slash 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 lights up on the slash that comes after a website name in a web address, the part that leads into the page path. Turning it up or down did not make the model talk about links, so the steering results do not show this pattern.
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.
-
8, July 6). 1929 Stock Market Crash and the Great Depression –⏎Documentary [Video File]. Retrieved from https://youtu.be/qlSxPouPCIM⏎The post 1929 Stock
-
the Labour Inspectorate, Department and Employment.⏎External link to legislation: http://www.irishstatutebook.ie/1996/en/act/pub/0016/index.html<|endoftext|>Glacier Park's cold winters, short growing
-
fem – la⏎3rd sing masc – lu⏎3rd mass masc – lo⏎see also: http://pidweb.ii.uam.es/coser/contenido.php?es&publicaciones⏎South-Central Italian (Rohlfs
-
the community in the Hollywood Gazette’s November 2019 issue — https://hollywoodgazette.com/2019/11/03/hollywood-history-the-heroes-and-the-history-of-liberia-and-d
-
object to the edge of another. Perhaps when hovering over the current alignment dot, a sub triple-dot "menu" appears allowing the object to be aligned left/right/center of the current dot you
-
for whatever reason, it is usually pronounced as '/iy/' or “ee”.⏎How do you pronounce daemon in Golden Compass?⏎http://dictionary.reference.com/browse/daemon Pronounces it DEE-mon Which is how I've always said it.
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:sOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
and prices for different fruit crops.⏎Information on economic returns for some tropical fruit crops may be found on EDIS (keyword search "marketing") and on the UF/IFAS Tropical Research and Education Center (TREC) Agricultural Economics Extension website (
-
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
-
look like.⏎Here is their PDF brochure for posterity: Bosch Rexroth_R999000288_PE_2013_12_AE_media<|endoftext|>Page 2/2 - Posted February 1
-
rib is the large central vein which extends from the base of the blade to its tip. Its job is to provide support to the leaf.⏎Image: Avalon_Studio/Getty Images⏎The stem is the main structure that
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
- L19:1016 strongly suppresses
- L20:344 strongly excites hyphen inside compound words
- L24:248 excites
- L22:271 suppresses
- L20:854 excites
- L0:2014 suppresses
- L22:265 excites
- L26:2003 suppresses hyphen joining two word parts
- L10:90 excites
- L2:1787 excites
- L25:36 suppresses
- L23:474 excites
- L4:421 excites
- L2:262 suppresses
- L19:697 suppresses
- L26:254 excites
- L2:964 suppresses
- L22:2384 suppresses
- L0:1214 excites
- L19:263 suppresses
- L3:297 excites
- L16:471 suppresses blank spot before a year
- L2:404 excites
- L25:118 weakly excites
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
- قيم
- ете
- 闻
- достиг
- 可惜
- dotyczą
- يض
- 执教
- 有一
- .only
Suppresses
- s
- O
- dem
- S
- C
- tw
- LB
- t
- care
- bash