Neuron Explorer

Qwen3-0.6B

"that" introducing a clause or "that is"

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 word "that" after a claim, especially in phrases like "positive rights; that is" or "prove that" or "believe that", where a new statement is about to start. Turning it up made the model repeat "the" over and over, so the steering results are not very informative.

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

    unitarians seek to bolster social capital and the institutions of civil society. . . .⏎Central to many communitarians' philosophy is the concept of positive rights; that is, rights or guarantees to certain things. These may include free education, affordable

  2. #2 peak

    home, or stay at his, it only takes one security incident to cast doubt over the safety of this model. The company created a way to check and prove that users are “real people.”⏎And this is where the Digital Identity story begins.⏎

  3. #3 peak

    , and laws limiting such things as pollution and gun violence.⏎A common objection is that by providing such rights, they are violating the negative rights of the citizens; that is, rights to have something not done to you. For example, taking money

  4. #4 peak

    “States should identify marine ecosystems exhibiting high levels of biodiversity and productivity and other critical habitat areas and should provide necessary limitations on use of these areas, …(and that) Priority should be accorded to seagrass beds…”;⏎Reminded by

  5. #5 peak

    responsible for the vast majority of federal death sentences. By addressing the unique geography, we identify a possible explanation for the racial distortions in the federal death penalty: that federal death sentences are sought disproportionately where the expansion of the venire from the county

  6. #6 peak

    even after filtering through Chamberland filter candles⏎, in 1892⏎. However, he remained convinced despite repeated failures to produce evidence, that bacteria were the infectious agents. In 1898, Martinus Be

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: Please be very very specific about who and who be back to be back at who be back at who. Please be very very specific. Please be very very very specific. Please be very very very very very very very very very very very very

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

    20|⏎|Order Information:||Web: http://www.tandfonline.com/pricing/journal/VECE20|⏎When requesting a correction, please mention this item's handle: RePEc:taf:jeduce

  2. #2 peak

    a woman's life to a contemporary meditation on the meaning of the story of Lot's wife.⏎This article is available to subscribers only. Please subscribe for full access—subscriptions begin at $2.95. Already have an online account? Log

  3. #3 peak

    MENA region participated in the workforce. Today, that number is 25 percent—still half the rate of the global average.1⏎Gayle Tzemach Lemmon, a senior fellow at the Council on Foreign Relations as well as

  4. #4 peak

    precautions to help control the spread of the flu and norovirus in their facilities.⏎Kim LaFreniere, Ph.D., associate research fellow at the Clorox Professional Products Company, offers cleaning professionals four infection control tips, provided below.⏎

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. L17:1282 excites
  2. L19:1100 suppresses
  3. L17:2684 suppresses
  4. L20:2430 excites
  5. L16:1303 excites 'that' starting a clause after a noun
  6. L3:1003 excites
  7. L20:982 excites
  8. L18:1389 weakly excites
  9. L0:684 weakly suppresses
  10. L21:191 weakly suppresses
  11. L20:1832 weakly excites 'as' introducing a reason
  12. L7:1098 weakly excites
  13. L18:645 weakly suppresses
  14. L17:253 weakly excites
  15. L18:598 weakly suppresses
  16. L6:2313 weakly excites
  17. L19:1972 weakly suppresses
  18. L18:2031 weakly suppresses
  19. L15:3031 weakly suppresses
  20. L21:172 weakly suppresses
  21. L7:313 weakly excites
  22. L17:2695 weakly suppresses
  23. L0:442 weakly suppresses
  24. L6:1646 weakly suppresses

Feeds later layers

  1. L27:1042 weakly excites
  2. L27:149 weakly suppresses
  3. L27:796 weakly suppresses
  4. L24:298 weakly suppresses 'would' in hypothetical or past statements
  5. L25:247 weakly excites
  6. L24:2151 weakly suppresses
  7. L26:1092 weakly suppresses
  8. L26:368 weakly suppresses
  9. L27:478 weakly excites
  10. L27:2250 weakly suppresses
  11. L26:1007 weakly excites
  12. L25:2123 weakly excites
  13. L26:2984 weakly suppresses
  14. L27:108 weakly suppresses
  15. L26:2385 weakly suppresses
  16. L24:306 weakly excites
  17. L25:1404 weakly suppresses
  18. L26:2490 weakly excites
  19. L26:693 weakly excites
  20. L26:127 weakly suppresses 'co' starting a word piece
  21. L24:261 weakly excites
  22. L27:2972 weakly suppresses
  23. L26:1891 weakly suppresses
  24. L27:121 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.

2 values between -13.29 and -10.965 values between -10.96 and -8.6311 values between -8.63 and -6.2924 values between -6.29 and -3.96129 values between -3.96 and -1.632648 values between -1.63 and 0.70638 values between 0.70 and 3.04198 values between 3.04 and 5.3790 values between 5.37 and 7.7051 values between 7.70 and 10.0320 values between 10.03 and 12.3719 values between 12.37 and 14.7016 values between 14.70 and 17.039 values between 17.03 and 19.366 values between 19.36 and 21.691 values between 21.69 and 24.036 values between 24.03 and 26.364 values between 26.36 and 28.693 values between 28.69 and 31.021 values between 31.02 and 33.363 values between 33.36 and 35.694 values between 35.69 and 38.021 values between 38.02 and 40.351 values between 40.35 and 42.682 values between 42.68 and 45.021 values between 45.02 and 47.352 values between 47.35 and 49.681 values between 49.68 and 52.012 values between 52.01 and 54.352 values between 54.35 and 56.681 values between 56.68 and 59.012 values between 59.01 and 61.342 values between 61.34 and 63.672 values between 70.67 and 73.001 values between 77.67 and 80.00offnegativepositiveoften

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. a
  2. those
  3. no
  4. laughter
  5. éné
  6. Prediction
  7. none
  8. а
  9. ging
  10. af

Suppresses

  1. 過來
  2. депут
  3. 有多大
  4. ạp
  5. ');</
  6. 로
  7. pline
  8. ожно
  9. ---</
  10. 请联系