Neuron Explorer

Qwen3-0.6B

non-English text, mid-phrase

A neuron in layer 23 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron mostly fires inside sentences written in French, Spanish or other non-English languages, often on a piece of a longer word. Its effect on the written output was hard to read: the steered text drifted between topics and languages with no 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

    roof generates more solar power than the school consumes, especially in the summer months and during weekends. The NPO APERe (Association pour la Promotion des Energies Renouvables) and Brussels grid operator Sibelga selected the school as a

  2. #2 peak

    Museum of Alberta, Musée national des arts et traditions populaires, Département de l'organisation des systèmes d'information

  3. #3 peak

    this milestone overtime because of our rigorous hiring process that filters professionals from average writers. We believe that exposing our customers to qualified writers is<|endoftext|>Investigación con células madre embrionarias humanas revela los primeros pasos en el desarrollo

  4. #4 peak

    adie pour 2014 – Rapport au Ministre chargé de la Sécurité Sociale et au Parlement sur l’évolution des charges et produits de l’Assurance Maladie au titre de 20

  5. #5 peak

    In the following example, hijos is a masculine, plural noun so the masculine plural direct object pronoun los is used to replace it in the second sentence.Son muy buenos hijos. No los cambiaría.⏎They are good

  6. #6 peak

    ics.org/incentives/020828-01.htm.⏎http://www.serna.gob.hn/ (secretaria de recursos naturales y de ambiente)<|endoftext|>SCAMwatch is aware that some consumers have

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: it's always so easy to set it as the only one, which is known as "first in" or "last in". It's also known as "active" or "active as it is". What's the difference between

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

    975) Theory of Reasoned Action (TRA) and Ajzens (Ajzen 1985) extended Theory of Planned Behavior (TPB) model. The TRA

  2. #2 peak

    85 CE), Japanese art flourished, and new forms of art, such as screen painting, began to emerge. The Kamakura period (1185 CE to 1333 CE) saw the rise of Zen Buddhism

  3. #3 peak

    frustrated and lose interest in it when it doesn’t show immediate commercial applications. I read just recently about how the science minister of Canada made statements about how the National Research Council isn’t going to pursue projects with immediate commercial applications and it makes me sad

  4. #4 peak

    To view this ebook on an iPhone, iPad or Android mobile device you will need the Adobe Digital Editions app, or BlueFire Reader or Txtr app. These are free, too. For more details see this article.⏎|Size: ||

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. L19:1422 strongly excites
  2. L20:807 strongly excites
  3. L21:607 excites
  4. L19:1743 excites
  5. L19:639 excites
  6. L20:240 suppresses
  7. L18:2391 excites
  8. L19:1424 excites
  9. L19:1427 suppresses
  10. L20:1842 suppresses
  11. L21:1376 suppresses
  12. L21:266 excites
  13. L8:253 suppresses
  14. L19:1695 excites
  15. L17:1583 excites
  16. L7:327 excites
  17. L9:1062 excites
  18. L18:444 suppresses
  19. L16:1418 excites
  20. L14:1390 excites
  21. L18:1433 excites
  22. L19:2131 excites
  23. L19:2382 suppresses
  24. L19:1909 excites

Feeds later layers

  1. L24:434 strongly excites
  2. L27:2186 strongly excites
  3. L25:335 strongly excites
  4. L25:812 strongly excites
  5. L24:360 suppresses
  6. L26:111 excites
  7. L27:60 suppresses
  8. L25:403 excites
  9. L24:109 suppresses
  10. L26:552 excites
  11. L27:477 excites
  12. L25:377 excites
  13. L24:419 excites
  14. L27:1010 excites
  15. L25:320 suppresses
  16. L24:489 suppresses
  17. L26:441 suppresses
  18. L27:1052 excites
  19. L25:464 suppresses
  20. L26:1017 excites
  21. L26:437 suppresses
  22. L27:227 excites
  23. L24:488 excites
  24. L27:307 excites

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 -2.69 and -2.501 values between -2.50 and -2.322 values between -1.95 and -1.772 values between -1.77 and -1.596 values between -1.59 and -1.4013 values between -1.40 and -1.2222 values between -1.22 and -1.0432 values between -1.04 and -0.8548 values between -0.85 and -0.67107 values between -0.67 and -0.49178 values between -0.49 and -0.30582 values between -0.30 and -0.122480 values between -0.12 and 0.07266 values between 0.07 and 0.2593 values between 0.25 and 0.4342 values between 0.43 and 0.6218 values between 0.62 and 0.809 values between 0.80 and 0.982 values between 0.98 and 1.172 values between 1.17 and 1.351 values between 2.08 and 2.271 values between 4.47 and 4.65offnegativepositiveoften

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. 們
  2. 家长们
  3. ackers
  4. majors
  5. 们
  6. ones
  7. 学生们
  8. targets
  9. 老师们
  10. (targets

Suppresses

  1. который
  2. щий
  3. ющая
  4. ющий
  5. которое
  6. Its
  7. метр
  8. енный
  9. ístico
  10. Its