Neuron Explorer

Qwen3-0.6B

quantity words like 'two' and 'several'

A neuron in layer 21 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 fires on number-like words such as 'two' or 'several' that come right after a verb or preposition and set up a count of things, as in 'separate two organizations'. Turning it down made the model repeat small quantity phrases like 'a few' and 'a bit', so the neuron seems tied to how much or how many.

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

    a drug.⏎Over the years, occasional compounds of this sort would show up in the literature, usually as part of an effort to, for example, bind to two receptors at once when they were thought to be near to

  2. #2 peak

    $1,500.⏎The same idea holds true for businesses. The difference between gross income and net income can be enormous, and the numbers can tell two wildly different stories as to the health of the business.⏎Too many small businesses focus

  3. #3 peak

    other defense proposals, threatens to further separate two organizations already experiencing strains on their relationship.⏎The EU’s shifting perspective toward defense comes

  4. #4 peak

    known as a chromosomal microarray analysis, which can detect large scale deletions and duplications of the genome.⏎Learn more about how genetic testing positively impacted two SPARK participants: Hope and Alex.⏎While these techniques were state-of-the-art

  5. #5 peak

    there are an abundance of activities children<|endoftext|>January 2022 Newsletter⏎Mosquitofish are small guppy-like fish that can eat several hundred mosquito larvae each day. The District uses mosquitofish as part of

  6. #6 peak

    that parents are concerned about.⏎4. In some urban school districts, there are very low levels of parent engagement, due to barriers that can include parents who work two jobs, kids who have foster parents or are being raised by grandparents, single parent

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: A two- a four a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a

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

    with you. If we have whetted your appetite for interesting food facts, here are some previous posts you may find interesting:⏎- Furikake: Everything you need

  2. #2 peak

    berries seem to affect the genes that are related to fat-burning and storage.⏎So what does this mean for you? Toss a handful on your yogurt, throw a few in a salad or add some to your morning cereal. Even though this

  3. #3 peak

    , but we’re still trying to imitate the real world.⏎Animation on the Web: a short history⏎Before we get onto more practical things, let’s take a look at how animation on the Internet came to its present (and very cool

  4. #4 peak

    mination érrimo-aSuperlative adjectives are formed with the suffix érrimo-a, a Latin termination occurring while it is completely correct obsolete. Here are some examples: pobre-pauperrimo, celebre-celeb

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. L19:1501 suppresses
  2. L19:1935 excites
  3. L18:2804 weakly suppresses
  4. L16:1333 weakly excites 'two' after 'divided into' or 'there are'
  5. L18:1265 weakly suppresses
  6. L20:36 weakly suppresses
  7. L19:1379 weakly excites
  8. L18:801 weakly excites
  9. L18:2021 weakly excites
  10. L18:2788 weakly excites
  11. L18:1942 weakly excites
  12. L18:857 weakly suppresses
  13. L20:2827 weakly suppresses
  14. L12:2053 weakly excites
  15. L12:2424 weakly excites
  16. L18:692 weakly excites
  17. L19:687 weakly suppresses
  18. L17:131 weakly suppresses
  19. L18:1076 weakly excites
  20. L7:1706 weakly suppresses
  21. L19:1748 weakly suppresses
  22. L16:1860 weakly suppresses
  23. L6:2583 weakly suppresses
  24. L10:1990 weakly excites

Feeds later layers

  1. L24:823 suppresses
  2. L27:797 suppresses
  3. L25:45 excites
  4. L25:1586 suppresses
  5. L24:2538 suppresses
  6. L27:663 suppresses
  7. L27:1124 suppresses
  8. L22:103 excites
  9. L22:926 suppresses
  10. L25:433 excites
  11. L26:213 excites
  12. L22:171 suppresses
  13. L25:2547 suppresses
  14. L23:2883 weakly suppresses
  15. L27:709 weakly suppresses
  16. L22:360 weakly suppresses
  17. L24:488 weakly excites
  18. L27:173 weakly excites
  19. L22:825 weakly suppresses 'for' before a span of time
  20. L27:231 weakly excites
  21. L26:484 weakly suppresses
  22. L26:411 weakly excites
  23. L25:2527 weakly suppresses
  24. L24:826 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.

3 values between -5.95 and -5.441 values between -5.44 and -4.932 values between -4.42 and -3.914 values between -3.91 and -3.406 values between -3.40 and -2.908 values between -2.90 and -2.3921 values between -2.39 and -1.8828 values between -1.88 and -1.3773 values between -1.37 and -0.86460 values between -0.86 and -0.352233 values between -0.35 and 0.16524 values between 0.16 and 0.67163 values between 0.67 and 1.1890 values between 1.18 and 1.6967 values between 1.69 and 2.2050 values between 2.20 and 2.7138 values between 2.71 and 3.2120 values between 3.21 and 3.7234 values between 3.72 and 4.2314 values between 4.23 and 4.7412 values between 4.74 and 5.2510 values between 5.25 and 5.766 values between 5.76 and 6.278 values between 6.27 and 6.786 values between 6.78 and 7.296 values between 7.29 and 7.807 values between 7.80 and 8.311 values between 8.31 and 8.811 values between 8.81 and 9.321 values between 9.32 and 9.833 values between 9.83 and 10.341 values between 10.34 and 10.854 values between 10.85 and 11.361 values between 11.87 and 12.381 values between 12.38 and 12.891 values between 13.91 and 14.41offnegativepositiveoften

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. lettes
  2. tings
  3. �재
  4. UDGE
  5. pequeño
  6. ovich
  7. arently
  8. lore
  9. bordel
  10. rowsers

Suppresses

  1. some
  2. a
  3. 一些
  4. Some
  5. some
  6. Some
  7. 几个
  8. 一个
  9. 几张
  10. an