Neuron Explorer

Qwen3-0.6B

comma after 'So' or intro phrase

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 a comma that comes right after an opening phrase such as 'So,' at the start of a new sentence, as in 'So, what exactly does...'. Turning it down made the model produce strings of commas and symbols, and turning it up gave loose number lists.

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

    where they have pinpointed the differences between boats and canoes kayaks. But they should’ve been doing was pinpointing the technique differences and explained why. So, I will do exactly what needs to be done.⏎To understand the difference of techniques

  2. #2 peak

    for anyone with mobility issues, these car journeys are necessary; therefore, the analysis accounted for the population of adults who reported mobility issues in the NTS).⏎So, car trips for 1 mile or under add up to 1.8 billion

  3. #3 peak

    , making it difficult to distinguish.⏎Perhaps you have found a less common variety that doesn’t match your preconceived notion of what obsidian looks like. So, what exactly does obsidian look like?⏎Obs

  4. #4 peak

    marker. Battleship U.S.S. Ohio (approx<|endoftext|>“Snow White,” the beloved, first, full-length animated feature film, created by the Walt Disney Studios, is celebrating its 75th anniversary this

  5. #5 peak

    and smaller space.⏎In almost all eukaryotes, histone-based chromatin is the standard, yet in bacteria, there are no histones. So, how and why have histones become so entrenched in our chromatin structures during evolution

  6. #6 peak

    titanosaur have been revealed by researchers who excavated the bones from a site in central Russia. The first titanosaur ever discovered in the country, it shows unique features never before seen in other sauropod specimens.⏎The bones

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:,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

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

    do everything in this world?⏎Close MATLAB with keyboard⏎Find the treasures in MATLAB Central and discover how the community can help you!⏎Choose a web site to get translated content where available and see local events and offers.

  2. #2 peak

    ⏎Polygon Interior Angle Sum⏎Bit to Gigabyte⏎Find the treasures in MATLAB Central and discover how the community can help you!⏎Choose a web site to get translated content where available and

  3. #3 peak

    2-year-old to respect others. Begin with the basics (saying "please" and "thank you"), then try these six other strategies.<|endoftext|>This archived Web page remains online for reference, research or recordkeeping purposes. This page will

  4. #4 peak

    5 direct from Amazon Amazon page|⏎Setup an account with your affiliations in order to access resources via your University's proxy server⏎Configure custom proxy (use this if your affiliation does not provide a proxy)⏎|Through your library|⏎References found in

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. L17:118 excites comma after a clause ending
  2. L20:863 excites
  3. L20:1298 suppresses
  4. L19:726 suppresses comma inside large numbers
  5. L9:77 excites commas in citations and name lists
  6. L18:48 suppresses
  7. L15:123 suppresses
  8. L14:35 suppresses
  9. L16:240 excites comma after 'sometimes' or 'the truth'
  10. L14:51 excites colon that introduces a statement
  11. L19:380 excites
  12. L5:5 suppresses digit after a label like paper or type
  13. L17:62 suppresses
  14. L9:7 excites
  15. L14:348 weakly suppresses
  16. L20:1026 weakly excites
  17. L15:44 weakly suppresses
  18. L17:38 weakly suppresses
  19. L16:1596 weakly suppresses
  20. L0:68 weakly suppresses
  21. L11:18 weakly excites
  22. L6:228 weakly excites comma after a filler word
  23. L20:8 weakly suppresses
  24. L20:128 weakly suppresses

Feeds later layers

  1. L27:63 excites closing parenthesis in data lists
  2. L27:139 excites
  3. L27:433 suppresses
  4. L27:954 suppresses
  5. L26:153 suppresses
  6. L26:193 suppresses
  7. L27:1332 suppresses
  8. L27:1749 suppresses hyphen joining word parts
  9. L22:215 suppresses
  10. L27:77 excites
  11. L23:184 excites
  12. L26:22 excites the 'res' fragment inside words
  13. L27:25 excites
  14. L26:80 excites
  15. L27:838 suppresses
  16. L25:112 suppresses
  17. L27:71 excites
  18. L27:143 excites
  19. L27:327 weakly suppresses
  20. L22:428 weakly suppresses
  21. L27:191 weakly suppresses
  22. L26:94 weakly excites
  23. L22:16 weakly excites comma in dates and birth lines
  24. L26:345 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 -6.61 and -5.296 values between -3.97 and -2.6520 values between -2.65 and -1.331581 values between -1.33 and -0.01895 values between -0.01 and 1.32297 values between 1.32 and 2.64221 values between 2.64 and 3.96212 values between 3.96 and 5.28150 values between 5.28 and 6.6095 values between 6.60 and 7.9261 values between 7.92 and 9.2541 values between 9.25 and 10.5732 values between 10.57 and 11.8912 values between 11.89 and 13.2111 values between 13.21 and 14.536 values between 14.53 and 15.8513 values between 15.85 and 17.1812 values between 17.18 and 18.5010 values between 18.50 and 19.8214 values between 19.82 and 21.1412 values between 21.14 and 22.4613 values between 22.46 and 23.7826 values between 23.78 and 25.1111 values between 25.11 and 26.439 values between 26.43 and 27.7516 values between 27.75 and 29.076 values between 29.07 and 30.3914 values between 30.39 and 31.7113 values between 31.71 and 33.0314 values between 33.03 and 34.3616 values between 34.36 and 35.6816 values between 35.68 and 37.0019 values between 37.00 and 38.324 values between 38.32 and 39.648 values between 39.64 and 40.9610 values between 40.96 and 42.294 values between 42.29 and 43.613 values between 43.61 and 44.934 values between 44.93 and 46.25offnegativepositiveoften

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. whereas
  2. 물을
  3. while
  4. ras
  5. Wikimedia
  6. 물
  7. {}".
  8. puis
  9. 何况
  10. #{@

Suppresses

  1. strument
  2. IRONMENT
  3. atomy
  4. alyzed
  5. ned
  6. adays
  7. alyze
  8. ishly
  9. icient
  10. xiety