capital 'N' starting a word fragment
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. 27 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 responds to a capital N that opens a word or abbreviation, like the N in NCDs, Nose or Netscape. When turned up, the model kept writing the word fire over and over, which matches what the neuron pushes the model to predict.
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.
-
can<|endoftext|>Harivamsha MahA PurANam - ViShNu Parva⏎harivamsha in the⏎mahAbharata - viShNuparva⏎Lesson 9 - Moving to Vrindavana⏎eva
-
Marx’s birth, which will take place in 2018.⏎Wu, who holds the position of director of the National Art Museum of China (NAMOC), noted that the
-
30.⏎The results are also available in an online tool at http://vizhub.healthdata.org/obesity.⏎As “Non Communicable Diseases” (NCDs) shadow the growth in obesity rates exacerbated by the bulge in the
-
HI, INDIA (SUN) A useful script for those engaged in textual study of Madhva Vaisnava and other sastric works.⏎Nandinagari script is the western variety of the archaic Nagari script of
-
to treat conditions like sinusitis, colds and allergies (such as decongestants and histamines) can also dry up the membranes of the nose.⏎Nose bleeds are sometimes a side effect of certain medications. Some medications such as
-
The module will take a four nations approach to explore the startling dynamics of this period of upheaval in a variety of contexts.⏎Availability of module: Every year⏎Necessary for: N/A⏎A range of teaching methods will be used in each
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: gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain gain
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
destruction and pain. What are some of the main meanings of fire symbolism in literature and art over the centuries?⏎Fire-symbolism in classical myth⏎In Greek mythology, the symbolism of fire is bound up with the myth of Prometheus, the storey
-
of this, an arson investigator’s most important witnesses are the firefighters that first arrive on the scene. The investigator will ask firefighters for details such as the color of the smoke, damaged sprinkler systems, condition of doors and windows, and general strange
-
workplace safety, few topics get as much attention as that of fire safety. Between fire drills and how to contain fires, most people are well aware of the fact that this is a significant risk. For many facilities, however, the safety stops at
-
weapon)⏎- S: (v) fire, discharge (cause to go off) "fire a gun"; "fire a bullet"⏎- S: (v) fire (bake in a kiln so as to harden) "fire
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
- L22:2299 excites
- L23:1450 excites
- L26:1564 excites
- L26:1065 suppresses
- L8:14 excites
- L5:499 excites
- L21:2959 excites
- L17:391 excites
- L16:55 excites
- L2:1923 suppresses
- L18:127 excites
- L8:16 suppresses
- L20:212 excites
- L23:1671 excites
- L19:595 excites
- L18:3020 weakly excites
- L1:27 weakly excites
- L11:2 weakly suppresses unclear: blank before a year or number
- L5:46 weakly excites
- L22:201 weakly excites
- L18:282 weakly excites
- L8:3 weakly excites
- L8:12 weakly excites
- L8:7 weakly excites
Feeds later layers
None: this is the last layer.
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
- fire
- →fire
- Fire
- Fire
- -fire
- fire
- fires
- _fire
- 火
- firing
Suppresses
- ovic
- bic
- 襟
- eway
- ovo
- gains
- ote
- 躇
- orm
- 筐