numbers and ranges in data and statistics
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. 9 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 spots just before a number in statistical or data-heavy text, like age ranges or years in a study. Turning it up did not produce a clear, consistent change in the writing.
Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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.
-
See<|endoftext|>Data Set Overview⏎|Data set||LBA-ECO LC-09 Land Cover Transitions Maps for Study Sites in Para, Brazil: 1970-2001|⏎|Project||
-
of a flash<|endoftext|>Undated (WXOW) - In 2009, the US Preventative Task force recommended against breast cancer screening for women aged 40 to 49.⏎It said the risks outweigh the small potential benefit
-
live, feed and breed.⏎The report also divides the Arctic Ocean into 12 regions. Using daily sea ice concentration data from satellite images taken by NASA between 1979 and 2013, changes in the dates of
-
) than were nonveterans (19 percent).⏎Among Gulf War-era II veterans in 2009, 31 percent of veterans aged 18 to 24 years were enrolled in school. As for older
-
Skip to Search Results⏎Identifying and analyzing spatial and temporal patterns of lightning-ignited wildfires in Western Canada from 1981-2018.Download⏎This study looked at the spatial and temporal patterns of lightning-ignited
-
, 2,019 waterbodies had 5 or more years of data. When the University of Florida data are added, the number jumps to 4,800 waterbodies. Of these, 17%
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: How many full 3-digit numbers are there that can be used to describe a person's state of being in a fair and peaceful situation? A person has 1000000 characters in their system. You have a
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
blank sheet. You can make as many columns as<|endoftext|>Johann Christian Heyl⏎Johann Christian Heyl (1
-
?⏎Use the alphabetical glossary listing to find a short definition to a medical or dental term. If you would like to know more, please remember to call us on (02) 9420 5050. Our
-
treat serious damage, where an endodontist does.⏎The type of issues an endodontist treats⏎While minor issues such as plaque buildup or even a minor chip in the tooth may only need the services of a general dentist, there are many
-
light is off.⏎Mirrors and Windshield⏎- Mirrors should be clean and adjusted properly from the inside.⏎- Windshield should be clean with no illegal
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
- L7:1111 excites
- L7:34 excites
- L7:2773 suppresses
- L7:277 suppresses
- L8:969 excites
- L7:275 suppresses
- L6:337 suppresses
- L6:263 suppresses
- L6:391 suppresses
- L8:1027 suppresses
- L5:151 excites
- L7:1550 suppresses
- L7:468 excites
- L6:262 weakly suppresses
- L8:383 weakly excites
- L7:231 weakly excites
- L2:2088 weakly excites
- L7:322 weakly suppresses
- L6:762 weakly suppresses
- L2:2320 weakly suppresses
- L6:665 weakly excites
- L7:619 weakly suppresses
- L5:2013 weakly excites
- L7:140 weakly excites
Feeds later layers
- L10:391 excites
- L11:924 excites
- L10:2012 excites
- L11:1330 excites
- L10:347 suppresses
- L11:2343 excites
- L11:330 suppresses
- L11:1030 excites
- L10:1436 suppresses
- L12:1172 excites
- L10:935 excites
- L11:2196 excites
- L11:1193 excites
- L10:948 suppresses
- L10:2998 excites
- L11:2825 excites
- L12:70 suppresses
- L10:736 excites
- L11:497 suppresses
- L11:3020 suppresses
- L21:1826 excites
- L17:2826 weakly excites
- L12:30 weakly suppresses
- L11:1828 weakly excites
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
- auté
- Seeder
- .INSTANCE
- 箔
- Wildcard
- instance
- Installation
- HeaderValue
- Gundam
- 种子
Suppresses
- attery
- tery
- db
- uzzi
- =db
- eleg
- ��
- proprietary
- manage
- _db