Neuron Explorer

Qwen3-0.6B

sports leagues and team names

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 lights up on sports words, especially league and team names like NFL, NBA, Negro League and baseball. It turns off on soccer-style words like 'corner' and 'the goal'. Turning it up did not produce any sports text, so the link is only in the reading of text.

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

    The Pressure to Perform⏎Chuck Knoblauch was an outstanding baseball player in the 1990s. He won Rookie of the Year” and multiple Gold Gloves (outstanding defense). Because of his excellence, he was offered a

  2. #2 peak

    in the black community. His namesake park boasts an aquatic complex and state-of-the-art sports facilities, with two solar trees near the well-equipped playground. Several NFL players, including Lavonte David of the Tampa Bay Buccaneers and Devonta Freeman of

  3. #3 peak

    14/08/18/out-at-home/⏎- Related link: Click here to view photos from the 17th annual Jerry Malloy Negro League Conference on August 14-16, 2014⏎

  4. #4 peak

    disciplined me has a great deal to do with the success I have enjoyed as a man,” Peterson said in a statement.⏎Among those coming to Peterson’s defense was NBA Hall of Famer Charles Barkley. On the CBS pregame show The NFL

  5. #5 peak

    enzel found that people tend to seek restorative justice rather than retributive justice when they perceive a shared identity with the offender. In the sense that National Football League fans, employees, and players share the “NFL family” identity, the

  6. #6 peak

    happened makes compelling reading; from “The National Pastime” of 1983, which I had created in the previous year for the Society for American Baseball Research.⏎Jerry Malloy (1946-2000)

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: If I had 6000 of the kind of power that can drive the economy, what would happen if we give more of that power out to the economy, and give the economy more control? If the economy is more powerful,

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

    will claim that a goal kick should be given and other will call for a corner.⏎Why does the kicker raise his hand?⏎Raising their hand acts as a

  2. #2 peak

    ank free throws off the back of the backboard. Quarterbacks throw passes that sail off into the air. Soccer players kick empty net goals over the top of the goal. All of these players seem to ‘forget’ how to do movements they

  3. #3 peak

    etc.⏎Why do soccer players raise their hands after a goal?⏎Players raise their hand to claim that their team should take the throw in. … In case of corner one team

  4. #4 peak

    the tournament for the first time as Dutch East Indies, now Indonesia, qualified while other Asian teams withdrew.⏎The tournament saw an average of 4.66 goals per match much to the delight of the spectators across the 10 host cities

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. L20:2469 strongly excites
  2. L21:1039 strongly suppresses
  3. L6:2728 suppresses
  4. L3:1552 excites
  5. L9:857 excites
  6. L8:2451 excites
  7. L3:877 suppresses
  8. L3:1202 excites
  9. L4:1812 suppresses
  10. L16:2303 suppresses
  11. L19:2619 weakly excites
  12. L3:2672 weakly suppresses
  13. L0:2525 weakly excites
  14. L14:175 weakly suppresses
  15. L8:1869 weakly suppresses
  16. L8:504 weakly excites
  17. L21:2155 weakly excites
  18. L14:2303 weakly suppresses
  19. L10:3043 weakly excites
  20. L7:1535 weakly suppresses
  21. L1:694 weakly excites
  22. L7:2335 weakly excites
  23. L20:1895 weakly excites
  24. L10:2350 weakly excites

Feeds later layers

  1. L23:1483 suppresses
  2. L23:2551 weakly suppresses baseball and sports mentions
  3. L25:2891 weakly suppresses
  4. L27:1546 weakly suppresses
  5. L23:2267 weakly suppresses
  6. L24:2314 weakly excites
  7. L27:1784 weakly excites
  8. L27:2772 weakly suppresses
  9. L25:2724 weakly suppresses
  10. L25:927 weakly suppresses
  11. L27:1249 weakly suppresses
  12. L25:129 weakly suppresses
  13. L25:3013 weakly suppresses
  14. L23:988 weakly suppresses
  15. L24:1608 weakly suppresses
  16. L27:1519 weakly suppresses
  17. L23:1031 weakly suppresses
  18. L24:2620 weakly excites
  19. L24:1939 weakly suppresses
  20. L23:987 weakly excites
  21. L26:1658 weakly suppresses
  22. L23:116 weakly suppresses
  23. L27:1766 weakly suppresses
  24. L23:93 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 -36.59 and -35.421 values between -16.67 and -15.491 values between -9.63 and -8.461 values between -7.29 and -6.121 values between -6.12 and -4.944 values between -4.94 and -3.779 values between -3.77 and -2.6020 values between -2.60 and -1.43826 values between -1.43 and -0.252984 values between -0.25 and 0.9238 values between 0.92 and 2.099 values between 2.09 and 3.268 values between 3.26 and 4.442 values between 4.44 and 5.611 values between 5.61 and 6.782 values between 9.13 and 10.30offnegativepositiveoften

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. Co
  2. (co
  3. Co
  4. co
  5. co
  6. -co
  7. /co
  8. -Co
  9. _co
  10. .assign

Suppresses

  1. nets
  2. .openg
  3. net
  4. emie
  5. aris
  6. 威胁
  7. threat
  8. giveaways
  9. ******************************************************************************⏎
  10. *******************************************************************************/⏎