Neuron Explorer

Qwen3-0.6B

unclear: 'if' opening a statement, 'local'

A neuron in layer 0 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 is hard to pin down. It fires most on the word 'if' when it starts a line of code or a condition, and on the word 'local' in ordinary sentences. When it was turned up, the model's writing became repetitive and stuck on the word 'sunny', which does not clearly relate to what the neuron responds to.

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

    headed things you can do - but look at the C++-tagged posts in this blog for examples of what the compiler thinks isn't a bad idea.)⏎if (x=0) init_subsystem();⏎So lots of books have been written

  2. #2 peak

    be achieved with the following code:⏎TYPE List = REF RECORD next: List; value: INTEGER; END;⏎VAR x := NEW(List, value:=1); BEGIN x.next :=

  3. #3 peak

    popular⏎foldr. This package provides that same function for the⏎Bool data type; a function called⏎bool function is a replacement for the build-in⏎if then else-syntax. However, since it is a function instead of syntax, it

  4. #4 peak

    different things I need to return. If I ask for something that isn't there, I get an exception, which is life as usual for a Python programmer.⏎if not server_response['successful']:⏎etag = server_response['file_handle'].info().

  5. #5 peak

    local traditions and folklore with just a touch of modern influence<|endoftext|>In the last post,

  6. #6 peak

    local taxes to make them more comparable to gasoline taxes in other countries. Others want these

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: I have been thinking about the history of the two main things that are about the history of the two main things that are about the history of the two main things that are about the history of the two main things that are about the history of

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

    ⏎Updated: Sep 24, 2018⏎Detoxification – you’ve heard this word many times over the last few years, but unless you have actually participated in a detoxification program, you probably don’t really understand

  2. #2 peak

    frequently if they had met the dog before.⏎Control tests were done to ensure that the dogs

  3. #3 peak

    an emergency.⏎Rest until you are fully recovered⏎after you leave the hospital. This may take 4 to 6 weeks. Your healthcare provider will tell you if you need to be on bedrest. This means you stay in bed and do

  4. #4 peak

    55 file extension⏎Recommended software programs are sorted by OS platform (Windows, Mac OS X etc

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

None: this is the first layer.

Feeds later layers

  1. L25:2536 weakly suppresses
  2. L3:363 weakly excites
  3. L1:1572 weakly suppresses
  4. L8:1715 weakly suppresses
  5. L21:303 weakly excites
  6. L20:1321 weakly suppresses
  7. L27:1814 weakly suppresses
  8. L27:2317 weakly suppresses
  9. L1:2368 weakly suppresses
  10. L1:2080 weakly excites
  11. L5:2960 weakly suppresses
  12. L4:362 weakly suppresses
  13. L6:724 weakly suppresses
  14. L17:2787 weakly excites comma after 'that' or a verb before an aside
  15. L4:1859 weakly suppresses
  16. L25:2756 weakly suppresses
  17. L23:13 weakly excites
  18. L19:2586 weakly suppresses
  19. L13:3063 weakly excites
  20. L27:920 weakly suppresses
  21. L2:458 weakly excites
  22. L20:1031 weakly suppresses
  23. L5:164 weakly excites
  24. L17:77 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 -0.95 and -0.911 values between -0.66 and -0.622 values between -0.58 and -0.531 values between -0.33 and -0.292 values between -0.29 and -0.244 values between -0.24 and -0.2027 values between -0.20 and -0.16103 values between -0.16 and -0.12249 values between -0.12 and -0.08461 values between -0.08 and -0.04799 values between -0.04 and 0.00898 values between 0.00 and 0.05686 values between 0.05 and 0.09422 values between 0.09 and 0.13175 values between 0.13 and 0.1760 values between 0.17 and 0.2111 values between 0.21 and 0.252 values between 0.25 and 0.291 values between 0.29 and 0.341 values between 0.34 and 0.381 values between 0.42 and 0.461 values between 0.67 and 0.71offnegativepositiveoften

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. nick
  2. found
  3. termin
  4. lix
  5. planning
  6. ElapsedTime
  7. affected
  8. related
  9. like
  10. Termin

Suppresses

  1. 婢
  2. 因为她
  3. sense
  4. 主义者
  5. auer
  6. 男篮
  7. 剃
  8. ciné
  9. <textarea
  10. ело