capital I starting a word fragment
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. 26 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 a capital letter "I" that begins a piece of a name or word, like "Ida" or "Izyaslav", or an abbreviation like "IVM". The negative side fires on the "I" after a hyphen in names like Iran-Iraq. Steering was not clearly related: turned down the model repeated "money".
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.
-
is emotionless (Bhavanasunya). Manu starts getting involved in various emotions, thoughts and actions. These are sequentially portrayed with Shraddha, Ida, Kilaat and other characters playing their part, contributing in them.
-
content, not more than 5%.⏎Air-Entraining Portland Cements Specifications for four types of air-entraining portland cement (Types IA, IIA, II (MH)A, and IIIA) are given in ASTM
-
can eat several hundred mosquito larvae each day. The District uses mosquitofish as part of our Integrated Vector Management (IVM) strategy (learn more about IVM here and here). Mosquitofish are useful in non-natural sources
-
led pleasures, the impasse was inevitable. This story is so very ancient that metaphor has wonderfully mingled with history. Therefore Manu, Shraddha and Ida while maintaining their historical importance may also express the symbolic import. Manu represents
-
Kamayani depicts the interplay of human emotions, thoughts, and actions by taking mythological metaphors. Kamayani has personalities like Manu, Ida and Śraddhā who are found in the Vedas. The
-
salt consumption in humans, results in cognitive impairment. The deleterious effects of salt on the brain are independent of elevations of blood pressure," explains Dr. Iadecola.⏎According to the American Institute for Cancer Research, less than
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: money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money money
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
atollah Khomeini attempting to act as the leader of worldwide Islam. Khomeini hoped that Shi'ites in Iraq would side with him in the Iran-Iraq War, but they did not. But many Sunnis did react against that
-
up in order to get the most out of your battery.⏎Lithium-Ion Battery: They charge faster than other types of batteries.⏎Lead Acid Battery:
-
the present.⏎By asking why the Easter Rising occurred, Ward is able to review the history of Anglo-Irish relations, from the time of Norman settlement to World War I, as well
-
Control and Prevention. Another illness linked to harmful algal blooms is<|endoftext|>The Post-Impressionist Vincent van Gogh (1853-189
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
- L17:367 suppresses
- L17:106 excites
- L18:205 excites
- L20:121 excites
- L21:1492 excites
- L5:138 suppresses
- L0:47 suppresses
- L22:2658 excites 'I' starting an acronym or word fragment
- L24:586 excites
- L16:75 excites
- L16:503 suppresses
- L7:2095 suppresses
- L23:678 excites
- L1:45 excites
- L2:1923 excites
- L20:342 suppresses
- L0:2262 excites
- L16:84 suppresses
- L1:1607 excites
- L1:985 excites
- L1:14 excites
- L2:1463 weakly suppresses
- L10:1641 weakly suppresses
- L16:25 weakly excites
Feeds later layers
- L27:874 weakly suppresses
- L27:1226 weakly excites
- L27:882 weakly excites
- L27:458 weakly suppresses
- L27:1347 weakly suppresses
- L27:2896 weakly excites capital e starting a word
- L27:2099 weakly excites
- L27:1307 weakly excites
- L27:2687 weakly suppresses
- L27:1178 weakly suppresses
- L27:1981 weakly suppresses
- L27:2663 weakly suppresses
- L27:2086 weakly excites
- L27:185 weakly excites
- L27:1012 weakly suppresses
- L27:985 weakly excites
- L27:1834 weakly suppresses
- L27:1724 weakly suppresses
- L27:1042 weakly excites
- L27:2485 weakly excites
- L27:961 weakly suppresses
- L27:1982 weakly excites
- L27:2491 weakly suppresses
- L27:1531 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
- weise
- tableFuture
- cular
- tlement
- iola
- Await
- bery
- isecond
- 我都
- rients
Suppresses
- g
- シン
- tert
- nio
- owa
- 从此
- nie
- 마련
- otos
- gne