Neuron Explorer

Qwen3-0.6B

'co' starting a word piece

A neuron in layer 26 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 fires on the syllable 'co' when it starts a word that has been split into pieces, as in 'coexist', 'coining' and 'coauthored'. The steering outputs did not clearly reflect this, so the pattern rests on the text snippets alone.

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

    Arabia’s Information Minister put his profile on Facebook and got more than 5,000 “friends.” And when the king fired a senior cleric for criticizing co-education at a new Saudi university, the cleric’s supporters struck back in cy

  2. #2 peak

    be members traveled near and far to get some of the mysterious liquid. In 1627 the Seneca people, brought French missionary, Joseph De La Roche d’Aillon, to the sacred spot, and he blessed the spring

  3. #3 peak

    ball⏎Here to Help Average Joes & Janes Play Better Basketball⏎A Basketball term meaning to embarrass some one usually while slamming the ball over them.⏎James Naismith wrote the original 13 rules of basketball.<|endoftext|>a sheet with

  4. #4 peak

    the father of chaos theory and discovered the Lorenz attractor that often occurs in chaotic systems.⏎Lorenz is perhaps best known for coining the term butterfly effect. Whilst a professor at the Massachusetts Institute of Technology,

  5. #5 peak

    Tunis where several African communities coexist, made news headlines.⏎Ksiksi remembers when Tunisia became the first

  6. #6 peak

    source philosophy in order to produce the premier software of their markets and many nonprofit organizations have coalesced in support of open source software. Even governments have heralded their support

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: a lot of a lot of a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot a lot

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

    Concrete poured and formed on the construction site by concrete contractors is obviously at the mercy of weather conditions at the time of curing. Precast concrete panel construction, therefore, results in a stronger, less porous, higher

  2. #2 peak

    It’s important to note core stability is a key component of pelvic and core health. You can find all of the core stability and pelvic floor exercises in the Prenatal and Postnatal Exercise Guide.⏎Perform these seated, sitting on a Swiss ball

  3. #3 peak

    similar to gray mold. The color of the decayed area is light to dark brown, and black in advanced stages. A second symptom of infection starts at lenticels. Symptoms are brown to black specks with a white to light tan center

  4. #4 peak

    Institution: Lutheran Children and Family Service, Philadelphia, PA.<|endoftext|>Updated: May 19, 2021⏎A great video to help explain Concussion Management by our friend, Dr. Mike Evans before we get into this:⏎

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. L1:460 suppresses
  2. L17:123 weakly excites
  3. L16:91 weakly suppresses
  4. L1:2342 weakly suppresses
  5. L20:59 weakly suppresses
  6. L1:110 weakly suppresses
  7. L0:7 weakly excites
  8. L5:1458 weakly suppresses
  9. L21:296 weakly suppresses
  10. L16:34 weakly suppresses
  11. L22:14 weakly excites
  12. L1:2159 weakly excites
  13. L0:2780 weakly excites
  14. L4:343 weakly suppresses
  15. L16:1659 weakly suppresses
  16. L1:81 weakly suppresses
  17. L6:19 weakly excites second half of proper names
  18. L4:122 weakly suppresses
  19. L20:452 weakly excites
  20. L0:2107 weakly excites
  21. L6:942 weakly suppresses
  22. L20:38 weakly suppresses
  23. L17:218 weakly suppresses
  24. L7:2696 weakly excites

Feeds later layers

  1. L27:131 weakly suppresses
  2. L27:802 weakly excites
  3. L27:2983 weakly excites
  4. L27:1132 weakly suppresses
  5. L27:2242 weakly excites
  6. L27:159 weakly excites
  7. L27:682 weakly excites
  8. L27:1962 weakly excites
  9. L27:1067 weakly suppresses
  10. L27:601 weakly suppresses
  11. L27:2143 weakly suppresses
  12. L27:1228 weakly suppresses
  13. L27:890 weakly suppresses
  14. L27:276 weakly excites
  15. L27:3061 weakly excites
  16. L27:2164 weakly suppresses
  17. L27:549 weakly suppresses
  18. L27:1452 weakly excites
  19. L27:1508 weakly excites
  20. L27:536 weakly excites
  21. L27:2714 weakly suppresses
  22. L27:2199 weakly excites
  23. L27:959 weakly suppresses
  24. L27:2861 weakly excites the letter f starting fume or fountain

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 -34.41 and -33.012 values between -26.00 and -24.602 values between -23.20 and -21.802 values between -21.80 and -20.403 values between -20.40 and -19.006 values between -19.00 and -17.603 values between -17.60 and -16.204 values between -16.20 and -14.8010 values between -14.80 and -13.4012 values between -13.40 and -11.9912 values between -11.99 and -10.5917 values between -10.59 and -9.1923 values between -9.19 and -7.7938 values between -7.79 and -6.3968 values between -6.39 and -4.99102 values between -4.99 and -3.59215 values between -3.59 and -2.19475 values between -2.19 and -0.792390 values between -0.79 and 0.61259 values between 0.61 and 2.0171 values between 2.01 and 3.4253 values between 3.42 and 4.8248 values between 4.82 and 6.2237 values between 6.22 and 7.6215 values between 7.62 and 9.0213 values between 9.02 and 10.4212 values between 10.42 and 11.826 values between 11.82 and 13.222 values between 13.22 and 14.622 values between 16.02 and 17.422 values between 17.42 and 18.821 values between 18.82 and 20.222 values between 20.22 and 21.63offnegativepositiveoften

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. lets
  3. 年底前
  4. stream
  5. ৎ
  6. MARK
  7. ände
  8. --
  9. wide
  10. <>

Suppresses

  1. .didReceiveMemoryWarning
  2. üzere
  3. antan
  4. ounder
  5. ilio
  6. earer
  7. igham
  8. rawler
  9. (PC
  10. onus