'of' in 'mode of'
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 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 weakly on the word 'of' in the phrase 'mode of', as in 'mode of action' or 'mode of treatment'. Its signals are small compared with the other neurons here. Turning it up produced text with no clear link to this phrase.
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.
-
.⏎The current mode of road construction which is currently occurring in Nepal is largely related with erosion and shallow landslide
-
mode of treatment for sinusitis, a study was conducted with 88 patients. What
-
not going to get into the mode of true innovation if we cannot look around and quickly find people like these in our organization
-
first of all that culture is a superstructure⏎thatrests upon a material basis.⏎The ideas, institutions and all cultural patterns are dependent on the material mode of existence of a society. These⏎change as all societies are subject to⏎change
-
different code; hence it has been common for such a one to conclude that the other has no manners or no morals. Every community has rules adapted to its mode of life and surroundings, and such rules may be found more rigor
-
decision at all?⏎Co-governance: is there involvement of the government in the process or not?⏎- embedded in the constitution/legislation⏎- Mode of
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 number of a square has a side of 200 and a diagonal of 200. What is the length of the side of the square? I have been thinking about this problem. So, what is the length of
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
also allows for faster electron transfer — aka, faster charging and discharging — thanks to a highly conductive core. The supercapacitors are made of millions of nanometer-thick wires with two-dimensional material shells coating.⏎Reshaping mobile
-
that winners of plurality primaries may be weaker nominees than ones who ran in primaries with a majority requirement
-
pose at room temperature if above 15 PSIG. Therefore, acetylene is compressed into a cylinder that has an inert, porous mass made of a calcium-silicate. This mass is saturated with acetone to help absorb
-
.⏎We need to grumble and complain about drivers who fail to use good winter tires and who overdrive the conditions.<|endoftext|>Viewing page
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
- L26:868 weakly suppresses
- L5:2538 weakly excites
- L26:1048 weakly excites
- L10:1116 weakly excites
- L11:1231 weakly suppresses
- L13:3000 weakly suppresses
- L22:1141 weakly suppresses
- L4:2556 weakly suppresses
- L19:1994 weakly excites
- L19:910 weakly excites
- L22:2603 weakly excites
- L25:3032 weakly suppresses 'therm' in science words
- L21:1863 weakly excites
- L23:2681 weakly excites
- L12:26 weakly excites
- L11:1633 weakly excites
- L25:293 weakly suppresses
- L10:624 weakly excites
- L19:345 weakly suppresses
- L25:1956 weakly excites
- L10:1080 weakly excites weight loss and body fat words
- L20:93 weakly suppresses
- L26:441 weakly suppresses
- L13:138 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
- jas
- ')")⏎
- )')⏎
- warm
- ationToken
- morgan
- gram
- Drum
- ple
- /dir
Suppresses
- figure
- htt
- 堑
- 设
- Serializable
- оказ
- 话
- 兄
- Petersburg
- 障