Neuron Explorer

Qwen3-0.6B

space right before an expected number

A neuron in layer 19 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 empty space just before a number, such as the gap after 'more than' or 'an estimated'. It seems to be getting ready for a figure to come next. Its negative side fires on the decimal point inside numbers like 0.001. Turning it up gave repetitive filler text, so the steering does not show this idea clearly.

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

    students. Astronomy Center is sponsored by the American Astronomical Society, the American Association of Physics Teachers, and the NSF-NSDL.⏎Science.gov searches over 60 databases and over 2200 selected websites from 15

  2. #2 peak

    in academic publishing, have you ever wondered how many academic journals are there in the world? Well, I do.⏎It is estimated that there are currently more than 30,000 academic journals, and the number continues to increase by

  3. #3 peak

    rates (~99%) with very low false-positive rates (~0.2%). In other words, these tests are about 99.0% sensitive and 99.8% specific.⏎With performance like that one might expect these to

  4. #4 peak

    encies. LibGuides: Commerce Research Guide: Where can find journal ...⏎Functioning as a research and publishing network, ScienceOpen offers open access to more than 28 million articles in all areas of science. Although you do need to register

  5. #5 peak

    -so-simple question leads to an even scarier follow-up: What happens to us?⏎Coral reefs are the richest ocean ecosystems we have — they support an estimated 25 percent of marine species. According to a 2014 report

  6. #6 peak

    vitamin D. Our sun beds produce UV light that is about 95% UVA and 5% UVB.⏎Note that no supplementation should be substituted for proper medical treatment

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 first two, three, or any, of the first, or, or, or, or, or, or, or, or, or, or, or, or, or, or, or, or, or, or

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

    mm) – this is one thousandth of a meter. In scientific notation, 1 millimeter is 1 – E3 m, which means 0.001 x 1

  2. #2 peak

    A single heat pulse will switch the magnetic orientation of a nanometer-sized magnetic grain in just a few picoseconds (1 picosecond is 0.000000000001 seconds). As a

  3. #3 peak

    % should take the following relationship: 1.000×25 / 100 , or , 1.000×0.25 .⏎Giving us a result of 250, this would be

  4. #4 peak

    ), the nickel ($0.05), the dime ($0.10), the quarter ($0.25), the half-dollar 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. L12:2885 weakly suppresses
  2. L11:71 weakly suppresses
  3. L18:1391 weakly suppresses
  4. L17:2453 weakly excites
  5. L10:287 weakly suppresses
  6. L9:1394 weakly suppresses number theory terms like 'number' and 'Gauss'
  7. L17:1773 weakly excites
  8. L1:1852 weakly suppresses
  9. L12:706 weakly excites
  10. L7:2699 weakly suppresses
  11. L2:431 weakly excites
  12. L12:1986 weakly excites
  13. L14:2166 weakly suppresses
  14. L12:496 weakly excites
  15. L6:3006 weakly excites
  16. L0:1807 weakly excites
  17. L17:2217 weakly excites
  18. L13:2726 weakly excites
  19. L6:2290 weakly suppresses
  20. L13:2928 weakly suppresses
  21. L5:1079 weakly excites
  22. L10:1423 weakly suppresses
  23. L10:2820 weakly excites
  24. L15:2758 weakly suppresses

Feeds later layers

  1. L22:1638 suppresses
  2. L21:1954 suppresses
  3. L22:1494 suppresses
  4. L21:1159 weakly suppresses
  5. L22:2597 weakly excites
  6. L23:929 weakly suppresses
  7. L23:2924 weakly suppresses
  8. L20:675 weakly excites
  9. L20:2351 weakly excites
  10. L25:783 weakly excites
  11. L23:478 weakly suppresses
  12. L23:2850 weakly excites
  13. L23:1534 weakly excites
  14. L22:1210 weakly suppresses
  15. L20:1691 weakly excites
  16. L26:2351 weakly suppresses
  17. L24:1960 weakly suppresses
  18. L20:1337 weakly suppresses
  19. L20:1103 weakly excites
  20. L22:2003 weakly suppresses
  21. L25:856 weakly suppresses
  22. L24:1701 weakly excites
  23. L23:2551 weakly excites baseball and sports mentions
  24. L23:2592 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.

3 values between -2.94 and -2.462 values between -2.46 and -1.981 values between -1.98 and -1.516 values between -1.51 and -1.03104 values between -1.03 and -0.551768 values between -0.55 and -0.071864 values between -0.07 and 0.4174 values between 0.41 and 0.8913 values between 0.89 and 1.3617 values between 1.36 and 1.847 values between 1.84 and 2.3213 values between 2.32 and 2.806 values between 2.80 and 3.285 values between 3.28 and 3.766 values between 3.76 and 4.234 values between 4.71 and 5.192 values between 5.19 and 5.673 values between 6.15 and 6.622 values between 6.62 and 7.101 values between 7.10 and 7.581 values between 7.58 and 8.061 values between 9.49 and 9.971 values between 12.36 and 12.841 values between 13.80 and 14.282 values between 14.28 and 14.751 values between 15.71 and 16.19offnegativepositiveoften

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. ViewItem
  2. isks
  3. 畎
  4. ਯ
  5. يت
  6. بك
  7. igrams
  8. makt
  9. дей
  10. level

Suppresses

  1. Labour
  2. labour
  3. getResource
  4. urement
  5. 羔
  6. ilot
  7. enco
  8. erson
  9. حاسب
  10. enerator