INDEX

Explanations

describing sizes and durations

names and abstract concepts

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 kaufen

0.46

 commentary

0.45

 advocated

0.44

 Rechts

0.44

eup

0.44

 equating

0.43

åt

0.43

 Verständ

0.43

 reiter

0.43

Schneider

0.43

POSITIVE LOGITS

 chiếc

0.54

рит

0.53

VICIO

0.47

 tình

0.45

 francisco

0.44

нг

0.43

 phong

0.43

 ярко

0.41

ريقه

0.40

 гаран

0.40

Activations Density 0.001%