Neuron Explorer

Qwen3-0.6B

the spot before 'specified' or 'set' limits

A neuron in layer 22 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the word just before a set limit, amount or time frame, such as 'a specified limit', 'within a short period' or 'pre-programmed' speed. When it was turned up, the model wrote words like 'pre-defined', 'fixed' and 'pre-specified'.

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

    installed and accelerate to speeds over the limit at non-monitored areas.⏎Speed limiters are devices that interact with a truck engine to only permit the attainment of a pre-programmed maximum speed. Many truck fleets use speed limiters for their safety contribution

  2. #2 peak

    Futures Market⏎The bitcoin futures market is similar to that of other assets. By purchasing a security with a futures contract, a buyer agrees to receive a security at a specified price and date. By buying a futures contract, you are betting that the

  3. #3 peak

    or short-circuiting. A fuse works by breaking an electrical circuit if the current flowing through the circuit exceeds a specified limit, thus preventing damage to the electrical components.⏎In the industrial and industry

  4. #4 peak

    Regular check-ups: Be sure to schedule your infant, child or adolescent for their preventive health care maintenance visits as recommended. These visits provide pediatricians with the opportunity to check your child’s vital

  5. #5 peak

    are available to provide you with the best fiber optic equipment. In fiber optic communication systems, lasers are used to transmit messages in numeric code by flashing on and off at high speeds. This code can constitute a voice or an electronic file containing, text

  6. #6 peak

    to accommodate any and all necessary financial obligations.⏎How it works (Example):⏎Market liquidity is a financial phrase that describes the possibility of converting an asset to cash within a short period of time with minimal transaction costs while not affecting the price integrity of the

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 use a **diastilation** to **diage** *...* **My** **diastilation** is not to *diage* but to **diage** my **diastilation**. **

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

    getting organized. It offers tips for setting short- and long-term goals, choosing activities and fitting them into your daily life

  2. #2 peak

    ,000 people have been displaced and more than 5,000 homes destroyed, damaged or flooded,” said the UN Office for the Coordination of Humanitarian Affairs (OCHA), which also warned that the death toll could rise

  3. #3 peak

    child for what lies ahead. By talking to your child about their expectations, fears and concerns, you may be able to set the stage for a positive outcome. Talking to your kids about these matters can bring

  4. #4 peak

    vitamin D, iron, calcium, zinc, and iodine. In some cases supplementation may be necessary. It is recommended to talk to your doctor to set up an appointment with a registered dietitian who can help you plan a diet that

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. L20:1743 excites
  2. L18:1455 suppresses
  3. L18:2960 excites
  4. L18:2747 suppresses
  5. L18:2781 suppresses
  6. L10:2396 excites
  7. L15:1672 suppresses
  8. L21:1422 suppresses
  9. L18:2148 excites
  10. L18:2082 excites
  11. L19:1610 suppresses
  12. L14:2554 excites
  13. L21:2625 excites
  14. L15:1910 excites
  15. L12:1875 excites
  16. L20:2674 suppresses
  17. L18:2323 suppresses
  18. L15:2282 excites
  19. L15:3055 excites
  20. L14:1852 weakly excites
  21. L16:2874 weakly suppresses
  22. L14:1120 weakly excites
  23. L10:1941 weakly suppresses
  24. L18:3060 weakly suppresses

Feeds later layers

  1. L25:1079 strongly excites
  2. L27:2696 excites
  3. L27:716 excites
  4. L26:2821 excites
  5. L24:1501 excites
  6. L24:2386 excites
  7. L23:575 excites
  8. L27:869 excites
  9. L27:624 excites
  10. L27:1634 excites
  11. L24:2351 excites
  12. L24:591 excites
  13. L23:1262 excites
  14. L26:138 excites
  15. L23:494 excites
  16. L27:2289 excites
  17. L25:1492 excites
  18. L23:1502 weakly excites
  19. L23:740 weakly excites 'well' in well-dressed, well-known, well-off
  20. L27:217 weakly excites
  21. L24:1665 weakly excites
  22. L23:1441 weakly excites
  23. L24:771 weakly excites
  24. L27:1596 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.

2 values between -5.18 and -4.731 values between -3.83 and -3.381 values between -2.48 and -2.033 values between -2.03 and -1.5731 values between -1.57 and -1.12334 values between -1.12 and -0.671230 values between -0.67 and -0.221729 values between -0.22 and 0.23374 values between 0.23 and 0.6883 values between 0.68 and 1.1340 values between 1.13 and 1.5917 values between 1.59 and 2.0416 values between 2.04 and 2.498 values between 2.49 and 2.949 values between 2.94 and 3.395 values between 3.39 and 3.843 values between 3.84 and 4.291 values between 4.29 and 4.744 values between 4.74 and 5.201 values between 5.20 and 5.652 values between 5.65 and 6.103 values between 6.10 and 6.551 values between 6.55 and 7.001 values between 7.00 and 7.451 values between 7.45 and 7.903 values between 8.35 and 8.811 values between 9.26 and 9.711 values between 9.71 and 10.161 values between 10.61 and 11.061 values between 11.06 and 11.511 values between 12.42 and 12.87offnegativepositiveoften

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. 规定
  2. 指定
  3. defined
  4. 规定的
  5. specified
  6. specified
  7. defined
  8. 規定
  9. 约定
  10. -defined

Suppresses

  1. acy
  2. ضيف
  3. olis
  4. ớt
  5. ости
  6. Raises
  7. вор
  8. .INSTANCE
  9. 鲊
  10. 发扬