Neuron Explorer

Qwen3-0.6B

'Old' as part of a name or phrase

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 lights up on the word 'Old' when it starts a name or phrase, like 'Old World' or 'Old Faithful', and on similar words like 'old' and 'heirs'. Its strongest negative cases are about inheritance and succession, like 'heirs' and 'testamentary'. Turning it up did not produce a 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

    , damask, Cherokee, moss, monthly, musk, and "wild roses."⏎Beloved for centuries for their practical uses as well as their beauty, Old World roses were brought to the New World by European settlers who desired a link with

  2. #2 peak

    woman, even reaching almost four children per woman in the late 1950s.⏎Developed countries have low fertility, and<|endoftext|>Another edition of The Old Farmer’s Almanac Monthly is here! January is full of fun and

  3. #3 peak

    duty of eliminating economic inequalities to ensure a reasonable and equitable standard of living, to all citizens of the state.⏎Education, health care, social insurance, housing, old age pensions and medical care have become the responsibility of welfare states. They even bother

  4. #4 peak

    NASA Remix Challenge. Credit: Cookieater2009 on Flickr.⏎Some people like an adventure, but don’t want to leave their home behind — like old Carl in the movie “Up.” So, if you wanted to go to space

  5. #5 peak

    Comment on the shape of the distribution. In Problem 28 from Section 3.1, we drew a histogram of the length of eruption of California's Old Faithful geyser and found that the distribution is symmetric. Draw a box

  6. #6 peak

    cables juts out of the sand next to a public walkway at First Landing State Park. Four small rods stick out horizontally from the sides.⏎Research scientists from Old Dominion University installed the contraption on the dunes of the campground beach last week

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 power of the will to be the only legacy to the children of the will. That's the way of the will. Her, the will, is the only legacy to the children of the will. In the spirit of the will,

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

    nearest surviving male relative to the French king after the three French Capetian kings, Louis X, Philip V, and Charles IV, all failed to produce male heirs. In order to prevent Edward from claiming the throne,

  2. #2 peak

    . With little knowledge of the hall prior to the Norman Conquest, documented history starts around the end of the 13 th century, when the hall was passed to the Earl of Chester, a nephew of William the Conqueror.⏎Since

  3. #3 peak

    of two categories – testamentary or living trusts. A testamentary trust is activated by a provision in the Settlor’s Will at the time

  4. #4 peak

    , returns from banishment. Because he and the Duchess had been together and had children, his son is believed to be the Duchess’s eldest son and therefore the heir to the throne of Amalfi (or Malfi, as the play

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. L9:2911 suppresses
  2. L10:2640 suppresses
  3. L4:2548 excites
  4. L15:931 weakly excites
  5. L6:1894 weakly excites
  6. L20:871 weakly suppresses
  7. L9:2035 weakly excites
  8. L12:12 weakly suppresses
  9. L11:925 weakly suppresses
  10. L18:1905 weakly suppresses
  11. L15:2329 weakly suppresses
  12. L3:2151 weakly excites
  13. L3:52 weakly excites
  14. L20:2423 weakly suppresses
  15. L18:2729 weakly suppresses
  16. L10:2762 weakly excites
  17. L11:2737 weakly excites
  18. L20:730 weakly suppresses action verbs in how-to steps
  19. L8:2889 weakly excites
  20. L14:1684 weakly excites
  21. L4:1624 weakly excites
  22. L11:2161 weakly excites
  23. L10:2328 weakly suppresses
  24. L11:1035 weakly excites

Feeds later layers

  1. L23:2059 suppresses
  2. L24:2075 weakly suppresses
  3. L24:2602 weakly suppresses
  4. L24:2256 weakly suppresses
  5. L25:1564 weakly suppresses word pieces before 'trans' or 'transfer'
  6. L25:2668 weakly suppresses
  7. L25:2169 weakly suppresses
  8. L25:1648 weakly suppresses
  9. L27:2338 weakly suppresses capital P starting a name
  10. L25:2851 weakly suppresses
  11. L23:2060 weakly excites
  12. L24:2433 weakly excites
  13. L22:2774 weakly excites
  14. L25:2843 weakly suppresses
  15. L25:1348 weakly suppresses
  16. L22:2342 weakly excites
  17. L23:923 weakly excites
  18. L25:2891 weakly excites
  19. L27:1512 weakly suppresses
  20. L26:2605 weakly suppresses
  21. L23:518 weakly suppresses
  22. L23:1262 weakly suppresses
  23. L24:2587 weakly excites
  24. L22:838 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 -18.73 and -18.141 values between -18.14 and -17.541 values between -17.54 and -16.942 values between -15.74 and -15.151 values between -13.95 and -13.352 values between -12.16 and -11.562 values between -8.57 and -7.971 values between -7.97 and -7.371 values between -7.37 and -6.773 values between -6.77 and -6.182 values between -6.18 and -5.582 values between -5.58 and -4.983 values between -4.98 and -4.385 values between -4.38 and -3.783 values between -3.78 and -3.1917 values between -3.19 and -2.597 values between -2.59 and -1.9926 values between -1.99 and -1.3955 values between -1.39 and -0.79476 values between -0.79 and -0.202576 values between -0.20 and 0.40677 values between 0.40 and 1.0033 values between 1.00 and 1.605 values between 1.60 and 2.202 values between 2.20 and 2.802 values between 2.80 and 3.391 values between 3.39 and 3.991 values between 4.59 and 5.19offnegativepositiveoften

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. OCI
  2. olygon
  3. isy
  4. Movie
  5. 濉
  6. aise
  7. Ai
  8. Hang
  9. rij
  10. ences

Suppresses

  1. 繼
  2. 继承
  3. succession
  4. 接
  5. 继
  6. contin
  7. sucess
  8. 传承
  9. Transmission
  10. 交接