Neuron Explorer

Qwen3-0.6B

'in the process of' and continuing actions

A neuron in layer 10 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 that set up an ongoing action, such as 'in the process [of]', 'being [built]' and 'want [to]'. Its strongest negative side is on these same phrases. Turning it up did not give a clear topic.

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

    , exascale computing requires breakthrough innovation and re-designing of processors. The country had crossed the petaflop computer range and needs exascale computers, which are machines capable of performing 1000 trillion operations a second."⏎

  2. #2 peak

    enough for the subject to overhear: "Excuse me. I'm a student in the psychology department. We're doing some experiments right now and need subjects. The whole thing takes about 15 minutes. I wonder if you would be

  3. #3 peak

    the Outlook window, choose the transferred particulars folder.⏎- Then open the email that you want to recall. You must double-click to open the communication. The Recall option isn’t

  4. #4 peak

    are busy trying to harness the power of the immune system in the fight against cancer. The HPV vaccine is just one of the first; expect to hear more and more about the topic in the future.⏎The HPV vaccine is a traditional vaccine in the

  5. #5 peak

    4) DOE expenditures to modify the Hanford facility in preparation of RTG assembly.⏎GAO found that: (1) DOE plans to continue RTG assembly and testing at its Mound facility through the completion of the National Aeronautics

  6. #6 peak

    manned and unmanned missions to the Moon, NASA is worried that the Apollo landing sites and others could be endangered by the next wave of lunar explorers. To prevent this, the space agency issued a set of guidelines that politely asks everybody to keep their

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: I am a new teacher in a small town, and I have been working with a group of students. I am the teacher who is taking a class in a small town. The class is taking a video of a teacher who is holding a

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

    007, a Protocol for the Convention on Illicit Trade in Tobacco Products is currently being negotiated. Meanwhile, the WHO Report on Global Tobacco Epidemic 200

  2. #2 peak

    , in the northern beaches area of Sydney (as well as the Western suburbs of Sydney), and the South Coast of NSW. The LDAA is in the process of mapping instances of Lyme disease and known locations of tick bites that led to Lyme disease

  3. #3 peak

    that established green building practices for its own public buildings and for new private developments. The Alexandria County Public School (ACPS) system in Virginia is in the process of modernizing the aging elementary facilities in its portfolio. The first school being built with

  4. #4 peak

    American Council of Learned Societies, Society of Architectural Historians, Mellon Foundation, Whiting Foundation, and the Ambrose Monell Foundation. She is currently writing a history of postwar preservation and urban renewal based upon the Philadelphia neighborhood of Society

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. L4:2065 strongly suppresses
  2. L5:2777 strongly suppresses
  3. L7:654 suppresses
  4. L8:1880 excites
  5. L7:1892 suppresses
  6. L6:1954 excites
  7. L8:2488 excites
  8. L7:2780 suppresses
  9. L7:1403 excites 'to' after attempt or seek verbs
  10. L7:1173 suppresses
  11. L6:2440 suppresses
  12. L7:1472 suppresses
  13. L7:2692 excites
  14. L6:1993 excites
  15. L6:1505 suppresses
  16. L5:2426 excites
  17. L8:255 excites
  18. L9:508 excites
  19. L9:2508 suppresses
  20. L7:951 weakly excites "in" right before a year
  21. L7:277 weakly suppresses
  22. L8:1416 weakly suppresses
  23. L4:2470 weakly suppresses
  24. L9:2989 weakly excites

Feeds later layers

  1. L23:1900 suppresses
  2. L18:2640 suppresses
  3. L11:2050 suppresses
  4. L11:1644 suppresses
  5. L18:2594 suppresses
  6. L22:2477 suppresses
  7. L12:1989 excites
  8. L22:1063 suppresses
  9. L17:3034 suppresses
  10. L12:1257 suppresses
  11. L21:2726 suppresses
  12. L12:952 suppresses
  13. L11:2589 excites
  14. L12:2761 suppresses
  15. L19:1348 suppresses
  16. L16:2226 suppresses
  17. L14:2297 suppresses
  18. L11:1669 excites
  19. L11:811 suppresses
  20. L15:1903 suppresses
  21. L25:376 suppresses
  22. L25:1193 weakly suppresses
  23. L13:2681 weakly suppresses
  24. L19:2705 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 -10.91 and -10.591 values between -8.99 and -8.671 values between -7.39 and -7.071 values between -5.15 and -4.832 values between -4.19 and -3.873 values between -3.87 and -3.554 values between -3.55 and -3.231 values between -3.23 and -2.913 values between -2.91 and -2.593 values between -2.59 and -2.2710 values between -2.27 and -1.9610 values between -1.96 and -1.6410 values between -1.64 and -1.3227 values between -1.32 and -1.0032 values between -1.00 and -0.6870 values between -0.68 and -0.36413 values between -0.36 and -0.042515 values between -0.04 and 0.28765 values between 0.28 and 0.6031 values between 0.60 and 0.922 values between 0.92 and 1.241 values between 1.24 and 1.562 values between 1.56 and 1.88offnegativepositiveoften

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. 窍
  2. vé
  3. bord
  4. 那次
  5. ��
  6. agan
  7. 网站地图
  8. ��
  9. icult
  10. ingham

Suppresses

  1. ngoing
  2. ongoing
  3. progress
  4. progressing
  5. →progress
  6. -progress
  7. advancing
  8. averse
  9. _PROGRESS
  10. progress