Neuron Explorer

Qwen3-0.6B

unclear: words inside software descriptions

A neuron in layer 24 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 connecting words inside descriptions of software and web tools, like 'free and open-source' or 'Mosaic' as an early browser. The pattern is fuzzy, and turning it up gave loosely related text about management and memory.

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

    can be configured to replicate across Zones to spread demand and avoid downtime from failures.⏎Officially launched in 2006, Amazon Web Services provide online services for other web sites or client-side applications. Most of these services are not exposed directly

  2. #2 peak

    figure out.⏎Perhaps the most popular early browser most people are familiar with and many of us used it, was Netscape Navigator. The browser that preceded that was Mosaic, which was co-founded by venture capitalist and millionaire, Marc Andreessen.

  3. #3 peak

    ier features to be used, but if you're not using these features, you can safely omit both.⏎Documents written in HTML are not WYSIWYG - Mosaic and other WWW clients will re-arrange the layout of your text so

  4. #4 peak

    Finally, the authors find that students in advanced economics classes significantly outperform students in regular classes.⏎If you experience problems downloading a file, check if you have the proper application to view it first. In case of further problems read the IDEAS help

  5. #5 peak

    8. Then Oracle obtained MySQL in 2010 and has owned it ever since.⏎For our purposes and it’s general definition, MySQL is a free and open-source database, and one of the most popular in the world.⏎Now,

  6. #6 peak

    elevated devil abundances following the extinction of the previous apex predator - the thylacine - approximately 80. years ago.<|endoftext|>Drupal is a free and open source Content Management Framework (CMF) that uses a PHP and a backend

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 child whose child is the child whose child is the child whose child is the child whose child whose child whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose whose

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

    a compilation of "best engineering⏎practices" which have proven successful in modelling large, complex systems.⏎UML succeeds the concepts of Booch, OMT and OOSE by fusing them into a single,⏎common and widely usable modelling language

  2. #2 peak

    of what is common practice now days, this is what I take out of this chapter.⏎Start your application by structuring the domain. Follow DDD principles, defining Bounded Contexts that relate to business concepts. Group your code units per conceptual

  3. #3 peak

    Understand how to create parallelized collections and external datasets⏎- Work with Resilient Distributed Dataset (RDD) operations⏎- Utilize shared variables and key-value pairs⏎Module 3 - Spark application programming⏎- Understand

  4. #4 peak

    outcomes. Each feature is captured as a “story”, which defines the scope of the feature along with its acceptance criteria. This article introduces the BDD approach to defining and identifying stories and their acceptance criteria.⏎Software delivery is about writing software to achieve

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. L5:2 suppresses
  2. L9:18 suppresses text boundary after a document break
  3. L15:921 excites
  4. L20:429 excites
  5. L21:269 excites
  6. L9:49 suppresses
  7. L10:32 excites
  8. L22:156 weakly excites
  9. L9:411 weakly excites
  10. L20:461 weakly suppresses
  11. L8:2395 weakly excites
  12. L20:96 weakly suppresses
  13. L18:1517 weakly suppresses
  14. L2:972 weakly suppresses
  15. L21:385 weakly suppresses
  16. L3:63 weakly excites
  17. L20:1783 weakly excites
  18. L19:552 weakly suppresses
  19. L7:3 weakly suppresses
  20. L20:66 weakly suppresses
  21. L18:140 weakly suppresses
  22. L0:9 weakly suppresses
  23. L3:551 weakly suppresses
  24. L8:67 weakly suppresses

Feeds later layers

  1. L27:1515 weakly excites
  2. L27:2064 weakly excites
  3. L27:1978 weakly suppresses
  4. L27:1684 weakly excites
  5. L27:1826 weakly excites
  6. L25:1808 weakly suppresses
  7. L25:2360 weakly excites
  8. L26:546 weakly suppresses
  9. L27:2899 weakly excites
  10. L26:2459 weakly excites letter k starting a unit or fragment
  11. L25:1708 weakly excites
  12. L27:787 weakly excites
  13. L26:2160 weakly excites
  14. L26:2485 weakly suppresses capital I starting a word fragment
  15. L27:1912 weakly suppresses
  16. L27:792 weakly excites
  17. L26:251 weakly excites
  18. L25:2751 weakly suppresses
  19. L27:1036 weakly suppresses
  20. L25:1712 weakly suppresses
  21. L25:1543 weakly excites
  22. L25:813 weakly suppresses
  23. L25:2774 weakly suppresses
  24. L25:323 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 -7.48 and -6.781 values between -5.38 and -4.682 values between -3.97 and -3.273 values between -3.27 and -2.5712 values between -2.57 and -1.8722 values between -1.87 and -1.16385 values between -1.16 and -0.462923 values between -0.46 and 0.24465 values between 0.24 and 0.9440 values between 0.94 and 1.6511 values between 1.65 and 2.3511 values between 2.35 and 3.0510 values between 3.05 and 3.755 values between 3.75 and 4.461 values between 4.46 and 5.161 values between 5.16 and 5.863 values between 5.86 and 6.561 values between 6.56 and 7.273 values between 7.27 and 7.971 values between 7.97 and 8.671 values between 8.67 and 9.371 values between 10.78 and 11.481 values between 12.18 and 12.881 values between 12.88 and 13.591 values between 14.29 and 14.991 values between 19.91 and 20.61offnegativepositiveoften

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. (man
  2. _Man
  3. MAN
  4. MAN
  5. -man
  6. management
  7. _man
  8. マン
  9. enza
  10. 管理

Suppresses

  1. 外
  2. �
  3. 名叫
  4. 另有
  5. khúc
  6. Outside
  7. Fore
  8. alker
  9. -roll
  10. thấy