Neuron Explorer

Qwen3-0.6B

vehicles, transport and movement words

A neuron in layer 7 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 responds to words about getting things or people from one place to another, such as passenger, transport, delivery and vehicle speed. It is strongly negative on words about turning or steering a vehicle or aircraft. Turning it up pushed the model toward unrelated stories and word lists.

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

    In many countries the maximum speed of commercial vehicles are restricted by the installation of Road Speed Limiters. Road Speed Limiters were introduced in Europe in commercial trucks and passenger service vehicles initially to curb the maximum speed as this was seen as a major threat

  2. #2 peak

    their PTI experience⏎- Increases driver’s experience and goes some way to reduce stress and pressure.⏎- Assists the driver to maintain on-time services by reducing passenger boarding dwell time, further helping to reduce safety related incidents.⏎Train and Station Operators

  3. #3 peak

    developing an electric car that will break – no, shatter – the existing E-1 class electric land speed record of 204 mph. By combining aerodynamics with new, cutting-edge green energy technology, the Electric Land Speed Record Car

  4. #4 peak

    ATIONAL AIR DELIVERY⏎Cargo delivery is carried out in two ways: air delivery or sea delivery. Air transportation is the most convenient, fast and reliable way of

  5. #5 peak

    the bridge over the canal. The first bridge was a transporter bridge, on of only a few ever built, and had a gondola which was raised and transported across the top span.⏎By the 1920s the bridge could

  6. #6 peak

    ant. Criteria for vehicle evaluation include safety, ability to reach diversion airports, adaptability to different sonic boom restrictions and technological maturation. Steel

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: a uniform circular motion. What is the relationship between the speed of the plane and the radius of the turn in the plane at constant speed? The answer is a combination of both the rate of rotation and the radius of the turn in the plane

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

    at all.⏎At the International Conference on Robotics and Automation tomorrow, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) will present a new lane-change algorithm that splits the difference. It allows for more aggressive lane changes than the simple

  2. #2 peak

    greater capability to more rapidly detect chemical, biological or nuclear hazards.”⏎Tailbot's tail can be maneuvered in the air to help stabilize the robot as it jumps from one surface to

  3. #3 peak

    . turn indicator-An instrument for indicating the existence and approximate magnitude of angular velocity about the normal axis of an aircraft. turnmeter-An instrument that measures the rate of turn of an aircraft about any predetermined axis. ultimate load-See LOAD, ULTIM

  4. #4 peak

    record data during test flights.⏎While the plane is still on the ground, you design and develop measuring devices that take temperature, noise, altitude, navigation, and maneuverability readings. Then you hard-wire the devices into place, and send 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. L6:2606 suppresses
  2. L6:18 excites
  3. L6:6 excites
  4. L5:2463 suppresses
  5. L6:2898 weakly suppresses
  6. L6:1185 weakly excites
  7. L3:1203 weakly excites
  8. L6:2549 weakly excites
  9. L4:2726 weakly suppresses
  10. L3:1202 weakly excites
  11. L6:2838 weakly suppresses
  12. L0:1301 weakly suppresses
  13. L6:1055 weakly excites
  14. L4:1903 weakly excites
  15. L6:2398 weakly suppresses
  16. L6:2852 weakly excites
  17. L3:2369 weakly suppresses
  18. L6:17 weakly suppresses
  19. L6:1966 weakly suppresses
  20. L6:1226 weakly suppresses
  21. L6:2619 weakly excites
  22. L6:1772 weakly suppresses
  23. L4:2729 weakly excites
  24. L0:1224 weakly suppresses

Feeds later layers

  1. L18:1592 suppresses
  2. L25:2367 suppresses
  3. L24:1026 suppresses
  4. L17:1817 suppresses
  5. L11:934 suppresses
  6. L10:962 suppresses
  7. L24:971 suppresses
  8. L20:2474 suppresses
  9. L16:1660 suppresses
  10. L9:1220 suppresses
  11. L17:1581 suppresses
  12. L21:629 suppresses
  13. L19:1621 weakly suppresses
  14. L21:1761 weakly suppresses
  15. L9:2837 weakly suppresses
  16. L25:2935 weakly suppresses
  17. L22:3007 weakly suppresses
  18. L25:1943 weakly suppresses
  19. L9:2310 weakly suppresses
  20. L16:1678 weakly suppresses
  21. L20:2330 weakly suppresses
  22. L24:2678 weakly suppresses
  23. L22:1213 weakly suppresses
  24. L19:2203 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 -2.13 and -2.031 values between -1.93 and -1.831 values between -1.62 and -1.521 values between -1.42 and -1.321 values between -1.22 and -1.121 values between -1.12 and -1.022 values between -0.92 and -0.823 values between -0.82 and -0.725 values between -0.72 and -0.626 values between -0.62 and -0.523 values between -0.52 and -0.4220 values between -0.42 and -0.3272 values between -0.32 and -0.22318 values between -0.22 and -0.121079 values between -0.12 and -0.021403 values between -0.02 and 0.09661 values between 0.09 and 0.19216 values between 0.19 and 0.2988 values between 0.29 and 0.3913 values between 0.39 and 0.495 values between 0.49 and 0.592 values between 0.59 and 0.691 values between 0.79 and 0.892 values between 1.19 and 1.291 values between 1.49 and 1.601 values between 1.60 and 1.701 values between 1.80 and 1.90offnegativepositiveoften

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. bsites
  2. ество
  3. tery
  4. eger
  5. 第二届
  6. AsStream
  7. ,LOCATION
  8. igroup
  9. 优势
  10. posable

Suppresses

  1. pitches
  2. 匀
  3. 转弯
  4. pitching
  5. maneuvers
  6. curvature
  7. manoe
  8. pitch
  9. yaw
  10. thrust