Neuron Explorer

Qwen3-0.6B

comma after 'for example'

A neuron in layer 15 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 the comma right after phrases like 'For example' or 'For instance', just before the example itself starts. Turning it up or down did not produce a clear, matching change in the written text.

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

    reason)". Often, you can use exceptions to handle this situation.⏎Sometimes you might want to return objects of different types based on arguments to the function. For instance, did the caller ask for the data in this format or that format?⏎Sometimes you

  2. #2 peak

    answers.⏎- Be specific. When setting goals, do some background work. Research your child’s favorite schools together to find out the criteria for admission. For example, if their school of choice requires a 3.5 GPA but your child is

  3. #3 peak

    0. Then we use a symbol like a triangle, circle, line, star or a box to show if that number has that many equal groups. For example, on our 10th day of school we found 5 groups of

  4. #4 peak

    all AVA’s are appellations, but not all appellations are AVA’s. For example, a state or county name may be used as an appellation on a wine label

  5. #5 peak

    of juvenile court involvement among a community's young mental health clients is substantial. For example, a study in one city found that adolescents in contact with the community's mental health

  6. #6 peak

    use the numeral converter for the number of reasons in their lives. For instance, you may want to list up the number of pages in your book using the Roman

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: (1) the problem of the (1) the (1) the (1) ) the (1) ) the (1) the (1) the (1) the (1) the (1) the (1)

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

    |• Land||96.6 sq mi (250.2 km2)|⏎|• Water||0.7 sq mi (1.7 km2)|⏎|

  2. #2 peak

    the same amount of time running at a pace of 5 miles per hour. A study in the May 1996 issue of the "Journal of the American Medical Association" also concluded that running on a treadmill is superior to using the

  3. #3 peak

    processing, and analyzing national security information from around the world. The CIA is an independent agency, overseen by the Director of National Intelligence, and reports to both the President of the United States and the United States Congress.⏎The CIA is an independent

  4. #4 peak

    a vegetarian diet causes weight loss, even in the absence of exercise or calorie counting, according to a new meta-analysis published as an online advance in the Journal of the Academy of Nutrition and Dietetics on Thursday, Jan. 22,

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. L10:29 suppresses
  2. L10:77 weakly suppresses
  3. L9:22 weakly excites
  4. L7:1123 weakly suppresses
  5. L13:414 weakly suppresses
  6. L7:2705 weakly suppresses
  7. L8:2149 weakly excites
  8. L2:1211 weakly suppresses
  9. L8:127 weakly suppresses
  10. L2:2940 weakly suppresses
  11. L14:692 weakly excites word right before a section number
  12. L8:1648 weakly excites
  13. L12:488 weakly suppresses
  14. L14:348 weakly suppresses
  15. L1:990 weakly excites
  16. L1:591 weakly suppresses
  17. L1:1062 weakly suppresses
  18. L7:1346 weakly excites
  19. L8:851 weakly excites
  20. L13:1862 weakly excites
  21. L6:2182 weakly suppresses
  22. L11:1158 weakly suppresses
  23. L14:542 weakly excites
  24. L8:1433 weakly suppresses

Feeds later layers

  1. L19:94 suppresses
  2. L16:71 weakly suppresses
  3. L16:87 weakly suppresses
  4. L16:110 weakly suppresses
  5. L16:1304 weakly suppresses
  6. L17:628 weakly excites
  7. L24:2033 weakly excites
  8. L20:2407 weakly suppresses
  9. L18:432 weakly suppresses
  10. L23:100 weakly excites comma or full stop ending a clause
  11. L16:946 weakly suppresses the word 'late' in time phrases
  12. L16:2461 weakly suppresses
  13. L22:612 weakly excites
  14. L16:287 weakly suppresses
  15. L18:598 weakly suppresses
  16. L20:2122 weakly excites
  17. L22:689 weakly excites comma closing an intro clause
  18. L26:1625 weakly suppresses
  19. L16:193 weakly suppresses 'wide' in 'a wide range of'
  20. L16:306 weakly suppresses
  21. L17:243 weakly suppresses
  22. L27:422 weakly excites
  23. L16:105 weakly suppresses
  24. L17:203 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 -2.40 and -2.191 values between -1.98 and -1.772 values between -1.56 and -1.354 values between -1.35 and -1.1415 values between -1.14 and -0.9421 values between -0.94 and -0.7347 values between -0.73 and -0.5289 values between -0.52 and -0.31404 values between -0.31 and -0.102117 values between -0.10 and 0.11831 values between 0.11 and 0.32181 values between 0.32 and 0.5253 values between 0.52 and 0.7340 values between 0.73 and 0.9428 values between 0.94 and 1.1522 values between 1.15 and 1.3615 values between 1.36 and 1.5710 values between 1.57 and 1.7811 values between 1.78 and 1.984 values between 1.98 and 2.192 values between 2.19 and 2.401 values between 2.40 and 2.611 values between 2.61 and 2.821 values between 3.03 and 3.241 values between 3.44 and 3.651 values between 3.86 and 4.072 values between 4.07 and 4.281 values between 4.28 and 4.491 values between 4.49 and 4.691 values between 5.74 and 5.95offnegativepositiveoften

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. environment
  2. Environment
  3. matic
  4. PCS
  5. abar
  6. pies
  7. 环境下
  8. consider
  9. accept
  10. +z

Suppresses

  1. aurant
  2. pendicular
  3. 斯塔
  4. etadata
  5. Stuart
  6. Vere
  7. steller
  8. ceed
  9. świ
  10. owntown