dots inside web addresses
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. 27 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 full stop inside web addresses and domain names, such as 'www.1914.org' or 'pidweb.ii.uam.es'. It is the strongest signal in this set. Turning it down made the output a string of dots, which matches the same punctuation theme.
Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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.
-
, IWM will help reconnect people with the First World War and keep the history and personal recollections of the war alive."⏎The exhibition can be viewed at www.1914.org/faces⏎- More about:⏎- First World War
-
rehoused as of end of December 2013 which has not happened yet.⏎If the cam comes back, the address should be https://www.avo.alaska.edu/webcam/Cleveland.php⏎Mount Cleveland seems to be a
-
979)⏎3rd sing fem – la⏎3rd sing masc – lu⏎3rd mass masc – lo⏎see also: http://pidweb.ii.uam.es/coser/contenido.php?es&publicaciones⏎South
-
the time and remote control for situations that the machine can’t quite do things,” he said.⏎For the rest of this article, click here: http://www.abc.net.au/news/rural/2017-05-2
-
Awards⏎What differences do you see in countries who have endured war compared to those who have not?⏎Featured image: children during Pol Pot provided by http://www.ridecambodia.com<|endoftext|>Crystallization is one of the most ancient
-
.G.McTavish. For fourteen years she had faithfully attended to his needs and during that time she had
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:................................................................................................................................................
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Dynasty Setting of Spring Dawn⏎Meaning for Line 1 春 : spring 眠 : sleep 不 : not 觉 : to feel, wake up 晓 : dawn⏎Meaning for Lien 2 处 : place
-
, time and space are connected to each other, so the watch of a moving person will have different pace than that of a resting person.⏎chronic adj. �
-
withdrawn his secessionist Missouri State Guards from Lexington under pressure from Major General John C. Fremont’s larger Federal Army of the West. Price moved southwest, seeking to
-
the doors of Pandit houses, asking them to convert, flee, or perish; as prominent Pandits were being murdered, the Chief Minister of the state Dr. Farooq Abdullah abandoned his people and ran away to London in order to let
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
- L20:962 strongly excites decimal points and dots in numbers
- L19:831 strongly suppresses
- L22:271 strongly suppresses
- L20:344 strongly excites hyphen inside compound words
- L25:118 excites
- L24:248 excites
- L16:432 suppresses period after month or initial
- L16:3 suppresses
- L23:109 excites
- L4:7 excites
- L25:422 excites
- L22:1055 suppresses
- L23:474 excites
- L11:132 excites period ending an opening topic sentence
- L19:380 excites
- L18:791 suppresses
- L1:259 suppresses
- L22:265 excites
- L19:581 suppresses blank slot before a number
- L20:961 suppresses
- L16:191 suppresses
- L21:1894 excites
- L25:222 excites
- L20:61 suppresses
Feeds later layers
None: this is the last layer.
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
- nhạc
- ้า
- 开玩笑
- ��
- 克斯
- icia
- ��
- ывать
- -bot
- спин
Suppresses
- ⏎⏎
- BY
- by
- ...
- on
- '
- ⏎
- "...
- "
- _