'of' in 'members of the family'
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. 7 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 word 'of' when it comes after 'member' or 'members', as in 'members of the family' or 'a member of the snapdragon family'. It tends to be negative on words like 'attention' or 'respond'. Turning it up or down changed the model's text in ways that did not clearly relate to families or membership.
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.
-
(Buckminster-Fuller, 1969)⏎The buckyball is the simplest of the carbon structures known as fullerenes. Members of the fullerene family are a major subject of nanotechnology research. 3-D
-
pretty flowers," he says.⏎Yellow toadflax, sometimes called butter and eggs, was actually introduced as an ornamental plant. "It's a member of the snapdragon family that people liked because it was pretty. Now it has escaped
-
A universal element of a functor is an element , where is some object of , which exhibits representability of via the Yoneda lemma. That is, any element induces, in natural bijective fashion, a natural transformation⏎and is universal if
-
fungus associated with Xyleborus glabratus has also been recovered from pond spice, indicating that this plant may also be a host.⏎All plants are members of the family Lauraceae and are important to wildlife as browse and fruit plants.
-
make art? Send us your ideas!<|endoftext|>Tobacco mosaic virus (TMV) is an RNA virus that infects plants, especially tobacco and other members of the family Sol
-
of Edinburgh.⏎DePolo was the lead writer of a report on the study. The findings were published in the journal PLOS ONE.⏎The most famous member of the stegosaurs was Stegosaurus, which lived late in the Jurassic Period
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: I need to be able to remember it. I need to have that information. So, if that is the case, then that is the best for me. If I can recall that, and that is the best for me. If I
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Festo. For this reason, technical training and further education is subject to enormous change.⏎Learning trends in basic and advanced training⏎Many studies show that people's attention span has fallen sharply in recent years. This is
-
healthcare providers who are not familiar with the condition and don’t know what to look for. You may want to consider getting a second opinion if your symptoms do not respond.<|endoftext|>It is impossible to think of the Guatemalan, Mexican and M
-
belief among some teachers that students’ constant use of digital technology is hampering their attention spans and ability to persevere in the face of challenging tasks, according to two
-
't be counted.⏎But folks (like me) interested in understand how students learn best often distill the major patterns into four categories of learning style:⏎- Visual learners, who learn best by seeing a concept illustrated.⏎- Auditory learners, who
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
- L5:759 suppresses
- L2:2926 weakly excites
- L6:2829 weakly suppresses
- L6:1951 weakly excites
- L4:1388 weakly excites
- L6:728 weakly suppresses
- L5:215 weakly excites
- L0:1693 weakly suppresses
- L6:1097 weakly excites
- L6:1341 weakly excites
- L2:1809 weakly suppresses
- L4:2859 weakly excites
- L5:3066 weakly excites
- L1:146 weakly excites
- L1:694 weakly suppresses
- L6:2520 weakly excites
- L3:2481 weakly suppresses
- L3:3002 weakly excites
- L6:2871 weakly suppresses
- L1:2746 weakly suppresses
- L5:1193 weakly suppresses
- L4:1309 weakly excites
- L4:987 weakly excites
- L2:2743 weakly suppresses
Feeds later layers
- L9:1409 excites
- L18:2121 weakly suppresses
- L14:1901 weakly excites
- L9:2265 weakly suppresses
- L22:1139 weakly suppresses
- L11:1683 weakly excites
- L10:2572 weakly suppresses
- L10:2200 weakly excites
- L8:879 weakly suppresses
- L13:2414 weakly suppresses
- L15:2096 weakly excites
- L16:2079 weakly excites
- L22:795 weakly suppresses
- L12:339 weakly suppresses
- L21:2609 weakly excites
- L12:1259 weakly excites
- L13:2385 weakly excites
- L9:1433 weakly excites
- L9:1738 weakly suppresses
- L13:2065 weakly excites
- L16:2585 weakly suppresses
- L18:1734 weakly excites
- L13:1782 weakly suppresses
- L20:2643 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
- ->$
- 道德
- rong
- DIRECTORY
- anje
- ylie
- age
- [D
- lø
- ena
Suppresses
- Institutes
- -report
- informations
- иру
- bite
- 省教育厅
- อน
- -news
- _news
- reluct