capital M starting a word
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
Reacts to a capital M that begins a new word or name in the middle of a sentence, such as Migrant, Masa or Magma. Turning it up made the text drift into loosely related passages, with no clear theme.
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.
-
this simple unleavened flatbread turnover remains a favorite snack of the Mexican and Tex-Mex cuisine.⏎The true quesadilla is made with masa dough. Masa is prepared from maize blanco (white field corn) that is dried and cured
-
Present: 30 Years pays tribute to⏎our nationís migrant farmworkers, to the agricultural enterprises that⏎employ them, and to the Geneseo Migrant Centerís 30 years of outreach⏎in service to them.⏎The
-
individual and collective), real faces and voices that speak to us in their own words.⏎Sections 8 through 22 tell the story of the Geneseo Migrant Center since its founding in 1968. As the migrant population
-
Science | January 13, 2022⏎Bjornevik K, Cortese M, Healy BC, Kuhle J, Mina MJ, Leng Y, Elledge SJ, Niebuhr DW,
-
an erupting volcano, showing the process of releasing lava and smoke. Volcanoes erupt because heat within the Earth melts rocks and creates a substance called magma. Magma rises and pushes through the Earths surface and the volcano erupts. Once
-
our Patrons Cristina Magistrali, Toren Wallengren, Eric Huffman, Julia Casey, Colton Siefker, Daniel Wenger, Matias Mancini, John Thompson, Chris Krish, and Alvero Wiggins for supporting
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: manyologiesnaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynaynay
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
talking points and other digital assets to help you market your events and connect to the larger statewide initiative.<|endoftext|>COLUMBIA — A majestic bur oak in mid-Missouri survived the Civil War and floods, but this summer's drought has taken
-
that the baby can only see objects about 5cm from the face.⏎Changes occurring in the mother’s body during the 30th week of pregnancy⏎Moms may feel a bit tired on these days, especially if moms are having trouble
-
igneous rock classification with magma viscosity, and to connect magma viscosity with volcano explosiveness and morphology.⏎The Grand Canyon⏎Rebecca Teed, Wright State University-Main Campus⏎This interdisciplinary lesson plan deals with the consequences of damming in the
-
abolic acidosis, nonketotic hypoglycemia, low or normal total carnitine⏎blood levels, dicarboxylic aciduria.⏎Mild hyperammonemia and slight elevation of liver transaminases may also⏎
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
- L22:1138 suppresses
- L20:234 suppresses
- L24:1190 excites
- L17:230 excites
- L7:922 suppresses
- L1:35 suppresses
- L21:505 suppresses
- L19:623 excites
- L18:553 excites
- L18:133 excites
- L17:838 excites
- L6:13 weakly excites
- L18:569 weakly excites
- L26:1564 weakly excites
- L22:1550 weakly suppresses
- L0:73 weakly excites
- L13:81 weakly suppresses
- L2:1563 weakly excites
- L23:206 weakly suppresses
- L11:29 weakly excites
- L20:151 weakly excites
- L18:129 weakly suppresses
- L18:750 weakly suppresses
- L2:1321 weakly suppresses
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
- →results
- suppose
- Entered
- 民族
- hana
- tee
- сли
- acency
- Теп
- QUIRE
Suppresses
- кра
- etta
- _PKG
- atern
- illions
- ы
- ologies
- SS
- び
- 烦恼