Neuron Explorer

Qwen3-0.6B

end-of-document boundary token

A neuron in layer 26 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 invisible marker that separates one web document from the next, and on the spot right before it. Turning it up made the text wander into unrelated topics and repeated words, so the steering results do not show a clear meaning.

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

    two different numbers cant have the same cube). Therefore, by Definition 1, is one-to-one....⏎View Full Document⏎- Spring '09<|endoftext|>part of speech: noun⏎definition 1: an upright line or pole with

  2. #2 peak

    analysis can be obtained through finite element analysis; in this research analytical solution approaches will be used instead. The steady-state temperature distribution inside the machine can be formulated as Poisson’s equation. Geometry is assumed to be circular and windings are modeled as

  3. #3 peak

    Chapter 6 Exercise 6.5, we have discussed the Pythagoras theorem. Other important topics related to the chapter are covered in the following exercises.<|endoftext|>What you will learn in this module⏎Understand different tasks that have to be

  4. #4 peak

    ions iv. Special Limits c. Will be seen later d. If none of the top three apply, try using the Squeeze Theorem....⏎View Full Document<|endoftext|>A Woman of Salt, by Mary Potter Engel⏎Midrash refers to the

  5. #5 peak

    can also practice Chemistry: An Atoms First Approach - Zumdahl Atoms 1st 2nd Edition practice problems.<|endoftext|>The legal system of imperial China developed from two schools of thought: Confucian

  6. #6 peak

    them use unknowns for the sequence terms starting with the first given term. Solve the resulting equation. See the hint at the end of Example 1b.<|endoftext|>This five-part series chronicles the history of the global anti-apartheid movement

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: human styles manifest as manifest styles manifest as manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles

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

    ' developed. It looks at intersection of algebraic geometry and computer algebra. It is concerned with the development of algorithms and software for studying and finding the properties of explicitly given algebraic varieties.⏎Much of the development of the main stream of algebraic

  2. #2 peak

    extended-spectrum beta-lactamase with additional resistance to carbenicillin, amoxicillin, and increasingly to ceftazidime.

  3. #3 peak

    algebra, especially commutative algebra, with the language and the problems of geometry.⏎The main objects of study in algebraic geometry are algebraic varieties, which are geometric manifestations of sets of solutions of systems of polynomial equations. Examples of the most

  4. #4 peak

    and contemporaries, including the Roman emperor Caesar Augustus, queen Cleopatra of Egypt and the Nabataean king Obodas III, issued coins emblazoned with their portraits.⏎But as the ruler of a Jewish kingdom where the

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. L20:293 suppresses
  2. L9:41 excites
  3. L18:408 excites
  4. L3:25 suppresses
  5. L5:78 excites
  6. L20:238 excites
  7. L3:461 excites
  8. L21:128 excites
  9. L16:566 suppresses unclear: mid-sentence word choices, mixed
  10. L10:276 excites
  11. L9:96 excites
  12. L10:291 suppresses
  13. L9:693 suppresses
  14. L9:65 suppresses
  15. L17:133 excites
  16. L23:115 excites
  17. L7:334 excites
  18. L8:77 excites
  19. L10:2217 suppresses
  20. L19:746 excites
  21. L9:117 excites
  22. L3:105 excites
  23. L21:345 suppresses
  24. L4:739 suppresses

Feeds later layers

  1. L27:182 weakly excites
  2. L27:2963 weakly suppresses the 'st' piece inside words
  3. L27:2810 weakly suppresses
  4. L27:1606 weakly suppresses
  5. L27:164 weakly excites
  6. L27:2572 weakly suppresses
  7. L27:2297 weakly suppresses
  8. L27:2619 weakly excites
  9. L27:546 weakly excites
  10. L27:1955 weakly suppresses
  11. L27:1804 weakly suppresses
  12. L27:536 weakly excites
  13. L27:2086 weakly suppresses
  14. L27:2898 weakly excites
  15. L27:2867 weakly excites
  16. L27:2374 weakly suppresses
  17. L27:510 weakly suppresses
  18. L27:2060 weakly suppresses
  19. L27:2274 weakly suppresses
  20. L27:1049 weakly suppresses
  21. L27:1799 weakly suppresses
  22. L27:3053 weakly suppresses
  23. L27:1404 weakly suppresses
  24. L27:2473 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.

3 values between -12.54 and -11.571 values between -11.57 and -10.602 values between -9.63 and -8.674 values between -8.67 and -7.703 values between -6.73 and -5.765 values between -5.76 and -4.793 values between -4.79 and -3.8214 values between -3.82 and -2.8524 values between -2.85 and -1.89104 values between -1.89 and -0.921857 values between -0.92 and 0.051682 values between 0.05 and 1.0297 values between 1.02 and 1.9941 values between 1.99 and 2.9624 values between 2.96 and 3.938 values between 3.93 and 4.9013 values between 4.90 and 5.8610 values between 5.86 and 6.832 values between 6.83 and 7.801 values between 7.80 and 8.771 values between 8.77 and 9.741 values between 10.71 and 11.682 values between 11.68 and 12.642 values between 12.64 and 13.611 values between 13.61 and 14.581 values between 15.55 and 16.521 values between 16.52 and 17.491 values between 25.24 and 26.20offnegativepositiveoften

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. ду
  2. stein
  3. Università
  4. cour
  5. orse
  6. воль
  7. ши
  8. ится
  9. 셀
  10. wróci

Suppresses

  1. ihm
  2. tlement
  3. วย
  4. .pipe
  5. .GetObject
  6. 的道理
  7. pipe
  8. 道理
  9. .Default
  10. rails