Neuron Explorer

Qwen3-0.6B

line breaks inside sentences

A neuron in layer 5 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 a line break that cuts a sentence in the middle, as when text is wrapped to a new line in the middle of a thought. Turning it up gave rambling text with no clear theme.

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

    largely worked. Lake Champlain is navigable along its west border,⏎and sloops can ascend Otter creek 7 miles to Vergennes. The railroad connecting⏎Bellows Falls with Burlington passes through this county. Organized in⏎17

  2. #2 peak

    these chemical signals, in⏎principle, are vulnerable to disruption. Scientists, for example,⏎have just begun to look at the chemical signals that mediate communication⏎between symbiotic organisms, such as nitrogen-fixing bacteria⏎and the roots of

  3. #3 peak

    sources,⏎some quite conspicuous and others completely unexpected.⏎of the synthetic chemical compounds are notorious because they⏎accumulate in our fatty tissue. These chemicals can then bioaccumulate⏎up the food chain, and can be passed on<|endoftext|>Dry, flaky skin

  4. #4 peak

    ⏎ask? In reality, as long as there are vehicular accidents involving children⏎(in any kind of vehicle), our nation will continue to debate, make recommendations,⏎and try to find ways to ensure the safe transport of our children.⏎On March

  5. #5 peak

    seen that expand to include estrogen⏎blockers, androgen blockers, progesterone blockers and compounds⏎that interfere with thyroid. This last one is especially important⏎because thyroid hormone is key to proper brain development.⏎not just the disruption of the

  6. #6 peak

    debt with the region of Kerala⏎in India, which used decentralized government programs to provide⏎education, welfare, and health care. The system worked astoundingly⏎well, leaving the region with the highest literacy and life expectancy⏎rates in India

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: 1, 2, 3, 4, 5, 6, 10, 3, 1, 1, 1, 3, 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

    Researchers in Iceland led by Decode Genetics' Kari Stefansson carried out a sequencing and population screening effort of people in the country at highest risk for SARS-CoV-2, and found that children under 10 years of age and women

  2. #2 peak

    that the haplotypes of the sequenced SARS-CoV-2 viruses were diverse and changed over time, but that the percentage of infected

  3. #3 peak

    ano technologies for disinfection⏎- Micro/nano technologies for SARS-CoV-2 nucleic acid testing⏎- Micro/nano technologies for SARS-CoV-2 antibody testing⏎- Micro/nano technologies for environmental monitorin⏎-

  4. #4 peak

    molecular test is being used as the gold standard method for detecting SARS-CoV-2, its complexity limits its use at the point of need. There is

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. L4:337 excites
  2. L2:1156 suppresses
  3. L2:235 weakly excites
  4. L2:1886 weakly excites
  5. L3:1613 weakly suppresses
  6. L2:2818 weakly suppresses
  7. L4:1204 weakly excites
  8. L0:799 weakly suppresses
  9. L0:55 weakly excites
  10. L4:1247 weakly suppresses
  11. L4:198 weakly suppresses
  12. L3:740 weakly excites
  13. L3:124 weakly suppresses
  14. L3:428 weakly excites
  15. L3:2857 weakly suppresses the word 'fiber' or 'fibre'
  16. L3:1676 weakly excites
  17. L3:1167 weakly suppresses
  18. L2:790 weakly suppresses
  19. L3:2625 weakly excites
  20. L4:71 weakly suppresses
  21. L3:1780 weakly suppresses
  22. L4:863 weakly suppresses
  23. L4:240 weakly suppresses
  24. L2:2078 weakly suppresses

Feeds later layers

  1. L10:2060 weakly suppresses
  2. L7:823 weakly suppresses
  3. L11:1726 weakly suppresses
  4. L15:438 weakly suppresses
  5. L15:541 weakly suppresses
  6. L11:2578 weakly suppresses
  7. L7:1856 weakly excites
  8. L8:2156 weakly excites
  9. L23:718 weakly suppresses
  10. L6:283 weakly excites
  11. L10:2194 weakly suppresses
  12. L6:2089 weakly suppresses
  13. L17:2169 weakly excites
  14. L11:2891 weakly suppresses
  15. L13:1896 weakly suppresses
  16. L15:2396 weakly excites
  17. L14:1547 weakly excites
  18. L15:753 weakly suppresses
  19. L6:2201 weakly suppresses
  20. L8:1171 weakly suppresses
  21. L12:309 weakly excites
  22. L14:2186 weakly suppresses
  23. L14:2495 weakly excites
  24. L16:308 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 -1.121 values between -1.12 and -1.051 values between -0.90 and -0.831 values between -0.83 and -0.761 values between -0.54 and -0.473 values between -0.47 and -0.4018 values between -0.40 and -0.3248 values between -0.32 and -0.25174 values between -0.25 and -0.18406 values between -0.18 and -0.11787 values between -0.11 and -0.031208 values between -0.03 and 0.04739 values between 0.04 and 0.11329 values between 0.11 and 0.19124 values between 0.19 and 0.2635 values between 0.26 and 0.3317 values between 0.33 and 0.404 values between 0.40 and 0.482 values between 0.48 and 0.551 values between 0.55 and 0.622 values between 0.62 and 0.691 values between 0.69 and 0.771 values between 1.06 and 1.132 values between 1.13 and 1.201 values between 1.42 and 1.491 values between 1.64 and 1.71offnegativepositiveoften

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. IGH
  2. prosec
  3. ihad
  4. gio
  5. WI
  6. oward
  7. statt
  8. pliant
  9. hips
  10. asto

Suppresses

  1. REV
  2. /Header
  3. ệt
  4. k
  5. .css
  6. -dess
  7. 麂
  8. ableView
  9. (InputStream
  10. ket