Neuron Explorer

Qwen3-0.6B

people nouns after a modifier

A neuron in layer 2 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 for groups of people, such as 'people', 'patients', 'parents' and 'children', especially when a word like 'closely', 'less time' or 'each year' comes just before them. Turning it up did not produce a clear, related change 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

    for Alzheimer's and dementia.⏎Heidi Wengreen, RD, PhD, Assistant Professor of Nutrition at Utah State University, and colleagues examined associations between how closely people adhered to the DASH diet and risk of cognitive decline and dementia among older

  2. #2 peak

    how long they will be sick, exposed to dialysis or ventilators. The less time patients spend in there the better off they are.”⏎For patients who spend several weeks in

  3. #3 peak

    established.<|endoftext|>The MOST Theological Collection: Mary in Our Life⏎"Questions for Discussion"⏎For private, class

  4. #4 peak

    an individual school and on a system-wide basis.⏎Some of the challenges to working with parents:⏎1. By nature, they are a moving target: each year parents enter and depart the school system as they sequence from elementary, to middle, to

  5. #5 peak

    barriers brought partly by COVID-19, the education system and its governing bodies can offer strategic lessons and good practices in almost any situation. Wherever education is disrupted for a protracted period, the three lessons mentioned below can actively ensure

  6. #6 peak

    traditions. Compliance with tribal norms held people together strongly, and enabled the tribe to deal effectively with external threats. If you

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 word "a" in the word "b" in the word "c" The word "d" The word "s" and the word "d" The word "s" and the word "d" The word "s

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

    able to share that with their children gave parents a feeling of normalcy, and made some feel more in control of the situation they were in.⏎“As health professionals, we

  2. #2 peak

    do the homework and don’t ask questions, or engage in discussion. Instead, they just watch or listen and let knowledge wash over them. The challenge of teaching passive learners is especially apparent for kindergarteners, who have not yet developed independent

  3. #3 peak

    6 Examination.⏎We had mentioned some tips for cracking the HSC Rajasthan Board exam here: HSC Study Tips to

  4. #4 peak

    patch to generate a faithful copy of the other.⏎Unfortunately, some changes to directory trees cannot be expressed using⏎current patch formats; also,⏎patch does not handle some of the⏎existing formats. These shortcomings motivate the following suggested⏎patch do not

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. L1:254 excites
  2. L1:386 excites
  3. L1:201 suppresses
  4. L1:1874 suppresses
  5. L1:266 weakly excites
  6. L1:958 weakly suppresses
  7. L1:1752 weakly suppresses
  8. L1:535 weakly excites
  9. L1:1055 weakly excites
  10. L1:1495 weakly excites
  11. L1:2352 weakly suppresses
  12. L1:384 weakly excites
  13. L1:1231 weakly suppresses
  14. L1:848 weakly suppresses
  15. L0:2129 weakly excites
  16. L1:969 weakly excites
  17. L1:861 weakly excites
  18. L1:127 weakly suppresses
  19. L0:2616 weakly suppresses
  20. L1:302 weakly excites
  21. L1:632 weakly excites
  22. L1:456 weakly suppresses
  23. L1:1870 weakly suppresses
  24. L1:807 weakly excites

Feeds later layers

  1. L3:472 excites
  2. L4:250 excites
  3. L3:195 excites
  4. L3:352 excites
  5. L3:126 excites
  6. L3:997 excites
  7. L7:310 suppresses
  8. L3:84 suppresses
  9. L15:2616 suppresses
  10. L8:917 excites
  11. L3:426 suppresses
  12. L4:286 excites
  13. L3:226 weakly suppresses
  14. L12:710 weakly excites
  15. L6:738 weakly excites
  16. L3:1124 weakly suppresses
  17. L7:401 weakly suppresses
  18. L3:637 weakly suppresses
  19. L13:78 weakly excites
  20. L27:932 weakly suppresses
  21. L18:2401 weakly suppresses
  22. L8:283 weakly excites
  23. L4:319 weakly excites
  24. L13:961 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 -1.19 and -1.131 values between -0.94 and -0.881 values between -0.88 and -0.813 values between -0.81 and -0.752 values between -0.75 and -0.691 values between -0.69 and -0.622 values between -0.62 and -0.567 values between -0.56 and -0.5014 values between -0.50 and -0.4315 values between -0.43 and -0.3744 values between -0.37 and -0.31122 values between -0.31 and -0.24186 values between -0.24 and -0.18335 values between -0.18 and -0.12554 values between -0.12 and -0.051082 values between -0.05 and 0.01863 values between 0.01 and 0.07417 values between 0.07 and 0.14160 values between 0.14 and 0.2053 values between 0.20 and 0.2621 values between 0.26 and 0.337 values between 0.33 and 0.392 values between 0.39 and 0.453 values between 0.45 and 0.522 values between 0.52 and 0.582 values between 0.58 and 0.645 values between 0.64 and 0.711 values between 0.71 and 0.771 values between 1.09 and 1.151 values between 1.28 and 1.34offnegativepositiveoften

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. 鹄
  2. 为客户
  3. -sc
  4. mistakes
  5. ;break
  6. ate
  7. hacks
  8. -bo
  9. Ac
  10. -de

Suppresses

  1. ements
  2. ures
  3. [obj
  4. ilver
  5. Blond
  6. seulement
  7. qx
  8. ured
  9. CMD
  10. proj