ampersand inside names and abbreviations
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. 23 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 "&" sign when it sits inside a name or abbreviation, like P&G, F&V or D&RGW. Turning it down pushed the model toward writing blank fill-in-the-line underscores.
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.
-
. It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ 1&2. To view this ebook on an iPhone, iPad or Android mobile device you
-
. It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ 1&2. To view this ebook
-
victor. How true, the losers are not around to defend their position.⏎In my early days at P&G, we had a saying, if there were three people walking up a hill
-
cost of repairing and upgrading the water supply infrastructure is estimated to be approximately R400-million (US$52.8 million).⏎World Cruise on P&O⏎Pass
-
70s. These writings are from publications by people outside of industry, mostly college professors. Their perspective is quite different than mine. I started working at P&G in 1951, inside industry. At that time, there
-
) as an environmental indicator. These studies principally showed that meat and dairy are among the largest contributors to GHGE; whereas high intake of fruits, vegetables (F&V), and legumes/pulses/nuts consumption is associated with the lowest GH
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 to view_ a___________________________________________
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
identifying organisms that uses two words to name the species is.⏎Their habitat and physical differences⏎In Aristoles system of classification, animals were classified on the basis of ____________________.⏎The greater the number of taxa two organisms have in common, the __________________
-
. What have you found?⏎2. 1/4 turn clockwise, forward 2, 1/4 turn anti-clockwise and then forward 3 _______.⏎3. Forward 2, 1/4 turn clockwise, forward
-
things? No! Your flexible brain is so good at making sense of patterns that it can see two images in one image.⏎The image at left is both a __________________ and __________________.⏎Your brain also solves mysteries in language. Imagine you find the crypt
-
0 in another(),so this prints 20. Thus final answer is 10 0 20. I hope i have been clear enough. ------------------ Come on in !! Drinks are on the house in the Big Moose Saloon
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
- L19:1654 excites
- L8:7 excites
- L21:195 excites
- L18:267 excites
- L6:7 suppresses
- L8:14 excites
- L10:33 suppresses
- L21:173 excites
- L1:8 excites
- L20:603 excites
- L17:38 weakly suppresses
- L20:1664 weakly excites
- L6:16 weakly excites
- L5:110 weakly suppresses
- L21:189 weakly suppresses document break token mid-sentence
- L8:10 weakly excites
- L16:170 weakly suppresses
- L6:13 weakly excites
- L8:2 weakly excites
- L3:187 weakly excites
- L0:2776 weakly suppresses
- L18:48 weakly suppresses
- L9:7 weakly excites
- L18:887 weakly excites
Feeds later layers
- L25:292 suppresses
- L27:432 excites
- L27:635 suppresses
- L26:98 suppresses
- L26:396 weakly suppresses
- L27:2159 weakly suppresses
- L24:206 weakly excites
- L24:138 weakly excites
- L26:575 weakly suppresses
- L25:1986 weakly suppresses
- L27:802 weakly suppresses
- L26:524 weakly suppresses
- L26:1750 weakly suppresses
- L26:2975 weakly excites
- L26:313 weakly excites
- L24:2740 weakly suppresses
- L27:1373 weakly suppresses
- L27:1692 weakly suppresses
- L24:659 weakly suppresses
- L27:2046 weakly suppresses
- L24:1662 weakly excites
- L25:161 weakly excites
- L27:921 weakly excites
- L26:1534 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
- dilig
- *,⏎
- *"
- **⏎
- *,
- *&
- isations
- ―
- -----------
- !");⏎
Suppresses
- _
- &
- #
- amp
- //
- ype
- o
- _VIEW
- hotmail
- &