Neuron Explorer

Qwen3-0.6B

trailing zero in decimal numbers

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 zero that pads out a decimal number, like the 0 in 1.000 or 10.00 or 0.750. Steering it produced rambling text that did not carry the number theme forward.

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

    if the marked price is 1,000 and has a discount rate of 25% should take the following relationship: 1.000×25 / 100 , or , 1.00

  2. #2 peak

    Human Development Index (HDI) that is regularly estimated by the United Nations Development Programme (UNDP) measures the progress. It was 0.750 in 2013 – a notable improvement from 0.55

  3. #3 peak

    a packed agenda to read aloud Boylan's request. The council members, struck by Boylen's youthful enthusiasm, passed the hat and collected $10.00.⏎Flush with funds, Boylan and Cook experimented with various rocket nozzle designs

  4. #4 peak

    1%), high (45.8%), and extra-high (67.1%) intensities, as well as clear-cut harvesting (100.0%), with non-harvesting as the control, based on permanent sample plots

  5. #5 peak

    tons)|⏎|Length:||205 ft (62.48 m)o/a|⏎|Beam:||33 ft (10.06 m)|⏎|Draught:||11.5 ft (

  6. #6 peak

    one serving there are 20 mg of Calcium and 1.08 mg of Iron. In addition, there are 610 mg of

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: What a good job do you have? The same thing, but not that much less? Maybe that, but not that much less. Let's see. I think the user might be. So, I need to check. Let me do

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

    is a quarter and three fourths is three quarters so tehrefore three fourths is bigger.⏎three tenths is more than a quarter⏎1 and a half is bigger⏎About a quarter of all stars are bigger than the sun, some

  2. #2 peak

    3.5 million African-American slaves, and now the Union Army began to recruit a second all

  3. #3 peak

    number of calories they get from food and beverages (energy IN) and increase their physical activity (energy OUT).⏎For a weight loss of 1–1 ½ pounds per week, daily intake should be reduced by 500 to

  4. #4 peak

    2.4 \times 100 \div 3.2$.⏎You could

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. L4:440 excites
  2. L4:2290 weakly suppresses
  3. L4:2366 weakly excites
  4. L0:1006 weakly suppresses
  5. L4:2780 weakly excites
  6. L3:918 weakly excites
  7. L4:338 weakly excites
  8. L4:1423 weakly suppresses
  9. L0:2355 weakly excites
  10. L2:2834 weakly suppresses
  11. L3:2282 weakly excites
  12. L2:2143 weakly excites
  13. L3:1299 weakly suppresses
  14. L3:1316 weakly excites
  15. L1:317 weakly suppresses
  16. L1:154 weakly excites
  17. L0:1392 weakly excites
  18. L3:576 weakly suppresses
  19. L3:2569 weakly excites
  20. L3:1813 weakly excites
  21. L0:379 weakly excites
  22. L1:2654 weakly excites
  23. L3:604 weakly suppresses
  24. L4:1594 weakly excites

Feeds later layers

  1. L15:1855 suppresses
  2. L9:1373 weakly excites
  3. L21:1589 weakly suppresses
  4. L16:2783 weakly excites
  5. L8:981 weakly suppresses
  6. L19:928 weakly excites
  7. L11:2468 weakly excites
  8. L12:2262 weakly suppresses
  9. L18:2419 weakly suppresses
  10. L15:1103 weakly excites
  11. L23:1109 weakly suppresses
  12. L11:1941 weakly excites
  13. L11:890 weakly suppresses
  14. L11:1106 weakly suppresses
  15. L20:3052 weakly suppresses
  16. L24:2386 weakly excites
  17. L21:3000 weakly suppresses
  18. L26:3 weakly suppresses
  19. L20:1764 weakly suppresses
  20. L12:2447 weakly suppresses
  21. L20:2072 weakly suppresses
  22. L16:2280 weakly excites
  23. L18:1699 weakly excites
  24. L10:1848 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 -2.45 and -2.361 values between -1.29 and -1.191 values between -1.19 and -1.091 values between -0.99 and -0.908 values between -0.80 and -0.706 values between -0.70 and -0.602 values between -0.60 and -0.519 values between -0.51 and -0.4121 values between -0.41 and -0.3188 values between -0.31 and -0.21371 values between -0.21 and -0.121199 values between -0.12 and -0.021637 values between -0.02 and 0.08431 values between 0.08 and 0.1889 values between 0.18 and 0.2723 values between 0.27 and 0.3710 values between 0.37 and 0.472 values between 0.47 and 0.564 values between 0.56 and 0.661 values between 0.66 and 0.761 values between 0.76 and 0.861 values between 0.86 and 0.951 values between 1.34 and 1.44offnegativepositiveoften

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. том
  2. MULT
  3. pháp
  4. 磋
  5. zespo
  6. venth
  7. ่ว
  8. _module
  9. alle
  10. forme

Suppresses

  1. ificates
  2. (wait
  3. Wait
  4. (inner
  5. etric
  6. exels
  7. rawer
  8. 相对
  9. lsx
  10. tık