INDEX

Explanations

everything you need

The neuron fires on promotional “all-inclusive” terms—words like “everything,” “features,” “amenities,” “you could want,” etc., that advertise a complete set of offerings.

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

 različ

-1.27

 różnych

-1.27

 разных

-1.13

 różne

-1.10

 برخی

-1.07

헹

-1.04

 olika

-1.02

 différents

-1.00

 некоторых

-1.00

 différentes

-0.99

POSITIVE LOGITS

you

2.41

 imaginable

2.22

 needed

1.90

any

1.84

 necessary

1.82

 need

1.61

 anyone

1.54

1.45

for

1.44

 what

1.43

Activations Density 0.042%