'of' after publisher or division nouns

What this neuron does

This neuron fires on the word 'of' when it follows words like 'reprint', 'imprint' or 'division', as in 'an imprint of Perseus Books'. It was quiet or negative on words that name a relation like 'replicas' or 'counterpart'. Turning it up gave text with no clear link to 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 activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

1. #1 peak

   fact and to convinced ideas of rationality, its scope, and its effects for motion and enterprise.⏎This Elibron Classics publication is a facsimile reprint of a 1901 variation by means of the Macmill<|endoftext|>This question

2. #2 peak

   arian publication is a facsimile reprint of the unique. because of its age, it could comprise imperfections similar to marks

3. #3 peak

   -reactive chemicals used the well-known actinic light activated cross linking or hardening<|endoftext|>Quick Guide to AI and ML Universe for Business Leaders⏎This article was published as a part of the Data Science Blogathon.⏎Ever since the advent of Globalisation

4. #4 peak

   (36 pairs)⏎Box size: 3 x 3 x 4¼ in.⏎This item is published by PomegranateKids®, an imprint of Pomegranate Communications, and is CPSIA compliant.⏎⚠️WARNING:

5. #5 peak

   observatory in a site monitoring mode. The sensor array includes advanced conductivity, turbidity, pressure (depth) and current. Aanderaa, a division of ITT Analytics, has

6. #6 peak

   counter to Charlotte Brontë's “Jane Eyre” — is considered an integral part of the literary canon, but what makes her so great?⏎In celebration of the Royal Society of Literature’s 200th birthday, and the

Steering

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.

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

1. #1 peak

   dynamic automated tutor. This tutor helps students, in a dialogue, to think through arguments in a strategic and systematic fashion.⏎In the TruthLab, the semantic counterpart to the ProofLab, students practice techniques for semantic analysis of formulae and arguments.

2. #2 peak

   between the divine and the human realm.⏎As expected, the goddess was given the name Iris, which is the Greek word for “rainbow.” Her name lives on in English in words like “ir<|endoftext|>Insight from WFPD: Defensive

3. #3 peak

   technologies (PAT) can help digitally clone manufacturing processes⏎There are a plethora of definitions and interpretations of what digital twins are. At their simplest, they are virtual replicas of physical entities that can be used to understand or predict real-world outcomes. These

4. #4 peak

   reveals how “ghost glaciers” have protected Greenland’s ancient bedrock landscapes from the island’s ice sheet.⏎“These ghost glaciers come and go, and leave very little evidence of their presence,” Corbett tells LiveScience. “There are indications that these

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Activation histogram

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 lens

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 layers instead.