Neuron Explorer

Qwen3-0.6B

"a" or "the" opening an object noun phrase

A neuron in layer 20 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 small words "the" or "a" right after a verb or preposition, where a noun is about to follow, such as "showed the" or "write the". When it was turned up, the text drifted into repeated "a" and "a certain" phrasing, with no clear topic.

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

    /2 fasting, which involves eating regularly for 5 days, with 2 non-consecutive fast days.⏎- Time-restricted fasting, which limits eating within a certain window of time each day. This method appeals to many people because you can

  2. #2 peak

    and a new study indicating that drinking diet soft drinks may increase risk for stroke only makes us more worried about the mysterious effects of those calorie-free cans. Researchers tracked the habits and health of about 2,500 adults in Manhattan over a

  3. #3 peak

    considering the carbon-farming legislation. Introduced by senators Mike Braun (R-Indiana) and Debbie Stabenow (D-Michigan), the bill sets up the US Department of Agriculture to certify independent consultants for farms that want to participate in carbon

  4. #4 peak

    with your counselor and discuss⏎- With your parent's permission, locate at least five Web sites that are helpful for⏎your Scouting or other activities. Write the Internet addresses of these sites in⏎your log. Talk with your counselor or a

  5. #5 peak

    figured out who it was — Italian Prime Minister Silvio Berlusconi.⏎Todorov's next example illustrated how facial recognition can have serious consequences. He showed the students a photo of Darrell Edwards, a man who is currently fighting a

  6. #6 peak

    were laser cut (bottom right).⏎Task: For this final project exercise, students were given the constraint of working with a common and uncommon design material. The team highlighted in this

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 new a A) a A) a A) a A) a A) a A) a A) a A) a A) a A) a A) a

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

    of people with particular medical conditions to determine DNA variants linked to those diseases, said University of Washington genome scientist Dr. John Stamatoyannopoulos, lead author of a study examining the connection between gene regulation and disease published Wednesday in the journal Science

  2. #2 peak

    718<|endoftext|>Women who have kids later on in life may live longer, according to the findings of a recent study.⏎Following the birth of a woman’s last child, certain measurements

  3. #3 peak

    a plant-based diet can help you lose weight without counting calories and without ramping up your exercise routine,â saysNeal Barnard, M.D., lead author of the study, president of the Physicians Committee, and an adjunct associate professor of medicine

  4. #4 peak

    defines work order, i.e., the sequencing of tasks): This is work management.<|endoftext|>Small businesses are starting smaller and staying smaller. That’s the conclusion of a study released by entrepreneurship advocate Kauffman Foundation.⏎“Starting

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. L18:2960 strongly excites
  2. L19:1050 suppresses
  3. L14:2473 excites
  4. L17:763 suppresses
  5. L17:2320 excites
  6. L17:2728 excites
  7. L15:2008 excites
  8. L18:1740 suppresses
  9. L15:826 excites
  10. L15:29 suppresses
  11. L17:1470 excites
  12. L18:2323 suppresses
  13. L19:2438 suppresses
  14. L15:1715 excites
  15. L19:1610 suppresses
  16. L19:2756 suppresses
  17. L19:1475 excites
  18. L16:691 excites
  19. L19:1352 excites
  20. L18:1991 suppresses
  21. L18:735 suppresses
  22. L15:2603 weakly excites
  23. L19:2227 weakly suppresses
  24. L17:1983 weakly excites

Feeds later layers

  1. L22:103 strongly excites
  2. L25:45 strongly excites
  3. L27:797 strongly suppresses
  4. L27:663 strongly suppresses
  5. L24:2538 strongly suppresses
  6. L25:1586 strongly suppresses
  7. L24:823 strongly suppresses
  8. L21:1457 strongly excites
  9. L26:484 strongly suppresses
  10. L26:213 excites
  11. L27:709 suppresses
  12. L25:871 suppresses
  13. L27:231 excites
  14. L27:173 excites
  15. L25:1132 excites
  16. L27:854 suppresses
  17. L25:549 suppresses
  18. L26:411 excites
  19. L21:1265 suppresses
  20. L27:1044 suppresses
  21. L27:2396 suppresses
  22. L25:837 suppresses
  23. L21:1725 suppresses
  24. L26:677 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.

2 values between -11.79 and -11.271 values between -8.15 and -7.631 values between -7.11 and -6.591 values between -6.59 and -6.072 values between -6.07 and -5.551 values between -5.55 and -5.034 values between -5.03 and -4.515 values between -4.51 and -3.996 values between -3.99 and -3.478 values between -3.47 and -2.957 values between -2.95 and -2.4318 values between -2.43 and -1.9128 values between -1.91 and -1.3965 values between -1.39 and -0.87349 values between -0.87 and -0.352370 values between -0.35 and 0.17820 values between 0.17 and 0.69154 values between 0.69 and 1.2124 values between 1.21 and 1.7317 values between 1.73 and 2.2512 values between 2.25 and 2.772 values between 2.77 and 3.296 values between 3.29 and 3.813 values between 3.81 and 4.341 values between 4.34 and 4.861 values between 8.50 and 9.02offnegativepositiveoften

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. ittel
  2. 娃
  3. ��
  4. ясн
  5. ernes
  6. parable
  7. الوا
  8. رام
  9. orges
  10. 率达到

Suppresses

  1. a
  2. an
  3. 某种
  4. another
  5. 一个新的
  6. something
  7. 一个
  8. 一种
  9. something
  10. 某个