Neuron Explorer

Qwen3-0.6B

'and' or 'or' inside noun lists

A neuron in layer 2 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 'and' or 'or' when it joins a list of things, like 'surveillance cameras, access controls and secure data centres'. When it was turned up, the model's writing fell into loops of 'the' and 'and' and stopped making sense.

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

    such as spiders, heights, flying or crowded places, even when there is no. For example, you may know a

  2. #2 peak

    nature, processes and methods of science…“⏎This needs to be embedded within our teaching of the

  3. #3 peak

    , including surveillance cameras, access controls and secure data centres. These measures safeguard against physical threats like theft, fire, and

  4. #4 peak

    , health and wellbeing considerations have matured into technical building components like indoor air quality, temperature and humidity. These improvements come with their own financial upsides. MIT’s Real Estate

  5. #5 peak

    of plants and animals). Other plants, animals and/or insects may use them for habitat or food. When natural areas are lost or

  6. #6 peak

    Avocado plays an important role in women’s health. It helps in estrogen production and reproductive hormonal balance.⏎- Nutrients like folic acid, potassium, fiber and other minerals make it super food for good health.⏎Turmeric or golden spice is

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 a student who used to be a bit of a long time ago. Now that is, I am a student who used to be a bit of a long time ago. Now that is, I am a student who used to

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

    points help you find the general shape of your graph so that you have a nice place to start.⏎To sketch the full parent graph of cotangent, follow these steps:⏎Find the vertical asymptotes so you can find the domain.⏎Because cotangent

  2. #2 peak

    by visiting ruins of sugar plantation sites and rock carvings.⏎Saint Croix is the largest of three main U.S. Virgin islands. This island is known for its historical sites, rainforest, beaches and hills. There are plantation villages of

  3. #3 peak

    not every native plant will thrive in every location. It's important to know your site conditions and the requirements of the native plants when making a selection

  4. #4 peak

    will inform you on how to perform the proper set-up of your AutoCAD settings. This set-up is crucial in order for you to produce a drawing that makes sense and is visually attractive.

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. L1:1100 weakly suppresses
  2. L0:1190 weakly excites
  3. L1:686 weakly suppresses
  4. L0:2808 weakly suppresses
  5. L0:21 weakly suppresses last digit of years like 1970
  6. L1:304 weakly excites
  7. L0:839 weakly excites
  8. L1:108 weakly suppresses
  9. L0:2529 weakly excites
  10. L1:2992 weakly excites
  11. L1:494 weakly suppresses
  12. L0:988 weakly excites
  13. L0:861 weakly suppresses
  14. L1:2100 weakly excites
  15. L1:2119 weakly excites
  16. L0:2535 weakly excites
  17. L0:2834 weakly suppresses
  18. L0:126 weakly excites
  19. L0:1053 weakly suppresses
  20. L1:1355 weakly suppresses
  21. L1:149 weakly excites
  22. L1:387 weakly excites "to" after "it's important"
  23. L0:1613 weakly suppresses
  24. L0:1626 weakly suppresses

Feeds later layers

  1. L3:177 suppresses
  2. L4:121 suppresses
  3. L6:930 weakly suppresses middle digits inside numbers
  4. L8:347 weakly suppresses
  5. L9:1073 weakly suppresses
  6. L8:516 weakly suppresses
  7. L6:2438 weakly suppresses
  8. L11:627 weakly excites
  9. L15:2979 weakly suppresses
  10. L8:168 weakly suppresses
  11. L17:1712 weakly suppresses
  12. L13:1147 weakly excites
  13. L7:292 weakly excites
  14. L3:747 weakly suppresses
  15. L11:3037 weakly suppresses
  16. L6:237 weakly suppresses
  17. L21:641 weakly excites
  18. L8:870 weakly suppresses "Disease" in Centers for Disease Control
  19. L5:22 weakly suppresses
  20. L8:976 weakly suppresses
  21. L13:2145 weakly excites
  22. L16:1484 weakly excites
  23. L15:3012 weakly excites
  24. L16:814 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.

1 values between -1.07 and -1.011 values between -0.90 and -0.852 values between -0.79 and -0.741 values between -0.63 and -0.571 values between -0.57 and -0.524 values between -0.52 and -0.462 values between -0.46 and -0.414 values between -0.41 and -0.3513 values between -0.35 and -0.3038 values between -0.30 and -0.24102 values between -0.24 and -0.19238 values between -0.19 and -0.13403 values between -0.13 and -0.08739 values between -0.08 and -0.031313 values between -0.03 and 0.03572 values between 0.03 and 0.09261 values between 0.09 and 0.14110 values between 0.14 and 0.1951 values between 0.19 and 0.2522 values between 0.25 and 0.3012 values between 0.30 and 0.368 values between 0.36 and 0.413 values between 0.41 and 0.471 values between 0.47 and 0.522 values between 0.52 and 0.581 values between 0.58 and 0.632 values between 0.80 and 0.851 values between 1.07 and 1.13offnegativepositiveoften

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. 'l
  2. '
  3. status
  4. ië
  5. '^
  6. ijkl
  7. .setStatus
  8. '|
  9. lul
  10. "},"

Suppresses

  1. -mouth
  2. trot
  3. 跑了
  4. -spacing
  5. -transitional
  6. (mapped
  7. transparent
  8. scape
  9. Convertible
  10. )*/⏎