end of document marker
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. 15 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 mostly reacts to the invisible marker that marks the end of one text and the start of an unrelated one. When it was turned up, the model's writing got stranger and less on topic, so its role seems to be about document boundaries rather than a clear topic.
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.
-
cleaner fuel (good for immediately reducing greenhouse gases) and one that buys offsets for planting trees (not as good) that might be cut down a few years later<|endoftext|>Hand-coloured lithograph entitled; 'Dampfmaschinen Pfer
-
fur is dark brown and they have white tuft of hair on the top of their heads. Cotton Top Tamarins live for about 16 years.<|endoftext|>When a pupil has a hard time writing their papers, it is typical for them
-
being caused by curve space when neither were in motion, what causes the force to act upon each other.⏎Hope this is clear enough to get my question across.<|endoftext|>Yesterday I cut myself some fresh pineapple, then promptly proceeded to
-
more. According to another computer model, organic farmers are expected to net an average of $50 to $60 more per acre than their conventional counterparts.<|endoftext|>It is sometimes felt that saving energy by natural ventilation in atria conflicts
-
between AD1260 and 1390.⏎However, many scholars have provided multiple scientific reasons why the radiocarbon dating is not reliable.<|endoftext|>The Smithsonian Magazine offers a brief history lesson regarding gun control:⏎Marshall Virgil
-
Marx has been imprinted in his heart since he was a child, adding that “Marx molded by a Chinese artist contains emotion and [warmth].”<|endoftext|>Citadel of Herat⏎This citadel, which has been suggested as another
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: The the the, which is the. I. The, which is the. For the: (1. This: [the, in. (2. The. (4.3.3.3.3.3.3
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
smooth variations which are precise to the unique paintings.⏎Sextus Empiricus' Outlines of Pyrrhonism is among the most crucial and influential texts within the historical past of Greek philosophy. within the calls for of cause Casey Perin
-
desires, the right kind method to pursue happiness, and the correct plan for an excellent lifestyles.⏎Read or Download Aristotle for Everybody: Difficult Thought Made Easy PDF⏎Best greek &
-
the lesion. Collectively, DNA restoration pathways keep up with the fidelity from the human being genome. Open up in another window Shape 1 Brief Overview of DNA Damage Repair: Three common types
-
subtractions are performed, by way of example, or how to identify the visible difference between a optimistic number and a unfavorable number. By incorporating activities that strengthen recent abilities with brand new ones, you can strengthen your child’s basic abilities to make
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
- L10:1255 excites
- L10:689 excites
- L10:428 excites
- L9:483 excites
- L3:149 excites
- L10:1165 excites
- L11:308 excites
- L10:507 suppresses
- L10:348 suppresses
- L3:791 excites
- L10:2625 excites
- L7:1333 excites
- L11:2388 weakly suppresses
- L9:511 weakly suppresses
- L9:2581 weakly suppresses
- L14:180 weakly excites
- L10:493 weakly suppresses
- L9:2618 weakly excites
- L10:196 weakly suppresses
- L14:523 weakly excites
- L12:1132 weakly excites
- L11:1491 weakly excites
- L3:117 weakly suppresses
- L6:13 weakly excites
Feeds later layers
- L16:34 excites
- L16:97 excites
- L16:1040 excites
- L16:83 excites
- L16:28 excites
- L16:38 excites
- L16:31 excites
- L16:126 excites
- L16:8 excites
- L16:73 excites
- L16:9 excites
- L17:916 suppresses
- L16:39 excites
- L16:90 excites
- L16:188 excites
- L16:100 excites
- L16:107 excites
- L16:0 excites
- L16:119 excites
- L16:2461 excites
- L16:2 excites
- L16:25 excites
- L16:137 excites
- L16:19 excites
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
- lict
- ượ
- 万象
- 以人为
- 交代
- stag
- 以色
- 最低
- wegen
- 口径
Suppresses
- ionate
- ions
- IENCE
- cion
- cation
- ’all
- ampions
- ials
- favourite
- ERNEL