'the' after across, around or of
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. 14 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 "the" in phrases about wide places or whole groups, like "around the world", "across the country" or "and the elderly". Turning it up made the model repeat simple sentences about the sun and weather, which is only loosely related.
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.
-
drives. But IBM’s ideas still rely on the use of a magnetic field to store that data.⏎Researchers from around the world are collaborating on solving the speed problem, and they’ve come up with a
-
ers during the French religious wars, Mercier was obliged to take refuge in Venice in 1567 and, after returning to France, he died of the plague. One of his best-known works was the Libellus de abbrevi
-
confusing, punctuation marks.⏎From Afghanistan to Zimbabwe, discover surprising and intriguing language facts from around the globe.⏎The definitions of ‘buddy’ and ‘bro’ in the O
-
can be mixed. This could be because of already mentioned approximate land type detection. In<|endoftext|>Nutrition plays a very important role in the lives of both young and old. However for seniors, it’s important to be aware of new studies on how
-
the article.⏎The latest study followed a group of nuns, monks and priests from across the United States over a period of 12 years. They were chosen as a
-
those who experience it. Unfortunately, despite efforts to raise awareness and implement preventative measures, instances of sexual harassment continue to occur in schools, colleges, and universities across the country.⏎In Pakistan, schools, colleges, and universities are considered to be safe
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 think of a system of equations of the first order and the system of equations of the first order of the system of equations of the first order of the system of equations of the first order of the system of equations of the first order of
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Whooper,"<|endoftext|>AttachmentsAccessible Version Accessible version of the tutorial content in PDF format⏎Source and Access Information⏎Aligned StandardsThis vetted resource aligns to concepts or skills in these benchmarks.⏎In this interactive tutorial, learn about the major factors
-
listen to and obey His voice⏎Jeremiah 7:28 "And you shall say to them, 'This is the nation that did not obey the voice of the LORD their God or accept correction; truth has perished and has been
-
at Mount Sinai when G-d descended upon the mountain and gave us His Torah. As it says: “Only beware and guard yourself carefully, lest you forget the things your eyes have seen and lest they stray from your heart all the days of your
-
great joy that will be for all the people. 11 For unto you is born this day in the city of David a Savior, who is Christ the Lord. 12 And this
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
- L11:1145 suppresses 'of' inside institution names
- L5:136 suppresses
- L5:636 weakly suppresses
- L1:376 weakly suppresses
- L13:1204 weakly suppresses
- L13:846 weakly suppresses
- L12:521 weakly suppresses
- L10:120 weakly suppresses
- L11:199 weakly suppresses
- L13:2746 weakly suppresses
- L11:1747 weakly suppresses
- L10:352 weakly suppresses
- L13:2267 weakly excites
- L8:15 weakly suppresses unclear: mid-phrase words across many languages
- L5:511 weakly suppresses
- L11:137 weakly suppresses
- L7:938 weakly excites
- L9:1351 weakly suppresses
- L12:1598 weakly suppresses
- L10:1388 weakly excites
- L13:1535 weakly suppresses
- L13:1952 weakly suppresses
- L1:1576 weakly excites
- L13:633 weakly suppresses
Feeds later layers
- L16:1465 suppresses
- L17:2783 suppresses
- L15:942 excites
- L18:1672 suppresses
- L19:2815 suppresses
- L15:435 excites
- L15:1902 suppresses
- L18:2370 suppresses
- L15:1226 suppresses
- L15:2547 suppresses
- L20:450 suppresses
- L15:2079 excites
- L16:749 excites
- L15:702 excites
- L18:1011 excites
- L19:241 excites
- L21:1422 suppresses
- L17:181 suppresses
- L15:1848 suppresses
- L18:898 excites
- L15:2893 suppresses
- L22:429 suppresses
- L16:1441 excites
- L17:2037 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
- DNS
- frey
- doom
- 戴上
- iband
- DDD
- eeper
- 的关注
- surrounded
- 的理解
Suppresses
- nants
- isting
- orian
- -login
- roup
- 我在
- bable
- noss
- ofApp
- unist