end-of-document boundary token
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. 26 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 invisible marker that separates one web document from the next, and on the spot right before it. Turning it up made the text wander into unrelated topics and repeated words, so the steering results do not show a clear meaning.
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.
-
two different numbers cant have the same cube). Therefore, by Definition 1, is one-to-one....⏎View Full Document⏎- Spring '09<|endoftext|>part of speech: noun⏎definition 1: an upright line or pole with
-
analysis can be obtained through finite element analysis; in this research analytical solution approaches will be used instead. The steady-state temperature distribution inside the machine can be formulated as Poisson’s equation. Geometry is assumed to be circular and windings are modeled as
-
Chapter 6 Exercise 6.5, we have discussed the Pythagoras theorem. Other important topics related to the chapter are covered in the following exercises.<|endoftext|>What you will learn in this module⏎Understand different tasks that have to be
-
ions iv. Special Limits c. Will be seen later d. If none of the top three apply, try using the Squeeze Theorem....⏎View Full Document<|endoftext|>A Woman of Salt, by Mary Potter Engel⏎Midrash refers to the
-
can also practice Chemistry: An Atoms First Approach - Zumdahl Atoms 1st 2nd Edition practice problems.<|endoftext|>The legal system of imperial China developed from two schools of thought: Confucian
-
them use unknowns for the sequence terms starting with the first given term. Solve the resulting equation. See the hint at the end of Example 1b.<|endoftext|>This five-part series chronicles the history of the global anti-apartheid movement
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: human styles manifest as manifest styles manifest as manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles manifest styles
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
' developed. It looks at intersection of algebraic geometry and computer algebra. It is concerned with the development of algorithms and software for studying and finding the properties of explicitly given algebraic varieties.⏎Much of the development of the main stream of algebraic
-
extended-spectrum beta-lactamase with additional resistance to carbenicillin, amoxicillin, and increasingly to ceftazidime.
-
algebra, especially commutative algebra, with the language and the problems of geometry.⏎The main objects of study in algebraic geometry are algebraic varieties, which are geometric manifestations of sets of solutions of systems of polynomial equations. Examples of the most
-
and contemporaries, including the Roman emperor Caesar Augustus, queen Cleopatra of Egypt and the Nabataean king Obodas III, issued coins emblazoned with their portraits.⏎But as the ruler of a Jewish kingdom where the
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
- L20:293 suppresses
- L9:41 excites
- L18:408 excites
- L3:25 suppresses
- L5:78 excites
- L20:238 excites
- L3:461 excites
- L21:128 excites
- L16:566 suppresses unclear: mid-sentence word choices, mixed
- L10:276 excites
- L9:96 excites
- L10:291 suppresses
- L9:693 suppresses
- L9:65 suppresses
- L17:133 excites
- L23:115 excites
- L7:334 excites
- L8:77 excites
- L10:2217 suppresses
- L19:746 excites
- L9:117 excites
- L3:105 excites
- L21:345 suppresses
- L4:739 suppresses
Feeds later layers
- L27:182 weakly excites
- L27:2963 weakly suppresses the 'st' piece inside words
- L27:2810 weakly suppresses
- L27:1606 weakly suppresses
- L27:164 weakly excites
- L27:2572 weakly suppresses
- L27:2297 weakly suppresses
- L27:2619 weakly excites
- L27:546 weakly excites
- L27:1955 weakly suppresses
- L27:1804 weakly suppresses
- L27:536 weakly excites
- L27:2086 weakly suppresses
- L27:2898 weakly excites
- L27:2867 weakly excites
- L27:2374 weakly suppresses
- L27:510 weakly suppresses
- L27:2060 weakly suppresses
- L27:2274 weakly suppresses
- L27:1049 weakly suppresses
- L27:1799 weakly suppresses
- L27:3053 weakly suppresses
- L27:1404 weakly suppresses
- L27:2473 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
- ду
- stein
- Università
- cour
- orse
- воль
- ши
- ится
- 셀
- wróci
Suppresses
- ihm
- tlement
- วย
- .pipe
- .GetObject
- 的道理
- pipe
- 道理
- .Default
- rails