Neuron Explorer

Qwen3-0.6B

commas before an elaboration

A neuron in layer 20 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 a comma or a word that opens a list or explanation, such as the comma after 'There are many ways to set up multi-factor authentication'. The peaks are modest. Turning it up made the model repeat the word 'and' over and over.

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

    221BC). Unlike most Chinese idioms, 毋忘在莒 strangely has actually two stories of origin, which gave this id

  2. #2 peak

    . A seizure is a short episode of symptoms caused by a burst of abnormal electrical activity in your brain. Different parts of the brain control different functions of your body, so the symptoms that occur during a seizure will depend on where the abnormal burst of

  3. #3 peak

    a copy of the report go to the NOAA news website www.noaanews.noaa.gov⏎– Erich Luening⏎Print this page<|endoftext|>Today is Sunday, Feb. 3, the 34th day of 201

  4. #4 peak

    chede and colleagues are optimistic that it could become an integral part of the semiconductor industry. Lithography is widely used in the industry to make two-dimensional circuit boards, but a convenient method for creating three-dimensional<|endoftext|>Embedding a Racial Equity Perspective

  5. #5 peak

    attach themselves to shell fragments so they have something to latch on to as they begin to grow their own shells. The farmer said any type of shell fragments will do, they don’t have to be oyster shells.⏎Once the young

  6. #6 peak

    mobile app, web app, or website as part of your business offering, multi-factor authentication is a must. There are many ways to set up multi-factor authentication, but some popular, secure methods include:⏎- Email – let the user receive a

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: with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous, with the previous,

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

    ��性病的⏎anachronism /əˈnækrəˌn

  2. #2 peak

    the data.⏎D. Types of Splits⏎There are two types of splits in decision trees: continuous and categorical. Continuous splits are based on a threshold value, such as the split in a regression task where all instances below the threshold value are

  3. #3 peak

    weather’ and ‘climate.’ Weather represents short-term atmospheric conditions in a specific area, while climate is the average of these conditions over extended periods. This distinction is key

  4. #4 peak

    use it to access patient records and share test results. Governments use it for elections, services and disaster alerts. The global financial system—bank transfers, stock trades—increasingly relies on it.⏎Because of abstraction and the distributed nature of cloud

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. L19:39 excites
  2. L2:422 excites
  3. L19:43 excites
  4. L16:36 suppresses
  5. L4:490 suppresses decade digit in years like 1930s
  6. L5:97 suppresses
  7. L18:569 excites
  8. L16:90 suppresses
  9. L9:77 excites commas in citations and name lists
  10. L0:579 suppresses
  11. L0:132 suppresses
  12. L16:177 suppresses zero digit inside dates and codes
  13. L2:202 suppresses "of" after "many" or "which"
  14. L16:7 suppresses
  15. L4:489 excites
  16. L19:118 excites
  17. L6:840 suppresses
  18. L5:1278 suppresses
  19. L17:215 weakly excites
  20. L0:304 weakly excites
  21. L19:1221 weakly suppresses
  22. L18:366 weakly excites comma inside numbers and dates
  23. L17:213 weakly excites
  24. L16:34 weakly excites

Feeds later layers

  1. L24:205 strongly excites
  2. L26:347 strongly excites the 're' in self-reliant and similar
  3. L27:289 strongly excites
  4. L26:65 strongly suppresses
  5. L22:791 strongly excites
  6. L23:379 strongly excites
  7. L27:172 excites
  8. L25:6 suppresses
  9. L21:739 excites 'both X and Y' pairing words
  10. L22:632 excites
  11. L25:1270 suppresses
  12. L27:1902 weakly suppresses
  13. L25:583 weakly excites
  14. L21:2699 weakly suppresses
  15. L26:434 weakly suppresses
  16. L22:286 weakly excites
  17. L27:1593 weakly suppresses
  18. L27:1121 weakly suppresses
  19. L24:921 weakly suppresses
  20. L25:120 weakly suppresses
  21. L22:2060 weakly excites
  22. L26:859 weakly excites
  23. L25:1786 weakly suppresses
  24. L26:1150 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.

1 values between -5.63 and -5.451 values between -4.73 and -4.551 values between -3.12 and -2.942 values between -2.40 and -2.231 values between -1.87 and -1.691 values between -1.33 and -1.156 values between -1.15 and -0.973 values between -0.97 and -0.796 values between -0.79 and -0.616 values between -0.61 and -0.4312 values between -0.43 and -0.2576 values between -0.25 and -0.073579 values between -0.07 and 0.11150 values between 0.11 and 0.2840 values between 0.28 and 0.4610 values between 0.46 and 0.646 values between 0.64 and 0.823 values between 0.82 and 1.001 values between 1.00 and 1.181 values between 1.18 and 1.362 values between 1.36 and 1.54offnegativepositiveoften

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. 和
  2. and
  3. 和
  4. 和个人
  5. và
  6. and
  7. 和地区
  8. 和技术
  9. 性和
  10. и

Suppresses

  1. /or
  2. ари
  3. ape
  4. นา
  5. elif
  6. istica
  7. rade
  8. obi
  9. erta
  10. ᴇ