Neuron Explorer

Qwen3-0.6B

'words' in 'in other words'

A neuron in layer 11 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 "words" in the phrase "In other words", where a writer restates an idea more plainly. Turning it up gave replies that drifted into repetitive lists, which does not show the phrase's meaning.

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

    not only includes controlling a specific cell, but it also allows for creating a secondary drop-down list that is completely dependent on the first or primary list. In other words, once the first list is displayed, and a parameter chosen, then a subset

  2. #2 peak

    many skeptics of Christianity take a simple approach to reading the Bible. They treat the entire Bible and all of its contents as a moral command textbook. In other words, every single sentence is to be read with an eye toward what moral behavior the

  3. #3 peak

    recognition that a sedentary lifestyle is being driven, at least in part, by environmental factors that affect individuals' physical activity choices and health behaviours. In other words, the environments in which we live, and with which we interact, have become

  4. #4 peak

    learning experience for students. As a community college, it is understood that Kirtland maintains somewhat of an obligation to connect the classroom to the community. In other words, service to the community performed by students is consistent with community college tradition. This

  5. #5 peak

    these rankings were used to grant a standard reward to a top group of teachers, not to make fine distinctions in the amount of compensation that teachers received. In other words, the results of value-added analysis, when examined over a period of two or

  6. #6 peak

    about the prospect, and the board as a whole made it clear that their tentative decision does not mean a regulation of this nature would ultimately be approved. In other words, as of the date of this writing, it is still illegal to kill g

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 fact that, as I have been told, and I have been told by the judge, and I have been told by the judge, and I have to be. So I have to be. This is a really important thing to be

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

    as being a List of Users.⏎Sometimes, it’s also handy to define a list of multiple Entities and Structures.⏎Until 184.108.40.2

  2. #2 peak

    uler or calipers with mm divisions⏎- Paper towel⏎- Microscope slide and cover slip⏎AQA: Topic 184.108.40.206 and 220.1<|endoftext|>Re export

  3. #3 peak

    Volume (Year): 42 (2011)⏎Issue (Month): 2 (June)⏎|Contact details of provider:|| Web page: http://www.tandfonline.com/VECE20|⏎|Order Information

  4. #4 peak

    to agriculture.⏎For more information, please contact:⏎Dr. Norihito Nakamichi⏎Biodynamics Research Team⏎RIKEN Plant Science Center⏎Tel: +81

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. L5:134 suppresses
  2. L9:396 excites colon after a metadata label
  3. L10:572 excites
  4. L6:1285 excites
  5. L4:946 suppresses
  6. L9:2604 weakly excites
  7. L9:2159 weakly excites
  8. L5:1526 weakly suppresses
  9. L5:1847 weakly suppresses
  10. L9:2590 weakly suppresses
  11. L6:1016 weakly suppresses
  12. L9:97 weakly excites
  13. L4:193 weakly excites
  14. L7:2354 weakly excites
  15. L6:74 weakly excites
  16. L9:84 weakly suppresses
  17. L8:749 weakly suppresses
  18. L0:3037 weakly suppresses
  19. L4:51 weakly suppresses
  20. L8:747 weakly excites
  21. L5:2364 weakly suppresses
  22. L9:797 weakly excites
  23. L2:2826 weakly excites
  24. L1:1931 weakly suppresses

Feeds later layers

  1. L17:2684 excites
  2. L14:2774 excites
  3. L16:2241 excites
  4. L17:2365 excites
  5. L15:598 suppresses
  6. L17:806 excites
  7. L13:895 excites
  8. L15:38 weakly excites
  9. L13:584 weakly suppresses unit conversions in parentheses
  10. L14:2156 weakly excites opening parenthesis after a term
  11. L20:1772 weakly excites
  12. L17:1229 weakly excites
  13. L12:1506 weakly excites
  14. L15:261 weakly excites
  15. L12:1820 weakly suppresses
  16. L12:1837 weakly excites
  17. L14:2610 weakly excites
  18. L17:2605 weakly suppresses
  19. L16:1394 weakly suppresses
  20. L12:959 weakly suppresses
  21. L17:2364 weakly excites
  22. L17:1988 weakly excites
  23. L16:656 weakly suppresses
  24. L13:626 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.

2 values between -1.29 and -1.133 values between -1.13 and -0.9614 values between -0.79 and -0.6350 values between -0.63 and -0.46120 values between -0.46 and -0.29476 values between -0.29 and -0.122088 values between -0.12 and 0.04871 values between 0.04 and 0.21136 values between 0.21 and 0.3864 values between 0.38 and 0.5531 values between 0.55 and 0.7115 values between 0.71 and 0.8813 values between 0.88 and 1.056 values between 1.05 and 1.214 values between 1.21 and 1.383 values between 1.38 and 1.553 values between 1.55 and 1.722 values between 1.72 and 1.881 values between 1.88 and 2.051 values between 2.05 and 2.221 values between 2.22 and 2.382 values between 2.55 and 2.721 values between 2.88 and 3.051 values between 5.22 and 5.39offnegativepositiveoften

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. dating
  2. antar
  3. doi
  4. amt
  5. śmy
  6. posite
  7. REST
  8. ressing
  9. \":{\"
  10. uspendLayout

Suppresses

  1. 썸
  2. 然后
  3. untu
  4. ))]⏎
  5. 슈
  6. 페
  7. atinum
  8. 冠军
  9. тогда
  10. iteration