Neuron Explorer

Qwen3-0.6B

period inside an abbreviation

A neuron in layer 15 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 in short forms like E. coli, Univ. Arizona, B.C. or U.S., where the dot does not end a sentence. It is quiet or negative on question marks and full stops that really do end a sentence. Turning it up gave rambling text that did not clearly 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

    er.2008 May 27⏎Credit: Phoenix, HiRISE, NASA, JPL-Caltech, Univ. Arizona⏎Explanation: The Phoenix lander's footpads are about the size

  3. #3 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

  4. #4 peak

    , observance of a pause in between was necessary. This we have already explained in the introduction to Surah Al Muddaththir; and in E. N. 5 of Surah Al-Muzzammil also we have explained

  5. #5 peak

    . 3-4.5m high. These are covered in deciduous woodland. On the east side the camp defences consist of an inner bank C. 0.5-1m high on inside fence, and well-preserved

  6. #6 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

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 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

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

    bullets, but few⏎of the horses escaped. Tender-hearted gunners took time out to<|endoftext|>Is PET Plastic Safe for Food and Beverage?⏎Is PET plastic safe?

  2. #2 peak

    eventual climate impacts<|endoftext|>Transforming Canada’s Food Guide: What’s missing?⏎In the aim to provide a full diet, what does Canada’s Food Guide leave out? Take

  3. #3 peak

    essential. This comprehensive guide aims to provide a clear understanding of what a bilan comptable entails, including its key components such as assets, liabilities, and equity. By delving into the importance of the bilan comptable for financial analysis and

  4. #4 peak

    minster Township, and it consisted of 105 acres of land, part of<|endoftext|>How do you bring an endangered plant species back from the brink of extinction? The answer might be found in zoo animals.⏎That’s the inspiration for Chicago Bot

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. L14:422 excites periods after initials and abbreviations
  2. L14:155 excites sentence-ending periods in prose
  3. L14:51 excites colon that introduces a statement
  4. L4:7 excites
  5. L5:155 suppresses
  6. L5:168 suppresses
  7. L11:132 excites period ending an opening topic sentence
  8. L9:2447 excites
  9. L6:305 weakly suppresses
  10. L8:541 weakly excites
  11. L9:7 weakly excites
  12. L6:250 weakly excites
  13. L7:64 weakly suppresses
  14. L14:634 weakly excites
  15. L2:9 weakly suppresses sentence-final periods vs dots in code
  16. L4:1546 weakly suppresses
  17. L10:612 weakly excites full stop ending boilerplate or a title
  18. L8:143 weakly suppresses
  19. L8:740 weakly excites
  20. L13:53 weakly excites period ending a claim before elaboration
  21. L13:95 weakly excites
  22. L0:2602 weakly excites
  23. L12:5 weakly excites
  24. L10:921 weakly excites

Feeds later layers

  1. L17:1772 suppresses
  2. L16:1545 suppresses
  3. L21:1086 suppresses end of sentence before a claim or statistic
  4. L20:473 suppresses
  5. L17:1105 suppresses
  6. L17:347 suppresses full stops ending informative sentences
  7. L19:384 suppresses
  8. L19:484 suppresses
  9. L25:198 suppresses
  10. L24:353 suppresses decimal point inside a number
  11. L17:231 excites
  12. L16:489 excites
  13. L17:990 suppresses
  14. L23:179 suppresses
  15. L18:227 excites
  16. L20:822 suppresses
  17. L27:133 suppresses
  18. L16:339 suppresses
  19. L27:1637 suppresses
  20. L20:296 suppresses
  21. L22:282 suppresses period then line break in lists
  22. L20:343 suppresses
  23. L20:940 suppresses
  24. L22:67 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 -9.87 and -9.552 values between -9.24 and -8.921 values between -8.92 and -8.611 values between -8.61 and -8.292 values between -8.29 and -7.981 values between -7.98 and -7.661 values between -7.35 and -7.031 values between -7.03 and -6.726 values between -6.72 and -6.404 values between -6.40 and -6.0910 values between -6.09 and -5.776 values between -5.77 and -5.465 values between -5.46 and -5.147 values between -5.14 and -4.834 values between -4.83 and -4.516 values between -4.51 and -4.207 values between -4.20 and -3.883 values between -3.88 and -3.565 values between -3.56 and -3.252 values between -3.25 and -2.932 values between -2.93 and -2.624 values between -2.62 and -2.302 values between -2.30 and -1.991 values between -1.99 and -1.674 values between -1.67 and -1.363 values between -1.36 and -1.045 values between -1.04 and -0.7311 values between -0.73 and -0.4161 values between -0.41 and -0.10766 values between -0.10 and 0.221788 values between 0.22 and 0.53930 values between 0.53 and 0.85229 values between 0.85 and 1.1622 values between 1.16 and 1.484 values between 1.48 and 1.791 values between 2.42 and 2.74offnegativepositiveoften

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. 缮
  2. .&
  3. МА
  4. wind
  5. об
  6. vron
  7. ma
  8. 肃
  9. RequestMethod
  10. .`

Suppresses

  1. ernet
  2. 不过
  3. -*-
  4. rän
  5. Each
  6. Also
  7. ään
  8. moreover
  9. ⏎→→⏎
  10. ⏎⏎