Neuron Explorer

Qwen3-0.6B

'of' in US agency names

A neuron in layer 10 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 reacts to the word 'of' inside the names of American health agencies, such as 'National Institute of' or 'National Center for'. Turning it up did not bring out that topic in the model's writing.

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

    approach to ?xml:namespace>⏎Producers based in Europe, the⏎Producers have put forward their

  2. #2 peak

    Rare Disorders⏎Canadian AIDS Society⏎CORD—Canadian Organization for Rare Disorders⏎NINDS progressive multifocal leukoencephalopathy information page. National Institute of Neurological Disorders and Stroke website. Available at http://www.ninds.nih.gov

  3. #3 peak

    by National Institutes of Health-funded Microbicide Trials Network , a worldwide collaborative clinical trials network focused on preventing the sexual transmission of HIV.⏎The NIH’s National Institute of Allergy and Infectious Diseases issued this statement about the

  4. #4 peak

    However, there is still no consensus on what exactly happens when Sars-CoV-2 invades the brain. Walter J., director of the US National Institute of Neurological Disorders and Stroke in Bethesda, Maryland. Korochetz said in a

  5. #5 peak

    health is awesome (and I’m not just saying that). Thanks to the achievements of public health, we live longer and healthier lives. According to the National Center for Health Statistics, “life expectancy at birth among U.S. residents increased by

  6. #6 peak

    ickle Cell Disease Association of America⏎This group provides education, advocacy, and other initiatives to promote awareness of and support for sickle cell disease programs.⏎National Institute of Neurological Disorders and Stroke (NINDS)⏎NINDS offers research information

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 the following two expressions: 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1

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

    sun-exposed crowns, such as leaf water potential and photosynthesis.⏎Both tree species showed significant drought effects<|endoftext|>Introduction to Kubernetes⏎Kubernetes, also known as K8s or Kube, is an open-source system that helps organizations manage

  2. #2 peak

    power N is cxN/(1-x) ≈ 40 ⋅ 0.97N.⏎Let us look at A0073

  3. #3 peak

    zipping the .tgz files using the terminal. Let’s get into these.⏎Note: For this particular tutorial, we will use the Ubuntu 22.04. However, the commands used here are same for all the distributions.⏎

  4. #4 peak

    , which looms on the horizon north of the major city of Yogyakarta, is on the island of Java.⏎President Susilo Bambang Yudhoyono has called the volcanic eruptions a crisis situation. He and several

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. L8:111 suppresses the space before a year in dates
  2. L8:25 weakly suppresses
  3. L7:26 weakly suppresses
  4. L8:1527 weakly suppresses
  5. L0:2893 weakly suppresses
  6. L9:13 weakly excites
  7. L8:2430 weakly suppresses
  8. L0:1510 weakly excites
  9. L8:749 weakly suppresses
  10. L8:447 weakly suppresses
  11. L8:861 weakly suppresses
  12. L9:746 weakly suppresses
  13. L8:1876 weakly excites
  14. L9:330 weakly suppresses gap where a number was removed
  15. L8:149 weakly suppresses
  16. L8:282 weakly suppresses
  17. L8:2006 weakly excites
  18. L6:4 weakly suppresses
  19. L7:8 weakly excites
  20. L3:492 weakly excites
  21. L6:428 weakly excites
  22. L9:401 weakly excites
  23. L8:1451 weakly suppresses
  24. L1:1474 weakly excites

Feeds later layers

  1. L17:876 excites
  2. L11:222 weakly suppresses
  3. L24:1486 weakly excites
  4. L16:1262 weakly suppresses
  5. L11:2138 weakly excites
  6. L13:52 weakly suppresses
  7. L18:2695 weakly excites
  8. L13:787 weakly suppresses
  9. L17:880 weakly suppresses
  10. L26:2872 weakly excites
  11. L21:2768 weakly excites
  12. L14:135 weakly suppresses
  13. L13:3027 weakly excites
  14. L13:1406 weakly excites
  15. L16:263 weakly suppresses
  16. L16:907 weakly suppresses
  17. L17:599 weakly suppresses
  18. L14:39 weakly suppresses
  19. L24:776 weakly suppresses
  20. L18:2869 weakly excites
  21. L25:598 weakly excites
  22. L14:2286 weakly suppresses
  23. L24:1328 weakly excites
  24. L23:1040 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.

1 values between -3.35 and -3.191 values between -2.88 and -2.731 values between -2.57 and -2.412 values between -2.26 and -2.101 values between -2.10 and -1.951 values between -1.95 and -1.795 values between -1.79 and -1.631 values between -1.63 and -1.489 values between -1.48 and -1.326 values between -1.32 and -1.176 values between -1.17 and -1.019 values between -1.01 and -0.8510 values between -0.85 and -0.7014 values between -0.70 and -0.5446 values between -0.54 and -0.39215 values between -0.39 and -0.23766 values between -0.23 and -0.071926 values between -0.07 and 0.08690 values between 0.08 and 0.24159 values between 0.24 and 0.4016 values between 0.40 and 0.5511 values between 0.55 and 0.714 values between 0.71 and 0.863 values between 0.86 and 1.021 values between 1.02 and 1.181 values between 1.18 and 1.331 values between 1.49 and 1.641 values between 1.80 and 1.961 values between 2.74 and 2.89offnegativepositiveoften

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. ***/⏎
  4. )").
  5. .getTransaction
  6. ".⏎
  7. .").
  8. ,",
  9. ")).
  10. ,['

Suppresses

  1. oldt
  2. ски
  3. ework
  4. ull
  5. ild
  6. ştir
  7. ้าง
  8. asset
  9. illis
  10. assin