Neuron Explorer

Qwen3-0.6B

mid-word fragments, bracket output

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

This neuron fires on word pieces that start a longer word, such as 'Spe' in 'Speaks' or 'Contr' in 'Contrary'. Turned up, the model wrote long runs of curly or square brackets, and turned down it got stuck on 'exposure'.

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

    ancient as the world and older than the flow of human blood in human veins.⏎My soul has grown deep like the rivers.⏎Langston Hughes, “The Negro Speaks of Rivers External.”⏎Hughes’ parents separated shortly after he was born,

  2. #2 peak

    The Negro Speaks of Rivers,” written the summer after he graduated from high school in Cleveland,

  3. #3 peak

    a venom that causes an immediate, localized pain. Within minutes, a red, raised spot usually develops, followed a day later by a white, pimple-like pustule that itches like crazy.⏎Because each ant can sting repeatedly — and

  4. #4 peak

    , often host concerts and other events. During the Christmas holiday season, approximately 100,000 Christmas lights sparkle from<|endoftext|>Disseminated Intravascular Coagulation (DIC) in Dogs⏎Clotting factors, such

  5. #5 peak

    Thus, the increasing demand for energy and exponential surge in consumption stimulates problems such as an increase in carbon emissions, the greenhouse effect and climate change. Contriving a means of saving energy is therefore crucial. Consequently, many governments and organizations

  6. #6 peak

    (four men and five postmenopausal women, ages 51-57).⏎"We are the first to document improved aortic hemodynamics in prehypertensive but otherwise healthy middle-aged men and women receiving therapeutic doses of watermelon

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: study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study study

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

    -2 hours depending on the size of your jar. A full melt pool will reach a depth of 1/2 inch. This will allow the fragrance to disperse more completely into the room. (Remember burn your candles for a maximum of

  2. #2 peak

    Neem oil. Add up to 5ml of the aromatherapy blend into a foot bath with warm, not hot, water, and agitate to disperse the oils. This bend is beneficial for fungal infestions.⏎Is Tag

  3. #3 peak

    urry, and the slurry will not be able to adhere tightly to the foamed plastic net. After demolding, it will have no strength and will easily disassemble. Hang up even if it has a certain strength, because the slurry

  4. #4 peak

    the locations of fractures predominantly on the left side of the posterior crania strongly suggest that these injuries were not accidental. The severity of fractures with large pieces of bone dislodged indicate that these were formed from violent blows in a standardised manner (

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. L24:2317 excites
  2. L26:2172 excites
  3. L25:2371 suppresses
  4. L19:2046 suppresses
  5. L25:1058 suppresses
  6. L24:581 excites
  7. L23:81 weakly suppresses
  8. L0:92 weakly suppresses
  9. L22:1728 weakly excites unclear: code assignments and parentheses
  10. L22:301 weakly excites
  11. L26:1026 weakly suppresses
  12. L9:2088 weakly suppresses
  13. L26:643 weakly excites
  14. L24:146 weakly suppresses
  15. L16:76 weakly excites
  16. L11:841 weakly suppresses
  17. L25:62 weakly suppresses
  18. L1:375 weakly suppresses
  19. L18:1123 weakly suppresses
  20. L19:1201 weakly suppresses
  21. L1:2282 weakly suppresses
  22. L5:1403 weakly suppresses
  23. L16:1645 weakly excites
  24. L5:1157 weakly suppresses

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 -8.06 and -7.262 values between -6.45 and -5.651 values between -4.04 and -3.233 values between -3.23 and -2.435 values between -2.43 and -1.6223 values between -1.62 and -0.822306 values between -0.82 and -0.011523 values between -0.01 and 0.7914 values between 0.79 and 1.607 values between 1.60 and 2.404 values between 2.40 and 3.211 values between 3.21 and 4.012 values between 4.01 and 4.821 values between 4.82 and 5.622 values between 5.62 and 6.431 values between 6.43 and 7.231 values between 7.23 and 8.041 values between 8.04 and 8.841 values between 8.84 and 9.652 values between 10.45 and 11.261 values between 11.26 and 12.061 values between 12.87 and 13.681 values between 14.48 and 15.292 values between 15.29 and 16.091 values between 21.73 and 22.531 values between 23.34 and 24.14offnegativepositiveoften

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. ancode
  2. ит
  3. 津
  4. .Generated
  5. rä
  6. \Module
  7. ridged
  8. ylan
  9. meis
  10. 研究人员