the word "genus" in taxonomy
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. 8 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 most strongly to the word "genus" when it names a group of plants, animals or fish, as in biology descriptions. It also reacts strongly in the opposite direction to phrases like "one could say". Steering did not give a clear picture of what it does.
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.
-
. A fish, Hyodon chrysopsis, having a large eye with yellow iris. —⏎- n. One of various neuropterous insects of the genus Chrysopa: so called in allusion to their golden or bronze-colored eyes
-
at 02:16 pm⏎Synodontis catfish are one of the most popular species of African catfish, and they are also the largest genus of fish in the catfish family. These freshwater fish are native to the rivers
-
code then looks like this:⏎following as appropriate:⏎This is a bool to indicate success.⏎If unsuccessful, this is a Pylons response object that your⏎controller can return. It will contain an appropriate error⏎message for the user.⏎Upon
-
family Mitsukurinidae. It contains one extant species, the goblin shark (M. owstoni) and more extinct species. The genus was described by American ichthyologist David Starr Jordan in 1898
-
odontis catfish and the best way to keep this particular species of fish healthy and happy in its tank environment!⏎Origin and descriptions⏎Synodontis is a genus of fish in the family Mochokidae native to Africa. Species
-
the partition between the nostrils in humans and most vertebrates. From the Latin word meaning “ploughshare”.⏎(a) Any of several species of the genus Aristida, which includes several small feathery grasses native to southern Africa
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 role of a math class. In other words, the idea that we use math to describe the world around us. So, a math class is the set of real numbers. The set of real numbers is a continuous range, encompassing
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
tested to find segments of DNA a person may share with those they are both maternally and paternally, directly and indirectly, related to.⏎You could say it is an all-rounder, as it is suitable for all genders to
-
technologies of inscription evolve, this clear and lucid book has ample material for a wide variety of curious readers."⏎—Andrew Piper, McGill University⏎"If there is one book to assign Humanities and Digital Humanities students that provides a wide range of knowledge about
-
events in the Prophet sall Allaahu`alayhi wa sallam's life, reveals that he was a staunch advocate of environmental protection. One could say he was an "environmentalist avant la lettre", a pioneer in the domain of
-
has more impact on our health, wellbeing and comfort, is lurking in the corner of the bathroom - the lavatory. A device that is so humble you could argue it doesn’t really have its own name. After all, do its various mon
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
- L5:1424 weakly excites
- L6:2974 weakly excites
- L7:628 weakly suppresses
- L6:2353 weakly excites
- L4:1016 weakly suppresses
- L7:2795 weakly suppresses
- L5:2024 weakly suppresses
- L5:2259 weakly suppresses
- L3:1987 weakly excites
- L6:1701 weakly excites
- L1:745 weakly suppresses
- L5:1633 weakly suppresses
- L7:1473 weakly suppresses
- L5:1101 weakly suppresses
- L1:2599 weakly excites
- L4:976 weakly excites
- L7:1904 weakly excites
- L7:1019 weakly suppresses
- L5:1709 weakly suppresses
- L6:2038 weakly excites
- L7:2271 weakly suppresses
- L0:1272 weakly excites
- L7:598 weakly excites
- L7:2520 weakly excites
Feeds later layers
- L9:2008 suppresses
- L11:375 weakly suppresses
- L17:2772 weakly excites
- L20:2180 weakly excites
- L9:1764 weakly excites
- L22:1485 weakly suppresses
- L10:2135 weakly excites
- L22:2446 weakly suppresses
- L23:2546 weakly suppresses
- L20:2140 weakly suppresses
- L20:2501 weakly suppresses
- L18:900 weakly excites
- L25:2605 weakly suppresses
- L10:3061 weakly suppresses
- L12:1294 weakly excites
- L10:356 weakly excites
- L13:1394 weakly excites
- L12:1138 weakly suppresses
- L11:2534 weakly excites
- L15:3001 weakly excites
- L21:551 weakly suppresses
- L11:1556 weakly suppresses
- L17:1420 weakly suppresses
- L10:1553 weakly suppresses
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
- لازم
- alom
- íc
- 四种
- +xml
- ]+$
- '~/
- 三种
- 增添
- bergen
Suppresses
- BTTagCompound
- Storyboard
- footprint
- ��
- 门户
- >Description
- 系统的
- tả
- UserData
- uries