'of' after 'one' or 'some' (partitives)
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. 12 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 'of' in phrases like 'one of the oldest', 'some of the most common questions' or 'some of these limitations', where a part of a group is named. It has only weak negative responses. Turning it up did not give a clear theme in the generated text.
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.
-
the world? 70% of all PET bottles are recycled here, compared to 31% in the USA. But before you give yourself a pat on the back for this achievement, can you name the invisible force behind it? It
-
of the message.⏎The Internet Acronym Server⏎This site can be used to look up the meaning of an acronym (abbreviations⏎constructed from the first letters of several words, such as ADA for⏎American Dental Association). You can also type
-
Hittites contribute to cultural diffusion?⏎How did the Hittites contribute to cultural diffusion of early Mesopotamian culture and ideas, and what was one of their important technological advances? The Hittites had learned to get iron from ore
-
sent NIH his grant application:⏎Boylan's letter made its way to Ernest M. Allen, head of NIH's fledgling Division of Research Grants and one of the key architects of the agency's peer-review grants system. "Ernest Allen
-
these criteria is nearly impossible. And the risks you’re taking by assuming it’s safety far outweigh the possible benefits of drinking raw milk.⏎Here are answers to some of the most common questions about raw milk.⏎What is raw milk?⏎Milk that
-
distances to primary markets, and vulnerability to climate change.⏎Although he confirmed the United Nations Industrial Development Organization (UNIDO) has been implementing various programmes to address some of these limitations - including introducing technical cooperation on waste management, giving assistance with building resilience
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: The function is, I think, a function that is a function from (x, ) to (0, ) is a vector, but I need to get the function. Let's see, let's take, say, (x)
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
in the Ussr. StudyMode.com. Retrieved 05, 2012, from http://www.studymode.com/essays/The-Struggle-For-Power-In-The-99432
-
flowers. Stems have vascular tissues that move food and water around the plant to help it grow. They are also used as a place to store food.⏎Image: amenic181/Getty Images⏎Roots grow underground and gather water and
-
mynursingpaper.com/wp-content/uploads/2020/06/cropped-15597
-
and culturally sensitive manner. Privacy is essential and the nurse must be aware of their own ...⏎You have reached your viewing limit for this book.⏎What people are saying - Write a review⏎Respiratory system 52 Thoracotomy⏎St
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
- L10:272 suppresses
- L4:16 weakly excites
- L8:1003 weakly suppresses "the" or "of" before a noun phrase
- L8:137 weakly suppresses
- L8:892 weakly excites
- L9:836 weakly excites
- L9:734 weakly excites
- L4:24 weakly excites
- L10:2717 weakly suppresses
- L11:2057 weakly excites
- L8:252 weakly suppresses 'of' after 'each' or 'one' or 'all'
- L7:2218 weakly suppresses
- L11:293 weakly suppresses
- L3:2116 weakly excites
- L8:1101 weakly suppresses
- L1:408 weakly excites
- L11:473 weakly suppresses
- L8:437 weakly suppresses
- L6:1169 weakly suppresses
- L2:2957 weakly excites
- L8:328 weakly excites
- L10:779 weakly excites
- L2:1036 weakly excites
- L4:1842 weakly suppresses
Feeds later layers
- L14:655 strongly excites
- L16:1219 excites
- L19:695 suppresses
- L18:977 excites
- L15:1283 excites
- L19:2385 excites
- L17:1250 excites
- L16:528 excites
- L15:1702 excites
- L17:2736 excites
- L15:1028 excites
- L17:2702 excites
- L13:844 excites 'of' after a percentage or share
- L18:552 suppresses
- L18:748 excites
- L18:520 excites
- L15:1033 excites
- L18:1199 excites
- L18:315 suppresses
- L18:867 excites
- L15:2740 excites
- L17:865 excites
- L13:2116 excites
- L14:1588 suppresses
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
- __(*
- /Gate
- пед
- legate
- .reserve
- __ ⏎
- ällt
- öße
- culator
- loat
Suppresses
- R
- ne
- 变化
- 灵活
- ur
- mer
- D
- N
- 렴
- in