INDEX

Explanations

proper nouns and brand names

The neuron is a “name‐spotter” that lights up on the distinctive suffix or latter segment of proper‐noun names—publishers, brands, programs, places, and other named entities.

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

ườ

-1.61

 akces

-1.57

tomatik

-1.43

 vermek

-1.41

schöne

-1.39

leid

-1.34

lē

-1.33

 sposo

-1.32

 toreb

-1.31

Would

-1.30

POSITIVE LOGITS

you

1.90

at

1.72

“

1.70

 that

1.68

on

1.63

),

1.52

MessageOf

1.51

 either

1.49

 these

1.48

1.43

Activations Density 0.118%