Neuron Explorer

Qwen3-0.6B

'of' after nouns like 'variation' and 'section'

A neuron in layer 2 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 'of' when it follows a noun like 'variation', 'section', 'variant' or 'author' (as in 'a variation of' or 'a section of'). Its negative side is a comma right after a person's role, such as 'a biologist,'. Turning it up did not produce a clear, consistent 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

    's especially ironic as the Caroline minuscule was invented under king Charlemagne as a way to counter the Merovingian cursive (in fact a variation of the Roman cursive) that

  2. #2 peak

    Last Winter (Agence France-Presse)⏎Beekeepers Sue EPA over Failure to Protect Bees from Pesticides (by Noel Br<|endoftext|>The following variation of a Bulls and Cows problem was given at the Fall 2008

  3. #3 peak

    content region. A view replaces the node to display a list or some other small variation of content.⏎Next you are asked to make the view more complicated. You scratch your

  4. #4 peak

    , after Filipino rebels seized control of a remote section of Malaysia’s Sabah state as part of an unresolved territorial dispute that stretches back centuries

  5. #5 peak

    xylem cells by the Ophiostoma fungus. Attacked trees exhibit wilted foliage with reddish or purplish discoloration within a section of the crown or throughout the entire crown. The foliage eventually turns brown and tends to

  6. #6 peak

    Terminal?⏎If you are using open-source packages, then sooner or later, you will encounter the .tgz or .tar.gz files. This is a variation of zipped files; most open-source packages use this format to zip their files.

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: I have been working on this week, and I have to have been thinking about... so I have to have been thinking about... and I have to have been thinking about... so I have to have been thinking about... and I have to

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

    temperatures, though it seems that these critters would be able to acclimatize to increasing temperatures in their natural habitat,” says Ricardo Nieves, an author of<|endoftext|>January 10, 2002⏎Parshat Va

  2. #2 peak

    e.g., infectious bacteria, would be harmless to other organisms, e.g., humans. In 1928, Sir Alexander Fleming, a Scottish biologist, observed that Penicillium notatum, a common mold, had destroyed st

  3. #3 peak

    cientific, while others took the time to address the main arguments. Alex Gilbert, an energy analyst, took to Twitter to explain the problems with the study in detail. While the authors

  4. #4 peak

    campaign on energy policy. “We desperately need to find a new form of energy.”⏎Colin Campbell, a retired geologist, says governments are too caught up in short-term issues to focus on the long-term

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. L1:878 suppresses
  2. L1:427 excites
  3. L1:192 weakly excites
  4. L1:295 weakly suppresses
  5. L1:653 weakly excites
  6. L1:2075 weakly suppresses
  7. L1:137 weakly suppresses
  8. L0:537 weakly excites the 'or' in 'and/or'
  9. L1:1861 weakly excites
  10. L1:898 weakly excites
  11. L1:385 weakly excites "as" after "known"
  12. L0:2009 weakly suppresses
  13. L0:588 weakly suppresses
  14. L0:2002 weakly excites
  15. L1:1007 weakly suppresses name pieces inside a multi-part word
  16. L0:1091 weakly excites
  17. L1:463 weakly excites
  18. L0:448 weakly excites
  19. L0:456 weakly suppresses
  20. L1:152 weakly suppresses
  21. L1:1385 weakly suppresses
  22. L0:2434 weakly suppresses
  23. L1:2408 weakly excites
  24. L1:947 weakly excites

Feeds later layers

  1. L6:54 suppresses
  2. L9:1674 suppresses
  3. L5:192 excites
  4. L5:692 suppresses
  5. L6:361 weakly excites
  6. L5:194 weakly excites 'a' or 'of' right after a verb
  7. L5:999 weakly excites
  8. L8:1677 weakly suppresses unclear: mixed, mostly math and instructions
  9. L17:258 weakly excites
  10. L6:695 weakly suppresses
  11. L5:1055 weakly excites
  12. L4:2010 weakly excites
  13. L6:426 weakly excites
  14. L4:1192 weakly excites
  15. L17:1974 weakly excites
  16. L5:2600 weakly excites
  17. L15:2146 weakly suppresses
  18. L20:2044 weakly suppresses
  19. L5:121 weakly suppresses
  20. L9:1139 weakly excites
  21. L4:139 weakly excites
  22. L8:498 weakly excites
  23. L10:1354 weakly excites
  24. L15:1961 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 -1.54 and -1.451 values between -1.20 and -1.121 values between -1.03 and -0.951 values between -0.95 and -0.871 values between -0.87 and -0.785 values between -0.70 and -0.626 values between -0.62 and -0.538 values between -0.53 and -0.4527 values between -0.45 and -0.3659 values between -0.36 and -0.28115 values between -0.28 and -0.20285 values between -0.20 and -0.11761 values between -0.11 and -0.031824 values between -0.03 and 0.05555 values between 0.05 and 0.14156 values between 0.14 and 0.2252 values between 0.22 and 0.3114 values between 0.31 and 0.3910 values between 0.39 and 0.478 values between 0.47 and 0.565 values between 0.56 and 0.645 values between 0.64 and 0.722 values between 0.81 and 0.892 values between 0.89 and 0.971 values between 0.97 and 1.061 values between 1.06 and 1.141 values between 1.14 and 1.231 values between 1.73 and 1.81offnegativepositiveoften

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. ially
  2. entially
  3. ologies
  4. esis
  5. ings
  6. aint
  7. IFO
  8. DbType
  9. iffs
  10. �

Suppresses

  1. keh
  2. 六个
  3. 五个
  4. 뉴
  5. herr
  6. yaptığı
  7. ję
  8. 找回
  9. DEN
  10. ousse