decimal point inside numbers
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. 15 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 lights up on the dot in a decimal number, like the dot in 18.5 or 4.0. Turning it up made the model's writing drift into looser, more rambling sentences about random topics.
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.
-
are a rich source of vitamins, minerals and fiber that support optimal health.⏎A 100-gram serving of tamales contains about 18.
-
: pages Publisher: W. The Death and. Register a free business account Paperback: pages Publisher: W. Choose the book you like when you register 4.⏎Discussion questions for ‘The Death and Life of the Great Lakes’⏎Skip to
-
depending on the problem.⏎Examples of Calculations with Gravity⏎Example 1: A ball is thrown up into the air at an initial velocity of 56.3m/s. Determine its velocity after 4.52s have
-
and humor throughout the book. Make sure to find them all!⏎Where Does Art Come From? by Leah McCurdy is licensed under a Creative Commons Attribution 4.0 International License, except where otherwise noted.⏎History of art<|endoftext|>Spread over the
-
and red flag.⏎Copyright (c) 2022 Proceedings Of International Conference On Communication Science⏎This work is licensed under a Creative Commons Attribution 4.0 International License.<|endoftext|>How do you explain how babies are made to an
-
A&M College⏎Rights© 2021 Prairie View A & M University⏎This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.⏎Date of Digitization⏎John B Coleman Library⏎City
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 possible impact of the social media and the current environment. of the society and the natural environment and the focus on today's environmental and social impact of the and 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.
-
of the conflict between two worlds, East and West, which determined the course of history.⏎Proving History is available online: https://2500years.culture.gov.gr/en.<|endoftext|>"Tell all the Truth"⏎Tell all the Truth
-
2021 bibliography, you will find 386 pages of resources to learn more about Caddo history and culture, https://scholarworks.sfasu.edu/cgi/viewcontent.cgi?article=4012&context
-
http://projects.fnb.nl⏎A great deal⏎of work relating to the presentation of Braille
-
. You can export a word list with a word frequency count to import into other visualization tools. http://www.antlab.sci.waseda.ac.jp/software.html⏎Voyant Tools: Voyant Tools
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
- L2:9 strongly suppresses sentence-final periods vs dots in code
- L8:143 strongly suppresses
- L4:7 strongly excites
- L0:36 strongly excites
- L5:168 suppresses
- L5:1304 excites
- L4:295 excites
- L5:155 suppresses
- L10:321 suppresses
- L14:422 excites periods after initials and abbreviations
- L5:454 suppresses
- L2:239 suppresses
- L1:130 suppresses
- L10:156 suppresses
- L0:258 excites
- L11:132 weakly excites period ending an opening topic sentence
- L2:504 weakly suppresses
- L2:1417 weakly suppresses
- L1:38 weakly suppresses
- L4:2532 weakly excites surname pieces split across tokens
- L10:15 weakly suppresses
- L0:670 weakly suppresses
- L1:1256 weakly excites
- L2:1283 weakly suppresses
Feeds later layers
- L17:642 strongly excites
- L18:713 strongly excites
- L16:432 strongly excites period after month or initial
- L17:2084 suppresses
- L17:1218 suppresses
- L19:831 excites
- L16:844 excites
- L18:1437 excites
- L17:347 excites full stops ending informative sentences
- L17:1772 excites
- L19:2497 excites
- L17:1155 excites opening parenthesis after a sentence
- L18:490 excites
- L19:1398 suppresses
- L16:718 excites
- L19:969 excites
- L17:2001 excites
- L20:505 excites
- L17:1105 excites
- L19:902 suppresses
- L20:2301 excites
- L18:1409 excites
- L16:1433 excites
- L23:509 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
- ernet
- lette
- ussions
- lettes
- plan
- Plan
- 령
- illé
- plier
- ors
Suppresses
- target
- challenge
- curacy
- loom
- parse
- target
- focus
- -output
- prim
- (Parser