Neuron Explorer

Qwen3-0.6B

verb right after 'will' or 'we'll'

A neuron in layer 19 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the action word that comes straight after a 'will', 'we'll' or 'and' in blog-style sentences, like 'we'll take a closer look' or 'will equip you'. It is quiet, and sometimes strongly negative, on verbs that follow 'to', such as 'try searching' or 'prefer to purchase'. Turning it up gave unrelated text, so the steering does not add much.

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

    all of this? Next time a sense of glum hits, roll up those sleeves, find an inspiring recipe from your favourite wellness guru and hit the kitchen to bake up a serious storm. After all, is there anything better

  2. #2 peak

    Tips to Crack HSC Exams.⏎We had also shared Important Questions Bank for HSC Examination 2016 and students have really appreciated it and showered us with love last year.⏎Today, we are posting the Economics– Important Questions

  3. #3 peak

    but they didn’t realize those were the earliest movies.⏎3. The invention of Kinetoscopes⏎Edison managed his plans to create a device that would do “for the Eye what the phonograph does for the Ear”. He invited the

  4. #4 peak

    hold the promise of transforming industries, revolutionizing the way we live and work, and addressing some of society’s most pressing challenges. In this blog, we’ll take a deep dive

  5. #5 peak

    Tsoy and Wilbert Swinkels has developed. Inspired by other DIY cube solvers comprised of LEGO and Fischertechnik, the Makers brought their creation

  6. #6 peak

    components such as assets, liabilities, and equity. By delving into the importance of the bilan comptable for financial analysis and decision-making, this article will equip you with the knowledge needed to make informed financial decisions.

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: the ability to in between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between between

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

    therapeutic dose is 15 mg/kg, 5 times daily for 5 days.⏎Best of luck---⏎Search for questions⏎Still looking for answers? Try searching for what you seek or ask your own question.⏎Posted 31 Aug

  2. #2 peak

    playing cards needed: SKIP, DRAW TWO, REVERSE, WILD⏎|This resource is also available for purchase on Teacher Pay Teachers. If you prefer to purchase it there, follow this link.|⏎How to assemble and use this resource:⏎

  3. #3 peak

    moisture meter is a great tool to help you keep track of your plant’s watering routine. Using one of these is a lot less hassle than trying to remember to water your succulents every day.⏎How to Water Succul

  4. #4 peak

    cool, but it requires both the sender and recipient to be running the app. In addition, it's not open with what it can send or where it can send it--it only works from phone to phone, and while there are many options

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. L18:1601 strongly suppresses
  2. L16:2227 suppresses
  3. L18:2153 suppresses
  4. L15:675 excites
  5. L11:153 suppresses
  6. L6:375 suppresses
  7. L13:2299 suppresses
  8. L14:149 suppresses
  9. L6:388 excites
  10. L14:526 suppresses
  11. L17:5 excites
  12. L5:146 excites
  13. L16:739 excites
  14. L18:2787 excites
  15. L2:491 excites
  16. L8:216 excites
  17. L12:336 suppresses
  18. L9:413 excites
  19. L17:790 suppresses
  20. L8:2000 suppresses
  21. L14:1148 excites
  22. L15:957 excites
  23. L2:2539 weakly excites
  24. L9:584 weakly suppresses

Feeds later layers

  1. L20:833 suppresses
  2. L23:570 suppresses
  3. L22:132 suppresses
  4. L20:2484 suppresses
  5. L20:1972 weakly suppresses 'of' in 'of course'
  6. L21:801 weakly suppresses
  7. L20:1974 weakly suppresses
  8. L25:1963 weakly excites
  9. L23:326 weakly suppresses
  10. L26:585 weakly suppresses
  11. L20:1061 weakly suppresses
  12. L22:353 weakly excites
  13. L21:140 weakly excites
  14. L21:1339 weakly excites
  15. L20:680 weakly suppresses
  16. L25:277 weakly suppresses
  17. L25:806 weakly suppresses
  18. L25:247 weakly suppresses
  19. L21:2342 weakly suppresses
  20. L25:802 weakly excites
  21. L24:859 weakly excites
  22. L20:1123 weakly excites
  23. L26:496 weakly excites
  24. L23:539 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 -21.38 and -20.701 values between -13.27 and -12.592 values between -9.89 and -9.211 values between -9.21 and -8.544 values between -8.54 and -7.867 values between -7.86 and -7.194 values between -7.19 and -6.518 values between -6.51 and -5.834 values between -5.83 and -5.168 values between -5.16 and -4.484 values between -4.48 and -3.8112 values between -3.81 and -3.1312 values between -3.13 and -2.4620 values between -2.46 and -1.7849 values between -1.78 and -1.10264 values between -1.10 and -0.432644 values between -0.43 and 0.25712 values between 0.25 and 0.9273 values between 0.92 and 1.6038 values between 1.60 and 2.2726 values between 2.27 and 2.959 values between 2.95 and 3.631 values between 3.63 and 4.301 values between 4.30 and 4.983 values between 4.98 and 5.65offnegativepositiveoften

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. (@(
  2. 跟她
  3. sWith
  4. Body
  5. okin
  6. {{--<
  7. 碼
  8. 我都
  9. 嗬
  10. ưu

Suppresses

  1. through
  2. ability
  3. between
  4. neider
  5. ние
  6. ibility
  7. according
  8. between
  9. ability
  10. 通过