Neuron Explorer

Qwen3-0.6B

'Aut' starts and colon prediction

A neuron in layer 27 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

It fires on the opening piece 'Aut' in words like 'autism' and 'autotransformer', and on some other word starts. Its output pushes the model toward writing a colon, and turning it up made the model write long strings of colons.

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

    they were able to share stories and discuss challenges. Although the ages of the children and levels of severity vary, members of the group have found much common ground.⏎Autism spectrum disorder

  2. #2 peak

    condition]⏎wiki: In the past, syndromes were often named after the physician or scientist who identified and described the condition in an initial publication⏎ex. 唐氏

  3. #3 peak

    become a foolhardy risk of life as well as a waste of time and energy.⏎So the strategy changes: evacuate the vulnerable, the impossible to defend. Defend that which can be defended, wait for the fire to burn itself out and

  4. #4 peak

    functional parts.⏎- 1 Power transformers⏎- 1.1 Laminated core⏎- 1.2 Toroidal⏎- 1.3 Autotransformer⏎- 1.4 Variable autotransformer⏎-

  5. #5 peak

    the production of silicon. The activity is an investigation of silicon: the sources, uses, properties, importance in the fields of photovoltaics (solar cells/renewable energy) and integrated circuits industries, and, to a limited extent,

  6. #6 peak

    ?⏎Suddenly, you would be a mess. You’d ask for help from your parents, and would likely be embarrassed that you had to ask them for help.⏎Autistic individuals are often in this situation. Autistic

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: sometimes it seems that way.",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".",".","

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

    and criteria as we use for PAA and CAF.⏎Through these three programs each<|endoftext|>发布时间:2017-03-27 来源:新

  2. #2 peak

    .⏎2 پژوهشگر دوره تخصصی دکتری معماری منظر، دانشکده معماری و شهرسازی، دانش

  3. #3 peak

    there is the risk of a major collapse in agricultural production in poor countries. This will increase malnutrition already at elevated levels in these countries.⏎Sr. Bénédicte Kikwengi, SNDdeN works with her students at

  4. #4 peak

    very low in calories, which makes<|endoftext|>A Study on Chinese Auxiliary Verbs for Thai Learners Based on the Comparative Analysis⏎- 人文学院-已发表论文 ⏎“助动词”在汉语中具有较大的作用,使用率高

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. L26:345 strongly excites
  2. L26:1329 strongly excites
  3. L25:737 strongly excites
  4. L26:2215 suppresses
  5. L24:1896 excites
  6. L24:500 excites
  7. L19:560 excites end of document, and 'it turned out'
  8. L26:122 suppresses
  9. L25:502 suppresses
  10. L26:582 excites
  11. L24:853 suppresses
  12. L21:165 excites
  13. L20:2106 excites
  14. L26:823 excites
  15. L23:236 excites
  16. L20:54 excites
  17. L14:989 weakly excites
  18. L24:2971 weakly suppresses
  19. L21:616 weakly excites
  20. L20:1313 weakly excites
  21. L3:2524 weakly excites
  22. L25:4 weakly excites
  23. L25:1050 weakly excites
  24. L21:2823 weakly excites

Feeds later layers

None: this is the last layer.

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 -3.19 and -2.444 values between -2.44 and -1.7059 values between -1.70 and -0.961236 values between -0.96 and -0.222518 values between -0.22 and 0.5356 values between 0.53 and 1.2714 values between 1.27 and 2.013 values between 2.01 and 2.763 values between 2.76 and 3.506 values between 3.50 and 4.244 values between 4.24 and 4.981 values between 4.98 and 5.731 values between 6.47 and 7.211 values between 7.95 and 8.701 values between 25.77 and 26.52offnegativepositiveoften

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. ってしまいます
  3. є
  4. uchen
  5. fortunate
  6. .","
  7. enerative
  8. afone
  9. .','
  10. pleaded