Neuron Explorer

Qwen3-0.6B

unclear: words mid-phrase before a fixed continuation

A neuron in layer 11 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 is hard to read. It turns up on words in the middle of a phrase, like 'their', 'two', 'mean' or 'peace' right before the rest of a set expression. It goes strongly negative on words like 'state' or 'language' that finish a noun phrase such as 'most populous state'. Turning it up or down did not produce a clear theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    ceptions of transcendental geometry the basis for an amusing story.⏎The very short article goes on to compare “Flatland” with “Through the Looking Glass” and their use of geometry as speculative and imaginative.⏎In trying to find more about this term

  2. #2 peak

    prosecution (sense 2)): from Outdated French, or from late Latin prosecutio(n-), from prosequi ‘pursue, accompany’ (see prosecute). Customary Application for Employment It’s our policy to adjust to all

  3. #3 peak

    like older women in the United Kingdom are using abortion like a “back-up” form of birth control even though preventing a pregnancy and killing a unique human life are two very different things. These women really need to be informed of the risks associated with

  4. #4 peak

    Antoine Pierre Joseph Marie (äNtwänˈ pyĕr zhōzĕfˈ märēˈ bärnävˈ) [key], 1761–93, French revolutionary. A

  5. #5 peak

    gods that others believe in, then we would regard them as false gods.⏎The second part of this declaration is that we believe that the Holy Prophet Muhammad, may peace and blessings of Allah be upon him, is the Messenger of Allah. Since he

  6. #6 peak

    what and how much you eat. Do you know the difference between serving and portion sizes? While the terms serving and portion often are used interchangeably, they actually mean different things.⏎A "serving" is the amount of food recommended in consumer

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 the course, and I have been working on the course. I have been working on the course. I have been working on the course. I have been working on the course. I have been working on the course

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

    before the Great Depression brought it all to an end. Florida's economy didn't fully recover until World War II. But today, it's the most populous state in the South besides Texas, and the fourth most populous in the United States.⏎

  2. #2 peak

    result in animals with calcium and phosphorus deficiencies. In young<|endoftext|>Like most states in the Midwest, Indiana has bountiful wind resources. The Hoosier State is 13th in the U.S. in installed wind energy capacity,

  3. #3 peak

    nearly every databases – MySQL, ORACLE, PostgreSQL, Sybase, Informix to name a few.⏎What is PHP?⏎- PHP is a server-side scripting language that means its code is executed on

  4. #4 peak

    explicitly non-GMO products for their in-store brands. Because so many Ohio growers depend upon certified non-GMO seed for their livelihood, concern is growing in the state

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. L5:1378 suppresses
  2. L10:2890 suppresses
  3. L9:590 excites
  4. L8:2143 suppresses
  5. L7:724 excites
  6. L7:1437 suppresses
  7. L9:608 excites
  8. L10:2920 excites
  9. L9:1390 excites
  10. L10:279 suppresses
  11. L8:2544 weakly suppresses
  12. L7:2811 weakly excites
  13. L10:2518 weakly excites
  14. L5:2566 weakly suppresses
  15. L7:2432 weakly excites
  16. L10:2094 weakly excites
  17. L10:1007 weakly suppresses
  18. L10:1631 weakly suppresses
  19. L10:2633 weakly suppresses
  20. L9:85 weakly excites
  21. L10:2062 weakly excites
  22. L10:2489 weakly suppresses
  23. L10:753 weakly suppresses
  24. L6:2082 weakly suppresses

Feeds later layers

  1. L12:1761 suppresses
  2. L13:2409 suppresses
  3. L12:1478 suppresses
  4. L15:1548 suppresses
  5. L15:2393 suppresses
  6. L12:1184 suppresses
  7. L14:1229 excites
  8. L14:2655 excites
  9. L13:2601 suppresses
  10. L13:904 suppresses
  11. L13:2333 excites
  12. L13:2538 suppresses
  13. L18:2141 excites
  14. L16:2057 excites
  15. L12:570 suppresses
  16. L12:1135 suppresses
  17. L14:1805 excites
  18. L14:2067 excites
  19. L13:1015 excites
  20. L13:2520 excites
  21. L13:1979 excites
  22. L13:2233 weakly suppresses
  23. L15:621 weakly excites
  24. L12:363 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 -4.95 and -4.771 values between -4.04 and -3.861 values between -3.68 and -3.501 values between -3.32 and -3.142 values between -2.96 and -2.781 values between -2.78 and -2.602 values between -2.60 and -2.424 values between -2.42 and -2.241 values between -2.24 and -2.066 values between -2.06 and -1.883 values between -1.88 and -1.708 values between -1.70 and -1.526 values between -1.52 and -1.3411 values between -1.34 and -1.1616 values between -1.16 and -0.9821 values between -0.98 and -0.8038 values between -0.80 and -0.6155 values between -0.61 and -0.43123 values between -0.43 and -0.25469 values between -0.25 and -0.072099 values between -0.07 and 0.11592 values between 0.11 and 0.29235 values between 0.29 and 0.4792 values between 0.47 and 0.6551 values between 0.65 and 0.8328 values between 0.83 and 1.0110 values between 1.01 and 1.1912 values between 1.19 and 1.377 values between 1.37 and 1.552 values between 1.55 and 1.735 values between 1.73 and 1.914 values between 1.91 and 2.091 values between 2.09 and 2.27offnegativepositiveoften

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. Pairs
  2. ấu
  3. чат
  4. drivers
  5. ters
  6. Families
  7. Regressor
  8. lesc
  9. _pairs
  10. tle

Suppresses

  1. ITY
  2. short
  3. CED
  4. 已久
  5. amic
  6. 之
  7. .vs
  8. extended
  9. extended
  10. Vers