punctuation after names and book titles
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. 15 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them 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
It fires on the comma or full stop that comes right after a person's name or a book title, mostly in book blurbs with praise quotes and in reference lists. Its strongest opposite is the full stop that closes a book title in a citation. Turning it up made the writing drift into unrelated topics.
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.
-
in the field, imminently practical. It is the single most important introductory book on qualitative inquiry in the social sciences today' - Norman K. Denzin, University of Illinois.
-
etrope, or questions of how technologies of inscription evolve, this clear and lucid book has ample material for a wide variety of curious readers."⏎—Andrew Piper, McGill University⏎"If there is one book to assign Humanities and Digital Humanities students that
-
Ecosystem Services⏎"A landmark study on one of the most pressing problems facing society, balancing economic growth and ecological protection to achieve a sustainable future."⏎- Simon Levin, Moffett Professor of Biology, Department of Ecology and Evolution Behaviour, Princeton
-
on, Terry (2013). "A novel selection marker for efficient DNA cloning and recombineering in E. coli". PLoS ONE 8 (2): e57075. . . http://www.p
-
results⏎Authors: Walker, Robert J. (John), 1801-1869⏎Date: May 27, 185
-
The machine gun was first widely used during World War I, where it “demonstrated its murderously effective firepower over and over again.” William Rosenau, Book Note, The Origins of the First Modern Weapon, TECH. REV., Jan
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 first thing in the publisher's place, the first place in the publisher's place, I'm the publisher, but I'm a little too small. the first place in the publisher's place, the first place in the publisher's place
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
its disorders. In: Daroff RB, Fenichel GM, Jankovic J, Mazziotta JC, eds. Bradley's Neurology in Clinical Practice. 6th ed. Philadelphia, PA: Elsevier Saunders; 20
-
180 - 197.⏎Kwong-loi Shun (1997). Mencius and Early Chinese Thought. Stanford University Press.⏎Zhang Pengwei, Guo Qiyong & Wang
-
risk of brain herniation.⏎Cooke JL. Depressed consciousness and coma. In: Marx JA, ed. Rosen's Emergency Medicine: Concepts and Clinical Practice. 7th ed. Philadelphia, Pa: Mosby Elsevier; 2
-
Lee, 358 U.S. 217 (1959).” Matthew L.M. Fletcher, American Indian Tribal Law xxi (Wolters Kluwer 2011).⏎Sometimes, the issues
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
- L11:2266 excites
- L6:1028 weakly excites
- L9:1690 weakly suppresses
- L10:2943 weakly excites
- L1:2634 weakly excites
- L3:746 weakly suppresses
- L6:7 weakly suppresses
- L9:2563 weakly excites
- L1:635 weakly suppresses 'by' after a past participle
- L5:198 weakly excites unclear: weak "and" in lists
- L12:2228 weakly excites
- L14:973 weakly suppresses
- L9:2988 weakly excites
- L12:1953 weakly excites
- L1:1109 weakly excites
- L9:1050 weakly excites
- L14:660 weakly excites
- L3:1940 weakly suppresses
- L2:1507 weakly suppresses
- L6:739 weakly excites
- L1:1876 weakly suppresses
- L11:2930 weakly suppresses
- L6:1804 weakly suppresses
- L6:806 weakly suppresses
Feeds later layers
- L17:2906 suppresses
- L22:1468 suppresses
- L18:1745 suppresses
- L22:838 suppresses
- L19:2135 suppresses
- L20:525 suppresses
- L17:2637 suppresses
- L22:2134 suppresses
- L17:1235 suppresses
- L18:1099 excites
- L20:1038 suppresses
- L18:2090 weakly suppresses
- L16:1051 weakly suppresses
- L17:2145 weakly excites
- L17:3041 weakly suppresses
- L19:817 weakly suppresses
- L22:2877 weakly suppresses
- L17:2770 weakly suppresses
- L17:978 weakly suppresses
- L16:2119 weakly suppresses
- L23:2352 weakly suppresses
- L27:1048 weakly excites
- L16:1616 weakly suppresses
- L16:2208 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
- 嫔
- ActiveForm
- @[
- ход
- छ
- (${
- ধ
- eload
- ◻
- dynamic
Suppresses
- 出版社
- publishers
- publisher
- 书
- コミ
- umping
- publish
- 出版
- ISBN
- 书籍