Neuron Explorer

Qwen3-0.6B

period in abbreviations and initials

A neuron in layer 3 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 full stops that come after a short abbreviation, such as 'c.' for 'circa' in dates or after a person's initial in a citation. It goes negative on the period after initials in author names. Turning it up or down did not clearly change the text.

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

    topic.⏎Outer Kingdoms of the First Millennium BC⏎MITANNI – c. 1150-950 BC⏎Mitanni was a multi

  2. #2 peak

    outside vascular bundles; seed solitary.⏎Distribution and occurrence: World: c. 65 genera, c. 900 species, cosmopolitan.

  3. #3 peak

    3. Demonstration on pest management technologies like pheromone traps, fruit bagging, and etc.<|endoftext|>Little Wound (c. 1835–Winter 1899;⏎His father Chief

  4. #4 peak

    a province within the Middle Assyrian Empire.⏎ Some historians place the Kassites and Hurrians within the bounds of Mitanni.⏎The KASSITES – c. 1184-763⏎When Matanni rose in northern

  5. #5 peak

    in range margins and community composition in response to global change. In the European Alps, plant species are shifting their range to a higher elevation with a speed of c. 3 – 5 m per year, and non-native species are moving even

  6. #6 peak

    a version counting back to 3076 BCE is still in use in parts of India today.⏎Astronomy flourished under the Gupta Empire (c. 320-550 CE) during which time Ujjain

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 need to be honested in the last. I have no idea what it is. So I have no idea what it is. But I have no idea what it is. I am not sure of what I am. I am not

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

    systems (H. Qammari, A. Venkatesan). A vacation on Mars - an artist's journey in a computer graphics world (C.A. Pickover). Part II. Attractors. Automatic generation of strange attractors

  2. #2 peak

    your brain's aging process. But what is up for debate is whether or not alcohol<|endoftext|>Using groundwater for hydro electric power generation⏎Bardsley, W.E. (1995). Using groundwater for hydro electric power generation. Journal of

  3. #3 peak

    , 1967)⏎- Mitsukurina owstoni Jordan, 1898 (Goblin shark)⏎- Jordan, D.S. (1898). "Description of a species of fish (Mits

  4. #4 peak

    Federal Death Penalty" Washington Law Review (2010).<|endoftext|>Last updated 2/27/15⏎Isothermal Community College⏎P.O. Box 804, Spindale NC 28160

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. L1:1795 excites
  2. L2:2614 weakly excites
  3. L1:331 weakly suppresses
  4. L2:2837 weakly suppresses
  5. L0:1336 weakly excites
  6. L0:1978 weakly suppresses
  7. L2:1321 weakly suppresses
  8. L0:2984 weakly suppresses
  9. L1:353 weakly suppresses
  10. L0:91 weakly suppresses
  11. L0:314 weakly excites
  12. L2:2100 weakly suppresses
  13. L0:1042 weakly suppresses "This" starting a labelled sentence
  14. L1:1676 weakly suppresses
  15. L0:12 weakly suppresses
  16. L1:128 weakly excites
  17. L0:553 weakly excites second half of a split name
  18. L2:666 weakly excites
  19. L0:76 weakly suppresses
  20. L1:223 weakly excites unclear: 'selected' and 'shape' in mixed text
  21. L0:2974 weakly excites
  22. L1:63 weakly suppresses
  23. L2:460 weakly suppresses
  24. L1:3049 weakly suppresses

Feeds later layers

  1. L4:204 weakly suppresses
  2. L6:2493 weakly excites
  3. L4:907 weakly excites
  4. L24:2695 weakly suppresses
  5. L4:2942 weakly excites 'yet' and 'haven't' after have
  6. L12:878 weakly excites
  7. L20:2181 weakly excites
  8. L4:744 weakly excites
  9. L5:2284 weakly excites
  10. L4:1006 weakly excites
  11. L6:2812 weakly excites
  12. L5:593 weakly excites
  13. L7:1522 weakly suppresses
  14. L15:1953 weakly excites unclear: document ends and number gaps
  15. L9:1943 weakly suppresses
  16. L14:2420 weakly suppresses
  17. L26:1155 weakly suppresses
  18. L6:2554 weakly excites
  19. L4:2496 weakly excites
  20. L19:446 weakly excites
  21. L9:2665 weakly excites
  22. L6:439 weakly excites
  23. L6:357 weakly excites
  24. L27:953 weakly suppresses

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 -5.75 and -5.601 values between -4.03 and -3.871 values between -2.14 and -1.992 values between -1.67 and -1.521 values between -1.20 and -1.041 values between -1.04 and -0.892 values between -0.73 and -0.573 values between -0.57 and -0.4234 values between -0.42 and -0.26538 values between -0.26 and -0.102516 values between -0.10 and 0.06737 values between 0.06 and 0.2166 values between 0.21 and 0.375 values between 0.37 and 0.53offnegativepositiveoften

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. udson
  2. ars
  3. .Diagnostics
  4. /rs
  5. sburg
  6. 血压
  7. setTitle
  8. .setUp
  9. veau
  10. .setHeight

Suppresses

  1. `'
  2. Ė
  3. ngo
  4. licas
  5. PageIndex
  6. board
  7. .%
  8. 弋
  9. pretending
  10. vox