Neuron Explorer

Qwen3-0.6B

'at' in 'at least' and 'minimum'

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 responds to the word 'at' when it comes before 'least', as in 'at least 70%', and it leans toward the idea of a minimum. When it was turned up, the model kept bringing up the word 'minimum'.

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

    of dog food, but all of them must meet certain requirements to be considered “real” dog food. For example, the food must have a meat content of at least 70% and it must not contain any artificial colors or flavors.

  2. #2 peak

    across disparate operating systems.⏎Kubernetes clusters are actually a set of nodes categorized into master nodes and worker nodes. Every cluster needs to have at least one of each. While the master node is meant to control the state of

  3. #3 peak

    least twenty-five years old upon assuming office. No person can serve as attorney general unless he or she has been an active member of the State Bar of Georgia for at least seven years.⏎The role of attorney general has been considered crucial from the outset

  4. #4 peak

    whose skeletal structures are still in the developmental stage. For them, insufficient vitamin levels can result in rickets and other skeletal malformations. All adults must get at least minimal levels of vitamin D to absorb sufficient calcium to maintain healthy bones. Post

  5. #5 peak

    drum is only one portion of being a drummer for most jazz, country and western, rhythm and blues, and rock and roll bands. A drum set generally features at minimum, one snare, two toms, a base drum and a high

  6. #6 peak

    : a minimum age of 16 for a learner's permit; a waiting period of at least six months before a driver with a learner's permit could apply for a provisional

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: how to make a complex yet beautiful, yet still open to change. What does that mean to you, or might it be unlimited, unlimited, unlimited? In the form of an article, can you make an article about this concept? In

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

    countries adopted the Paris Agreement, an international treaty on climate change, a precursor to the Glasgow conference. These countries have agreed on a goal of limiting global warming to an increase of 2 degrees Celsius, and preferably 1.5 degrees Celsius,

  2. #2 peak

    test the motor windings for continuity. If the windings do not have continuity, replace the evaporator fan motor.<|endoftext|>A fraction is a number that has a specific value. The percentage value

  3. #3 peak

    regulated by the U.S. Environmental Protection Agency (EPA). In 1986, EPA established a maximum allowable concentration for fluoride in drinking water of 4 milligrams per liter (mg/L), a guideline designed to prevent the public

  4. #4 peak

    2015). Loma Linda University Electronic Theses, Dissertations & Projects. Paper 294.⏎Loma Linda University Electronic Theses and Dissertations⏎Loma Linda University. Del E. Webb Memorial Library.

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. L20:2348 strongly excites "at least" and minimum-quantity phrases
  2. L16:2990 excites
  3. L20:425 suppresses
  4. L19:1171 excites
  5. L15:1910 excites
  6. L15:1963 excites
  7. L17:2887 excites
  8. L12:2847 excites
  9. L21:1558 suppresses
  10. L12:2102 suppresses
  11. L22:999 suppresses
  12. L21:1977 suppresses
  13. L21:725 excites
  14. L13:1711 suppresses
  15. L22:1104 weakly suppresses
  16. L21:1324 weakly suppresses
  17. L17:2568 weakly excites
  18. L21:1001 weakly excites
  19. L22:3027 weakly suppresses
  20. L11:1048 weakly excites
  21. L17:2446 weakly suppresses
  22. L21:544 weakly excites
  23. L18:2148 weakly excites
  24. L7:1387 weakly excites

Feeds later layers

  1. L27:863 strongly excites
  2. L24:812 strongly excites
  3. L25:2221 excites
  4. L27:1127 excites
  5. L25:1182 excites
  6. L24:748 suppresses
  7. L26:1951 suppresses
  8. L27:1711 weakly excites
  9. L24:1197 weakly excites
  10. L27:1465 weakly excites
  11. L27:2173 weakly excites
  12. L26:910 weakly excites
  13. L27:1060 weakly suppresses
  14. L26:2537 weakly excites
  15. L24:2125 weakly excites
  16. L24:814 weakly suppresses
  17. L27:1525 weakly excites
  18. L25:2378 weakly excites
  19. L24:3005 weakly suppresses
  20. L25:2870 weakly suppresses
  21. L27:2665 weakly excites
  22. L26:2034 weakly excites
  23. L25:982 weakly suppresses
  24. L25:2700 weakly excites

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 -5.04 and -4.4821 values between -1.70 and -1.15281 values between -1.15 and -0.591933 values between -0.59 and -0.041548 values between -0.04 and 0.5296 values between 0.52 and 1.0811 values between 1.08 and 1.636 values between 1.63 and 2.191 values between 2.19 and 2.742 values between 2.74 and 3.302 values between 3.30 and 3.851 values between 3.85 and 4.412 values between 4.97 and 5.521 values between 7.74 and 8.301 values between 12.74 and 13.301 values between 16.63 and 17.19offnegativepositiveoften

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. min
  2. min
  3. minimum
  4. -min
  5. 最小
  6. _min
  7. (min
  8. Min
  9. .min
  10. ,min

Suppresses

  1. strup
  2. unlimited
  3. maxi
  4. 宏大
  5. ionate
  6. getMax
  7. ecture
  8. opic
  9. maximize
  10. peaked