Neuron Explorer

Qwen3-0.6B

words introducing an example or aside

A neuron in layer 8 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 small connecting words that introduce an example or a side comment, such as 'for example', 'for instance', 'like', 'though' and 'or'. Turning it up did not produce a clear change in the model's writing style.

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

    we should be including in the language guide.⏎Throughout EmbraceU we have decided to use medical words (like mouth, anus, vulva, vagina, and penis) to describe body parts

  2. #2 peak

    was to break out a screwdriver and fiddle around with the device. Even then, there were only a few different things that could be changed. Now, though, there are hundreds

  3. #3 peak

    which is addictive. Public health advocates warn that e-cigarette makers are using the same strategies that tobacco companies employed to attract young people — candy flavorings, for example. Lax state-level regulations also make it easy for young people to buy them

  4. #4 peak

    inflation rate over the last seven months. This exercise reveals whether the increase in inflation since last fall is due primarily to a few outliers—food and energy, for instance—or whether it reflects a broader acceleration in prices.⏎A “Heat-Map”

  5. #5 peak

    , called added sugars, are viewed as a detriment to a healthy diet. Maximizing sugar’s benefits requires balancing the healthier and less wholesome sources.⏎Sucrose, or table sugar, is the main source of sugar in most American diets. It consists

  6. #6 peak

    collaborative coding environment can be conveniently captured by a graph. The vertices (or things) are people, organizations, projects, and tickets. The edges (or

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 number 14. So, the number 14 is in the equation. Now, the problem is to find the value of the equation. I will also consider the time limit of 120 seconds. The equation is

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

    berspace – being important part of the implementation of human rights online.⏎One of the most popular types of network communities today is knowledge exchange communities (practice communities) that represent social communities, members of which are involved in collaboration and the crucialcondition here

  2. #2 peak

    . To January 1, 1944, the Bartlesville sand has produced 7,566,000 barrels of oil from 5,610 productive acres and 871 oil wells.

  3. #3 peak

    precipitation for the year in Mount Surprise is 31.0" (787.4 mm). The month with the most precipitation on average is January with 8.0" (203.2 mm) of precipitation.

  4. #4 peak

    was crowned as the first king of Jerusalem on Christmas Day, 1100, by Dagobert of Pisa, the Patriarch, who would have preferred to turn the kingdom into a theocracy. In the spring 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. L7:337 excites
  2. L6:1389 excites
  3. L3:372 excites
  4. L7:1222 suppresses
  5. L3:487 suppresses
  6. L7:2811 excites
  7. L2:225 excites
  8. L7:65 excites
  9. L3:353 excites
  10. L2:263 suppresses
  11. L6:1387 excites
  12. L7:497 suppresses
  13. L7:3071 weakly excites
  14. L7:2503 weakly excites
  15. L7:1170 weakly suppresses
  16. L7:84 weakly excites
  17. L5:1067 weakly suppresses
  18. L6:327 weakly excites
  19. L7:506 weakly excites
  20. L7:658 weakly suppresses
  21. L0:843 weakly excites
  22. L6:1006 weakly excites
  23. L5:974 weakly excites
  24. L5:1623 weakly suppresses

Feeds later layers

  1. L9:1370 suppresses
  2. L10:1985 suppresses
  3. L10:1459 suppresses
  4. L9:1113 suppresses
  5. L10:271 excites
  6. L14:602 suppresses
  7. L10:1392 excites
  8. L12:982 suppresses
  9. L12:2240 suppresses
  10. L14:1181 suppresses
  11. L13:637 suppresses opening parenthesis after a term
  12. L12:2882 weakly suppresses
  13. L14:2820 weakly suppresses "and" or "or" linking two counted items
  14. L13:2071 weakly excites
  15. L11:1758 weakly suppresses
  16. L10:894 weakly suppresses
  17. L14:2964 weakly suppresses
  18. L9:671 weakly suppresses
  19. L10:1580 weakly suppresses
  20. L15:1370 weakly suppresses
  21. L21:2324 weakly suppresses "in" after "shown" (figure references)
  22. L9:630 weakly suppresses
  23. L17:2896 weakly suppresses
  24. L13:691 weakly 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 -3.37 and -3.241 values between -3.24 and -3.111 values between -2.98 and -2.851 values between -2.72 and -2.591 values between -2.08 and -1.951 values between -1.95 and -1.822 values between -1.82 and -1.693 values between -1.69 and -1.561 values between -1.56 and -1.434 values between -1.43 and -1.308 values between -1.30 and -1.185 values between -1.18 and -1.055 values between -1.05 and -0.9210 values between -0.92 and -0.7918 values between -0.79 and -0.6631 values between -0.66 and -0.5336 values between -0.53 and -0.4057 values between -0.40 and -0.27187 values between -0.27 and -0.14949 values between -0.14 and -0.021852 values between -0.02 and 0.11485 values between 0.11 and 0.24163 values between 0.24 and 0.3752 values between 0.37 and 0.5021 values between 0.50 and 0.635 values between 0.63 and 0.761 values between 0.76 and 0.892 values between 0.89 and 1.022 values between 1.02 and 1.151 values between 1.27 and 1.401 values between 1.53 and 1.661 values between 1.66 and 1.79offnegativepositiveoften

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. raf
  2. phia
  3. alink
  4. ongan
  5. etus
  6. olatile
  7. 这张
  8. fcc
  9. vfs
  10. льн

Suppresses

  1. 荡
  2. endum
  3. ぎ
  4. 頓
  5. towards
  6. endcode
  7. 보
  8. �
  9. upon
  10. Sho