Neuron Explorer

Qwen3-0.6B

mid-word fragments in names

A neuron in layer 12 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 the middle piece of a longer word, often a scientific or technical name split into parts, such as "ribofl[avin]", "subf[amilies]" or "list[eria]". The examples are varied, so the pattern is only loosely defined.

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

    Files?⏎We can zip and unzip files in Windows 10 easily. To zip files:⏎Step

  2. #2 peak

    and recombineering in E. coli". PLoS ONE 8 (2): e57075. . . http://www.plosone.org/article/info%3Adoi%2F10.137

  3. #3 peak

    500,000 species in existence.⏎The superfamily is presently divided into 19 separate families with more than 90 subfamilies recognised. Currently

  4. #4 peak

    or raised to move the cutting blade to or away from the material to be cut.⏎A drill press with a moveable arm that can be swung to various positions on the work

  5. #5 peak

    cheese⏎Cottage cheese has a low fat and calorie content. It is full of calcium, phosphorus, selenium, vitamin B12, riboflavin (vitamin B2

  6. #6 peak

    full potential by passively listening to the teacher and filling in worksheets they must engage in This is called a mixed number which is a mixture of a whole bbc bitesize to add the two fractions together print out the worksheet below from twinkl or copy

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 question is about the same day, and the same day is the same day." The same day. The same day. The same day. The same day. The same day. The same day. The same day. The the

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

    immediately. Just as I say, “ Imagine the tip of your left pinky finger… where the nail meets the skin…” Do you think of it? Did

  2. #2 peak

    , because the impairments in shifting gaze and attention that we found in 7-month-olds may be a fundamental problem in autism,” said Robert T. Schultz, PhD, director of the Center for Autism Research at CHOP and a co-author on

  3. #3 peak

    grounds. They will make the return dry season journey in June, to take advantage of the permanent water of the Tarangire River.⏎Numbers of wildebeest have fallen from 50,000 to 6,00

  4. #4 peak

    and young adults,” said Jennifer Duke, PhD, senior research public health analyst and co-author of the study. “And unfortunately

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. L8:1135 strongly excites
  2. L10:1219 excites
  3. L11:1385 excites
  4. L8:391 excites
  5. L9:804 excites
  6. L11:1336 suppresses
  7. L9:1840 excites
  8. L10:1231 excites
  9. L7:845 excites
  10. L11:640 excites
  11. L9:814 excites
  12. L11:849 suppresses
  13. L10:2456 excites
  14. L9:1048 suppresses
  15. L11:844 suppresses
  16. L11:1119 suppresses
  17. L9:581 weakly suppresses
  18. L9:968 weakly suppresses
  19. L11:60 weakly excites
  20. L11:1500 weakly excites
  21. L5:468 weakly excites
  22. L10:1674 weakly excites
  23. L10:882 weakly excites
  24. L8:2691 weakly excites

Feeds later layers

  1. L13:1567 excites
  2. L17:764 excites
  3. L17:2851 excites
  4. L13:1661 suppresses
  5. L20:1426 excites
  6. L13:2538 excites
  7. L13:1892 suppresses
  8. L14:1724 excites
  9. L13:2227 weakly excites
  10. L16:1461 weakly excites
  11. L13:2528 weakly excites
  12. L18:2449 weakly excites
  13. L18:1009 weakly suppresses
  14. L13:2026 weakly excites
  15. L24:2664 weakly suppresses
  16. L14:2069 weakly suppresses
  17. L17:1647 weakly excites
  18. L15:677 weakly suppresses
  19. L14:2377 weakly excites
  20. L15:2709 weakly excites
  21. L13:1285 weakly excites
  22. L13:1536 weakly excites
  23. L14:1091 weakly suppresses
  24. L16:988 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.

3 values between -2.69 and -2.554 values between -2.55 and -2.413 values between -2.41 and -2.273 values between -2.27 and -2.132 values between -2.13 and -1.993 values between -1.99 and -1.852 values between -1.85 and -1.707 values between -1.70 and -1.568 values between -1.56 and -1.4210 values between -1.42 and -1.2816 values between -1.28 and -1.1416 values between -1.14 and -1.0024 values between -1.00 and -0.8643 values between -0.86 and -0.7173 values between -0.71 and -0.5781 values between -0.57 and -0.43122 values between -0.43 and -0.29221 values between -0.29 and -0.15853 values between -0.15 and -0.011478 values between -0.01 and 0.13495 values between 0.13 and 0.28218 values between 0.28 and 0.42124 values between 0.42 and 0.5643 values between 0.56 and 0.7020 values between 0.70 and 0.8412 values between 0.84 and 0.985 values between 0.98 and 1.125 values between 1.12 and 1.275 values between 1.27 and 1.411 values between 1.41 and 1.552 values between 1.55 and 1.693 values between 1.69 and 1.831 values between 1.83 and 1.971 values between 1.97 and 2.111 values between 2.82 and 2.96offnegativepositiveoften

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. 诸如
  2. OTA
  3. Wolfe
  4. 相对
  5. rst
  6. aversal
  7. azed
  8. ivity
  9. 不论是
  10. ATURE

Suppresses

  1. /W
  2. /T
  3. /P
  4. 毛孔
  5. /S
  6. /R
  7. /E
  8. jo
  9. /AFP
  10. /V