Neuron Explorer

Qwen3-0.6B

"at" in "at first" and "initial" words

A neuron in layer 21 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 on the word "at" in phrases like "at first" or "at the time", and it leans against the word "initial". When turned up, the model's writing got stuck repeating "initial", "initially" and "at first start", which matches what the neuron seems to push toward.

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

    . We need to prepare an M-file which defines the equation and then call the subroutine (ode45) to do the integration. This method may seem mysterious at first because you call a subroutine, which in turn calls your M-file. Thus

  2. #2 peak

    as iron and calcium deficiency.⏎How can Celiac Disease be treated?⏎The only treatment required is a gluten free diet. It might seem difficult at first to enjoy a complete and balanced diet without gluten, but in fact, providing

  3. #3 peak

    different from a normal essay.” As students researched their four locations, they also had to locate them on a 21st century Google map. This was challenging at times as many locations no longer exist. Students had to figure out the most accurate

  4. #4 peak

    Countries⏎As you can see, the between online dating a girl through your country and dating a guy coming from another countries is great. It can be very confusing at first, but it’s wise to understand the different civilizations that exist just<|endoftext|>Why

  5. #5 peak

    online in both Greek and English, offers the visitor a fascinating tour in a simple and understandable way, utilizing modern audiovisual media. Even small fragments which seem insignificant at first glance, create an opportunity to reconstruct the facts,

  6. #6 peak

    matter what the circumstances were, and did what was right no matter what!⏎George Washington was indispensable in launching the U.S government. Although many opposed his political views at the time, Washington's leadership capabilities were unquestion<|endoftext|>Random changes in genes, rather

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 have a 3D ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ

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

    This tiny difference of starting values caused the equations to spit out a completely different set of answers. To mathematicians this effect is known as sensitive dependence on initial conditions and it is one of the hallmarks of mathematical chaos. It means that

  2. #2 peak

    Shareholders' equity, also referred to as net worth or owner's equity, represents the residual interest in the company's assets after deducting liabilities. It includes initial investments, retained earnings, and any additional capital contributed by shareholders. It

  3. #3 peak

    of understanding, completing 4 weeks of focused work, and then reassessing to check that progress has been made in key areas.⏎- Week 1: Initial assessment;⏎- Weeks 2 -5: One English, maths, and science

  4. #4 peak

    Generally, however, "people with disabilities" is the preferred initial reference.⏎Copywriters should portray people with disabilities as they would anyone else - with

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. L19:457 strongly excites
  2. L20:2200 strongly excites
  3. L18:2514 strongly excites
  4. L17:2372 excites
  5. L20:1687 excites
  6. L16:885 suppresses
  7. L14:2632 excites
  8. L17:2719 excites
  9. L17:2441 excites
  10. L18:1899 excites
  11. L20:1798 suppresses
  12. L17:2922 suppresses
  13. L15:2918 suppresses
  14. L13:2977 excites
  15. L15:2881 suppresses
  16. L19:2013 excites
  17. L18:2532 excites
  18. L13:1143 excites
  19. L13:2367 suppresses 'at the beginning/start' time phrases
  20. L13:386 suppresses
  21. L11:1081 excites 'first' and 'inaugural' as starting points
  22. L16:1576 excites
  23. L11:1819 suppresses
  24. L19:2601 excites

Feeds later layers

  1. L26:1259 strongly excites
  2. L25:1396 strongly excites
  3. L24:1775 strongly excites
  4. L24:2617 strongly excites
  5. L27:910 excites
  6. L26:2398 excites
  7. L23:2062 excites
  8. L23:964 excites
  9. L23:938 excites
  10. L25:869 excites
  11. L23:3063 excites
  12. L25:1422 excites
  13. L22:1551 excites
  14. L22:2172 excites
  15. L23:1703 excites
  16. L23:743 excites
  17. L22:1518 excites
  18. L27:1004 weakly excites
  19. L22:2713 weakly excites
  20. L22:1505 weakly excites
  21. L23:958 weakly excites
  22. L25:2229 weakly excites
  23. L24:1815 weakly suppresses
  24. L23:289 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 -17.23 and -16.471 values between -15.70 and -14.931 values between -14.93 and -14.161 values between -11.85 and -11.081 values between -7.23 and -6.462 values between -6.46 and -5.691 values between -5.69 and -4.922 values between -4.92 and -4.152 values between -4.15 and -3.383 values between -3.38 and -2.612 values between -2.61 and -1.8430 values between -1.84 and -1.071071 values between -1.07 and -0.312550 values between -0.31 and 0.47158 values between 0.47 and 1.2342 values between 1.23 and 2.0016 values between 2.00 and 2.776 values between 2.77 and 3.544 values between 3.54 and 4.315 values between 4.31 and 5.084 values between 5.08 and 5.851 values between 7.39 and 8.162 values between 8.16 and 8.931 values between 8.93 and 9.701 values between 12.78 and 13.55offnegativepositiveoften

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. initial
  2. initial
  3. 初始
  4. Initial
  5. Initial
  6. initially
  7. 初期
  8. 一开始
  9. first
  10. →initial

Suppresses

  1. ┻
  2. olvable
  3. ugged
  4. тельного
  5. (condition
  6. ┴
  7. archy
  8. >|
  9. ently
  10. 況