Neuron Explorer

Qwen3-0.6B

'World War II' and the 'II' after 'War'

A neuron in layer 1 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 "II" in "World War II", and only weakly on related war terms. Turning it up or down did not give a clear shift in the written text.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    she finds herself displaced to the 1940s Japanese-American internment camp that her late grandmother, Ernestina, was forcibly relocated to during World War II.⏎These displacements keep occurring until

  2. #2 peak

    of the war dwarfed anything seen before in the islands. In recognition of the close ties forged between the United States and the people of Solomon Islands during World War II

  3. #3 peak

    in its water Apra Harbor grave. The Tokai Maru was a Japanese passenger-cargo ship before World War II, but after the outbreak of World War II it became a transport for the Japanese Imperial Navy. In early 194

  4. #4 peak

    75th anniversary of the Battle of Guadalcanal in Solomon Islands, which was a critical turning-point for Allied forces in the Pacific theater of World War II.⏎The two countries established diplomatic relations following Solomon Islands' independence in 19

  5. #5 peak

    to wane following the arrival of the automobile, the Interstate Commerce Commission, World War I, and the Great Depression. A brief growth spurt during World War II was short-lived, and it was followed by the collapse of several major railroads

  6. #6 peak

    It was in a roundabout way that six Challengers ordered by the UP ended up hauling coal for the Clinchfield Railroad. In the midst of World War II, the War Production Board refused the Rio Grande's request to order new articulateds

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 am not a Christian, I do not think that the United States is a nation-state of the world. I am not a member of the Church of the People, I am not a member of the Church of the People. I am

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

    admitted by the workhouse due to their social standing. However, the Society still needed funds to support the organisation and the small number of inhabitants who lived within the Poorhouse. By commissioning Lanyon to create a survey, they were possibly

  2. #2 peak

    context, references and techniques. The original⏎text appears after the notes. For "deep explorers" of Anglo-Saxon poetry, there⏎is an advanced Language section written by Bob Zisk, followed by a list of⏎sources for further

  3. #3 peak

    to all use restrictions contained in the Terms and Conditions of Use of the BioOne website.⏎Vol. 65 •

  4. #4 peak

    some states more economically mobile than others? This study doesn’t seek to specifically address the causes of relative mobility between states, but there are some important factors that the Economic Mobility Project has found are essential to promoting economic mobility overall, and these include things

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. L0:212 strongly suppresses
  2. L0:115 excites
  3. L0:138 excites
  4. L0:300 suppresses
  5. L0:130 suppresses unclear: word right after a modifier
  6. L0:51 excites
  7. L0:1224 weakly suppresses
  8. L0:1301 weakly suppresses
  9. L0:337 weakly excites
  10. L0:178 weakly suppresses
  11. L0:470 weakly suppresses
  12. L0:36 weakly suppresses
  13. L0:486 weakly suppresses
  14. L0:2229 weakly excites
  15. L0:447 weakly suppresses
  16. L0:0 weakly excites
  17. L0:385 weakly excites
  18. L0:1855 weakly suppresses
  19. L0:301 weakly excites
  20. L0:179 weakly excites
  21. L0:535 weakly suppresses
  22. L0:222 weakly suppresses
  23. L0:350 weakly suppresses
  24. L0:369 weakly excites

Feeds later layers

  1. L4:0 strongly suppresses
  2. L2:882 suppresses
  3. L6:2114 suppresses
  4. L9:1190 suppresses
  5. L4:1963 suppresses
  6. L3:2055 suppresses
  7. L2:1342 suppresses
  8. L20:474 suppresses
  9. L7:29 suppresses
  10. L11:1194 suppresses
  11. L4:176 excites
  12. L2:932 suppresses
  13. L5:1935 suppresses
  14. L16:482 suppresses
  15. L17:1993 suppresses
  16. L2:104 suppresses
  17. L16:158 suppresses
  18. L10:1432 suppresses
  19. L3:1057 suppresses
  20. L2:1597 suppresses
  21. L11:1822 excites
  22. L13:1733 excites
  23. L8:219 suppresses
  24. L3:169 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 -2.54 and -2.411 values between -1.90 and -1.771 values between -1.77 and -1.641 values between -1.64 and -1.523 values between -1.52 and -1.392 values between -1.39 and -1.262 values between -1.26 and -1.142 values between -1.14 and -1.014 values between -1.01 and -0.888 values between -0.88 and -0.7510 values between -0.75 and -0.6316 values between -0.63 and -0.5020 values between -0.50 and -0.3728 values between -0.37 and -0.24115 values between -0.24 and -0.121885 values between -0.12 and 0.011538 values between 0.01 and 0.14205 values between 0.14 and 0.2736 values between 0.27 and 0.3916 values between 0.39 and 0.527 values between 0.52 and 0.652 values between 0.65 and 0.782 values between 0.78 and 0.901 values between 0.90 and 1.031 values between 2.43 and 2.56offnegativepositiveoften

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. unj
  2. itations
  3. واجه
  4. 】,
  5. ировки
  6. ahi
  7. amedi
  8. tog
  9. ере
  10. 搅

Suppresses

  1. "--
  2. 瓒
  3. право
  4. Guidance
  5. icrosoft
  6. Committee
  7. orem
  8. beros
  9. scription
  10. Caucas