Neuron Explorer

Qwen3-0.6B

'from' after 'range' or 'everything'

A neuron in layer 1 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 word 'from' when it starts the first item of a list of things, as in 'ranging from X to Y' or 'everything from A to B'. When it was turned up, the writing did not clearly change theme, so its effect on text is hard to see.

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

    , branching out into numerous fields outside the personal computer.⏎Microprocessors have revolutionized the world, especially in the area of electronics. A myriad of modern items ranging from cell phones to digital watches, elevators to washing machines contain microprocessors. It

  2. #2 peak

    is the future of content creation. In this blog post, we will explore some ways AI can help you create better content for your blog. We will cover everything from finding your niche to writing and editing your content. By the end of this post

  3. #3 peak

    depends on chemical processes that take longer to perfect than the systems that brought cheap electronics, finding a solution to the shortcomings of batteries has the potential to revolutionize everything from transportation to hand-held gadgets.⏎“We picked the path with the biggest risks and the

  4. #4 peak

    of erythropoietin used in these experiments did not elevate red blood cell counts. Such a treatment has obvious benefits for a wide range of health problems ranging from Alzheimer's to obesity, including mental health disorders for which increased physical activity is known

  5. #5 peak

    exams and research is crucial to your success at university. Our wide range of resources will help you achieve your goals. Our resources will provide you with strategies for everything from reading and note-taking to time management and exams.⏎Whether you've never thought about

  6. #6 peak

    venture beyond rudimentary and essentialist arguments and representations. These approaches range from arguing that religion and/or nationalism were the main factors leading to the Armenian Genocide

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 have to do with my friend, and I have to do with my friend. So, I have to do with my friend, and I have to do with my friend. I have to do with my friend, and I have to

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

    to be controlled by a series of impersonal adults, she assumes a tough exterior to protect herself against further emotional pain. Her abusive behavior and rejection of friendly overtures keep others at a distance; she avenges her own pain by

  2. #2 peak

    .⏎Plotinus, having been a proponent of Platonism, naturally concurs with Plato, postulating that each of us is comprised of an everyday, embodied experience (the physical body), and an immortal, incorporeal spirit (the

  3. #3 peak

    a global dialogue on open science⏎UNESCO is leading a global dialogue on Open Science, to ensure that Open Science practices meet their potential in bridging the world’s STI gaps and enabling sustainable development. The Member States of the Organisation have examined

  4. #4 peak

    need to do our part now in order to pass it on intact to our future generations.⏎Europeans should be proud of their ancestors who have done many wondrous things including those who painted in caves and on cliffs in Europe.⏎The photo of the

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. L0:2895 weakly suppresses
  2. L0:1301 weakly suppresses
  3. L0:1224 weakly suppresses
  4. L0:178 weakly suppresses
  5. L0:462 weakly suppresses
  6. L0:722 weakly excites
  7. L0:1835 weakly excites
  8. L0:2406 weakly suppresses
  9. L0:1919 weakly excites
  10. L0:686 weakly suppresses sentence-starting 'This' at text start
  11. L0:2967 weakly excites word after sentence-initial 'You'
  12. L0:1889 weakly suppresses
  13. L0:839 weakly excites
  14. L0:1821 weakly excites
  15. L0:334 weakly excites
  16. L0:513 weakly suppresses
  17. L0:1302 weakly excites
  18. L0:1212 weakly suppresses
  19. L0:1453 weakly excites
  20. L0:2741 weakly suppresses
  21. L0:680 weakly excites
  22. L0:2784 weakly suppresses
  23. L0:2771 weakly excites
  24. L0:2026 weakly excites

Feeds later layers

  1. L9:2434 suppresses 'of' in department and division names
  2. L11:843 excites
  3. L11:1072 weakly excites
  4. L3:579 weakly excites
  5. L15:2875 weakly suppresses
  6. L17:2951 weakly suppresses
  7. L24:3027 weakly suppresses
  8. L13:1330 weakly excites
  9. L9:1290 weakly suppresses
  10. L25:988 weakly suppresses
  11. L7:278 weakly suppresses
  12. L11:1686 weakly suppresses
  13. L14:760 weakly excites
  14. L17:516 weakly excites
  15. L10:2717 weakly excites
  16. L6:2722 weakly excites
  17. L13:3062 weakly suppresses
  18. L13:689 weakly suppresses
  19. L13:1049 weakly excites
  20. L11:351 weakly excites
  21. L17:1401 weakly suppresses
  22. L14:2332 weakly suppresses
  23. L10:888 weakly suppresses
  24. L11:2677 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.

1 values between -1.29 and -1.183 values between -1.18 and -1.073 values between -1.07 and -0.965 values between -0.96 and -0.858 values between -0.85 and -0.744 values between -0.74 and -0.6315 values between -0.63 and -0.5235 values between -0.52 and -0.4148 values between -0.41 and -0.30107 values between -0.30 and -0.19413 values between -0.19 and -0.081834 values between -0.08 and 0.031085 values between 0.03 and 0.14264 values between 0.14 and 0.2536 values between 0.25 and 0.3613 values between 0.36 and 0.4710 values between 0.47 and 0.588 values between 0.58 and 0.695 values between 0.69 and 0.801 values between 0.80 and 0.912 values between 0.91 and 1.021 values between 1.02 and 1.131 values between 1.24 and 1.352 values between 1.57 and 1.681 values between 1.68 and 1.791 values between 1.79 and 1.901 values between 2.01 and 2.121 values between 3.00 and 3.11offnegativepositiveoften

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. !"⏎
  2. !'⏎
  3. 建设项目
  4. prech
  5. 总局
  6. にお
  7. Boards
  8. ","⏎
  9. !"⏎⏎
  10. 意见反馈

Suppresses

  1. ограничен
  2. hỏ
  3. 拓
  4. みました
  5. себе
  6. Inspectable
  7. достат
  8. Formatting
  9. limited
  10. stown