Neuron Explorer

Qwen3-0.6B

newline starting a new block

A neuron in layer 0 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 responds to line breaks, especially the first line break at the start of a new piece of text or after a title. Turning it up made the model repeat phrases and ramble, so it does not seem to shape the topic.

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

    0⏎network => 172.16.0.0/15

  2. #2 peak

    ⏎1. n. Any one of numerous species of marine gastropods belonging to

  3. #3 peak

    ⏎- SOA services have self-describing interfaces in platform-independent XML documents. Web

  4. #4 peak

    ⏎Lin CC, Lin XM, Zou Y, Zhang L (200

  5. #5 peak

    ⏎Some people use their fridge to store everything in their kitchen. Your fridge is not

  6. #6 peak

    ⏎know much about it beyond that and couldn't find much information about⏎it on

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 lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately

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

    Empirical data suggest substantial variations in bone<|endoftext|>Missing Persons Alert⏎mental_floss is seeking information that will lead to the discovery of eight prominent people

  2. #2 peak

    an acid taste, if desired, but the acid<|endoftext|>5 Tips For Having Fun With Shadows Page 2⏎Silhouettes offer another opportunity to use shadows. A silhouette is a shadow--

  3. #3 peak

    Infinitely old, Static and Uniform⏎- Continually expanding, Finite age, of a finite size and of a finite

  4. #4 peak

    ten objects from museum collections in Greater Manchester⏎A compass used in a PoW escape, a railway station sundial and part

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. L2:2891 excites
  2. L10:295 suppresses
  3. L1:2228 weakly excites
  4. L4:80 weakly suppresses
  5. L2:980 weakly excites
  6. L14:356 weakly excites
  7. L2:477 weakly suppresses line break in the middle of a sentence
  8. L10:123 weakly suppresses
  9. L12:707 weakly excites
  10. L7:2775 weakly excites
  11. L19:629 weakly excites
  12. L25:1141 weakly excites
  13. L15:1844 weakly suppresses
  14. L10:969 weakly excites
  15. L18:1596 weakly excites
  16. L5:521 weakly suppresses
  17. L2:1290 weakly suppresses
  18. L13:475 weakly excites
  19. L24:1497 weakly suppresses
  20. L7:1386 weakly excites
  21. L16:1109 weakly excites
  22. L18:991 weakly excites
  23. L11:2856 weakly excites
  24. L16:3027 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 -1.16 and -1.111 values between -1.11 and -1.061 values between -1.06 and -1.011 values between -1.01 and -0.965 values between -0.91 and -0.863 values between -0.81 and -0.763 values between -0.76 and -0.713 values between -0.71 and -0.662 values between -0.66 and -0.6110 values between -0.61 and -0.569 values between -0.56 and -0.515 values between -0.51 and -0.466 values between -0.46 and -0.416 values between -0.41 and -0.368 values between -0.36 and -0.313 values between -0.31 and -0.2610 values between -0.26 and -0.2126 values between -0.21 and -0.1639 values between -0.16 and -0.11114 values between -0.11 and -0.06457 values between -0.06 and -0.011447 values between -0.01 and 0.041099 values between 0.04 and 0.09524 values between 0.09 and 0.14115 values between 0.14 and 0.196 values between 0.19 and 0.241 values between 0.34 and 0.391 values between 0.39 and 0.441 values between 0.44 and 0.491 values between 0.78 and 0.83offnegativepositiveoften

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. epy
  2. mới
  3. 财力
  4. 几何
  5. sparing
  6. deepest
  7. dành
  8. juste
  9. baru
  10. tym

Suppresses

  1. reas
  2. __': ⏎
  3. ? ⏎
  4. marvin
  5. __':⏎
  6. bell
  7. _ci
  8. ugen
  9. rogate
  10. (secret