INDEX

Explanations

in conference proceedings

numbers in different languages

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 amendments

0.48

 pleasure

0.46

 therapeutic

0.44

意的

0.43

 yesterday

0.41

 gardener

0.41

 subscriber

0.41

 malpractice

0.41

 diarrhoea

0.40

oplus

0.40

POSITIVE LOGITS

 Бі

0.57

 Один

0.54

 Energía

0.52

 линий

0.50

 שני

0.50

 Леон

0.50

 успі

0.50

 Inicial

0.50

 Fácil

0.50

 ഇരു

0.49

Activations Density 0.004%