Neuron Explorer

Qwen3-0.6B

'what' as a free relative

A neuron in layer 3 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 word 'what' when it means 'the thing that', as in 'built what he called' or 'painted what he felt'. Turning it up did not bring out that theme in the model's replies.

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

    six stories and broke his back. He died the next day.⏎Astrodome Capsule Stunt⏎Karel Soucek was a Canadian stuntman who built what he called a cushioned capsule to make a successful plunge over Niagara Falls. He

  2. #2 peak

    gives rise to the question of what is the actual rate of price inflation. They don’t use the CPI ratio put out by the Census Bureau. They use what is called a “chained” price deflator from the GDP report which means

  3. #3 peak

    bre colours behind him. He still differed quite a bit from the Impressionists, though: Impressionists painted what they saw, but Van Gogh painted what he felt.⏎Van Gogh and Georges Seurat became very good friends, and

  4. #4 peak

    students often struggle just to keep up with the pace and the pressure. In order to complete what can be a daunting plate of assignments, students stay up late into the night and

  5. #5 peak

    The idea was that legislators would convene at a safe remove from voters and, thus insulated from the din of narrow interests and widespread but ephemeral passions, do what was in the long-term interest of their constituents and of the nation. Now information

  6. #6 peak

    the Drone, French philosopher Grégoire Chamayou seeks to comprehend how drones have revolutionised warfare and to deconstruct the various narratives at the heart of what has become a conceptual and legal landmine.⏎The drone’s philosophy can be summed

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 no data. The map is a, and the rest. The rest is a and open, but no longer than system, with some combination, and no clear map. The map is a, and the adjacent. The rest of

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

    -emerge. This phenomenon is described in leaflet 8: Direction of cure.⏎Other reasons for giving only one medicine⏎at

  2. #2 peak

    en, 52 to 48 percent. Clark's defeat was attributed to intense anti-Clark campaigning conducted by direct mail, mailgrams, and leaflet distribution during the final weeks of the campaign, attacking Clark for his positions on abortion

  3. #3 peak

    ever had an allergic reaction to a medicine.⏎How to take oxcarbazepine⏎- Before you start the treatment, read the manufacturer's printed information leaflet from inside the pack. It will give you more information about oxcarbazep

  4. #4 peak

    help our planet.⏎You will need:⏎- Scrap paper⏎- Old books/newspapers/leaflets etc⏎- Needle and thread⏎- Something pokey, like a screwdriver

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. L2:111 weakly suppresses
  2. L0:2728 weakly suppresses
  3. L2:561 weakly excites numbers right after 'these' or 'first'
  4. L1:461 weakly suppresses
  5. L1:1155 weakly excites
  6. L1:1280 weakly excites
  7. L1:1469 weakly excites
  8. L1:2336 weakly excites
  9. L2:1030 weakly suppresses
  10. L2:549 weakly suppresses
  11. L2:49 weakly suppresses
  12. L1:1513 weakly suppresses
  13. L2:686 weakly suppresses
  14. L2:2395 weakly suppresses
  15. L1:1390 weakly suppresses
  16. L2:1086 weakly suppresses
  17. L2:722 weakly suppresses
  18. L2:2692 weakly excites
  19. L1:1913 weakly excites
  20. L0:31 weakly suppresses
  21. L1:1801 weakly excites
  22. L2:228 weakly excites
  23. L0:71 weakly excites
  24. L2:1481 weakly excites

Feeds later layers

  1. L4:2586 suppresses
  2. L9:75 weakly excites 'what' or 'how' after see or understand
  3. L4:3019 weakly suppresses
  4. L22:1146 weakly suppresses blank before a time or count
  5. L9:656 weakly excites
  6. L23:2694 weakly suppresses
  7. L6:502 weakly excites
  8. L10:2017 weakly suppresses
  9. L4:2380 weakly suppresses
  10. L15:3024 weakly excites
  11. L10:2893 weakly excites
  12. L17:625 weakly excites
  13. L12:2213 weakly suppresses
  14. L5:2609 weakly suppresses
  15. L4:1279 weakly suppresses
  16. L10:161 weakly excites
  17. L4:3061 weakly suppresses
  18. L16:1765 weakly suppresses
  19. L16:2993 weakly suppresses
  20. L6:1377 weakly suppresses
  21. L5:1251 weakly suppresses
  22. L8:1564 weakly excites
  23. L21:2054 weakly suppresses
  24. L11:421 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.

2 values between -4.77 and -4.341 values between -1.27 and -0.8310 values between -0.83 and -0.392648 values between -0.39 and 0.051222 values between 0.05 and 0.497 values between 0.49 and 0.934 values between 0.93 and 1.372 values between 1.37 and 1.801 values between 1.80 and 2.241 values between 2.68 and 3.123 values between 4.00 and 4.442 values between 5.31 and 5.751 values between 5.75 and 6.191 values between 9.26 and 9.701 values between 10.14 and 10.571 values between 11.01 and 11.451 values between 12.33 and 12.77offnegativepositiveoften

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. ordova
  2. uetype
  3. APON
  4. ..."); ⏎
  5. ibern
  6. itous
  7. 是一款
  8. звуч
  9. 您的孩子
  10. ."); ⏎

Suppresses

  1. 一刻
  2. duced
  3. acias
  4. nada
  5. ziehung
  6. siempre
  7. >[]
  8. DRAW
  9. Draw
  10. identally