Neuron Explorer

Qwen3-0.6B

periods after initials and abbreviations

A neuron in layer 14 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 full stop that follows a single capital letter or a short abbreviation, like the dot in 'D. C.', 'E. coli', or 'J. Murray'. It doesn't clearly change the model's writing when turned up, so the steering results don't reflect this pattern.

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

    the dilemma than Boone County, where major urban streams such as Flat Branch, which recently tested for seven times the safe limit for E. coli, remain unprotected.⏎The county has 94 treatment systems that would have

  2. #2 peak

    attempting to counter the claim that articulation has a one-to-one correspondence with burial. I was responding to Hovers, E. and Belfer-Cohen, A. “Insights into early mortuary

  3. #3 peak

    in Washington, D. C. (Image via Zach Montellaro on Flickr, CC BY-NC

  4. #4 peak

    This dogma has been described in the Vyāsabhāsya on the Yogasutras of Patañjali (200 B. C.). The word ‘ekabhavika’ has been introduced and discussed here

  5. #5 peak

    from the arthropod realm are summarized and their potential use in allergy diagnostics discussed.<|endoftext|>GRĪW, a Middle Iranian word (in Manichean Mid. Pers. script gryw, rarely gryyw; in Sogd

  6. #6 peak

    LANGUAGE TEACHING AND RESEARCH|⏎|DOI - pysyväislinkit|⏎|Tila||Julkaistu - 1 syysk. 2021|⏎|OKM-julkaisutyyppi

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: the we have, we have, we we, we, we we, we we we, we we we, we we we, we we we, we we we, we we we, we we we, we we we, we

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

    -12, college, and post-college.⏎To view this DRM protected ebook on your desktop or laptop you will need to have Adobe Digital Editions installed. It is a free software. We also strongly recommend that you sign up for an

  2. #2 peak

    solving of nonmathematical problems as mathematical problems.⏎To view this DRM protected ebook on your desktop or laptop you will need to have Adobe Digital Editions installed. It is a free software. We also strongly recommend that you sign up for an

  3. #3 peak

    be God, the all-loving creator who meets every human need. In Psalms we read, “Give thanks to the Lord, for he is good; his love endures forever. Let the redeemed of the Lord tell their story. .

  4. #4 peak

    Animal Adaptations". Please register for the second session through Eventbrite as well, if you are interested in attending both: https://scisatadapt2.eventbrite.ca⏎Please contact us if you have any questions & we hope to

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. L4:7 excites
  2. L2:9 suppresses sentence-final periods vs dots in code
  3. L8:143 suppresses
  4. L5:155 suppresses
  5. L5:168 suppresses
  6. L1:130 suppresses
  7. L6:305 suppresses
  8. L6:972 excites "well" after "how"
  9. L5:1304 excites
  10. L4:1546 suppresses
  11. L11:132 excites period ending an opening topic sentence
  12. L6:250 weakly excites
  13. L2:1473 weakly excites
  14. L4:300 weakly excites
  15. L5:454 weakly suppresses
  16. L5:125 weakly excites "of" after all three, both, ten
  17. L0:36 weakly excites
  18. L9:7 weakly excites
  19. L1:1 weakly excites
  20. L11:1981 weakly suppresses
  21. L10:33 weakly suppresses
  22. L9:240 weakly excites
  23. L3:272 weakly excites
  24. L0:1042 weakly suppresses "This" starting a labelled sentence

Feeds later layers

  1. L15:344 excites
  2. L15:403 excites period inside an abbreviation
  3. L17:642 excites
  4. L16:432 excites period after month or initial
  5. L17:347 excites full stops ending informative sentences
  6. L17:1770 excites
  7. L18:791 excites
  8. L17:1772 excites
  9. L15:241 excites
  10. L15:117 suppresses
  11. L15:313 excites decimal point inside numbers
  12. L19:831 excites
  13. L16:1545 excites
  14. L18:2497 excites hyphen inside compound words
  15. L17:1622 suppresses
  16. L27:133 excites
  17. L27:12 excites
  18. L18:713 excites
  19. L17:1155 excites opening parenthesis after a sentence
  20. L20:962 excites decimal points and dots in numbers
  21. L20:505 weakly excites
  22. L21:1237 weakly excites
  23. L22:67 weakly excites
  24. L16:797 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.

69 values between -0.87 and -0.561325 values between -0.56 and -0.242249 values between -0.24 and 0.07127 values between 0.07 and 0.3928 values between 0.39 and 0.7020 values between 0.70 and 1.0113 values between 1.01 and 1.3312 values between 1.33 and 1.6416 values between 1.64 and 1.9512 values between 1.95 and 2.2714 values between 2.27 and 2.586 values between 2.58 and 2.896 values between 2.89 and 3.213 values between 3.21 and 3.521 values between 3.52 and 3.832 values between 4.46 and 4.771 values between 4.77 and 5.092 values between 5.09 and 5.401 values between 6.65 and 6.961 values between 11.35 and 11.66offnegativepositiveoften

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. ugal
  2. мот
  3. 間
  4. рай
  5. alternative
  6. let
  7. 朋友们
  8. 례
  9. ев
  10. Nature

Suppresses

  1. MainThread
  2. ��
  3. lambda
  4. abilia
  5. egin
  6. backs
  7. nable
  8. ifies
  9. �
  10. yor