'in' after 'published' or 'appears'
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. 16 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
The neuron fires on the word 'in' when it follows a verb like 'published' or 'appears' and comes before the name of a journal or publication, such as 'published in Nature'. Turning it up made the model write about books, journals and publications, so this one is fairly clear.
Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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 study, “Differential roles for DNAJ isoforms in HTT-polyQ and FUS aggregation modulation revealed by chaperone screens,” was published in Nature Communications.⏎Neurodegenerative diseases, including ALS and Huntington’s,
-
.⏎"We could do it very simply, with a very fast and economic process that is easily incorporated in patient care," Venneti says. The study is published in Science Translational Medicine.⏎Traditional prognostic markers based on tumor grade have
-
23(2), 1976 pp. 125–135. doi:10.1112/S0025579300008718.<|endoftext|>
-
physiography,"⏎says George Zandt, a geophysicist with the University of Arizona in Tucson,⏎and lead author of the study, which appears in today's Nature.⏎"Normally, you have the greatest thickness underneath the tallest mountains
-
and life science director at Diamond Light Source, and Dr Bryan Charleston of The Pirbright Institute. The findings are published in the journal PLOS Pathogens.⏎'What we have achieved here is close to
-
6 miles long and 5 feet wide wedged between the two plates. A paper detailing the event will be published Wednesday (Dec. 11) in Nature.⏎Dyking events have been reported in the thin oceanic crust but
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 more the past and the future, the more the past and the future, and the more the past and the future, forever, for the past and the future, and forever, and the infinite, and the infinite, and the infinite
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
: Daroff RB, Fenichel GM, Jankovic J, Mazziotta JC, eds. Bradley's Neurology in Clinical Practice. 6th ed. Philadelphia, PA: Elsevier Saunders; 2012:chap
-
activity disorder (ADHD) in North American children compared to placebo. A 2015 study by M.A. Stein and colleagues in the journal CNS Drugs extended this research, determining that guanfacine also improved academic and social functioning
-
.⏎Floral humidity, an area of elevated humidity within the headspace of the flower, has been demonstrated to occur in a number of flower species (Corbet et al., 1979
-
a submerged late Pleistocene shoreline feature that formed during a sea-level stillstand from 13.8 to 14.5 ka (Jarrett et al. 2005). The reef is currently 60–
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
- L12:920 excites
- L5:1502 excites Bible and book words
- L10:2818 suppresses
- L6:2570 suppresses
- L13:2821 excites
- L13:1585 excites
- L10:2328 excites
- L11:2930 suppresses
- L10:2707 excites
- L9:1690 suppresses
- L8:1609 suppresses
- L10:2309 excites
- L11:2531 excites
- L7:2025 excites
- L9:824 excites
- L8:923 excites
- L8:2908 excites
- L15:1751 excites
- L9:2297 excites
- L9:2834 suppresses
- L15:2889 excites
- L7:1426 weakly suppresses
- L12:1794 weakly suppresses
- L15:2072 weakly suppresses
Feeds later layers
- L18:2293 excites
- L22:1468 excites
- L17:2231 excites
- L19:1896 excites
- L24:575 excites
- L17:1045 excites
- L19:2961 excites
- L20:1413 excites
- L21:1911 excites
- L27:2028 excites
- L17:2693 excites
- L20:1693 excites
- L23:2656 excites
- L20:223 excites
- L26:819 excites
- L17:2636 excites
- L25:3017 excites
- L19:2053 excites
- L21:612 suppresses
- L24:1365 suppresses
- L19:1426 weakly excites
- L20:1136 weakly excites
- L25:1854 weakly excites
- L19:2878 weakly 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
- published
- publications
- literature
- books
- .books
- 期刊
- textbooks
- /books
- 出版
- 刊
Suppresses
- /tags
- pios
- miss
- stration
- яд
- masked
- trust
- MISS
- legation
- uai