Neuron Explorer

Qwen3-0.6B

court case names and citations

A neuron in layer 22 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the dot or name in court case citations, like 'Jones v. City of Opelika' or 'Johnson v. California'. When it was turned up, the model drifted into talk about courts and laws, and turning it down made it repeat phrases about women and people.

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

    mirroring any modern-day gun regulation, even for weapons that the people of that time could not have imagined existing?⏎Justice Breyer, joined by Justices Sotomayor and Kagan, focuses his dissent on the patent ludicrousness

  2. #2 peak

    with the study of literature.<|endoftext|>The two Supreme Court cases, Jones v. City of Opelika 316 U.S. 584

  3. #3 peak

    land… In short, every Indian nation is free to adopt its own laws and be ruled by them, to paraphrase the United States Supreme Court. Williams v. Lee, 358 U.S. 217 (19

  4. #4 peak

    system in an effort to survive inside the system; some inmates who would not separate themselves by race outside of a correctional institution. While the case of Johnson v. California challenged the practice of segregating inmates by using race as a classification, the

  5. #5 peak

    )⏎The two Supreme Court cases, Jones v. City of Opelika 316 U.S. 584 (1942) and 319 U.S. 103 (1943

  6. #6 peak

    4] Terre de Kerguelen, Appellee par M. Cook, Isle de la Desolation... [and] Plan du

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: for a good, well-EW, and even for a good, well-EW, and even for a good, well-EW, and even for a good, well-EW, and even for a good, well-EW,

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

    Mr. Darcy while triggered by anger showed a level of independence that left him shocked and stunned.⏎The freedom she exhibited in finally accepting him in direct defiance of Lady Catherine and knowing her father would disapprove was unusual even for Austen. In

  2. #2 peak

    Cachefly for the bandwidth for this show.<|endoftext|>The Scarlet Letter is an 1850 romantic work of fiction in a historical setting, written by Nathaniel Hawthorne, and is considered to be his magnum opus.

  3. #3 peak

    —attended a gathering with her future husband, Percy Bysshe Shelley, her stepsister Claire Clairmont, Lord Byron and John Polidori. To pass the time, the group read a

  4. #4 peak

    her novel “Wide Sargasso Sea” — a creatively daring, strongly feminist, and brazenly anti-colonial counter to Charlotte Brontë's “Jane Eyre” — is considered an integral part of the literary canon, but

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. L19:2006 suppresses
  2. L10:1278 excites
  3. L20:1020 suppresses
  4. L18:560 suppresses
  5. L20:2289 suppresses
  6. L9:3036 weakly suppresses
  7. L6:1161 weakly excites
  8. L4:1252 weakly excites
  9. L19:1333 weakly suppresses
  10. L20:2215 weakly suppresses
  11. L20:1377 weakly excites
  12. L2:2597 weakly suppresses
  13. L9:1293 weakly excites
  14. L2:2279 weakly suppresses
  15. L8:2911 weakly suppresses
  16. L20:551 weakly excites
  17. L10:1293 weakly suppresses
  18. L13:1134 weakly excites
  19. L18:1389 weakly suppresses
  20. L20:1089 weakly suppresses
  21. L18:361 weakly suppresses
  22. L6:1848 weakly excites
  23. L2:3013 weakly excites
  24. L13:1974 weakly excites

Feeds later layers

  1. L24:985 weakly excites
  2. L24:2502 weakly excites
  3. L25:2049 weakly excites
  4. L23:2727 weakly suppresses
  5. L23:939 weakly suppresses
  6. L23:2230 weakly excites
  7. L26:1772 weakly suppresses
  8. L26:1211 weakly excites
  9. L24:2994 weakly excites
  10. L25:1760 weakly suppresses
  11. L25:2854 weakly suppresses
  12. L25:920 weakly excites
  13. L25:2062 weakly suppresses
  14. L24:2439 weakly excites
  15. L25:1372 weakly excites
  16. L23:617 weakly excites
  17. L23:1430 weakly suppresses
  18. L25:1268 weakly suppresses
  19. L23:1719 weakly suppresses
  20. L26:2623 weakly suppresses
  21. L24:667 weakly suppresses
  22. L26:2509 weakly excites
  23. L27:835 weakly suppresses
  24. L23:2546 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 -24.08 and -23.121 values between -22.17 and -21.211 values between -18.35 and -17.391 values between -12.62 and -11.662 values between -11.66 and -10.714 values between -10.71 and -9.751 values between -9.75 and -8.802 values between -8.80 and -7.841 values between -7.84 and -6.891 values between -6.89 and -5.935 values between -5.93 and -4.985 values between -4.98 and -4.0211 values between -4.02 and -3.0714 values between -3.07 and -2.1134 values between -2.11 and -1.161288 values between -1.16 and -0.202401 values between -0.20 and 0.7582 values between 0.75 and 1.7119 values between 1.71 and 2.662 values between 2.66 and 3.629 values between 3.62 and 4.576 values between 4.57 and 5.533 values between 5.53 and 6.487 values between 6.48 and 7.441 values between 7.44 and 8.401 values between 9.35 and 10.311 values between 10.31 and 11.261 values between 11.26 and 12.222 values between 13.17 and 14.13offnegativepositiveoften

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. icina
  2. ega
  3. 铟
  4. /antlr
  5. aminer
  6. Studio
  7. ano
  8. studio
  9. issippi
  10. orta

Suppresses

  1. ftar
  2. ージ
  3. imens
  4. ême
  5. ,ev
  6. ]]></
  7. acle
  8. crunch
  9. conde
  10. hid