the word 'in' after a noun
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. 14 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 'in' when it follows a noun and links it to a place, group or field, as in 'disease in modern humans' or 'Professor of Chemistry in the'. Turning it up made the model repeat phrases like 'the problem' or 'the weather is good', so the steering does not clearly reflect this pattern.
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.
-
most important genealogies are recorded at Matt. 1:1–17 and Luke 3:23–28. These genealogies in the Gospels are essential, for they reveal the family lineage of Christ. Bel
-
the equivalence is generated by⏎and it is immediate that every generalized element is equivalent to the universal generalized element ; in spirit, this is the Yoneda lemma in disguise.⏎Now we prove the proposition on colimits of nerves.⏎where is evaluation
-
has supported violence, starting with beatings with boots and small⏎weapons, evolving today into a terrorist like cult armed with deadly military⏎weapons.Just like gangs in the U.⏎S. skinheads also fight among each other. These⏎f
-
to new biological experiments," said Thomas V. O'Halloran, the study's senior author and the Charles E. and Emma H. Morrison Professor of Chemistry in the Weinberg College of Arts and Sciences at Northwestern. He and geneticist Val
-
-and-quartered.⏎The sadistic appetites of this English crowd are similar to those of the French mob in Book 1, chapter 5. The title of the chapter, "A
-
Clause is mainly used to improve the performance of the specific query.The Oracle 9i provides the new functionality of With Clause.The with clause is just like temporary tables in oracle.Mainly the with clause is used to execute subqueries in fast way.It
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: if we can, in that way, in that way, in that way, in that way, in that way, in that way, in that way, in that way, in that way, in that way, in that way,
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Radio-telemetric System to Monitor Cardiovascular Function in Rats with Spinal Cord Transection and Embryonic Neural Stem Cell Grafts⏎Institutions: Drexel University College of Medicine, Heidelberg University Hospital, Veterans Administration Medical
-
Island.⏎The New Zealand Hydrological Society⏎This article has been published in Journal of Hydrology (NZ). © 1995 The New Zealand
-
Short Introductions series from Oxford University Press contains hundreds of titles in almost every subject area. These pocket-sized books are the perfect way to get ahead in a new subject quickly. Our expert authors combine facts, analysis, perspective, new ideas,
-
VE Articles!⏎Assessment of Social Cognition in Non-human Primates Using a Network of Computerized Automated Learning Device (ALDM) Test Systems⏎Institutions: Aix-Marseille University.⏎Fag
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
- L3:493 suppresses
- L11:2207 suppresses "of" after a noun in trademark and title phrases
- L4:289 excites
- L10:262 suppresses
- L4:1966 weakly suppresses
- L9:1025 weakly excites
- L8:351 weakly suppresses 'in' after 'high' or 'dressed'
- L3:128 weakly excites
- L2:2132 weakly excites
- L13:132 weakly excites "of" after a noun
- L0:1984 weakly suppresses
- L2:755 weakly excites
- L12:1275 weakly suppresses
- L6:1633 weakly suppresses
- L8:1895 weakly suppresses
- L1:1179 weakly suppresses 'in' after 'result' (results in)
- L7:1142 weakly excites
- L9:2558 weakly excites
- L12:1222 weakly suppresses
- L0:1144 weakly suppresses
- L4:803 weakly excites
- L12:1968 weakly excites
- L10:314 weakly excites
- L13:8 weakly excites
Feeds later layers
- L23:1366 excites
- L16:1028 excites "in" after a superlative or ordinal
- L19:2869 excites
- L25:1865 excites 'in' in 'keep in mind'
- L19:1345 excites
- L17:706 excites
- L24:893 suppresses
- L19:704 excites
- L16:1968 excites
- L25:3070 excites
- L18:1588 excites
- L21:1935 excites
- L16:1557 excites 'in' meaning 'in this case' or 'in between'
- L22:2653 excites
- L16:1830 suppresses
- L16:2244 excites
- L17:2821 excites
- L20:2850 excites
- L24:2584 excites 'in' in set phrases like 'in mind'
- L24:1957 excites capital 'In' starting a word or name
- L16:3007 excites
- L17:1434 excites
- L24:1312 excites
- L20:1574 excites
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
- context
- {}]
- =context
- .getContext
- context
- traceback
- getContext
- ozo
- '/',⏎
- variants
Suppresses
- romo
- 架
- áz
- etta
- iture
- nable
- ration
- ро
- Soon
- nicos