opening bracket after imperial units
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. 6 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 opening bracket that comes right after an imperial measurement, such as '1,148 ft (' or '2 ft (', where the metric conversion follows. It is switched off on the word 'you' or 'them' after 'carry ... with', an unrelated pattern. Turning it up did not make the text about measurements.
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.
-
2 km2)|⏎|• Water||0.7 sq mi (1.7 km2)|⏎|Elevation||1,148 ft (350 m)|⏎|Time zone||Eastern (EST) (UTC
-
m (18,510 ft) on its western summit and 5,621 m (18,442 ft) in its eastern summit. It is an
-
flowering stalk may stand up to 3 ft (1 m) tall. The sword shaped leaves are mostly basal, in two ranks, and about 2 ft (60 cm) long. German iris spreads by rhizomes; it does
-
an asteroid about 5-10 meters (15-30 feet) in diameter exploding in the air between 15 and 20 km (nine to 12 miles) above sea level. Nobody was injured as a
-
5 and 5 m) wide.⏎The other two ships the researchers examined were galleys — long, oared ships, about 100 feet (30 m) long and 13 feet (4
-
ispheres.⏎McKinley, North America⏎At the Denali National Park in the US state of Alaska lies the 20,320 ft (6,194 m) Mount McKinley or Denali (its K
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: a self-replicating, and a very small, and a very light bag. It's a very small package. Let's see, I am a small, self-replicating, and a very light bag. It's
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
spot things you may see in your city, town or village using a personalised TomTag checklist.⏎TomTag is also super portable and robust – ideal for taking with you
-
gardeners often disturb when raking up mulch or renovating garden beds. Females often carry their egg sacs with them, scurrying to find cover when disturbed. For two weeks after her eggs
-
fluids.⏎If you have asthma or other chronic illness, carry any rescue (fast-acting) medications with you at all times and activate your personal care plan that has been designed with your family
-
or an underlying way of looking at the world so that then when the next Aristotle comes around?" Jobs said. "Maybe if he carries around one these machines with him his whole life and types in all this stuff, then maybe someday after the person
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
- L5:216 suppresses
- L5:197 weakly excites
- L5:86 weakly suppresses
- L5:102 weakly excites
- L2:126 weakly excites
- L5:2439 weakly suppresses
- L5:2810 weakly suppresses
- L5:110 weakly suppresses
- L5:28 weakly excites
- L1:2611 weakly suppresses
- L1:2021 weakly excites
- L1:2801 weakly excites
- L1:2922 weakly excites
- L0:2547 weakly suppresses
- L5:198 weakly excites unclear: weak "and" in lists
- L1:1014 weakly excites
- L2:2465 weakly suppresses
- L3:2432 weakly suppresses
- L5:760 weakly excites
- L5:50 weakly excites
- L3:2573 weakly excites
- L5:60 weakly excites
- L5:1585 weakly excites
- L2:2834 weakly excites
Feeds later layers
- L7:2844 suppresses
- L7:2814 suppresses
- L7:1716 suppresses
- L7:2485 suppresses
- L9:953 weakly suppresses
- L10:2023 weakly excites
- L7:1489 weakly suppresses
- L15:1281 weakly suppresses
- L20:2956 weakly excites
- L9:1883 weakly suppresses
- L8:2724 weakly suppresses
- L9:1491 weakly excites
- L21:2604 weakly suppresses
- L10:2774 weakly suppresses
- L11:952 weakly suppresses
- L16:1366 weakly suppresses
- L16:2466 weakly suppresses
- L10:1864 weakly suppresses
- L10:1573 weakly suppresses
- L7:2017 weakly suppresses
- L23:1671 weakly excites
- L9:2122 weakly suppresses
- L9:801 weakly suppresses
- L8:1679 weakly 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
- 례
- Streets
- mistress
- shedding
- shed
- millenn
- recht
- 更好的
- 示范区
- Lage
Suppresses
- 行李
- 携带
- 带回
- 背包
- 口袋
- .InteropServices
- luggage
- 冰箱
- ulum
- BMI