, sunlight, tight-fitting clothing, etc.)<|endoftext|>ERIC Number: ED348648⏎Record Type: Non-Journal⏎Publication Date: 1992-Apr⏎Reference Count: N/A⏎Finding

  3. #3 peak

    area.⏎Neil burn Drive was created.⏎Canmore place was erected.<|endoftext|>ERIC Number: ED379961⏎Record Type: RIE⏎Publication Date: 19

  4. #4 peak

    at 361-396-2831. Or if you need to make a claim with Flint Hills you can do so at 1-800-<|endoftext|>This is good book for TPR Excellent book that

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. L16:2810 suppresses
  2. L17:1872 excites
  3. L19:2786 excites
  4. L0:2685 suppresses
  5. L0:1075 weakly excites
  6. L19:2473 weakly excites
  7. L18:476 weakly suppresses
  8. L13:1545 weakly suppresses
  9. L19:2444 weakly excites
  10. L18:560 weakly suppresses
  11. L19:1333 weakly suppresses
  12. L16:899 weakly suppresses
  13. L2:1502 weakly excites
  14. L17:1246 weakly excites
  15. L19:2242 weakly suppresses
  16. L1:179 weakly suppresses
  17. L7:1034 weakly suppresses
  18. L11:223 weakly suppresses
  19. L2:94 weakly suppresses
  20. L15:855 weakly excites
  21. L16:2908 weakly suppresses
  22. L11:1484 weakly excites
  23. L1:2546 weakly excites
  24. L6:605 weakly suppresses

Feeds later layers

  1. L24:600 strongly excites
  2. L23:176 strongly suppresses
  3. L24:401 strongly suppresses 'it' referring back to a noun
  4. L24:1080 excites
  5. L24:120 suppresses
  6. L25:586 suppresses
  7. L23:625 excites
  8. L27:650 excites
  9. L21:1108 suppresses
  10. L25:2055 excites
  11. L23:1210 excites
  12. L24:241 excites
  13. L26:72 excites 'they' or 'we' after where or who
  14. L21:275 suppresses
  15. L21:468 suppresses
  16. L26:2947 excites
  17. L22:448 excites
  18. L23:1071 excites
  19. L26:874 excites
  20. L23:287 excites 'are' after a plural subject
  21. L23:564 excites non-English text, mid-phrase
  22. L24:314 suppresses
  23. L23:311 excites
  24. L27:577 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 -7.32 and -6.821 values between -5.84 and -5.342 values between -5.34 and -4.851 values between -4.35 and -3.863 values between -3.86 and -3.374 values between -2.87 and -2.381 values between -2.38 and -1.885 values between -1.88 and -1.3939 values between -1.39 and -0.90378 values between -0.90 and -0.402746 values between -0.40 and 0.09637 values between 0.09 and 0.5942 values between 0.59 and 1.0812 values between 1.08 and 1.579 values between 1.57 and 2.076 values between 2.07 and 2.563 values between 2.56 and 3.055 values between 3.05 and 3.553 values between 3.55 and 4.041 values between 4.04 and 4.541 values between 4.54 and 5.031 values between 5.52 and 6.021 values between 6.02 and 6.511 values between 7.01 and 7.501 values between 7.50 and 7.992 values between 8.49 and 8.982 values between 11.94 and 12.44offnegativepositiveoften

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. 都
  2. 都有
  3. _are
  4. are
  5. 都很
  6. 都可以
  7. 都没
  8. 都没有
  9. 都不
  10. 都非常

Suppresses

  1. blr
  2. AILS
  3. manages
  4. astes
  5. heiro
  6. recognizes
  7. ости
  8. Handles
  9. needs
  10. speaks