least common multiples and factors
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. 4 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
This neuron is strongly negative on school maths pages about least common multiples (LCM) and common factors. Its positive side is weak and does not form a clear pattern. When it was turned down, the model started listing multiples and factors.
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.
-
.⏎Giving us a result of 250, this would be 25% of 1,000 or what is equal to the part of
-
als. The quadratic residue Gauss walk exhibits a more complex behavior whose shape is roughly determined by the remainder of modulo 24.⏎Contributed by: Erqian Wang (June
-
) and current I(t)=Imax×sin(ωt+ΦI) where Vmax, Imax are the peak values, ω is the common angular pulsation and ΦV, ΦI are the instant phase of
-
quadrilateral. What do you notice about this quadrilateral as the⏎dynamic image changes? Can you prove your conjecture?⏎It is impossible to trisect an angle using
-
in the Photocatalytic Synthesis Group at the University of Twente. The project is part of the Advanced Research Center Chemical Building Blocks Consortium (ARC CBBC; www.arc-cbbc.nl). Within the ARC CBBC, the University of
-
a subgroup of the N numbers.⏎Another mystery was solved in researching this class. At the Brid
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: 12, 24, 36, 48, 60, 72, 9, 120, 126, 144, 16, 1
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
multiplication.⏎Fractions - Adding Mixed Numbers⏎Learn how to Add mixed numbers with like and unlike denominators.⏎Fractions - Adding Unlike Denominators - LCM⏎Least common multiples are very useful in finding the least common denominator when adding and
-
regroup or to borrow with mixed numbers? Take a deep breath and let us take it one step at a time. Click and begin.⏎GCF and LCM by Prime Factorization⏎A straight forward way to find the greatest common factor (
-
Math - What to Expect⏎Are you considering earning your GED? Take away some of the fear of math by knowing what to expect.⏎How to Find Common Factors⏎An Efficient Method to Find Common Factors⏎How to Simplify Expressions⏎
-
9 (these are positive⏎numbers less than 1000). This will generate your test numbers. If you⏎determine that num is not divisible by 5 and 7, then increase⏎count by one.⏎Share your code
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
- L3:1562 suppresses
- L3:703 suppresses
- L1:2414 suppresses
- L2:1657 weakly excites
- L3:52 weakly suppresses
- L2:2328 weakly suppresses
- L3:59 weakly excites
- L3:50 weakly excites
- L2:356 weakly excites
- L2:527 weakly excites
- L3:421 weakly suppresses
- L2:2468 weakly excites
- L2:2043 weakly excites
- L2:49 weakly suppresses
- L3:2958 weakly excites
- L0:2948 weakly suppresses
- L2:549 weakly suppresses
- L3:122 weakly excites
- L3:272 weakly excites
- L2:1899 weakly excites
- L3:1045 weakly suppresses
- L3:456 weakly suppresses
- L2:1434 weakly suppresses
- L3:202 weakly suppresses
Feeds later layers
- L18:2628 suppresses
- L18:1925 weakly suppresses
- L11:1383 weakly excites
- L5:1399 weakly excites
- L12:3001 weakly suppresses
- L9:679 weakly suppresses
- L8:1812 weakly suppresses
- L17:3040 weakly suppresses
- L24:2702 weakly suppresses
- L9:2555 weakly suppresses
- L22:978 weakly excites
- L15:874 weakly excites
- L5:211 weakly suppresses
- L10:2571 weakly suppresses
- L6:2857 weakly excites
- L17:1520 weakly suppresses
- L6:1935 weakly excites
- L24:598 weakly excites
- L7:913 weakly suppresses
- L6:1228 weakly excites
- L16:1818 weakly suppresses "two" and the link between two things
- L12:1907 weakly suppresses
- L5:860 weakly suppresses
- L7:1153 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
- [self
- [state
- alan
- aupt
- experimental
- south
- StateManager
- [Any
- [count
- owards
Suppresses
- 辜负
- 瘤
- verständlich
- icast
- ials
- eg
- ake
- ег
- 帮助
- '):⏎