Neuron Explorer

Qwen3-0.6B

third digit of years and numbers

A neuron in layer 21 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 middle zero in years and long numbers, like the 0 in 2005 or 1690. Where it works in reverse, it is more about words like 'root' and 'core', and turning it down made the writing repeat 'root' and 'core' over and over.

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

    aimed at the asteroid 2005 YU55 a few days ago, when it passed the Earth at

  2. #2 peak

    Jacques Bovedere, said the possible applications of the Magnifier tube in the astronomical field in letter to Galileo Galilei, and in 1609 Galileo built several glasses without having seen models Dutch, using a convex lens

  3. #3 peak

    nearshore areas in French Polynesia, determined using ultrasonic telemetry⏎Centre IRD de Montpellier, BP 5045, 34032 Montpellier cedex 1, France⏎2 Centre IRD

  4. #4 peak

    very bright stars with a view for follow up observations by other telescopes.⏎Kepler, which cost $600 million, was launched in 2009. It is one of the most successful missions in NASA’s history, having

  5. #5 peak

    the categories of the three majors to be your treatments.)<|endoftext|>During this week's close encounter with Earth, astronomers tracked the aircraft-carrier-sized asteroid 2005 YU55 with radar instruments, infrared cameras, visible-light telescopes

  6. #6 peak

    the recognition of<|endoftext|>The association between British colonial rule and lower levels of corruption is often emphasized in legal origins literature (La Porta et al., 2008). However, given the historical context of Africa, we hypothesize that Britain

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: at the root of the core of the root of the root of the root of the root of the root of the root of the root of the root of the root of the root of the root of the root of the root of the root

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

    .⏎Blockchain and Decentralization⏎Blockchain technology burst onto the scene with the advent of cryptocurrencies like Bitcoin. However, its applications extend far beyond digital currencies. At its core, blockchain is a decentralized and immutable ledger that can record transactions and information in

  2. #2 peak

    employees to wear it, many facilities<|endoftext|>The Israel-Palestine conflict – often referred to as the ‘world’s most intractable conflict’ – is rooted in a dispute over land claimed by Jews as their biblical birthright and by the

  3. #3 peak

    do with science. But the journey from evidence to policy is far from simple, always lengthy, frequently divisive and often ineffective. Why is that? To get to the bottom of this question we first need to ask another — what is evidence?⏎Evidence

  4. #4 peak

    religion, but the most important of the tenets. It is the very soul and the base of all the Islamic ideas and beliefs. We can say that the roots, as well as the branches of Islam take

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. L19:2094 excites
  2. L9:1632 excites
  3. L19:2298 weakly suppresses
  4. L16:2820 weakly excites
  5. L14:985 weakly excites
  6. L8:2915 weakly excites
  7. L6:90 weakly suppresses
  8. L5:1794 weakly excites
  9. L16:2082 weakly suppresses
  10. L8:2120 weakly excites
  11. L19:1545 weakly excites
  12. L18:222 weakly excites
  13. L19:2520 weakly excites
  14. L19:3012 weakly excites
  15. L17:2653 weakly suppresses
  16. L4:702 weakly suppresses
  17. L17:2716 weakly excites
  18. L13:2890 weakly excites "still" or "as long as" meaning continuing
  19. L18:1497 weakly excites
  20. L15:2807 weakly suppresses
  21. L7:30 weakly suppresses
  22. L14:1057 weakly excites
  23. L19:2486 weakly suppresses
  24. L11:246 weakly suppresses

Feeds later layers

  1. L26:1543 suppresses
  2. L23:2966 suppresses
  3. L25:2036 suppresses
  4. L27:2290 suppresses
  5. L22:1612 suppresses
  6. L25:2502 suppresses
  7. L24:2635 suppresses
  8. L25:2040 suppresses
  9. L27:1376 suppresses
  10. L22:2708 suppresses
  11. L25:1980 suppresses
  12. L22:1701 weakly suppresses
  13. L23:2470 weakly excites
  14. L22:3052 weakly suppresses
  15. L22:927 weakly suppresses
  16. L24:2468 weakly suppresses
  17. L27:2037 weakly suppresses
  18. L22:1344 weakly excites
  19. L24:2970 weakly excites
  20. L23:2529 weakly suppresses
  21. L27:1327 weakly suppresses
  22. L26:2488 weakly suppresses
  23. L25:942 weakly suppresses
  24. L22:1172 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 -16.13 and -15.601 values between -15.60 and -15.071 values between -13.50 and -12.971 values between -12.44 and -11.921 values between -9.81 and -9.291 values between -9.29 and -8.761 values between -8.76 and -8.241 values between -7.71 and -7.182 values between -6.13 and -5.615 values between -5.61 and -5.083 values between -5.08 and -4.555 values between -4.55 and -4.036 values between -4.03 and -3.505 values between -3.50 and -2.989 values between -2.98 and -2.4527 values between -2.45 and -1.9232 values between -1.92 and -1.4087 values between -1.40 and -0.87452 values between -0.87 and -0.352198 values between -0.35 and 0.18932 values between 0.18 and 0.71106 values between 0.71 and 1.2315 values between 1.23 and 1.769 values between 1.76 and 2.283 values between 2.28 and 2.812 values between 2.81 and 3.341 values between 3.86 and 4.391 values between 4.39 and 4.91offnegativepositiveoften

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. migrationBuilder
  2. secutive
  3. ivities
  4. 过得
  5. 平均
  6. ặp
  7. 掖
  8. ulia
  9. اة
  10. icopter

Suppresses

  1. root
  2. 根
  3. 根本
  4. 根基
  5. root
  6. Root
  7. Root
  8. fundamental
  9. roots
  10. core