the word 'entry' in reference text
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. 12 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
This neuron responds to the word 'entry' when it means a dictionary or encyclopedia entry, and to similar phrases like 'post to' or 'search for' that point at a written item. Turning it up made the model drift into repetitive, loosely related 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.
-
the phrase has shifted since then, referring not to patrons and their clients, but 'a collective political preference exercised by a particular interest group'. See also the entry in Wikipedia, which adds that the meaning of Srinivas' expression was first
-
, Third Edition,⏎published by Thomas Nelson, 1897. Public Domain, copy freely.⏎Bibliography InformationEaston, Matthew George. "Entry for Tamarisk". "Easton
-
direct from Amazon (22% off) Amazon page|⏎|Call number||QC7.H62 2005|⏎|External links||This entry has no external links. Add one.|⏎|Through your library||Configure|⏎
-
dose is 15 mg/kg, 5 times daily for 5 days.⏎Best of luck---⏎Search for questions⏎Still looking for answers? Try searching for what you seek or ask your own question.⏎Posted 31 Aug 2
-
use, masterpieces of calligraphic art were produced, e.g., the Irish Book of Kells (8th cent.; Trinity College, Dublin; see under Ceanannus Mór) and the English Lindisfarne Gospels
-
the dictionary. Looking at the entry, it seems that Mr. Fresh was a bit of a lexical trail-blazer
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 other three, but this is the end of the other three... but this is the end of the other three... but this is the end of the other three... but this is the end of the other three... but this is the
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Rare Disorders⏎NINDS progressive multifocal leukoencephalopathy information page. National Institute of Neurological Disorders and Stroke website. Available at http://www.ninds.nih.gov/disorders/pml/pml.htm. Updated February 1
-
24 16:52:29 EST 2010⏎On Wednesday 24 November 2010 22:12:37, Russ Abbott wrote:⏎> Cool. I wasn't
-
network connection. Here are the steps to create an ad hoc network in Windows 7.⏎- Click on Start (Windows icon) and type wireless. Click on Manage wireless networks.⏎- Click on Add to add a network.⏎- Click on Create
-
Select General >> Empty Project and type in the project name.⏎- Under folder Source, right-click Select Add >> New Item.⏎- Under Code, Select C++ File (.cpp), however, when you type in the name use a program name with
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
- L11:2022 excites
- L10:1534 suppresses
- L10:176 weakly excites
- L6:1331 weakly excites
- L10:2521 weakly suppresses
- L11:1073 weakly excites
- L9:2985 weakly suppresses
- L10:2017 weakly excites
- L1:768 weakly suppresses
- L2:2075 weakly excites
- L7:1836 weakly excites
- L2:2027 weakly suppresses
- L7:1263 weakly excites
- L5:2645 weakly suppresses
- L4:382 weakly excites
- L9:2311 weakly suppresses
- L3:362 weakly excites
- L10:2980 weakly suppresses
- L9:602 weakly excites
- L4:982 weakly suppresses
- L6:69 weakly suppresses
- L3:600 weakly excites
- L7:1827 weakly suppresses
- L7:2156 weakly suppresses
Feeds later layers
- L15:2436 suppresses 'out of' in ratios and rankings
- L14:2007 excites
- L16:701 suppresses
- L17:867 excites unclear: punctuation before an explanation
- L16:2299 excites
- L17:2405 weakly excites
- L14:1423 weakly suppresses
- L23:2561 weakly suppresses
- L18:1701 weakly excites
- L14:1204 weakly suppresses
- L16:630 weakly excites
- L17:828 weakly suppresses
- L15:2741 weakly suppresses
- L13:486 weakly excites
- L14:1728 weakly suppresses "of" after a numbered section or title
- L13:2135 weakly suppresses
- L19:2453 weakly excites
- L16:2016 weakly suppresses
- L13:2310 weakly suppresses
- L17:2770 weakly suppresses
- L15:291 weakly suppresses
- L22:2477 weakly excites
- L18:2361 weakly excites
- L13:1405 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
- dedicate
- pee
- pped
- agnostics
- homelessness
- 련
- Propagation
- ##
- reeze
- 一个职业
Suppresses
- iev
- amongst
- IEL
- among
- antal
- among
- agal
- Tale
- iset
- ity