Neuron Explorer

Qwen3-0.6B

first piece of a split name or term

A neuron in layer 18 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 fires on the first fragment of a longer word or name that the tokenizer splits into pieces, such as "Inter" in "Interchange" or "Ass" in "Association". The examples are mostly institutional or scientific names. Turning it up gave repetitive or off-topic text, so the steering doesn't add much.

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

    Vashishth has found a 92 percent correlation between the cross-links and collagen stiffness. The bones for the study were obtained from the National Disease Research Interchange (NDRI).⏎The details of what happens to the collagen are tricky.

  2. #2 peak

    In early December, the East African Paediatric Association (EAPA) and Kenya Paediatric Assocation (KPA) in collaboration with the American Academy of Pediatrics and Royal College

  3. #3 peak

    waves created by the explosion allowed researchers Elizabeth Silber and Peter Brown of the Meteor Infrasound Group at the Univ. of Western Ontario to determine that the explosion

  4. #4 peak

    the scientific literature on cow-calf separation to provide perspective and, hopefully, guidance. The findings of the committee were presented at the 2019 Dairy Cattle Welfare Symposium in Tampa, Fla., followed by an active discussion among the conference

  5. #5 peak

    is the school's Early Years Policy and information regarding books and apps suitable for our Foundation Stage children as well as a recommended list of Early Years apps. Our Curriculum Overviews can be found on our class pages of this website.⏎Getting Ready for School

  6. #6 peak

    finding, the researchers have clarified the way in which this oscillator adjusts its activity throughout the day. They show that the three proteins studied, the Pseudo-Response Regulators PRR5, PRR7 and PRR9, associate with promoter

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: and the the, and the, and the, and the, and the, and the, and the, and the, and the, and the, and the, and [a, and [a, and [a, and [

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

    crucial to learning, however you need to search for worksheets that show the very same skills in a selection of ways. This not only stops monotony setup in, but additionally strengthens the principles by urging recognizing too. The importance of reading

  2. #2 peak

    . v is detached from the original object it was pointing to and now LOCAL VAR. v and vh point to same object.Since vh is a new obj. instance var vh.i = 10 and HENCE v.i=10. LOCAL

  3. #3 peak

    and merchandising. Artists themselves are looking for new ways to make money, they have to think outside of the box. Since the music became a digital product record sales give the wall to the royalties. The audience is changing step by step,

  4. #4 peak

    creating sort of territorial⏎reference points was probably one of the aims for such locations, considering that sacred mountains strategical position marks the⏎northern limit of the Po valley, which was, at that time

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. L16:1323 suppresses
  2. L16:1335 excites
  3. L16:627 excites
  4. L17:603 excites inside organisation and product names
  5. L15:1410 suppresses
  6. L15:1811 excites
  7. L14:2831 excites
  8. L14:553 suppresses
  9. L15:1701 excites
  10. L15:1022 excites
  11. L15:658 excites
  12. L15:2774 suppresses
  13. L14:817 excites
  14. L16:847 excites
  15. L15:713 suppresses
  16. L15:1580 excites
  17. L15:1653 suppresses
  18. L14:459 excites
  19. L15:2081 suppresses
  20. L13:1492 suppresses
  21. L9:392 suppresses "and" inside organisation names
  22. L17:1940 suppresses
  23. L16:1017 excites
  24. L8:982 suppresses

Feeds later layers

  1. L20:463 suppresses
  2. L20:353 suppresses
  3. L20:398 suppresses
  4. L19:211 excites
  5. L24:78 excites
  6. L19:155 suppresses
  7. L19:1009 weakly excites
  8. L21:243 weakly suppresses
  9. L19:667 weakly suppresses
  10. L20:225 weakly suppresses
  11. L20:494 weakly suppresses
  12. L20:328 weakly suppresses
  13. L19:1860 weakly suppresses
  14. L27:895 weakly excites
  15. L22:1295 weakly suppresses 'as' before 'well' or 'as much'
  16. L21:720 weakly suppresses
  17. L22:1619 weakly excites
  18. L20:1108 weakly suppresses
  19. L22:255 weakly suppresses
  20. L26:962 weakly suppresses
  21. L23:1985 weakly excites
  22. L19:608 weakly suppresses
  23. L25:334 weakly suppresses
  24. L25:62 weakly excites

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 -6.79 and -6.521 values between -5.97 and -5.691 values between -5.69 and -5.423 values between -5.15 and -4.872 values between -4.87 and -4.602 values between -4.60 and -4.332 values between -4.33 and -4.051 values between -4.05 and -3.782 values between -3.50 and -3.233 values between -3.23 and -2.965 values between -2.96 and -2.6811 values between -2.68 and -2.4110 values between -2.41 and -2.1314 values between -2.13 and -1.8612 values between -1.86 and -1.5930 values between -1.59 and -1.3144 values between -1.31 and -1.0458 values between -1.04 and -0.7792 values between -0.77 and -0.49174 values between -0.49 and -0.22717 values between -0.22 and 0.061252 values between 0.06 and 0.331043 values between 0.33 and 0.60328 values between 0.60 and 0.8848 values between 0.88 and 1.1510 values between 1.15 and 1.437 values between 1.43 and 1.709 values between 1.70 and 1.977 values between 1.97 and 2.254 values between 2.25 and 2.521 values between 2.52 and 2.807 values between 2.80 and 3.072 values between 3.07 and 3.341 values between 3.34 and 3.621 values between 3.62 and 3.892 values between 3.89 and 4.16offnegativepositiveoften

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. aken
  2. uncate
  3. FILENAME
  4. nameof
  5. الدولية
  6. ObjectType
  7. igious
  8. ánd
  9. rompt
  10. enerative

Suppresses

  1. (SS
  2. блок
  3. cek
  4. ces
  5. iards
  6. гля
  7. 的孩子
  8. (style
  9. (cert
  10. (car