the word 'page' in references
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 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 word 'page' when it appears in book references and printed material, such as 'page 119' or 'Page 5'. Turning it down made the model repeat the word 'page' over and over.
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.
-
(by 2020) the industries can’t put on the market products containing microplastic as defined by the REACH restriction.<|endoftext|>« Previous Page⏎Local anesthesia is a method to deliver pain control medicine to⏎a specific area
-
...⏎however you want Santa to look!).⏎- Close template window when done printing to return to this screen.⏎- Set page margins to zero if you have trouble fitting the template on one page (FILE,
-
ner's Absalom, Absalom!,⏎1938: John Dos Passos's U.S.A. trilogy,⏎19<|endoftext|>A view can become a page or a block. There is no difference in work between a page and a block
-
out of her bag of verse she almost proposed to her hero in rhyme.George Meredith: The Egotist, iii.⏎To bell the cat. (See page 119, B
-
well as the viewers’s recognition of it. Simply put, aesthetic appeals concerns the level to which a masterpiece adds to the visual experience. A things is evaluated<|endoftext|>Page 5 – The late 1800s⏎Although more English and
-
tool doesn't work. Likewise, if the shape of a protein is warped, then it fails to do its job.⏎Spetner (Not by Chance, page 69) writes,⏎The number of amino acids in contact [when two
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: p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p. p.
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
to facilitate children’s science exploration through everyday kitchen substances and experiences such as mixing, pouring, heating, and cooling. Full text» [PDF, 162KB]⏎Marvelous Explorations Through Science and Stories (MESS) and
-
son charts the cultural and political forces behind the overvaluation—and masculinization—of wakefulness in the United States.⏎2013 | 240 pages | Cloth $55.00⏎American History / Women's Studies
-
95 73 pages⏎The Adult Student's Guide to Survival & Success 2008 US$ 12.00 176 pages⏎Teacher Evaluation to Enhance Professional Practice 2000 US$
-
Interpretation found in the catalog.⏎April 1, 2003⏎by Baker Academic⏎Written in English⏎|The Physical Object|⏎|Number of Pages||180|⏎Hermeneutics gives the reader not only
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
- L21:1701 excites
- L20:1875 suppresses
- L22:1765 suppresses
- L20:967 excites
- L22:2574 excites
- L22:1233 excites
- L21:912 suppresses
- L11:2091 weakly excites
- L18:951 weakly excites
- L6:1545 weakly excites
- L19:1920 weakly suppresses
- L7:2480 weakly excites
- L19:2060 weakly suppresses
- L20:307 weakly suppresses
- L20:1532 weakly excites
- L10:2153 weakly suppresses
- L16:1926 weakly excites
- L22:2046 weakly suppresses
- L23:2352 weakly suppresses
- L21:998 weakly suppresses
- L4:97 weakly suppresses
- L3:1654 weakly suppresses
- L23:465 weakly excites
- L19:2053 weakly excites
Feeds later layers
- L27:1788 strongly suppresses
- L25:394 suppresses
- L26:600 suppresses
- L25:584 suppresses
- L25:2569 suppresses
- L26:1591 weakly suppresses
- L26:1360 weakly suppresses
- L27:736 weakly suppresses
- L26:2349 weakly suppresses
- L27:420 weakly suppresses
- L27:417 weakly suppresses
- L27:2160 weakly suppresses
- L27:346 weakly suppresses
- L25:1155 weakly suppresses
- L25:841 weakly suppresses
- L26:1720 weakly suppresses
- L25:453 weakly excites
- L26:2696 weakly excites
- L25:1134 weakly suppresses
- L27:2983 weakly suppresses
- L27:2533 weakly suppresses
- L27:418 weakly suppresses
- L27:2774 weakly excites
- L26:382 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
- earrings
- elong
- спор
- aea
- ública
- squad
- فجر
- SND
- نس
- ureka
Suppresses
- page
- page
- 页
- Page
- pages
- Page
- _page
- (page
- →page
- pages