Neuron Explorer

Qwen3-0.6B

word fragments before a name, 'ma' bias

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 the opening pieces of words split across tokens, often names or nouns, such as 'bul' in bulge or 'Sel' in Selma. Turned up strongly, the model began writing 'Ma Ma Ma', so it seems to push toward the sound 'ma'.

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

    Japanese scifi movie...Scientists have spotted stellar fossils in the center of the Milky Way, globular clusters orbiting in the central bulge which seem to have come from somewhere else. Like all fossils these are vital

  2. #2 peak

    Hungarian king Kalman, Jews were permitted to live only in cathedral towns and on bishops estates. In 1241 the Mongols (also known as Tartars) invaded Hungary, wreaking havoc and destruction. Jewish merchants made a major

  3. #3 peak

    content was digitised by double rekeying. All rights reserved.⏎The Plague of 1348—The Origin of the Charterhouse—Sir Thomas More there—Cromwell's Commissioners—Prior Houghton—The Depart

  4. #4 peak

    lle, o méle, lo méle, u mméla “honey”⏎lo latte, o latte, lə lattə “milk”⏎ru cane “dog”⏎ru peʃʃu “singular fish”⏎

  5. #5 peak

    case studies, demographics<|endoftext|>Tom Houck is a lifelong U.S. civil rights activist. In 1965, the white high school student traveled to Selma, Alabama, to participate in the march there, led by the Rev.

  6. #6 peak

    May morning to view the medows gay,⏎I met a pretty dairy-maid all in the new mown hay.⏎* 4 *⏎As I was

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: Can an algorithm be pointed out so that an algorithm can be pointed out again? 问题是问题是这个问题的问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题问题

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

    which at the time of the Three States (Samguk 三國) was occupying the South-Western part of the Korean peninsula 3. The state of Paekche, because of the continuous pression exerted upon it by K

  2. #2 peak

    , “being Buddha nature and non-being Buddha nature.” I like the use of “being” in this reference. In the first sentence when referencing all tathagatas, he is putting forth the meaning of “non-being Buddha nature”. In

  3. #3 peak

    an cheese and roast in the oven for a crispy and flavorful side dish.⏎- Broccoli and mushroom omelette: Saute broccoli florets and sliced mushrooms and fold them into a fluffy egg omelette for a filling and nutritious breakfast.

  4. #4 peak

    Nature Beyond Conditions and a commentary by Rev. David Xi-Ken Astor, Sensei⏎BUDDHA NATURE BEYOND CONDITIONS⏎“All tathagatas are without Buddha nature, but at the same time, previously they have fully

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. L26:789 excites
  2. L8:10 excites
  3. L26:1564 excites
  4. L25:2769 suppresses
  5. L16:17 weakly suppresses
  6. L22:859 weakly suppresses
  7. L3:2084 weakly excites
  8. L7:57 weakly excites
  9. L3:2630 weakly suppresses
  10. L23:2194 weakly excites
  11. L19:546 weakly excites
  12. L2:2142 weakly suppresses
  13. L22:1902 weakly excites
  14. L17:1140 weakly suppresses
  15. L8:7 weakly excites
  16. L25:1021 weakly suppresses
  17. L26:1624 weakly excites
  18. L15:984 weakly suppresses
  19. L8:13 weakly suppresses
  20. L0:2459 weakly suppresses
  21. L8:14 weakly excites
  22. L19:911 weakly excites
  23. L4:205 weakly excites the digit 2 inside citations
  24. L7:64 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 -8.05 and -7.551 values between -5.56 and -5.069 values between -2.08 and -1.5833 values between -1.58 and -1.08244 values between -1.08 and -0.581673 values between -0.58 and -0.081802 values between -0.08 and 0.41109 values between 0.41 and 0.9110 values between 0.91 and 1.416 values between 1.41 and 1.911 values between 1.91 and 2.404 values between 2.40 and 2.902 values between 2.90 and 3.407 values between 3.40 and 3.902 values between 4.89 and 5.391 values between 5.39 and 5.891 values between 6.88 and 7.381 values between 10.86 and 11.361 values between 11.36 and 11.86offnegativepositiveoften

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. Ma
  2. ma
  3. Ma
  4. ma
  5. MA
  6. MA
  7. sa
  8. (pa
  9. Pa
  10. pa

Suppresses

  1. 枝
  2. etcode
  3. onian
  4. olvers
  5. illator
  6. LICENSE
  7. ibil
  8. 问题是
  9. .PerformLayout
  10. IBLE