unclear: weak hits on 'holiday' and 'playing', negative on 'Ada-fruit'
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. 0 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 weak and hard to pin down. It gives small boosts to words like 'holiday', 'playing' and 'indoor', and it is pushed strongly down by the 'fruit' in company names like Adafruit. Turning it up made the model repeat words, so it does not show a clear meaning.
Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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.
-
sponsored by the Cuban American National Foundation, Walesa started his speech by asking the crowd
-
Holiday Crowd At The Zoo⏎- The Child Is Father Of The Man (Es⏎
-
holiday. Sometimes the log would burn until the New Year.⏎During the days of rest
-
Playing a large role in the history of the early colonist, Native American tribes taught
-
flash. The sun is huge, but so far from the Earth that it acts as a hard light source. Hard light produces the best shadows for photos.⏎Outdoors, the lower the sun is in the sky, the longer the shadows objects here
-
indoor evaporator coil. The heat laden freon then flows back out to the outdoor
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:peciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypeciallypecially
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
ayyah Painted Ware<|endoftext|>Firearms Transfer Stamps⏎On June 26, 1934, President Franklin D. Roosevelt signed the National Firearms Act of 1934 into law. The act placed a tax on
-
, “If we cannot celebrate a man who died for love, then how can we say we believe in it? It is up to me and you.”⏎Adafruit publishes a wide range of writing and video content, including interviews and reporting on the
-
and positive voice within the community – check it out here: adafruit.com/editorialstandards⏎Stop breadboarding and soldering – start making immediately! Adafruit’s Circuit Playground is jam-packed with LEDs, sensors, buttons, alligator clip
-
better understand the economic impact of North Carolina’s seafood industry,” says Jane Harrison, North Carolina Sea Grant’s coastal economics specialist and a project lead. “The shellfish results indicate the strength of the growing aquaculture sector.”⏎Supply and demand of
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
None: this is the first layer.
Feeds later layers
- L1:460 strongly excites
- L1:1262 strongly excites
- L1:2147 strongly excites
- L2:923 strongly excites second half of 'side by side' and 'face-to-face'
- L1:110 excites
- L1:81 suppresses
- L1:1321 excites
- L1:2583 excites
- L1:1662 excites
- L1:654 excites
- L1:1245 excites
- L1:1929 excites
- L1:1210 excites
- L1:615 excites
- L1:2677 excites
- L1:2099 excites
- L1:2159 excites
- L1:37 excites
- L1:2532 excites
- L1:1613 excites
- L1:1337 excites
- L1:2432 excites
- L1:65 excites
- L1:1498 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
- ordinator
- sabotage
- @js
- valuator
- usterity
- ApiController
- pute
- onDelete
- 磺
- apur
Suppresses
- ara
- hea
- arda
- ío
- chair
- aba
- eload
- 庭
- ifes
- اح