Neuron Explorer

Qwen3-0.6B

"two" and the link between two things

A neuron in layer 16 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 lights up on "two," "both," and the words right after them when the text compares or relates two things, like "the two words" or "the two regions." It goes strongly negative on words about bridging or connecting things, such as "bridge" or "linking." Turning it up did not produce a clear 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 an email to the maker with the neuron already filled in).

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

    . A group of conditions resulting from damage to the central nervous system. Do not assume that a person with cerebral palsy also has mental retardation; the two do not necessarily or typically occur together.⏎Congenital Disability. A physical impairment existing

  2. #2 peak

    the category of symptoms referred to as dementia, and can have a significant impact on cognitive abilities and physical performance of daily activities. While symptoms of the two conditions can often overlap,

  3. #3 peak

    same slope, m = 2, and we know that,⏎Lines with the same slope are parallel.⏎Hence, the given lines are parallel.⏎Let's consider another set

  4. #4 peak

    thalmologist? Ophthalmologists treat eye and visual problems. Neurologists deal with problems with the brain. A neuro-ophthalmologist straddles these two fields, handling brain issues that affect vision.⏎Does neuro optometry work?⏎A

  5. #5 peak

    of how we respond. In consumer research, it's been hard enough to demonstrate that emotions are different than feelings. The two words are often used interchangeably and leading to much confusion in discussions. So let

  6. #6 peak

    in the Indo-Pacific region.⏎- Although they possess a smaller number of species the corals of the Atlantic are still unique, with few common species between the two regions .⏎

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, or, one, or, or, or, or, or, or, or, or, or... one...? one... or...? or... or...? or... or... or...... or... or...

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

    ambos is afraid that if the last thin strip of ice connecting the cracking mass breaks off, about half the total ice shelf area could be lost

  2. #2 peak

    books/⏎They also say…. ‘ Students with special learning needs may offer the most concrete evidence of the benefits of tablets in the classroom. Advocates, desperate to bridge the educational gaps in children with learning issues

  3. #3 peak

    first public message by telegraph, from Washington to Baltimore.⏎In 1883 the Brooklyn Bridge, linking Manhattan and Brooklyn, was opened.⏎In 1935 the major leagues

  4. #4 peak

    lessness and bravery," during which he had distinguished himself by breaking through the Spanish line which separated Pinar del Río Province from Havana Province. He fought alongside General Calix

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. L9:149 strongly suppresses
  2. L10:883 strongly excites
  3. L10:788 suppresses
  4. L15:1126 excites
  5. L15:1844 suppresses
  6. L11:2938 excites
  7. L4:3045 suppresses
  8. L5:2950 excites
  9. L6:1284 suppresses
  10. L9:2396 excites
  11. L15:2433 suppresses
  12. L13:2813 suppresses
  13. L5:1850 suppresses
  14. L14:914 excites
  15. L14:2535 suppresses
  16. L10:3050 excites
  17. L15:3016 excites
  18. L10:712 suppresses
  19. L10:1152 suppresses
  20. L14:2783 suppresses the word 'same' in 'the same'
  21. L6:281 excites
  22. L12:1244 excites
  23. L7:699 suppresses
  24. L8:1540 excites

Feeds later layers

  1. L18:2373 strongly suppresses
  2. L17:2331 strongly suppresses
  3. L18:1738 strongly suppresses
  4. L22:1760 strongly suppresses
  5. L17:605 suppresses
  6. L17:2769 suppresses
  7. L24:741 suppresses
  8. L20:1267 suppresses
  9. L27:1198 suppresses
  10. L17:1901 suppresses
  11. L17:3040 suppresses
  12. L17:1812 suppresses
  13. L17:2925 suppresses
  14. L23:3045 suppresses
  15. L20:2209 suppresses
  16. L21:2200 suppresses
  17. L20:1842 suppresses
  18. L22:1221 suppresses
  19. L19:2131 suppresses
  20. L17:1834 suppresses
  21. L18:3057 suppresses
  22. L23:2303 suppresses
  23. L20:1156 suppresses
  24. L23:1137 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.

1 values between -11.20 and -10.822 values between -7.37 and -6.981 values between -6.60 and -6.221 values between -6.22 and -5.832 values between -5.83 and -5.451 values between -5.45 and -5.073 values between -5.07 and -4.681 values between -4.68 and -4.305 values between -4.30 and -3.923 values between -3.92 and -3.532 values between -3.53 and -3.155 values between -3.15 and -2.765 values between -2.76 and -2.389 values between -2.38 and -2.0012 values between -2.00 and -1.6123 values between -1.61 and -1.2334 values between -1.23 and -0.8594 values between -0.85 and -0.46396 values between -0.46 and -0.082675 values between -0.08 and 0.31607 values between 0.31 and 0.6919 values between 0.69 and 1.071 values between 1.07 and 1.462 values between 1.46 and 1.841 values between 1.84 and 2.221 values between 3.37 and 3.762 values between 3.76 and 4.14offnegativepositiveoften

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. OwnProperty
  2. ique
  3. ustum
  4. otation
  5. unavailable
  6. icated
  7. .optional
  8. inan
  9. ible
  10. gué

Suppresses

  1. between
  2. between
  3. Between
  4. Between
  5. BETWEEN
  6. 之间的
  7. 之间
  8. giữa
  9. между
  10. zwischen