INDEX

Explanations

fact that, Q &, associated Arena, not really

describing features and concepts

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 художника

0.48

の一つ

0.46

ラの

0.44

 manos

0.44

 cadenas

0.44

рья

0.44

 términos

0.43

ğrafl

0.43

浈

0.43

下の

0.42

POSITIVE LOGITS

 unlike

0.53

 Unlike

0.52

IL

0.47

PED

0.47

 erwart

0.47

Unlike

0.46

iding

0.46

ared

0.45

ORD

0.45

unlike

0.45

Activations Density 0.001%