hyphen 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. 10 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 a dash that sits between two numbers, like the dash in a phone number, a score range, or a date range. When it was turned up, the model's writing became scrambled and didn't show a clear theme.
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.
-
taught that you cannot compress liquids, so the the short answer to your question is yes, liquids have a definite volume :0).⏎Interested in booking a 1-1 lesson with me?Click here to view my profile and send me a message
-
was no simple, standardized way of describing hurricanes and their damaging effects, like the way the Richter scale is used to describe earthquakes. He created a 1-5 scale based on wind speed and sent it off to Simpson, who expanded<|endoftext|>
-
in 1942-44, with improvements based on lessons learned from the UP's 4-8-8-4 Big Boys. In service, the Challengers often complement
-
to perform basic statistical functions using both a scientific calculator and Microsoft Excel.⏎Saturday, Aug. 23, 2-4:30 p.m.⏎Basic Statistics with SPSS⏎If your major
-
and the climate conditions of Earth during that time with geochemist Tina Lüdecke.⏎You’ll find out about Tina’s research studying early humans from 1-4 million years ago. We investigate what happened during the paleolithic era. How
-
you surely will be one of those who lose. (all spiritual goodnesses)⏎THE HOLY QURAN - S39: 65⏎6- THE BRANCHES OF MON
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: what is the time of the day, and the number of days of the week? It's one of the day of the week, or the 31-day of the month, or the 31 week-end of the day,
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
oltage AC power supply⏎- Two capacitors, 0.1 µF each, non-polarized (Radio Shack catalog # 272-135)⏎- Two 2
-
axis cnc machining consists of linear interpolation movement of feed axes X, Y, Z and any 5 axes of X, Y, Z rotation axes A, B, and C.⏎Different processing characteristics⏎3-axis machining: The direction of the
-
The newest data sets include POLDER and GOES-7 images, soil carbon data, and information about the study sites.<|endoftext|>2. Verb. (third-person singular of surround) ¹⏎¹ Source: wiktionary.com
-
is distributed in a year;⏎n– The years to maturity; and⏎YTM– The yield to maturity.⏎Let's take Bond A, issued by Company Alpha, as an example. It has the following data:⏎- Face value: $
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
- L7:1759 excites
- L7:828 suppresses
- L9:708 excites
- L9:867 excites
- L5:271 suppresses
- L9:802 excites
- L9:940 suppresses
- L4:290 excites
- L5:3047 suppresses
- L9:190 excites
- L9:145 suppresses
- L9:1836 excites
- L6:94 suppresses
- L9:897 excites
- L8:781 excites
- L6:1098 suppresses
- L8:1468 excites
- L6:58 excites
- L7:1362 excites
- L6:182 suppresses
- L9:2308 suppresses
- L7:223 excites
- L4:1861 excites
- L6:145 suppresses
Feeds later layers
- L16:1872 suppresses
- L15:1665 excites
- L11:967 excites
- L15:762 excites
- L11:112 excites
- L12:453 excites
- L11:1834 suppresses
- L11:25 excites
- L18:318 weakly excites
- L16:1319 weakly excites
- L16:1465 weakly excites
- L11:325 weakly excites
- L12:2520 weakly excites
- L12:315 weakly excites
- L15:1690 weakly excites
- L11:1109 weakly excites
- L13:458 weakly suppresses
- L12:916 weakly excites
- L15:890 weakly excites
- L12:1377 weakly excites
- L13:3054 weakly excites
- L16:637 weakly excites
- L11:453 weakly suppresses
- L17:22 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
- orden
- ordon
- imit
- keit
- lear
- "=
- ead
- BarItem
- ordinates
- eyed
Suppresses
- 秘书长
- igli
- xDB
- 恺
- 窗外
- nerg
- 釜
- 艇
- tram
- 奶