Neuron Explorer

Qwen3-0.6B

end of document token

A neuron in layer 13 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up when one web page or document ends and the text is about to switch to something unrelated. It is mostly quiet elsewhere. Turning it up made the model's writing wander and repeat itself, but that does not clearly connect to the trigger.

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

    , but pointers can be uninitialized, which is a nasty security hole. An eagerly evaluating functional language without hacks, would only allow for acyclic graphs of pointers.<|endoftext|>The Death And Life Of The Great Lakes Pdf⏎- and pdf⏎- Tuesday

  2. #2 peak

    the other hand, are designed to handle complex logic functions that require high performance and low latency.⏎Developing software for microcontrollers is relatively easy compared to FPGA development<|endoftext|>The deepest human need is to engage with the world: confidently, productively,

  3. #3 peak

    work, including popular decision trees⏎- Get a comprehensive view of the machine learning ecosystem in R⏎- Explore the powerhouse of tools available in R’s caret package<|endoftext|>I just discovered a blog post by Ann M. Little, a historian at Colorado

  4. #4 peak

    manage operations in your warehouse.⏎- Lay out a proper strategy for handling damaged products and returns from customers.⏎- Implement pest control measures to protect your products.⏎-<|endoftext|>Nucleation properties of respirable coal dust (terminal progress report).⏎Physics Department

  5. #5 peak

    00 training examples, you can choose to pick 100 random samples (batch size=100) for each EPOCH (Training cycle)<|endoftext|>|Origin:||co

  6. #6 peak

    will discuss a basic MySQL installation on a Linux virtual server, and some of the basic MySQL commands.⏎Full documentation can be found on the MySQL.com website here.<|endoftext|>The law applicable to military strategic use of outer space⏎New technologies and the law

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 make a community that is open, and how to make the community members feel safe and connected. This is what I am working on in my community. I have been thinking about the importance of these values. This is what I have

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

    hook them to books.⏎- Read in front of your child. Be a role model for them. Kids are the best observers, and they tend to imitate there parents.⏎- Show the various options to your child, discuss books with them,

  2. #2 peak

    the entire crisis on p-values.⏎I think to develop a better understanding of why there is a “crisis”, we need to step back and look across differend fields of study. There is one key question you should be asking yourself: Is

  3. #3 peak

    your life. And you are to make them known to your children and to your children’s children – the day you stood before the L-ord, your G-d, at Sinai” (Deuteronomy 4:9-10

  4. #4 peak

    uscule script and its splendid and complex derivative, known as Gothic, were the principal calligraphic styles from the 9th to the 14th cent.⏎The humanistic handwriting style of the Renaissance, a deliberate imitation of Caroling

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. L6:1524 strongly excites
  2. L4:1203 suppresses
  3. L10:214 suppresses
  4. L6:689 excites
  5. L10:2199 suppresses
  6. L7:114 excites
  7. L8:1728 suppresses
  8. L8:492 suppresses
  9. L9:1593 excites
  10. L12:992 excites
  11. L11:2145 excites
  12. L7:220 suppresses
  13. L2:281 suppresses
  14. L11:345 excites
  15. L11:710 suppresses
  16. L11:258 excites
  17. L4:1963 suppresses
  18. L5:24 excites digit inside a date or number
  19. L12:661 excites
  20. L12:1334 suppresses
  21. L11:1348 excites
  22. L12:701 suppresses
  23. L11:639 excites
  24. L10:2250 suppresses

Feeds later layers

  1. L18:467 strongly suppresses
  2. L14:1200 strongly suppresses
  3. L18:897 suppresses
  4. L14:523 suppresses
  5. L16:1004 suppresses
  6. L17:279 suppresses
  7. L14:1332 suppresses
  8. L19:402 suppresses
  9. L18:1186 suppresses
  10. L19:47 excites
  11. L14:1590 suppresses
  12. L17:848 suppresses
  13. L17:523 suppresses
  14. L18:214 excites end of document marker
  15. L15:1616 suppresses word fragments mid-word
  16. L16:1251 excites
  17. L16:387 suppresses
  18. L16:822 suppresses
  19. L18:592 suppresses
  20. L15:2087 suppresses
  21. L15:1065 excites
  22. L16:444 suppresses
  23. L15:1404 suppresses
  24. L18:283 suppresses

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 -3.19 and -3.051 values between -1.98 and -1.851 values between -1.71 and -1.582 values between -1.04 and -0.912 values between -0.91 and -0.772 values between -0.77 and -0.646 values between -0.64 and -0.5116 values between -0.51 and -0.3750 values between -0.37 and -0.24269 values between -0.24 and -0.101895 values between -0.10 and 0.031214 values between 0.03 and 0.17329 values between 0.17 and 0.3088 values between 0.30 and 0.4321 values between 0.43 and 0.576 values between 0.57 and 0.702 values between 0.70 and 0.841 values between 0.97 and 1.101 values between 1.37 and 1.511 values between 2.04 and 2.18offnegativepositiveoften

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. am
  2. ishly
  3. ran
  4. sympath
  5. mlin
  6. amic
  7. amins
  8. amel
  9. ucker
  10. translation

Suppresses

  1. "]->
  2. _Msk
  3. ")->
  4. gard
  5. eña
  6. \")
  7. 糠
  8. ㏄
  9. >")⏎
  10. ()")⏎