Neuron Explorer

Qwen3-0.6B

period ending a contact or help sentence

A neuron in layer 22 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 full stop or exclamation mark that ends a sentence inviting the reader to get in touch or ask questions, such as 'contact us today.' Steering had no clear effect on the topic of the text.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens an email to the maker with the neuron already filled in).

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

    Contact Air Care Heating And Air Conditioning⏎For more information on the best humidity for your home, or to schedule an HVAC appointment, contact us today by clicking here.<|endoftext|>[authors: Barbara Reedy; Michael Reedy]⏎“The history of the

  2. #2 peak

    sufficient.⏎Of course the best way to see where the Ganges river is by using a map of the River Ganges.⏎I hope this answers your question Rebecca. You can find out more information on the River Ganges by looking at more information

  3. #3 peak

    you interested in visiting places that have been abandoned? Or is it too creepy? Have you been to any that you can add to the list? Please share below!<|endoftext|>How to tell if lines

  4. #4 peak

    transaction that comes their way. If you cheat, your transaction does not get forwarded.⏎I hope that explains the concept of how Bitcoin works in regards to your question. Just to be clear:⏎"Does each bitcoin has a unique identifier,

  5. #5 peak

    atures LibGuide. Contact Sue Samson, firstname.lastname@example.org, with questions or for more information on how this database can help you with your research or teaching.<|endoftext|>Iris is technically the Greek goddess of messages, the sky, and the

  6. #6 peak

    supported by Campus Ambassadors, Students Against Slavery and Willamette University's Religious Studies Department.⏎For more information contact Jessica Sternbach at firstname.lastname@example.org.<|endoftext|>St. Peter's Cathedral is one of the most significant buildings inside the Vatican

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 first and the second term of the sequence. But I can't think of a term which is a prime number. But wait, I have a prime number in the first term. But the first term is 2, which is a

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

    The second generation consist of the muon (μ−⏎) and muon neutrino (νμ⏎) The third generation consist of the tau (τ−⏎) and the tau neutrino (ντ⏎). Each type of neutr

  2. #2 peak

    least developed countries as regards multilateral negotiations;⏎• Formulate a sound negotiation strategy;⏎• Analyze the significance of data required for a multilateral trade negotiation; and⏎• Assess various issues that may arise during negotiation

  3. #3 peak

    Python.⏎Assign strings to variables and use them to print output at certain stages of the program. For example:⏎>>> >>> s1 = 'String in single quotes' >>> s1 'String in single quotes' >>> >>> s2 = "String

  4. #4 peak

    to the bumper itself. …⏎- Valance Panel. …⏎- Inner Fender. …⏎- Fender. …⏎- Cowl Panel. …⏎- Door. …⏎- Lower Door Skin.⏎Why are car bumpers

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. L21:660 suppresses
  2. L20:561 excites end of document marker
  3. L19:1069 excites
  4. L21:217 excites
  5. L19:1479 suppresses
  6. L18:2222 suppresses
  7. L19:1106 suppresses
  8. L20:338 suppresses
  9. L19:732 suppresses unclear: end of document token
  10. L21:466 excites
  11. L18:1113 suppresses
  12. L17:1568 excites
  13. L15:2070 suppresses
  14. L15:410 suppresses
  15. L17:644 suppresses
  16. L19:2591 suppresses
  17. L7:1466 weakly excites
  18. L20:515 weakly excites period right before a line break
  19. L17:1043 weakly excites
  20. L18:1148 weakly suppresses
  21. L19:789 weakly suppresses
  22. L17:1088 weakly excites
  23. L16:2583 weakly suppresses
  24. L21:396 weakly suppresses

Feeds later layers

  1. L26:156 strongly excites
  2. L24:96 strongly suppresses
  3. L24:303 strongly excites
  4. L25:207 strongly excites
  5. L26:133 strongly excites
  6. L27:179 strongly suppresses
  7. L26:1180 strongly excites
  8. L25:1274 strongly excites
  9. L26:288 strongly excites
  10. L26:640 strongly suppresses
  11. L23:194 strongly suppresses
  12. L27:256 strongly suppresses
  13. L26:16 strongly suppresses
  14. L27:1065 strongly excites
  15. L27:309 strongly excites
  16. L25:144 strongly suppresses
  17. L27:445 excites
  18. L27:136 excites
  19. L26:2289 suppresses
  20. L24:2894 excites
  21. L26:749 excites
  22. L27:207 excites
  23. L26:230 excites
  24. L27:1053 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 -6.72 and -6.311 values between -5.51 and -5.112 values between -3.51 and -3.103 values between -2.70 and -2.302 values between -1.90 and -1.506 values between -1.50 and -1.1040 values between -1.10 and -0.70197 values between -0.70 and -0.302783 values between -0.30 and 0.11620 values between 0.11 and 0.5192 values between 0.51 and 0.9136 values between 0.91 and 1.3135 values between 1.31 and 1.7125 values between 1.71 and 2.1114 values between 2.11 and 2.519 values between 2.51 and 2.925 values between 2.92 and 3.3216 values between 3.32 and 3.728 values between 3.72 and 4.123 values between 4.12 and 4.521 values between 4.52 and 4.921 values between 5.32 and 5.731 values between 5.73 and 6.133 values between 6.13 and 6.531 values between 7.33 and 7.732 values between 7.73 and 8.131 values between 8.94 and 9.34offnegativepositiveoften

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. ,⏎⏎
  3. ",⏎⏎
  4. 、⏎⏎
  5. ,⏎⏎
  6. (),⏎⏎
  7. ),⏎⏎
  8. ],⏎⏎
  9. \⏎⏎
  10. },⏎⏎

Suppresses

  1. ?⏎
  2. .⏎
  3. ...⏎
  4. !⏎
  5. ...⏎
  6. …⏎
  7. 。⏎
  8. ?⏎
  9. ....⏎
  10. ?⏎