Neuron Explorer

Qwen3-0.6B

the digit 0 in dates and numbers

A neuron in layer 18 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 digit 0 when it sits inside a year, a measurement or a date range, such as the 0 in 1600 or in 12.0 m. Turning it up gave answers that wandered into unrelated topics, so it does not clearly change the model's style.

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

    00–500 BC ) and because of the large number of new foundations in Hellenism (323–30 BC), the Mediterranean world remained urban for centuries, although the majority of the People

  2. #2 peak

    -e to produce a Sino-Japanese painting style which was dominated by the Kanō school from the sixteenth century. In the Edo period (1600-1868) the Rimpa, Zen and Bunjinga

  3. #3 peak

    a genre of Japanese art that flourished during the Edo period (1603-1868). It is a form of woodblock printing that

  4. #4 peak

    at⏎= 56.3m/s + (-9.81m/s2)(4.52s)⏎vf = 12.0 m/s⏎This value is still positive, but smaller. The ball is slowing

  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

    is derived from the good old Spring and Autumn period (772BC to 403BC) and the Warring States period right after (403BC to

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 use of a student who is a non-bone, but it can be given to a student. If that's possible, and it is, then a good and appropriate number of them can be given to. Then, we can have

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

    %) compared to those of white-nosed coati (9.05%), being this difference significant in summer 2012 (P < 0.00001), summer (P =

  2. #2 peak

    2.91% of sampled diarrhea patients. The seasonal distribution of norovirus infections was different from non-norovirus patients (p < 0.001), with a half-year peak. Higher proportions of males (p

  3. #3 peak

    and staying healthy, both physically and mentally.<|endoftext|>Tech moves fast! Stay ahead of the curve with Techopedia!⏎Join nearly 200,000 subscribers who receive actionable tech insights from Techopedia.⏎A vulnerability management system

  4. #4 peak

    3, 95% CI = 1.110-1.529), local citizens (p < 0.001) and officials/clerks (p = 0.0

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. L17:213 suppresses
  2. L2:1795 suppresses
  3. L16:177 excites zero digit inside dates and codes
  4. L17:128 excites
  5. L4:498 suppresses
  6. L5:1546 suppresses
  7. L16:827 suppresses
  8. L10:50 excites
  9. L10:1052 suppresses
  10. L16:215 suppresses
  11. L15:12 excites
  12. L0:44 weakly suppresses
  13. L13:33 weakly excites
  14. L0:2302 weakly excites
  15. L16:374 weakly excites
  16. L15:78 weakly excites
  17. L16:461 weakly excites
  18. L15:96 weakly suppresses
  19. L16:724 weakly excites
  20. L2:1334 weakly suppresses
  21. L7:1808 weakly excites
  22. L10:722 weakly suppresses
  23. L8:20 weakly suppresses
  24. L17:512 weakly excites

Feeds later layers

  1. L20:626 excites
  2. L19:843 excites
  3. L19:311 suppresses
  4. L19:246 suppresses
  5. L19:2478 suppresses
  6. L19:321 suppresses
  7. L20:121 excites
  8. L20:1852 suppresses
  9. L19:129 suppresses
  10. L20:1793 suppresses
  11. L19:974 suppresses
  12. L20:72 suppresses
  13. L19:287 suppresses
  14. L19:197 weakly suppresses
  15. L19:171 weakly suppresses
  16. L20:221 weakly suppresses
  17. L19:2751 weakly suppresses
  18. L24:1676 weakly excites zeros in large round numbers
  19. L20:110 weakly suppresses
  20. L21:280 weakly suppresses
  21. L20:133 weakly suppresses
  22. L19:217 weakly suppresses
  23. L20:217 weakly suppresses
  24. L20:189 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 -9.96 and -9.142 values between -6.67 and -5.851 values between -5.03 and -4.213 values between -4.21 and -3.395 values between -3.39 and -2.576 values between -2.57 and -1.7428 values between -1.74 and -0.92680 values between -0.92 and -0.103137 values between -0.10 and 0.7218 values between 0.72 and 1.543 values between 1.54 and 2.362 values between 3.19 and 4.012 values between 4.83 and 5.651 values between 5.65 and 6.472 values between 7.29 and 8.122 values between 8.12 and 8.941 values between 8.94 and 9.764 values between 10.58 and 11.401 values between 12.22 and 13.053 values between 13.05 and 13.872 values between 13.87 and 14.691 values between 15.51 and 16.331 values between 16.33 and 17.151 values between 17.98 and 18.801 values between 22.09 and 22.91offnegativepositiveoften

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. fq
  2. orable
  3. 模范
  4. oration
  5. حوال
  6. OfDay
  7. aurus
  8. واقع
  9. ◻
  10. 年代

Suppresses

  1. antha
  2. tures
  3. ∅
  4. istics
  5. реакци
  6. issant
  7. двиг
  8. แต
  9. /entity
  10. gest