Neuron Explorer

Qwen3-0.6B

words before 'signs' or 'symptoms'

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 lights up on the word just before a word like 'signs', 'symptoms' or 'reasons' is likely to come next, mostly in health and advice articles. Turning it up did not produce a clear change in what the model wrote.

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

    paragraph, with a sentence that is split over several lines in HTML. </BODY> </HTML>When this is displayed by Mosaic, it will look like:⏎As you can probably guess, commands are enclosed in angle-brackets <>, so

  2. #2 peak

    easy noise filtering and yields excellent transient response. The components connected around the ADP1613<|endoftext|>Safed Kabbalist (1542–1620), chief disciple of Isaac Luria, the Ari, whose

  3. #3 peak

    , and the effect is just as severe in them. German Shepherds are prone to becoming obese, and in this article, you will find out the common reasons why your GSD is becoming overweight, warning signs to watch out for the tips

  4. #4 peak

    or moving around. The pain may eventually radiate to the lower right side of the abdomen, where the appendix is located.⏎Loss of Appetite⏎One sign that you may have appendicitis is a loss of appetite. This happens because the

  5. #5 peak

    of the spine)⏎- Herniated disk⏎- Viral or bacterial infection⏎What are the symptoms of torticollis?⏎The following are the most common symptoms of torticollis. However, each person may experience symptoms differently. Symptoms

  6. #6 peak

    watch for signs of wilting. Leaves that are shriveled, papery or wilted are all signs of dehydration. Likewise, mealy bugs are a sign of overwatering.⏎A moisture meter is a great tool to help you keep

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: what is the value of the number in the form of a two-digit number in the form of a two-digit number in the form of a two-digit number. So, I think. So, I think. So, I think. So

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

    says. A solar storm would still pose a risk for wandering astronauts, he adds, but there would be time to return to a shielded base.⏎To continue reading this article, subscribe to receive access to all of newscientist.com, including

  2. #2 peak

    4.COM) on the Representative List of the Intangible Cultural Heritage of Humanity⏎- Chinese paper-cut⏎Present throughout China and in various ethnic groups, paper-cut is a popular art integral

  3. #3 peak

    -clients after their HTML page has loaded ? Or what to do if we would like to selectively update only parts of a page. For example only the stock item in an HTML TABLE whose rate has changed ?⏎Let's assume that we have a Java

  4. #4 peak

    laboratory on uncommon ground.''⏎The idea for PLUG came from Singh's frustration over ''trying to conduct space-age science with stone<|endoftext|>Washington, August 7 (ANI): A team of U.S. scientists has identified phosphorylated signalling proteins in

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. L11:310 suppresses
  2. L9:164 excites
  3. L6:295 excites
  4. L8:308 excites
  5. L10:841 suppresses
  6. L9:64 suppresses
  7. L7:305 excites
  8. L10:850 excites
  9. L11:935 excites
  10. L11:420 suppresses
  11. L11:2095 excites
  12. L8:323 suppresses
  13. L11:600 excites
  14. L10:448 suppresses
  15. L8:1000 suppresses names of agencies and institutions
  16. L10:483 suppresses
  17. L10:192 excites
  18. L11:598 excites
  19. L10:98 weakly suppresses
  20. L11:2135 weakly excites
  21. L10:2541 weakly suppresses
  22. L11:1905 weakly suppresses
  23. L11:1749 weakly excites
  24. L8:46 weakly suppresses

Feeds later layers

  1. L14:1651 strongly excites
  2. L15:878 suppresses
  3. L14:2104 suppresses
  4. L13:1667 excites
  5. L14:2639 excites
  6. L16:1237 excites
  7. L13:2144 suppresses
  8. L14:625 excites
  9. L15:741 suppresses
  10. L18:2589 excites
  11. L14:1477 suppresses
  12. L13:1277 suppresses
  13. L16:2332 excites
  14. L16:861 suppresses
  15. L13:1679 suppresses
  16. L15:1641 suppresses
  17. L13:2783 excites
  18. L13:1933 suppresses
  19. L13:1015 excites
  20. L18:406 suppresses
  21. L13:1216 suppresses
  22. L13:2676 excites
  23. L18:2947 excites
  24. L14:1417 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 -2.03 and -1.901 values between -1.90 and -1.775 values between -1.64 and -1.518 values between -1.51 and -1.3812 values between -1.38 and -1.2515 values between -1.25 and -1.1232 values between -1.12 and -1.0057 values between -1.00 and -0.8791 values between -0.87 and -0.74146 values between -0.74 and -0.61215 values between -0.61 and -0.48302 values between -0.48 and -0.35437 values between -0.35 and -0.22561 values between -0.22 and -0.09699 values between -0.09 and 0.04399 values between 0.04 and 0.16268 values between 0.16 and 0.29215 values between 0.29 and 0.42127 values between 0.42 and 0.5589 values between 0.55 and 0.6873 values between 0.68 and 0.8143 values between 0.81 and 0.9428 values between 0.94 and 1.0727 values between 1.07 and 1.1913 values between 1.19 and 1.3210 values between 1.32 and 1.4510 values between 1.45 and 1.588 values between 1.58 and 1.716 values between 1.71 and 1.842 values between 1.84 and 1.972 values between 1.97 and 2.101 values between 2.10 and 2.221 values between 2.35 and 2.481 values between 2.48 and 2.611 values between 2.74 and 2.871 values between 3.00 and 3.13offnegativepositiveoften

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. cob
  2. 指定
  3. olon
  4. .Handler
  5. incident
  6. don
  7. events
  8. 常委会
  9. 领导干部
  10. besch

Suppresses

  1. ignment
  2. Replies
  3. yps
  4. Fahrenheit
  5. 一个问题
  6. eus
  7. iage
  8. 他们是
  9. reeNode
  10. ArrayType