'and' joining parallel phrases
A neuronOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in layerOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. 13 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featuredA neuron picked for the full treatment: steering outputs and a written explanation. About 3,000 of the 86,000 neurons are featured., so it has steering outputs.
What this neuron does
This neuron fires on the word 'and' when it links two matching parts of a sentence, like two verbs or two nouns in an official or instructional text. When it was turned up, the model fell into repeating 'and' and similar filler words.
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 activationThe number a neuron produces for a token. Large positive means it fired hard; negative means it pushed the other way. was highest. Amber marks positive activation; the outlined tokenA piece of text the model reads at once: often a word, sometimes part of a word or a punctuation mark. is the peak. Hover a token for its strength.
-
ston. For more information about this artist and her work, visit Inuit Gallery of Vancouver.<|endoftext|>1. The problem statement, all variables and given/known data If an athlete jumped 2 feet high and left the ground
-
. Matuga⏎Measuring Housing Affordability⏎U.S. Department of Housing and Urban Development⏎The views expressed in this article are those of the author and do not represent the official positions or policies of
-
, to retain the full citation information.⏎- This article was written and prepared by U.S. Government employees on official time, and is therefore in the
-
⏎PortraitDyan Mazurana is senior research fellow at the Feinstein International Famine Center, Tufts University. Angela Raven-Roberts is director of research and programs at the Feinstein Famine Center, Tufts University. Jane Parpart is
-
of speech: nouns/pronouns, verbs, adjectives, adverbs, conjunctions, and prepositions. Each section:⏎- Defines the part of speech and its usage⏎- Gives examples to illustrate the definition⏎- Provides a range of
-
-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the page “GNU Free Documentation License”.⏎The copyright and license notices on this page only apply to
SteeringAdding a push along a neuron while the model writes, to see how its output changes. Turning it up or down.
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.
Here is something I have been thinking about lately: I have to of the people in the of the of the of of of of of of of of of... I OF... THIS, THE, THE, THE, THE, THE, THE, THE, THE, THE, ... THE,
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
our institutions, particularly those dedicated to learning? Hitherto clear-cut boundaries to educational institutions - those that defined the very nature of schooling - are being challenged and some are becoming more permeable. The massive uptake of mobile technologies has challenged the physical
-
paths (-<|endoftext|>Simply begin typing or use the editing tools above to add to this article.⏎Once you are finished and
-
iseáil go slán. Ní dhéanann muid sonraí cártaí creidmheasa a stóráil agus níl aon rochtain againn ar shonraí a
-
factors compromise responses to climate changes – both ‘normal’ variation and shocks as well as new variability and increased uncertainty.⏎Third, local support systems are changing, and so is people’s ability to make use of them. The report shows that changes in
Connections
Show on mapWhich neuronsOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in earlier layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.
Fed by earlier layers
- L9:92 strongly suppresses "and" or "or" joining two items
- L10:74 strongly excites
- L4:57 strongly suppresses
- L12:51 excites
- L1:1250 suppresses
- L5:880 excites
- L9:2010 suppresses
- L5:548 excites
- L10:1643 suppresses
- L4:108 excites "or" offering a choice in web text
- L4:2270 excites
- L8:70 excites 'and' between stacked superlatives
- L11:154 excites commas after citation numbers
- L0:313 excites "and" joining paired nouns
- L6:13 excites
- L0:114 suppresses
- L12:2732 weakly excites
- L1:1029 weakly excites
- L6:7 weakly suppresses
- L0:2146 weakly suppresses
- L10:554 weakly excites
- L8:13 weakly suppresses
- L8:7 weakly excites
- L5:75 weakly excites
Feeds later layers
- L16:376 strongly excites 'or' joining alternatives
- L17:572 strongly excites
- L18:870 strongly excites
- L17:2332 strongly excites 'and' joining two clauses
- L20:885 strongly excites 'and' joining paired words
- L19:2080 excites ampersand joining two items
- L22:198 excites
- L25:1694 excites 'and' joining paired names or fields
- L17:1856 excites
- L17:1537 excites
- L16:1669 excites
- L15:2765 excites
- L14:1116 excites 'and' inside lists
- L18:868 excites 'and' joining two nouns in a pair
- L18:2345 excites
- L15:727 excites
- L26:1007 excites
- L19:291 excites 'and' joining two full clauses
- L17:3061 excites
- L16:2049 excites
- L16:2577 suppresses
- L19:1754 excites
- L16:2442 excites
- L14:801 excites 'and' or 'or' joining list items
Activation histogramA chart of how often a neuron takes each activation value.
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.
Logit lensReading which output tokens a neuron pushes toward or away from, straight off its output weights, before the later layers have their say.
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 layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. instead.
Promotes
- /or
- /OR
- vice
- hence
- rogen
- thus
- rog
- rei
- thus
- 其它问题
Suppresses
- 理解和
- 和支持
- :“
- 企业和
- ary
- 能力和
- 生产和
- оп
- 质量和
- ylvania