Neuron Explorer

Qwen3-0.6B

street and urban place names

A neuron in layer 6 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 responds to words for streets, avenues, districts and neighbourhoods, such as 'Avenues', 'Street' and 'shopping district'. When it was turned up, the model's text drifted toward city streets, street vendors and neighbourhoods in two of the four prompts.

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

    Ten store, and the destruction and looting began to spread east and west on 125th Street, targeting white-owned businesses between Fifth and Eighth Avenues. Some stores posted

  2. #2 peak

    more for products purchased in a shopping district with trees. Those same shoppers also say they are willing to stay longer and rate

  3. #3 peak

    , 77 Jackson Street, Saint Paul, Minnesota (Razed)⏎|Edit with form|⏎|Address:||77 Jackson Street|⏎|Neighborhood/s:||Downtown, Saint Paul, Minnesota|⏎|

  4. #4 peak

    825 by free working class African-Americans in uptown Manhattan. The area from W82nd to 88th Streets between Seventh and Eighth Avenues (click here for a larger version of the map) was

  5. #5 peak

    area. In 1917, an administration building was built on the block east of the temple and in 1972, the twenty-eight story LDS Church Office Building, which was, for many years, the tallest building in

  6. #6 peak

    1935, an employee at the Kress Five and Ten store at 256 W. 125th Street (just across the street from the Apollo Theater) caught 16-year-old black Puerto Rican Lino

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 first time the moon has been able to be seen. I think that's something about the problem. I think it's difficult to say. I think the best to go through, I can't help with, but the moon's been

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

    Vintage, 1996.⏎Silver, Larry. “Nature and Nature’s God: Landscape and Cosmos of Albrecht Altdorfer.” Art Bulletin 81.2 (June 1999): 19

  2. #2 peak

    1,000 to 22,000 people.⏎If You Go⏎Doors to the Flagler College Auditorium — across from City Hall on Granada Street — open at 6 p.m. for lobby displays,

  3. #3 peak

    decades later, Houck is sharing that knowledge with visitors to Atlanta, taking them to places where prominent African

  4. #4 peak

    that if you take the average intelligence for all men and all women<|endoftext|>A metal pole anchored down with cables juts out of the sand next to a public walkway at First Landing State Park. Four small rods stick out horizontally from the sides.⏎

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. L5:787 strongly excites
  2. L5:1422 excites
  3. L3:2077 weakly suppresses
  4. L1:2739 weakly suppresses
  5. L4:182 weakly suppresses
  6. L5:434 weakly excites
  7. L5:66 weakly excites
  8. L3:2073 weakly suppresses
  9. L3:1984 weakly excites
  10. L2:1657 weakly excites
  11. L2:2106 weakly suppresses
  12. L5:102 weakly excites
  13. L5:94 weakly suppresses
  14. L5:28 weakly excites
  15. L2:1477 weakly excites
  16. L2:49 weakly suppresses
  17. L1:2811 weakly excites
  18. L5:1354 weakly suppresses
  19. L4:150 weakly suppresses
  20. L5:169 weakly excites
  21. L4:1462 weakly excites
  22. L5:1401 weakly excites
  23. L4:1158 weakly excites
  24. L2:2692 weakly excites

Feeds later layers

  1. L10:1563 excites
  2. L9:2988 excites
  3. L8:2481 suppresses
  4. L7:410 excites
  5. L20:2861 weakly excites
  6. L10:2943 weakly excites
  7. L10:2255 weakly excites
  8. L7:1465 weakly excites
  9. L10:2909 weakly excites
  10. L16:2543 weakly excites
  11. L7:3023 weakly excites
  12. L8:2076 weakly suppresses 'make' before decisions or predictions
  13. L9:2586 weakly suppresses
  14. L7:1890 weakly suppresses
  15. L8:2132 weakly suppresses
  16. L8:2632 weakly suppresses
  17. L26:2247 weakly excites
  18. L7:2321 weakly excites
  19. L10:521 weakly excites
  20. L10:769 weakly suppresses
  21. L7:1945 weakly suppresses
  22. L7:657 weakly excites
  23. L7:1203 weakly excites
  24. L23:555 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 -1.25 and -1.141 values between -1.03 and -0.927 values between -0.81 and -0.6910 values between -0.69 and -0.5837 values between -0.58 and -0.47110 values between -0.47 and -0.36262 values between -0.36 and -0.25654 values between -0.25 and -0.141146 values between -0.14 and -0.021131 values between -0.02 and 0.09390 values between 0.09 and 0.20100 values between 0.20 and 0.3128 values between 0.31 and 0.4311 values between 0.43 and 0.541 values between 0.54 and 0.653 values between 0.65 and 0.764 values between 0.76 and 0.871 values between 0.87 and 0.984 values between 0.98 and 1.101 values between 1.21 and 1.321 values between 1.32 and 1.431 values between 1.43 and 1.541 values between 2.21 and 2.331 values between 2.33 and 2.441 values between 2.89 and 3.001 values between 3.11 and 3.22offnegativepositiveoften

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. street
  2. Snackbar
  3. Grove
  4. floors
  5. 位
  6. InSection
  7. 楼
  8. 街区
  9. getView
  10. -comm

Suppresses

  1. olidays
  2. öh
  3. omens
  4. TestData
  5. othy
  6. akening
  7. jdbc
  8. persistent
  9. .tbl
  10. Turks