unclear: mixed, mostly everyday health and advice prose
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 mixed: they peak on different tokens. 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
The neuron fires on a grab bag of words in ordinary sentences, such as 'status' in a web error, 'you' in 'glad you did', and 'is' after 'a healthy diet'. There is no single clear trigger. Turning it up gave unrelated text, so the meaning is not clear.
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.
-
not index a URL with 406 Not Acceptable response status, and consequently, URLs that have been indexed in the past but are now returning this HTTP status code will be removed from
-
. This can make land unsuitable for agriculture, lead to crop stress and attract new pests that thrive in warmer temperatures⏎- Warmer temperatures and drought are also contributing to animal stress, decreasing production and making animal welfare an issue.⏎These impacts are
-
your caving library. -- Danny A. Brass Underground Movement If you are at all interested in bats, get a copy of this book. You will be glad you did! Book Bargains and Previews A well-written, interesting primer on bats
-
) in 1873 and into medical literature by Babington's translation of Hecker's Der Schwarze Tod in 1833.⏎The source of the infection of bubonic plague was traced in the 1890
-
scientific name keeps it easier to classify. Scientific names dont change and show relationships and classification.⏎Why is it better for scientist to
-
," Powell says. "It's that significant. He is one of the most important figures that we've seen in the last 25 years or so."⏎Powell interviewed Tupac a number of times for Vibe magazine, including while the
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 been thinking about this I have been thinking about this I have been thinking about this I have been thinking about this I have been thinking about this I have been thinking about this I have been thinking about this
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
3) diplomatic activities, (4) intelligence activities, and (5) cryptology. The latter two areas might be considered to be special parts of the first three areas. That is, intelligence and cryptology are "service" functions for the
-
ARD syndrome and Noonan syndrome are both caused by mutations in the PTPN11 and RAF1 genes.⏎The symptoms and physical characteristics associated with LEOPARD syndrome are highly variable. However, most affected individuals tend to exhibit characteristic abnormalities of
-
anagrams for endoplasm found in this word list.⏎Words formed by adding one letter before or after endoplasm (in bold), or to adelmnops in any order:⏎s - endoplasms⏎All words formed from end
-
come to speak to her classes about their careers. Across the country, just 10 percent of high schools offer computer science classes, according to a website promoting Computer Science Education Week, which began Monday.⏎And yet, experts predict more than
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
- L11:990 suppresses
- L11:1782 excites
- L11:1377 suppresses
- L8:105 suppresses
- L8:1377 excites
- L11:1001 excites
- L9:557 weakly excites
- L5:985 weakly suppresses
- L2:768 weakly excites
- L5:197 weakly suppresses
- L1:1435 weakly excites 'as' in 'just as'
- L10:279 weakly suppresses
- L11:1705 weakly suppresses
- L11:170 weakly suppresses
- L11:341 weakly excites
- L11:401 weakly excites
- L11:163 weakly excites
- L10:1452 weakly suppresses
- L8:621 weakly suppresses
- L10:2061 weakly suppresses
- L10:2416 weakly excites
- L1:2596 weakly excites
- L4:1052 weakly suppresses
- L0:1483 weakly excites
Feeds later layers
- L13:801 excites
- L13:1547 suppresses
- L15:1734 suppresses
- L13:1923 suppresses
- L13:1665 excites
- L13:1123 suppresses
- L13:2538 suppresses
- L13:143 excites
- L13:1680 suppresses
- L14:1768 suppresses
- L13:2002 suppresses
- L16:1398 suppresses
- L13:1716 suppresses
- L14:1199 suppresses
- L14:228 excites
- L13:1018 suppresses
- L14:108 excites
- L13:1761 suppresses
- L13:1913 suppresses
- L13:650 suppresses
- L13:644 suppresses
- L15:2673 suppresses
- L15:2526 weakly suppresses
- L17:807 weakly 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
- left
- shared
- set
- 共享
- thư
- library
- SET
- public
- legal
- إليه
Suppresses
- igers
- 碣
- �回
- 朋友们对
- 遑
- ี้
- displayName
- getProperty
- PageIndex
- IFICATIONS