INDEX

Explanations

university named locations

The neuron detects the token “at” when it appears in institution/campus affiliation phrases (e.g. “University of X at Y”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

それにしても

-1.77

そういえば

-1.73

two

-1.70

うーん

-1.68

ちなみに

-1.66

after

-1.59

 meubles

-1.55

兩種

-1.55

about

-1.54

POSITIVE LOGITS

to

2.31

mafia

1.62

 that

1.58

 yıl

1.57

↵↵

1.49

 whom

1.48

舷

1.45

 though

1.45

ющим

1.43

 yerde

1.42

Activations Density 0.049%