Neuron Explorer

Qwen3-0.6B

unclear: words before 'and' or 'or'

A neuron in layer 18 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 fires on words that sit in the middle of a list or contrast, often just before 'and' or 'or', as in 'multidisciplinary engineering' or 'exercise or health program'. The pattern is loose. Turning it up made the writing repeat itself and drift.

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

    the American Society of Mechanical Engineers, ASME is a not-for-profit professional organization promoting the art, science and practice of mechanical and multidisciplinary engineering and allied sciences. ASME develops codes and standards that enhance public safety,

  2. #2 peak

    |Math Topics:||Graphing of Data|⏎© 1994- Drexel University. All rights reserved.⏎The Math Forum is a research and educational enterprise of the Drexel University School of Education.<|endoftext|>Maybe you’ve

  3. #3 peak

    (more or less). Thus began the era of dancing babies and other horrors best forgotten<|endoftext|>PROTEOMICS is the study of many proteins at once- or a description of proteins in a cell, tissue or cellular organelle on a global

  4. #4 peak

    of the theory, linking it to best practices in the classroom.??<|endoftext|>I conducted this project as part of my doctoral dissertation in 2018 and early 2019 in an art immersion school in Western Canada.⏎The goal

  5. #5 peak

    not intended to substitute for the diagnosis, treatment, or advice of a medical professional. Consult your medical professional for guidance before changing or undertaking any diet, exercise or health program, or before using any prescription or over-the-counter drug. Neither the author

  6. #6 peak

    . Sellers and buyers can share hundreds and thousands of kilometers. Any distance is always costs, logistics. And it is very difficult to understand exactly what exactly and in what volumes buyers of another region want

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 problem is a simple one, but the first question is a simple question that asked it at first, which was originally a simple question, but it was a question that was initially a simple question. For example, I was once asked 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

    morality of God, and see him as their leader and prime spiritual guide.⏎Although the overwhelming majority of Sufis, both pre-modern and modern, were and are adherents

  2. #2 peak

    actionable insights. By understanding the correlation between CPPs and product’s critical quality attributes (CQAs) – measured in real time in-line, on-line and /or at-line by suitable analytical instruments, the process can be controlled in real-time

  3. #3 peak

    Sensitive Person.⏎- This may mean that you fall on the right side of the bell curve as a gifted person.⏎- This may mean you do have or are mislabeled as having a mental illness such as Emotion Regulation Disorder, ADHD

  4. #4 peak

    they are, and that the issues they bring to therapy are not restricted to (or located) within the clients themselves, but rather are⏎What are the key concepts of

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. L16:2994 strongly suppresses
  2. L10:163 suppresses
  3. L10:74 excites
  4. L17:1575 excites
  5. L16:678 suppresses
  6. L15:1443 excites
  7. L16:1388 suppresses
  8. L15:784 excites
  9. L13:1554 excites
  10. L15:217 suppresses
  11. L17:2221 excites
  12. L14:745 suppresses
  13. L9:92 suppresses "and" or "or" joining two items
  14. L13:131 excites 'and' joining parallel phrases
  15. L4:57 suppresses
  16. L17:1537 suppresses
  17. L9:212 excites
  18. L10:1643 suppresses
  19. L14:1249 suppresses
  20. L16:1937 suppresses 'and' joining two time or quantity ranges
  21. L13:372 suppresses
  22. L16:1659 excites
  23. L16:582 excites
  24. L15:1459 excites

Feeds later layers

  1. L21:735 excites 'or' in ranges and alternatives
  2. L19:883 excites
  3. L19:1431 excites
  4. L19:1849 weakly excites
  5. L19:2711 weakly excites
  6. L25:104 weakly excites
  7. L19:1823 weakly excites 'and' inside organisation or paired names
  8. L19:2311 weakly excites
  9. L19:2562 weakly excites
  10. L22:1113 weakly suppresses
  11. L19:1675 weakly excites
  12. L23:2414 weakly excites
  13. L25:1694 weakly excites 'and' joining paired names or fields
  14. L21:1402 weakly excites
  15. L20:867 weakly excites
  16. L22:2115 weakly suppresses
  17. L26:124 weakly suppresses
  18. L21:1540 weakly excites
  19. L27:49 weakly suppresses
  20. L21:510 weakly suppresses
  21. L20:1458 weakly excites
  22. L20:2027 weakly excites
  23. L19:1673 weakly excites
  24. L23:2356 weakly 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.

3 values between -7.29 and -6.912 values between -6.91 and -6.531 values between -6.53 and -6.152 values between -6.15 and -5.771 values between -5.77 and -5.391 values between -5.01 and -4.634 values between -4.63 and -4.255 values between -4.25 and -3.864 values between -3.86 and -3.4812 values between -3.48 and -3.107 values between -3.10 and -2.7214 values between -2.72 and -2.3420 values between -2.34 and -1.9620 values between -1.96 and -1.5823 values between -1.58 and -1.2047 values between -1.20 and -0.82359 values between -0.82 and -0.431467 values between -0.43 and -0.051235 values between -0.05 and 0.33291 values between 0.33 and 0.71140 values between 0.71 and 1.0995 values between 1.09 and 1.4760 values between 1.47 and 1.8535 values between 1.85 and 2.2323 values between 2.23 and 2.6116 values between 2.61 and 3.005 values between 3.00 and 3.386 values between 3.38 and 3.761 values between 3.76 and 4.143 values between 4.14 and 4.524 values between 4.52 and 4.901 values between 5.66 and 6.041 values between 7.57 and 7.95offnegativepositiveoften

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. rew
  2. 蔼
  3. éro
  4. 伺
  5. люч
  6. 잉
  7. fé
  8. ış
  9. loser
  10. 其他的

Suppresses

  1. is
  2. iani
  3. eties
  4. etri
  5. ');?>"
  6. ։
  7. ationale
  8. Official
  9. internal
  10. angel