Neuron Explorer

Qwen3-0.6B

comma right after 'and' in lists

A neuron in layer 19 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

The neuron fires on a comma that comes straight after the word 'and' in a list, like 'routine and, mental well-being'. It is pushed the other way in computer code, where it reacts to the equals sign and type names. Steering did not show a clear theme.

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

    child's development and, on the 'Words for Life' website, there are some useful tips about how

  2. #2 peak

    the international humanitarian system and, of course, refugees themselves. A new RAND report details how technology can be part

  3. #3 peak

    facilitate habits and, for some, sustained compulsivity and addiction. Like other complex illnesses and disorders,

  4. #4 peak

    culture, history, and society.⏎About the authors⏎Elaine Treharne is Roberta Bowman Denning Professor of Humanities and Professor of English (and, by courtesy, of German Studies) at Stanford University, as well as Director of

  5. #5 peak

    the other code (too tired to check who it was that posted it, and I can't see just now anyway :P)⏎void genFactorial(int x, int t)⏎int total = t; // you probably wouldn't even need total

  6. #6 peak

    personal lifestyle of a person includes a daily diet, working conditions, exercise routines and, mental well-being. This examination helps the medical practitioner to identify key symptoms that are

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: there, there, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing, nothing,

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

    1.DataSource, DataSet) ' Get the DataTableCollection. Dim tablesCollection As DataTableCollection = thisDataSet.Tables ' Get a specific table by name. Dim table As DataTable = tablesCollection("Suppliers") Console.WriteLine(table.TableName) End Sub⏎Available

  2. #2 peak

    ster March 3 2012 import random verb = ['erratic', 'unbelievable', 'unfortunate', 'inconsequential'] fruit = ['raspberry', 'pomegranite', 'bananana',

  3. #3 peak

    to check who it was that posted it, and I can't see just now anyway :P)⏎void genFactorial(int x, int t)⏎int total = t; // you probably wouldn't even need total now⏎void main(void)

  4. #4 peak

    memchr ( const void *, int, size_t );⏎instead of the two overloaded versions provided in C++.⏎/* memchr example */⏎int main ()⏎char * pch;⏎char str = "Example string";⏎pch = (char

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. L18:2702 suppresses
  2. L18:1948 excites
  3. L13:19 excites
  4. L12:1163 suppresses
  5. L17:1946 weakly suppresses
  6. L5:110 weakly suppresses
  7. L8:7 weakly excites
  8. L18:1367 weakly suppresses
  9. L15:2495 weakly suppresses
  10. L14:2423 weakly suppresses
  11. L8:3 weakly excites
  12. L17:503 weakly suppresses
  13. L12:2418 weakly suppresses
  14. L16:2987 weakly suppresses
  15. L18:2775 weakly excites
  16. L14:2923 weakly suppresses
  17. L10:81 weakly suppresses
  18. L3:456 weakly suppresses
  19. L18:741 weakly suppresses
  20. L12:177 weakly excites
  21. L12:1921 weakly suppresses
  22. L3:2140 weakly suppresses
  23. L12:1681 weakly suppresses
  24. L14:441 weakly excites

Feeds later layers

  1. L20:1479 strongly suppresses
  2. L22:539 strongly suppresses
  3. L22:1710 strongly suppresses
  4. L20:1768 suppresses
  5. L24:334 suppresses
  6. L25:2646 suppresses 'there' after a because or if clause
  7. L23:434 suppresses
  8. L22:1300 suppresses
  9. L20:333 excites
  10. L23:1295 suppresses
  11. L20:1985 excites
  12. L26:2533 suppresses
  13. L20:218 suppresses
  14. L20:46 excites
  15. L24:1318 suppresses
  16. L21:1871 suppresses
  17. L20:1873 suppresses
  18. L25:1229 suppresses
  19. L20:126 excites
  20. L25:976 suppresses
  21. L20:1960 suppresses
  22. L26:406 suppresses
  23. L22:1562 weakly suppresses
  24. L23:67 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 -1.08 and -0.972 values between -0.86 and -0.7516 values between -0.75 and -0.6427 values between -0.64 and -0.5252 values between -0.52 and -0.41104 values between -0.41 and -0.30244 values between -0.30 and -0.19533 values between -0.19 and -0.081411 values between -0.08 and 0.04948 values between 0.04 and 0.15324 values between 0.15 and 0.26145 values between 0.26 and 0.3763 values between 0.37 and 0.4826 values between 0.48 and 0.595 values between 0.59 and 0.713 values between 0.71 and 0.822 values between 0.82 and 0.931 values between 2.05 and 2.161 values between 3.27 and 3.39offnegativepositiveoften

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. except
  2. stär
  3. ypress
  4. except
  5. plementary
  6. vron
  7. zeug
  8. 久了
  9. ><?=$
  10. until

Suppresses

  1. none
  2. None
  3. NULL
  4. null
  5. None
  6. null
  7. nothing
  8. Nothing
  9. none
  10. .NONE