Neuron Explorer

Qwen3-0.6B

table pipe characters

A neuron in layer 13 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 vertical bar symbol used to build tables in web pages and wiki text, which marks the start of a new cell. It is negative on quote markers and forum quoting symbols. Turning it up gave repetitive text unrelated to tables.

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

    odhājīva to the Buddha. S.iv.308.<|endoftext|>The New International Encyclopædia/Kinkajou⏎|←Kinic acid||The New International Encyclopædia⏎|Edition of

  2. #2 peak

    , we could:⏎|Vitamin B12 Need||94%||Confirm|⏎|Lup

  3. #3 peak

    No discussions found⏎Start your discussion⏎Share your thoughts or question with experts in your field⏎Start a discussion⏎Found 2 matching solutions for this experiment⏎|Samples can be flash frozen (-80C) before use, this will likely

  4. #4 peak

    Back to 3⏎|5||Leaves mostly alternate or scattered; inflorescence a raceme

  5. #5 peak

    you will need the Adobe Digital Editions app, or BlueFire Reader or Txtr app. These are free, too. For more details see this article.⏎|Size: ||359 KB|⏎|Date published: ||

  6. #6 peak

    which it appears.”⏎For more information, The report card can be found online: www.nationsreportcard.gov.MORE IN Vermont News⏎- Most Popular⏎- Most Emailed⏎- MEDIA GALLERY<|endoftext|>Chim's striking image depicts

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:What is the type of climate change is most likely to be caused byhuman activities is the change that is caused byhuman activities. I think the most likely to be caused byhuman activities is thechange that is caused byhuman activities is the

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

    une Function, 2017, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5707683/#:~:text=It%20is%20a%20potent%

  2. #2 peak

    EST 2010⏎On Wednesday 24 November 2010 22:12:37, Russ Abbott wrote:⏎> Cool. I wasn't aware of that notation. It doesn't quite get to

  3. #3 peak

    did so far:⏎Originally Posted by MrFujin⏎Enter a 5-charcter string: hgout

  4. #4 peak

    2 km. The infrasound data is collected at 20 hz, while the meteorologic data is recorded at 1 hz.⏎Figure 4. Array layout. Blue dots indicate the 7 elements, while the purple

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. L8:18 weakly suppresses
  2. L12:1145 weakly excites
  3. L11:249 weakly excites
  4. L7:3038 weakly excites
  5. L11:769 weakly suppresses
  6. L2:2714 weakly excites
  7. L11:687 weakly excites
  8. L10:157 weakly suppresses end of document and web boilerplate
  9. L4:2772 weakly excites
  10. L3:2073 weakly excites
  11. L1:558 weakly suppresses
  12. L5:1019 weakly excites
  13. L7:2098 weakly suppresses
  14. L0:1420 weakly excites
  15. L7:2831 weakly suppresses
  16. L8:1329 weakly suppresses
  17. L4:1146 weakly excites
  18. L0:2406 weakly excites
  19. L9:215 weakly suppresses
  20. L3:1161 weakly excites first digit after 'more than'
  21. L12:118 weakly excites
  22. L4:1646 weakly suppresses 'fixed' as a finance or cost term
  23. L2:2156 weakly excites
  24. L4:1529 weakly excites

Feeds later layers

  1. L14:140 suppresses
  2. L16:723 excites
  3. L17:132 suppresses
  4. L18:76 suppresses blank spot before a number
  5. L25:30 suppresses
  6. L24:39 suppresses
  7. L27:33 suppresses
  8. L22:14 suppresses
  9. L20:59 suppresses
  10. L14:309 excites
  11. L17:15 weakly excites
  12. L26:21 weakly excites
  13. L19:509 weakly suppresses
  14. L24:77 weakly suppresses opening parenthesis after a name
  15. L16:396 weakly suppresses
  16. L27:117 weakly suppresses
  17. L20:1013 weakly excites
  18. L23:60 weakly suppresses
  19. L23:80 weakly suppresses blank slot before a year or number
  20. L27:97 weakly suppresses
  21. L16:1421 weakly suppresses
  22. L18:211 weakly excites
  23. L14:102 weakly excites
  24. L26:32 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 -0.54 and -0.511 values between -0.51 and -0.482 values between -0.48 and -0.453 values between -0.42 and -0.395 values between -0.39 and -0.374 values between -0.37 and -0.346 values between -0.34 and -0.316 values between -0.31 and -0.2811 values between -0.28 and -0.2521 values between -0.25 and -0.2239 values between -0.22 and -0.2053 values between -0.20 and -0.1767 values between -0.17 and -0.14113 values between -0.14 and -0.11174 values between -0.11 and -0.08170 values between -0.08 and -0.05252 values between -0.05 and -0.03344 values between -0.03 and 0.00330 values between 0.00 and 0.03366 values between 0.03 and 0.06348 values between 0.06 and 0.09332 values between 0.09 and 0.12272 values between 0.12 and 0.15238 values between 0.15 and 0.17184 values between 0.17 and 0.20158 values between 0.20 and 0.23122 values between 0.23 and 0.2688 values between 0.26 and 0.2947 values between 0.29 and 0.3249 values between 0.32 and 0.3435 values between 0.34 and 0.3721 values between 0.37 and 0.4014 values between 0.40 and 0.439 values between 0.43 and 0.467 values between 0.46 and 0.487 values between 0.48 and 0.514 values between 0.51 and 0.544 values between 0.54 and 0.571 values between 0.57 and 0.60offnegativepositiveoften

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. fi
  2. bean
  3. Bean
  4. front
  5. scr
  6. nations
  7. congregation
  8. O
  9. video
  10. rive

Suppresses

  1. //**⏎
  2. 成功
  3. conde
  4. rias
  5. >*/⏎
  6. >`⏎
  7. awah
  8. plevel
  9. ()}}⏎
  10. /")⏎