Neuron Explorer

Qwen3-0.6B

number right after a year or range word

A neuron in layer 22 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 blank spot right after a word like 'from', 'to', or 'between' where a number, usually a year, should come next. When it was turned down, the model began writing strings of zeros and ones.

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

    the health and living habits of 15,000 residents in six Brazilian capitals over a 10-year period between 2008 and 2018, researchers from the Longitudinal Study of Adult Health (

  2. #2 peak

    ics that gave the suspect shots<|endoftext|>Classic DACB CollectionAll articles created or submitted in the first twenty years of the project, from 1995 to 2015.⏎Jacob Osasona was born in the 18

  3. #3 peak

    oted"? Overwhelmingly for Obama. The "popular vote" (that is, the share of the overall votes cast) is running 85% to 15% in favour of the Democratic candidate. At the time of writing,

  4. #4 peak

    S., & Ekman, S.L. (2002). Caregiver singing and background music in dementia care. Western Journal of Nursing Research, 24(2), 195-216. My approach

  5. #5 peak

    be explored in depth in an Education Program Session today at 7:30 a.m. and again at 2:00 p.m. in rooms 393-396 of the Ernest N. Morial Convention Center

  6. #6 peak

    105 of whom had been exposed to gestational diabetes (mean age, 5.9 years; 50.5% boys) and 38 who were unexposed (mean age, 6.8 years

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: 00-10-00-10-00-10 00-10-00-10-00-10 00-10-00-10

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

    at his code, x would never be zero, that wasn't the issue. The fact is that it is pointless even having the x if you set total to 0 at the start because total will never increase, and that's even if you

  2. #2 peak

    elements of the array in reverse, we need to use a loop that will iterate from length - 1 to 0. We can use a for loop or a while loop to write this

  3. #3 peak

    to check who it was that posted it, and I can't see just now anyway :P)⏎void genFactorial(int x, int t)⏎int total = t; // you probably wouldn't even need total now⏎void main(void)

  4. #4 peak

    that can run all the commands with one message from you.⏎% filename simple.m⏎ini=1.; %set the variable ini to 1.⏎y0=ini; %set the initial condition to ini

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. L20:567 excites
  2. L20:205 excites
  3. L19:3047 suppresses
  4. L20:1479 excites
  5. L18:115 suppresses
  6. L14:827 suppresses open bracket after a value or citation
  7. L19:248 excites
  8. L13:287 suppresses
  9. L21:1595 suppresses
  10. L21:3007 weakly suppresses
  11. L13:818 weakly suppresses
  12. L18:1259 weakly suppresses
  13. L18:2519 weakly suppresses
  14. L18:930 weakly excites
  15. L20:947 weakly excites
  16. L3:385 weakly suppresses
  17. L10:272 weakly suppresses
  18. L20:1493 weakly excites
  19. L14:1272 weakly excites
  20. L21:253 weakly suppresses
  21. L8:2755 weakly excites
  22. L20:780 weakly excites 'of' before a measured value
  23. L16:1142 weakly excites
  24. L14:442 weakly excites

Feeds later layers

  1. L25:46 excites
  2. L23:198 suppresses
  3. L27:396 suppresses
  4. L26:329 excites
  5. L26:611 suppresses
  6. L27:287 suppresses
  7. L27:113 excites
  8. L25:473 suppresses zeros inside large round numbers
  9. L23:250 suppresses
  10. L27:153 excites the spot just before a historical number
  11. L26:413 suppresses
  12. L27:691 suppresses
  13. L26:2953 excites
  14. L27:48 excites
  15. L25:95 excites
  16. L25:253 suppresses
  17. L27:2172 excites
  18. L27:163 excites
  19. L26:137 excites
  20. L27:238 suppresses
  21. L25:854 weakly suppresses
  22. L26:216 weakly suppresses blank slot before a year or number
  23. L23:478 weakly excites
  24. L26:667 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 -11.94 and -11.282 values between -4.00 and -3.341 values between -3.34 and -2.682 values between -2.68 and -2.026 values between -2.02 and -1.36175 values between -1.36 and -0.702260 values between -0.70 and -0.031279 values between -0.03 and 0.6379 values between 0.63 and 1.2927 values between 1.29 and 1.9524 values between 1.95 and 2.6112 values between 2.61 and 3.274 values between 3.27 and 3.935 values between 3.93 and 4.603 values between 4.60 and 5.266 values between 5.26 and 5.927 values between 5.92 and 6.582 values between 6.58 and 7.242 values between 7.24 and 7.902 values between 7.90 and 8.561 values between 9.23 and 9.891 values between 9.89 and 10.553 values between 11.21 and 11.871 values between 11.87 and 12.531 values between 12.53 and 13.191 values between 13.19 and 13.851 values between 13.85 and 14.52offnegativepositiveoften

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. ovation
  2. angkan
  3. -awesome
  4. lients
  5. .Flag
  6. 赘
  7. изации
  8. agus
  9. arium
  10. rage

Suppresses

  1. 0
  2. -
  3. 1
  4. false
  5. "-
  6. ;-
  7. false
  8. 0
  9. true