Neuron Explorer

Qwen3-0.6B

limits, thresholds and how long it takes

A neuron in layer 13 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 words that lead into an amount, limit or time span, such as 'how long it will take', 'the maximum mass that' or 'until about 25 degrees'. When it was turned up, the model's replies became repetitive and drifted off topic, so the steering result tells us little about the neuron itself.

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

    over time, even after being recharged 30,000 times. Conversely, lithium-ion batteries suffer in performance after about 1,500 charging cycles

  2. #2 peak

    least, a poorly planned wheel assembly can shut down operations, causing unnecessary delays and need for repair.⏎Safe working load⏎A safe working load is the maximum mass that a product can carry without

  3. #3 peak

    percentage such as 0.10 percent, which indicates that one-tenth of one percent of a person’s blood is alcohol. The amount of alcohol it takes to reach a concentration of .08 percent varies on a person’s size,

  4. #4 peak

    ), with no profit indicated under existing conditions. The latter area would show a profit equal to the first-mentioned area only with an increase in price of crude oil of forty-five cents per barrel.⏎Subsurface waters at depths of 1,4

  5. #5 peak

    .458 inch.⏎Bullet weight: 250 grains.⏎Bullet ballistic coefficient: .175.⏎Rifle zero: 135 yards, 2″ mid-range rise.⏎|RANGE||MUZZLE||5

  6. #6 peak

    fabric.⏎- SPF: On the other hand, SPF (Sun Protection Factor) serves as a timetable of sorts, allowing you to know about how long it will take before your skin begins to redden while using a product that features it, such

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: the user. the assistant. the assistant. the assistant. the assistant. the assistant. the assistant. the assistant. the assistant. the assistant. the person. the person. the person. the person, the person, the person,

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

    in the air. The oxidation of the combustible material creates heat. If this heat cannot be dissipated, it will build up in the combustible material until ignition occurs. Generally, the build up of heat to ignition point occurs when the material

  2. #2 peak

    in salt into their models of the ocean, they found the situation changed from what was previously thought. With enough salt, liquid in Ganymede can become dense enough to sink to the very bottom of the seafloor, below Ice VI.

  3. #3 peak

    materials in this archive are copyright © The Manitoba Library Association. Reproduction for personal use is permitted only if this copyright notice is maintained. Any other reproduction is prohibited without permission Copyright information for reviewers⏎Young Canada Works<|endoftext|>From the Publisher:⏎Over roughly

  4. #4 peak

    assume that we have a more general problem. There are n questions, and a(n) is the smallest number of times we need to take the test to guarantee that we can figure out the answers. First we can try all combinations of answers.

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. L12:2492 excites
  2. L7:1644 excites 'to' after a noun that needs a purpose clause
  3. L12:1088 excites
  4. L11:1048 excites
  5. L12:1136 excites
  6. L12:1470 suppresses
  7. L12:3017 excites
  8. L11:1212 suppresses
  9. L12:517 suppresses
  10. L8:1641 suppresses
  11. L7:2108 excites
  12. L11:524 excites
  13. L11:1367 excites
  14. L10:212 excites
  15. L11:346 excites
  16. L12:282 excites
  17. L12:2062 weakly excites
  18. L12:2555 weakly suppresses
  19. L10:1512 weakly excites
  20. L8:2748 weakly suppresses
  21. L9:1346 weakly excites
  22. L12:2519 weakly suppresses
  23. L12:1970 weakly suppresses
  24. L10:2026 weakly suppresses

Feeds later layers

  1. L18:2148 excites
  2. L14:2911 excites
  3. L14:1151 excites
  4. L20:2529 excites
  5. L15:2308 excites
  6. L14:987 excites
  7. L14:1854 excites
  8. L15:1911 excites
  9. L18:1704 excites
  10. L16:2574 excites
  11. L14:1540 excites
  12. L21:2204 suppresses
  13. L21:2055 excites
  14. L14:2668 weakly excites
  15. L17:2246 weakly excites
  16. L17:1529 weakly suppresses
  17. L19:2132 weakly excites
  18. L18:2757 weakly excites
  19. L14:2767 weakly suppresses
  20. L14:2796 weakly excites
  21. L15:2032 weakly excites
  22. L19:2139 weakly excites
  23. L14:2197 weakly excites
  24. L14:2612 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 -1.01 and -0.814 values between -0.81 and -0.6141 values between -0.61 and -0.42365 values between -0.42 and -0.221741 values between -0.22 and -0.021527 values between -0.02 and 0.18176 values between 0.18 and 0.3718 values between 0.37 and 0.578 values between 0.57 and 0.776 values between 0.77 and 0.966 values between 0.96 and 1.162 values between 1.16 and 1.364 values between 1.36 and 1.562 values between 1.56 and 1.753 values between 1.75 and 1.951 values between 2.34 and 2.541 values between 3.92 and 4.121 values between 5.10 and 5.301 values between 6.68 and 6.88offnegativepositiveoften

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. 条件
  2. ines
  3. 阈
  4. Threshold
  5. bec
  6. ือ
  7. .Accept
  8. ȋ
  9. 才能
  10. 條件

Suppresses

  1. @update
  2. IVEN
  3. 猢
  4. costs
  5. anonymous
  6. 之情
  7. unce
  8. vais
  9. lycer
  10. 之心