Neuron Explorer

Qwen3-0.6B

leading digit of metric conversions

A neuron in layer 12 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 lights up on the first digit of a number given in metric units inside brackets, like the 2 in '87 hectares (217 acres)' or '83 square miles (210 km2)'. When it was turned up, the written text did not show any clear theme, so the steering tells us little about it.

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

    state of Tabasco which was almost unknown until a group of experts began excavating it three months ago.⏎The archaeological site, which is 87 hectares (217 acres) in size, is located adjacent to the ranching community of

  2. #2 peak

    Norman cathedral. It is the seat of the Bishop of Coutances and Avranches, previously the Bishops of Coutances. Standing 80 metres (295 ft) tall, it dominates the town and can be seen from as

  3. #3 peak

    710 km2), of which 576 square miles (1,490 km2) is land and 83 square miles (210 km2) (13%) is water. It is the second

  4. #4 peak

    town, and there were no organized churches.⏎According to the United States Census Bureau, the town has a total area of 97.3 square miles (252 km2), of which, 96

  5. #5 peak

    is 31.0" (787.4 mm). The month with the most precipitation on average is January with 8.0" (203.2 mm) of precipitation. The month with the least precipitation on

  6. #6 peak

    the distance down the lane from the<|endoftext|>Clare, New York⏎|Clare, New York|⏎|• Total||97.3 sq mi (251.9 km2)|⏎

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:... this action, my action, my action, my action, my action, my action, my,. this action. my. this. a. a. a. f. a. f. a . a. f

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

    more complex sentences, main idea, narrative development, and inferencing.⏎- Martin Luther King Jr Holiday 5 Resources for AAC Learners⏎- Visual Supports, Core Words, and Martin Luther King⏎- Core Values: Suggestions for Language Building with

  2. #2 peak

    front grille. …⏎- Bumper End. Bumper ends are an important addition to the bumper itself. …⏎- Valance Panel. …⏎- Inner Fender. …⏎- Fender. …⏎- Cowl Panel.

  3. #3 peak

    12 nations: Fiji, Papua New Guinea, Vanuatu, Tonga, Samoa, Solomon Islands, Kiribati, Tuvalu, Nauru, Palau, Federated States

  4. #4 peak

    .S. portion of the Pacific Flyway alone.⏎Pacific Flyway States⏎Alaska | Arizona | California | Idaho | Nevada | Oregon | Utah | Washington | Colorado | Montana | New Mexico | Wyoming<|endoftext|>Please boost your Plan to download papers⏎

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. L11:1633 strongly excites
  2. L11:689 excites
  3. L11:1659 excites
  4. L10:1271 suppresses
  5. L10:951 suppresses
  6. L10:715 excites
  7. L10:514 suppresses
  8. L10:1417 excites
  9. L11:1220 excites
  10. L11:5 excites
  11. L5:603 weakly suppresses
  12. L10:110 weakly suppresses
  13. L5:463 weakly excites
  14. L11:902 weakly excites
  15. L4:2176 weakly suppresses
  16. L5:374 weakly suppresses
  17. L11:4 weakly excites
  18. L1:2095 weakly excites
  19. L0:2093 weakly excites
  20. L1:1802 weakly excites
  21. L7:3024 weakly suppresses
  22. L7:282 weakly excites
  23. L10:761 weakly excites
  24. L8:462 weakly suppresses

Feeds later layers

  1. L13:1771 strongly suppresses
  2. L13:15 strongly excites
  3. L13:2215 strongly excites
  4. L14:2141 strongly suppresses
  5. L13:1113 strongly excites
  6. L13:1139 suppresses
  7. L16:1868 excites
  8. L13:2412 suppresses
  9. L13:2606 excites
  10. L13:2948 suppresses
  11. L13:2036 excites
  12. L17:670 excites
  13. L14:1752 excites
  14. L13:1757 excites
  15. L16:900 suppresses
  16. L14:1247 suppresses
  17. L13:2769 excites
  18. L13:674 weakly suppresses
  19. L22:2341 weakly suppresses
  20. L13:1104 weakly excites percentages and shares in statistics
  21. L14:2118 weakly suppresses
  22. L15:1682 weakly suppresses
  23. L14:2654 weakly excites
  24. L14:324 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.

5 values between -0.73 and -0.5952 values between -0.59 and -0.44205 values between -0.44 and -0.30786 values between -0.30 and -0.151691 values between -0.15 and -0.01948 values between -0.01 and 0.13133 values between 0.13 and 0.2837 values between 0.28 and 0.4219 values between 0.42 and 0.573 values between 0.57 and 0.719 values between 0.71 and 0.853 values between 0.85 and 1.003 values between 1.00 and 1.142 values between 1.14 and 1.294 values between 1.29 and 1.431 values between 1.43 and 1.572 values between 1.57 and 1.722 values between 2.01 and 2.151 values between 2.15 and 2.291 values between 2.87 and 3.021 values between 4.89 and 5.03offnegativepositiveoften

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. 窝
  2. head
  3. ibel
  4. onto
  5. oids
  6. head
  7. arcy
  8. (head
  9. amento
  10. frage

Suppresses

  1. ioned
  2. .LatLng
  3. Fuse
  4. ),$
  5. prite
  6. nominee
  7. 使用者或
  8. '],['
  9. ripping
  10. renom