Neuron Explorer

Qwen3-0.6B

unclear: mid-phrase words in mixed prose

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 fires on a word in the middle of a phrase where the next word is fairly predictable, such as the 'of' in 'memorandum of points' or 'without' in 'prohibited without permission'. It shows no single topic, and turning it up or down did not produce a clear change in the writing.

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

    who described a lawyer as 'a person who writes a 10,000 word document and calls it a brief.'⏎A brief usually contains a memorandum of points and authorities. Points and authorities explain why the law authorizes the judge to

  2. #2 peak

    materials in this archive are copyright © The Manitoba Library Association. Reproduction for personal use is permitted only if this copyright notice is maintained. Any other reproduction is prohibited without permission Copyright information for reviewers⏎Young Canada Works<|endoftext|>From the Publisher:⏎Over roughly

  3. #3 peak

    etermine value using Look-up table method⏎Create Volcano (or Atoll) martix. It is an extension of Bullseye matrix problem.⏎Who has power to do everything in this world?⏎Close MATLAB with keyboard⏎Find the treasures in

  4. #4 peak

    ’s assume that we have a more general problem. There are n questions, and a(n) is the smallest number of times we need to take the test to guarantee that we can figure out the answers. First we can try all combinations of answers

  5. #5 peak

    pointing in toward the nose.⏎crossroad: a road that crosses another road.<|endoftext|>If you are pondering how to make your computer virus, you’ve come towards the right place. There are many tutorials and steps on making a malware. However

  6. #6 peak

    which has more advantages. It is cheaper and more practical. This study aimed to examine the effect of cognitive therapy restructuritation in decreasing depression score of the mental disorder patient of Banyumas public hospital. This study

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 bit of a philosophical problem of the sort of thing as it is actually taught in schools... the matter of fact being that a man is not all the more fully formed from the world at large in fact to be the same as he

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

    be compared to a wise man who built his house on the rock……⏎Revelation 3:20 'Behold, I stand at the door and knock; if anyone hears My voice and opens the door, I will come

  2. #2 peak

    a sapphire stone, in appearance in the likeness of a throne, appeared above them. And He spoke to the man clothed in linen and said, “Enter between the whirling wheels under the cherubim…⏎…Then I

  3. #3 peak

    giveth of his bread to the poor.” – Proverbs 22:9⏎2 Corinthians 9:6-8 – “But this I say, He which soweth sparingly shall reap also sparingly; and he which

  4. #4 peak

    and statutes written down for Abraham, Isaac, and Jacob to keep, and yet they did the will of God because of

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. L9:1625 suppresses
  2. L3:159 weakly suppresses
  3. L8:277 weakly excites
  4. L3:615 weakly suppresses
  5. L9:115 weakly excites
  6. L9:17 weakly excites
  7. L6:1110 weakly suppresses
  8. L3:2075 weakly excites
  9. L8:254 weakly excites
  10. L3:1869 weakly excites
  11. L7:705 weakly excites
  12. L7:359 weakly suppresses
  13. L3:2761 weakly excites
  14. L0:1412 weakly excites
  15. L8:786 weakly suppresses
  16. L2:299 weakly excites
  17. L4:2228 weakly suppresses
  18. L8:2049 weakly excites
  19. L9:1293 weakly suppresses
  20. L7:2120 weakly suppresses
  21. L0:2625 weakly suppresses
  22. L1:49 weakly suppresses
  23. L0:137 weakly excites
  24. L1:1956 weakly suppresses

Feeds later layers

  1. L15:646 suppresses
  2. L11:393 suppresses
  3. L12:887 suppresses
  4. L13:1513 suppresses
  5. L11:1384 suppresses
  6. L18:819 suppresses
  7. L12:576 suppresses 'of' in fixed phrases like table of contents
  8. L12:434 suppresses
  9. L16:200 weakly suppresses
  10. L16:592 weakly suppresses
  11. L18:867 weakly suppresses
  12. L13:450 weakly excites
  13. L15:227 weakly suppresses
  14. L24:1787 weakly excites
  15. L13:1929 weakly suppresses
  16. L11:796 weakly suppresses
  17. L20:272 weakly suppresses
  18. L25:643 weakly excites
  19. L19:1982 weakly suppresses
  20. L18:775 weakly excites
  21. L22:157 weakly excites
  22. L15:364 weakly suppresses
  23. L14:2288 weakly suppresses
  24. L22:490 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 -1.62 and -1.562 values between -1.10 and -1.041 values between -1.04 and -0.971 values between -0.97 and -0.903 values between -0.77 and -0.712 values between -0.71 and -0.643 values between -0.64 and -0.581 values between -0.58 and -0.515 values between -0.51 and -0.459 values between -0.45 and -0.3814 values between -0.38 and -0.3224 values between -0.32 and -0.2566 values between -0.25 and -0.19158 values between -0.19 and -0.12313 values between -0.12 and -0.06930 values between -0.06 and 0.01889 values between 0.01 and 0.08639 values between 0.08 and 0.14422 values between 0.14 and 0.21238 values between 0.21 and 0.27103 values between 0.27 and 0.3457 values between 0.34 and 0.4012 values between 0.40 and 0.4710 values between 0.47 and 0.534 values between 0.53 and 0.601 values between 0.92 and 0.99offnegativepositiveoften

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. 声
  2. 精神
  3. PT
  4. NK
  5. gt
  6. 软
  7. (uint
  8. PDO
  9. (',',$
  10. 最后

Suppresses

  1. STRU
  2. ,))⏎
  3. contemplated
  4. esda
  5. ,");⏎
  6. testified
  7. ----------- ⏎
  8. ...,
  9. MainThread
  10. ..."⏎