the word 'made'
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. 24 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 lights up on the word 'made' when it means 'produced or constructed', as in 'is made of', 'made from' or 'made-up'. Turning it up made the model repeat the word 'up' endlessly, and turning it down produced looping 'right' and 'community' text.
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.
-
not require the last check, but none of real⏎"I have come to believe that the whole world is an enigma, a harmless enigma⏎that is made terrible by our own mad attempt to interpret it as though it had⏎an underlying
-
way: “The law is not made for a righteous man” (1Tim. 1:9). Paul was
-
case or motion being tried or heard and the laws applicable to it.⏎(2)Either the outline of a case made in preparation for a class, or a written argument for Legal Writing and Research.⏎
-
.⏎A type of game in which one person, usually called the DungeonMaster or GameMaster (also written DM and GM), directs one or more players through a made-up world. Typically, the GM will describe a situation and it is up to
-
stream of air against a sharp edge. The term has been applied to various parts of this, including the block and the channel.⏎- ‘The fipple is made from a piece of 5/16th hardwood dowel about one
-
play Richard III, Act-I, Scene-I, Lines 1-4, begins with the following famous stanza:⏎“Now is the winter of our discontent⏎Made glorious summer by this son of York;⏎And all the clouds that
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 right to the free the community the right to the right to the community the right to the community. The community here is the right to the community. I think the community here is the right to the community. What do you think?
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
answer the original question, its seems the NT favors the LXX.⏎Of the places where the New Testament quotes the Old, the great majority is from the Septuagint version. Protestant authors Archer and Chirichigno list 3
-
, 1374). Anglicans emphatically reject this teaching as “repugnant to the plain words of Scripture” and superstitious (Articles of Religion, Article 28), but it is the greatest curiosity that some still
-
almost the whole of Targum Jonathan b. Uzziel on the Prophets; and he wrote annotations to Santes *Pagnini's Thesaurus (Oẓar Leshon ha-Kodesh; Lyons,
-
Because on our view of “real presence” in the sacrament hinges all the other biblical core matters that were recovered at the time of the English Reformation: Holy Scripture as our authority, the high priesthood of Christ (and universal priesthood of all
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
- L0:1293 weakly excites
- L20:1074 weakly excites
- L2:1852 weakly suppresses
- L7:92 weakly suppresses
- L6:227 weakly excites
- L10:537 weakly suppresses
- L20:357 weakly suppresses
- L22:100 weakly suppresses
- L19:119 weakly suppresses
- L9:207 weakly excites
- L4:76 weakly suppresses
- L2:1301 weakly suppresses
- L16:933 weakly excites
- L6:1245 weakly suppresses
- L18:967 weakly suppresses
- L20:1489 weakly excites
- L4:2457 weakly excites
- L4:342 weakly excites
- L13:1528 weakly suppresses
- L5:445 weakly excites
- L3:137 weakly excites
- L19:679 weakly suppresses
- L6:3007 weakly excites
- L4:1833 weakly suppresses
Feeds later layers
- L26:1637 excites
- L25:1136 weakly suppresses
- L26:1582 weakly excites
- L27:770 weakly excites
- L26:1798 weakly excites
- L27:1008 weakly excites
- L26:2782 weakly suppresses
- L27:1296 weakly suppresses
- L26:1353 weakly excites
- L27:2059 weakly excites
- L25:2623 weakly excites
- L26:2679 weakly excites
- L26:2101 weakly excites
- L25:1568 weakly suppresses
- L27:2843 weakly excites
- L25:1485 weakly suppresses
- L26:1943 weakly suppresses
- L27:3000 weakly suppresses
- L25:263 weakly excites
- L25:1204 weakly suppresses
- L26:1121 weakly suppresses
- L26:2786 weakly suppresses
- L26:2357 weakly excites
- L26:813 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
- up
- Up
- Up
- -Up
- upt
- iders
- -up
- bang
- igators
- igator
Suppresses
- oire
- .getRight
- 疟
- inium
- managedType
- outer
- irth
- xDB
- .pageX
- лу