INDEX

Explanations

codes and identifiers

how the structure itself

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 pleural

0.48

če

0.47

 anunciou

0.46

 pengendalian

0.45

रेन

0.44

 adhes

0.43

 glamorous

0.43

﹚

0.43

 añadi

0.43

ብስብ

0.42

POSITIVE LOGITS

KO

0.42

稿

0.41

Split

0.41

蚕

0.41

0.40

۶

0.40

ft

0.40

 Providence

0.40

olique

0.40

 Chur

0.40

Activations Density 0.001%