Neuron Explorer

Qwen3-0.6B

sentence-ending periods in prose

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

The neuron lights up on full stops that close a sentence in ordinary essay-style writing, often at the end of a clause that introduces a topic. It goes quiet or negative on colons and line breaks in headings and short labels. Turning it up made the writing repetitive but did not show a clear theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    been thinking deeply about the alphabet—about which letter-shaped path of recovery the world will follow out of its recession. The economy could rebound sharply, tracing a V; it could trudge along the bottom a while longer before rising slowly into a U

  2. #2 peak

    , but it’s always happening.⏎Animation is no longer a novelty for web designers…it’s becoming the basis of effective interaction design.⏎Animation is change, and motion. It’s the closest we get to capturing

  3. #3 peak

    imaginative state of Greece in her best period, represented wealth, commerce, and conquest, in a greater degree, so were her arts, and with these the lyric. In her best state her nobles danced, Appius Claudius excelled,

  4. #4 peak

    up a depot for the provision of ships under the auspices of The Dutch East India Company, were of Dutch stock. Many later settlers were of German origin and after the revocation of the Edict

  5. #5 peak

    family members, or student partners.⏎- The students' stories can be compiled into a slide show or movie.<|endoftext|>Hyperion: A Fragment 1818 (1820

  6. #6 peak

    ⏎to his Life he was also called Kung-tê-jih (i.e., merit-sun; in Sanskrit, Puṇyaditya?). For he was not only a philosopher, but a preacher and an organiser,

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: that the "2011,2012" "2013" "2015" "2017" "2019" "2011" "2011

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

    Seeking Help:⏎Ben Locke, Ph.D., a mental health expert serving at Pennsylvania State University,

  2. #2 peak

    news articles may be copied or forwarded for individual use only. No other reproduction or distribution is permitted without prior written consent.⏎Please login or Register to leave a comment.<|endoftext|>We’ve long known that racism and discrimination negatively impact the mental health and well

  3. #3 peak

    elements of a successful speech. Not only do they get to become better-informed...⏎7th - 11th Social Studies & History⏎President Obama's State of the Union Address⏎Students watch the State of the Union

  4. #4 peak

    Therefore, children may need explicit instruction in recognizing alternative interpretations in their writing. (A sample story and four figures of data are included.) (RS)⏎Publication Type: Speeches/Meeting Papers; Reports - Research⏎Education Level: N/A⏎

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. L11:132 excites period ending an opening topic sentence
  2. L5:168 suppresses
  3. L10:612 excites full stop ending boilerplate or a title
  4. L6:250 excites
  5. L9:16 suppresses unclear: periods inside numbers and names
  6. L4:7 excites
  7. L8:541 excites
  8. L9:12 suppresses
  9. L9:2447 excites
  10. L9:2302 excites
  11. L8:740 excites
  12. L13:345 excites
  13. L12:2141 suppresses
  14. L10:478 suppresses
  15. L4:102 excites
  16. L3:259 suppresses
  17. L11:1194 excites
  18. L9:819 suppresses
  19. L5:155 weakly suppresses
  20. L10:945 weakly excites
  21. L4:51 weakly excites
  22. L11:1272 weakly excites unclear: sentence ends in factual prose
  23. L4:562 weakly suppresses blank slot between a date's day and year
  24. L2:9 weakly suppresses sentence-final periods vs dots in code

Feeds later layers

  1. L17:990 excites
  2. L16:1545 excites
  3. L17:1155 excites opening parenthesis after a sentence
  4. L19:94 suppresses
  5. L15:403 excites period inside an abbreviation
  6. L19:2097 excites sentence-end punctuation in speech
  7. L16:489 suppresses
  8. L17:1105 excites
  9. L16:894 excites full stops ending informational sentences
  10. L15:1207 excites
  11. L21:524 excites
  12. L17:1772 excites
  13. L17:231 suppresses
  14. L16:949 excites
  15. L15:21 suppresses
  16. L17:347 excites full stops ending informative sentences
  17. L17:1044 excites
  18. L19:2697 excites
  19. L19:909 excites
  20. L18:1465 excites semicolon joining clauses
  21. L16:436 excites
  22. L19:433 excites
  23. L16:1608 suppresses
  24. L16:797 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 -2.06 and -1.841 values between -1.20 and -0.9925 values between -0.99 and -0.77412 values between -0.77 and -0.561406 values between -0.56 and -0.351283 values between -0.35 and -0.13420 values between -0.13 and 0.0896 values between 0.08 and 0.3052 values between 0.30 and 0.5133 values between 0.51 and 0.7217 values between 0.72 and 0.9419 values between 0.94 and 1.1510 values between 1.15 and 1.3611 values between 1.36 and 1.5816 values between 1.58 and 1.7910 values between 1.79 and 2.015 values between 2.01 and 2.226 values between 2.22 and 2.438 values between 2.43 and 2.657 values between 2.65 and 2.869 values between 2.86 and 3.086 values between 3.08 and 3.299 values between 3.29 and 3.503 values between 3.50 and 3.727 values between 3.72 and 3.937 values between 3.93 and 4.146 values between 4.14 and 4.367 values between 4.36 and 4.573 values between 4.57 and 4.792 values between 4.79 and 5.006 values between 5.00 and 5.211 values between 5.21 and 5.431 values between 5.64 and 5.861 values between 5.86 and 6.071 values between 6.07 and 6.281 values between 6.28 and 6.50offnegativepositiveoften

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. 正当
  2. 気が
  3. I
  4. This
  5. 在这个
  6. A
  7. trl
  8. 据了解
  9. itu
  10. 随着

Suppresses

  1. %",⏎
  2. ;",⏎
  3. !",⏎
  4. :",⏎
  5. orns
  6. $",
  7. /she
  8. /",⏎
  9. \",\
  10. '{$