Neuron Explorer

Qwen3-0.6B

the word 'late' in time phrases

A neuron in layer 16 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 word 'late' when it is used to point at a period of time, like 'the late 19th century', 'in late 2009' or 'late summer'. Its negative side picks up odd spots inside words in other scripts. Turning it up or down produced text that did not show any clear theme.

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

    derived from living matter were applied to wounds to destroy infection, the fact that a microorganism is capable of destroying one of another species was not established until the latter half of the 19th cent. when Pasteur noted the antagonistic

  2. #2 peak

    Boston, he was a declared music major. Late in his collegiate career, he discovered a love of biology – particularly the brain and

  3. #3 peak

    combination of structural weaknesses of the Greek economy along with a decade long pre-existence of overly high structural deficits and debt-to-GDP levels on public accounts. In late 2009, fears of a sovereign debt crisis developed among investors concerning

  4. #4 peak

    000 acres of pastures for cattle and sheep that, starting in the late 19th century, gave way to development of the cities of Long Beach

  5. #5 peak

    dots to the naked eye. Despite their size, our arborists warn that mites can be a huge problem for homeowners, especially in the hot days of late summer.⏎When temperatures are over 80 degrees, oak spider mites reproduce

  6. #6 peak

    especially the United States, which retained ties with Chiang in Taiwan, recognising his as the legal Chinese government.⏎Since embarking on landmark reforms beginning in the late 1970s, the country has emerged from isolation to become the

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 people who have been in a place of power, or control over control over control, in the world. That’s a huge number, not just in the US or the United States. We can't understand how that's been

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 from the words 'Gur (ਗੁਰ)' (a reference to the Sikh Gurus) and 'Dwara (ਦੁਆਰਾ)' (gateway in Gurmukhi), together meaning

  2. #2 peak

    The Court of Appeals of Alabama reversed the conviction, and the Supreme Court of Alabama decided that the licensing provisions were constitutional and reversed the appellate court. The Jehovah’s Witnesses from Alabama appealed to the U.S. Supreme Court, which consolidated the

  3. #3 peak

    recognizably in the Hebrew name of The Jewish Agency – הַסּוֹכְנוּת הַיְּהוּדִית, or simply הסוכ

  4. #4 peak

    of nose bleeds. Because the membrane and blood vessels inside the nose are so sensitive, any irritation caused by the picking can lead to bleeding. Frequent, forceful nose-blowing can also be a factor. People who are prone to

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. L15:1587 strongly excites
  2. L15:1900 strongly suppresses
  3. L15:1404 strongly suppresses
  4. L15:792 strongly suppresses
  5. L15:381 strongly excites
  6. L15:222 strongly excites
  7. L15:1135 strongly suppresses
  8. L15:1616 strongly excites word fragments mid-word
  9. L15:435 strongly excites
  10. L15:201 strongly excites
  11. L15:451 strongly excites
  12. L15:2803 strongly suppresses
  13. L15:1164 strongly excites
  14. L15:787 strongly excites
  15. L15:12 suppresses
  16. L14:1090 excites
  17. L15:1300 excites
  18. L14:126 excites
  19. L14:169 excites
  20. L15:966 suppresses
  21. L15:227 excites
  22. L15:1043 suppresses
  23. L15:1394 suppresses
  24. L15:554 suppresses

Feeds later layers

  1. L19:2126 excites
  2. L19:1470 excites
  3. L27:775 suppresses
  4. L20:2064 excites
  5. L18:168 excites
  6. L19:833 suppresses
  7. L19:2724 suppresses
  8. L19:1016 excites
  9. L26:2022 excites
  10. L22:470 suppresses
  11. L26:1374 weakly suppresses
  12. L26:3023 weakly excites
  13. L26:1698 weakly excites
  14. L21:1736 weakly suppresses
  15. L27:606 weakly suppresses
  16. L18:3058 weakly excites
  17. L22:2027 weakly excites
  18. L26:1015 weakly suppresses
  19. L22:423 weakly excites
  20. L20:603 weakly suppresses
  21. L27:1007 weakly excites
  22. L20:2097 weakly suppresses
  23. L26:1861 weakly suppresses
  24. L26:1545 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 -8.33 and -7.921 values between -7.10 and -6.691 values between -6.69 and -6.282 values between -6.28 and -5.872 values between -5.87 and -5.476 values between -5.47 and -5.0610 values between -5.06 and -4.6513 values between -4.65 and -4.2418 values between -4.24 and -3.8339 values between -3.83 and -3.4250 values between -3.42 and -3.0160 values between -3.01 and -2.60109 values between -2.60 and -2.19154 values between -2.19 and -1.78156 values between -1.78 and -1.38224 values between -1.38 and -0.97335 values between -0.97 and -0.56506 values between -0.56 and -0.15651 values between -0.15 and 0.26466 values between 0.26 and 0.67349 values between 0.67 and 1.08233 values between 1.08 and 1.49177 values between 1.49 and 1.90121 values between 1.90 and 2.3169 values between 2.31 and 2.7150 values between 2.71 and 3.1234 values between 3.12 and 3.5330 values between 3.53 and 3.9416 values between 3.94 and 4.3510 values between 4.35 and 4.765 values between 4.76 and 5.174 values between 5.17 and 5.582 values between 5.58 and 5.992 values between 5.99 and 6.401 values between 6.80 and 7.211 values between 7.62 and 8.03offnegativepositiveoften

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. ning
  2. res
  3. anced
  4. stab
  5. ell
  6. 목
  7. oted
  8. 恙
  9. haste
  10. plate

Suppresses

  1. SBATCH
  2. sylvania
  3. orial
  4. ATIONAL
  5. '})
  6. )})
  7. ска
  8. )});⏎
  9. Frm
  10. stakes