Neuron Explorer

Qwen3-0.6B

prepositions opening a phrase ('after', 'on')

A neuron in layer 21 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 fires on short connecting words like 'on', 'after' and 'of' that lead into a phrase, such as 'Based on' or 'and each day after that'. When it was turned up, the model kept producing the word 'this' over and over.

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

    or does not have an aneuploidy. Recent recommendations from the American College of Obstetricians and Gynecologists (ACOG) are quite clear on that issue.⏎In those same recommendations, ACOG also states

  2. #2 peak

    .⏎- Have them develop an explicit formula for each sequence.⏎2. A company is offering to pay you one penny on your first day at work and each day after that your salary would triple. Assume you work five days a week.⏎a.

  3. #3 peak

    movements and synthesize the speed, so that the movement required for tooth surface cutting is transformed from the rotary movement to the linear movement as far as possible. Based on the analysis of the tooth surface cutting movement, the revolution and rotation of the blade

  4. #4 peak

    you would need to⏎pick out 15 sweets because the first 9 could be mice and the next 5⏎you pick out could be frogs and after that you're definitely going⏎to pick out a worm. As well as that

  5. #5 peak

    But, a bill in Congress would do just that. And if it passes, not just farmers and ranchers would be affected

  6. #6 peak

    operate kindergarten-through–12th grade (K–12) schools across the United States is creating safe and secure environments that support effective teaching and learning. To help them meet this obligation, the U.S

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 way in a **pre-sim** (pre-**sim************) **** **** **** **** **** **** **** **** **** **** **** **** **** **

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

    and our research aims to use statistical techniques and other methods such as artificial intelligence to interpret large data sets.⏎“While we have this mathematical and statistical knowledge, what’s missing is the framework that can capture all of this complex

  2. #2 peak

    This pressure causes them to forget even the things they know. Students who face math anxiety can use find the limit calculator for help.⏎- Poor Foundation⏎Foundation skills are very important when it comes to learning and solving hardcore mathematical problems. If a student

  3. #3 peak

    attention to the connectivity of the components in a circuit diagram. Arrows, lines and other symbols should be used to show how different components are connected. This information is crucial for figuring out the operation of the circuit.⏎By following the above circuit diagram

  4. #4 peak

    . It is possible to send and receive bitcoins using a Bitcoin wallet.⏎Even when the value of an asset drops, it’s possible to make a profit, which is counterintuitive

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. L18:2923 excites
  2. L19:1450 excites
  3. L19:1204 weakly suppresses
  4. L18:769 weakly excites
  5. L19:2138 weakly excites "of" in "matter of" phrases
  6. L17:1374 weakly excites
  7. L9:1995 weakly suppresses
  8. L15:3019 weakly suppresses
  9. L19:2052 weakly excites
  10. L20:593 weakly excites
  11. L18:2153 weakly suppresses
  12. L19:1759 weakly excites
  13. L3:2858 weakly suppresses
  14. L19:2558 weakly excites
  15. L19:1181 weakly suppresses
  16. L19:858 weakly suppresses
  17. L13:2159 weakly suppresses
  18. L11:163 weakly suppresses
  19. L16:1059 weakly excites
  20. L19:786 weakly excites
  21. L16:89 weakly excites
  22. L3:2629 weakly excites
  23. L6:828 weakly excites
  24. L19:1914 weakly suppresses

Feeds later layers

  1. L26:405 strongly excites
  2. L27:528 strongly excites
  3. L22:1469 strongly excites
  4. L24:803 excites
  5. L26:586 excites
  6. L26:1191 suppresses
  7. L26:64 weakly suppresses
  8. L23:1068 weakly suppresses
  9. L23:2734 weakly excites
  10. L24:306 weakly excites
  11. L24:859 weakly suppresses
  12. L26:1052 weakly suppresses
  13. L26:2261 weakly excites
  14. L27:1055 weakly excites
  15. L27:616 weakly excites
  16. L22:735 weakly excites
  17. L24:343 weakly excites
  18. L25:185 weakly excites comma after a city in an address
  19. L24:1975 weakly excites
  20. L25:2653 weakly excites
  21. L22:2794 weakly suppresses
  22. L23:334 weakly excites
  23. L22:1031 weakly excites
  24. L24:2045 weakly suppresses acronym introductions

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 -7.10 and -6.671 values between -6.24 and -5.811 values between -5.37 and -4.942 values between -4.94 and -4.515 values between -4.08 and -3.659 values between -3.65 and -3.2210 values between -3.22 and -2.7814 values between -2.78 and -2.3523 values between -2.35 and -1.9235 values between -1.92 and -1.4961 values between -1.49 and -1.06122 values between -1.06 and -0.62480 values between -0.62 and -0.192190 values between -0.19 and 0.24793 values between 0.24 and 0.67101 values between 0.67 and 1.1017 values between 1.10 and 1.5415 values between 1.54 and 1.979 values between 1.97 and 2.404 values between 2.40 and 2.831 values between 2.83 and 3.263 values between 3.26 and 3.693 values between 3.69 and 4.131 values between 4.13 and 4.563 values between 4.99 and 5.422 values between 5.42 and 5.851 values between 8.01 and 8.451 values between 9.74 and 10.17offnegativepositiveoften

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. this
  2. 此
  3. this
  4. This
  5. .this
  6. [this
  7. 这个
  8. =this
  9. This
  10. (this

Suppresses

  1. assin
  2. Serviço
  3. üş
  4. 说自己
  5. 一所
  6. 的画面
  7. reeNode
  8. artz
  9. 的日
  10. ******/⏎⏎