INDEX

Explanations

equivalents, usually not, row r

consider areas like

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

 direttore

0.52

 mesaj

0.52

 marcó

0.51

 comté

0.51

 tém

0.50

 confirmó

0.50

 टॉप

0.50

Zá

0.50

annonce

0.50

立て

0.49

POSITIVE LOGITS

 تحریک

0.46

xv

0.44

ikala

0.44

CAL

0.42

skraft

0.42

ma

0.42

xs

0.39

ks

0.39

Activations Density 0.001%