Neuron Explorer

Qwen3-0.6B

second digit of years like 2020

A neuron in layer 23 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 "0" in the year 2020 or 2021 when it comes right after the "2". It also fires on years like 2002 or 2001 with a negative sign. Turning it up made the output fall into repeated "2222" strings, and turning it down gave repeated zeros.

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

    are. You may even have heard of names like Bitcoin or maybe Ethereum to name a view. There are many most popular cryptocurrencies which can be used in 2020.⏎Maybe you’ve heard the terms digital gold or halving if you

  2. #2 peak

    post and review on that article. https://t.co/D9NIorgXuc— Bruce Rogers (@bjrmd) August 30, 2021⏎PACING: Dist

  3. #3 peak

    their data protection measures to safeguard their data and hardware assets. As a guide, here are the five biggest IT security threats that you should prepare for as 2016 comes.⏎1. Internet of Things (IoT) security⏎The

  4. #4 peak

    healthcare industry today is the most targeted industry in terms of cyber security threats. A recent study by Critical insights (A cyber security company) shows that in 2021 alone a whopping number of 45 million individuals were affected by cyber

  5. #5 peak

    seconds or use hand sanitiser with a minimum of 60% alcohol base.⏎9. Wash<|endoftext|>Date Updated: 12/17/2020⏎Strep throat is a bacterial infection that can make your throat feel

  6. #6 peak

    households in Nigeria do not have sufficient access to soap and water for handwashing and that there is low adherence to hygiene protocols (National Bureau of Statistics, 2020; Premium Times, 2020).⏎The most recent Afro

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 number of just-type meaning-type (0000000000000000000000000000000000000000

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

    the success of the struggle for equal rights for women and minorities<|endoftext|>The No Child Left Behind Act, signed into law by President Bush January 8, 2002 includes as one of its measures the standard to staff our public schools with

  2. #2 peak

    9⏎rothbard on the utility of quantitative easing⏎Economist Murray Rothbard ("America‟s Great Depression", Ludwig von Mises Institute 2000) made extensive analysis of why the inflationary policies of the US Federal

  3. #3 peak

    2010⏎In 2009, Gulf War-era II veterans—those who had served in the Armed Forces sometime since September 2001 and had returned to civilian life—were more likely to have some college

  4. #4 peak

    local law enforcement in northern Texas and developed this early warning system to help find abducted children. National coordination of state and local AMBER Alert programs began in 2003.⏎The AMBER Alert Plan is activated when a child 17

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. L17:1476 excites
  2. L14:1700 excites
  3. L18:1976 weakly suppresses
  4. L19:843 weakly suppresses
  5. L14:2966 weakly excites
  6. L21:2497 weakly suppresses
  7. L7:2258 weakly excites
  8. L11:1975 weakly suppresses
  9. L20:1520 weakly excites
  10. L2:2517 weakly suppresses
  11. L13:859 weakly suppresses
  12. L19:1145 weakly excites
  13. L1:416 weakly suppresses
  14. L16:1677 weakly excites
  15. L0:1071 weakly excites
  16. L4:2138 weakly excites
  17. L21:2106 weakly excites
  18. L12:2450 weakly excites
  19. L16:2238 weakly suppresses
  20. L8:2323 weakly excites
  21. L4:26 weakly excites
  22. L2:59 weakly excites
  23. L3:405 weakly excites
  24. L0:121 weakly excites

Feeds later layers

  1. L27:189 strongly suppresses
  2. L26:834 strongly excites
  3. L27:352 excites
  4. L25:379 excites
  5. L27:992 excites
  6. L25:1478 excites
  7. L27:294 suppresses
  8. L25:224 suppresses
  9. L25:473 suppresses zeros inside large round numbers
  10. L27:862 suppresses
  11. L27:153 excites the spot just before a historical number
  12. L25:2298 excites
  13. L27:396 suppresses
  14. L26:413 suppresses
  15. L24:1446 excites
  16. L27:1157 excites
  17. L26:848 suppresses
  18. L24:677 excites
  19. L27:1202 suppresses
  20. L27:1153 weakly suppresses
  21. L26:598 weakly excites spot just before a year or century
  22. L24:342 weakly excites unit conversions in parentheses
  23. L26:905 weakly suppresses
  24. L27:205 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 -41.25 and -39.461 values between -32.32 and -30.531 values between -19.81 and -18.031 values between -16.24 and -14.463 values between -10.88 and -9.102 values between -9.10 and -7.313 values between -7.31 and -5.526 values between -5.52 and -3.7418 values between -3.74 and -1.95393 values between -1.95 and -0.163452 values between -0.16 and 1.6215 values between 1.62 and 3.415 values between 3.41 and 5.202 values between 5.20 and 6.982 values between 6.98 and 8.771 values between 19.49 and 21.271 values between 26.63 and 28.421 values between 28.42 and 30.20offnegativepositiveoften

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. 2
  2. 二十
  3. 二十五
  4. twenty
  5. twenty
  6. 二十四
  7. 2
  8. 十八
  9. 十七
  10. kır

Suppresses

  1. tery
  2. ấy
  3. conomy
  4. 0
  5. auté
  6. uisine
  7. луш
  8. tür
  9. ldb
  10. чная