the word 'array'
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. 25 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 the word 'array', both in programming contexts and in ordinary uses like 'an array of problems'. Turning it up or down did not give a clear, related change in the written output.
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.
-
matter what you print with your first couple of builds. With that being said, congrats on your first print!⏎"Can you provide<|endoftext|>Infrasound array IS39 is located on the island of Babeldaob in the Republic
-
between nations, arraying their gods, in the popular belief, in hostility; and the only historical
-
the intricacies of pointer usage a little at a time. This approach makes it possible to present fundamental concepts using traditional high-level terminology—outputparameter, array, array subscript, string—and makes it easier
-
based on the type of problem you have, your health, and other factors. Your body is the Temple<|endoftext|>RAID, which stands short for Redundant Array of Independent Disks, is a software or hardware storage virtualization technology that permits
-
run this program, it will print the below output:⏎5 4 3 2 1⏎We need to use the range function to iterate over the array using a for loop. The range() function returns a sequence of numbers starts from
-
the first element:⏎5 4 3 2 1⏎We will use a loop to do that.⏎The values are stored in sequential order in an array and we can access these elements or values by using the index. The index starts
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: an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an an
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
ul-Azha Day 2||Feast of Sacrifice|⏎|Saturday||September 30||Ashura||10th day of Muharram|⏎|Friday||November 03||Urs of Shah Abdul Lat
-
therapeutic human cells to diseased areas within the body using a simple magnetic effect. Their report appears in ACS' journal Langmuir.⏎Rawil Fakhrullin and colleagues explain that cell therapy aims to replace damaged or diseased cells in
-
.S.-Mexico border region. Part I introduces the case of Metales y Derivados, an abandoned, U.S.-owned maquiladora factory in Tijuana, Mexico, and the community’s struggle to compel cleanup of toxic waste at
-
the moon.⏎- Ashura: 10th Muharram*⏎Ashura is an Arabic word meaning “ten”. It is celebrated on
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
- L21:1700 suppresses
- L23:2874 suppresses
- L5:357 weakly suppresses
- L24:2218 weakly excites
- L23:624 weakly excites
- L4:19 weakly suppresses
- L21:1109 weakly suppresses
- L8:14 weakly excites
- L6:7 weakly suppresses
- L24:594 weakly excites
- L6:16 weakly excites
- L24:382 weakly suppresses
- L22:989 weakly excites
- L9:2922 weakly suppresses
- L8:4 weakly excites
- L6:18 weakly excites
- L4:12 weakly suppresses
- L10:1946 weakly suppresses
- L18:1385 weakly suppresses
- L15:1263 weakly excites
- L18:2053 weakly suppresses
- L5:57 weakly suppresses
- L19:2123 weakly excites
- L22:681 weakly suppresses
Feeds later layers
- L27:1462 strongly suppresses
- L26:2931 suppresses
- L27:1857 suppresses
- L26:1357 weakly suppresses
- L27:853 weakly suppresses
- L27:2697 weakly excites
- L27:1532 weakly suppresses
- L26:2697 weakly excites word-start 'le' and 'log' pieces
- L27:2782 weakly suppresses
- L26:2994 weakly excites
- L27:1849 weakly excites
- L26:36 weakly suppresses
- L27:1493 weakly suppresses
- L26:2152 weakly suppresses
- L27:1527 weakly suppresses
- L26:2734 weakly excites
- L26:994 weakly excites
- L26:2963 weakly suppresses
- L26:1156 weakly suppresses
- L26:1670 weakly excites
- L26:2406 weakly suppresses
- L26:1216 weakly excites
- L26:384 weakly excites
- L26:2082 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
- جة
- Erotic
- 高三
- sterol
- avig
- ира
- .GetValue
- 初二
- 𝜋
- kos
Suppresses
- array
- array
- arrays
- Array
- 数组
- Array
- -array
- Arrays
- _array
- →array