INDEX

Explanations

contest

words preceding specific follow-ups

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 City

0.51

Cen

0.46

 Stadt

0.45

یب

0.45

 Lily

0.44

Lit

0.44

 [...]

0.44

 Light

0.43

 Kinder

0.43

شوند

0.43

POSITIVE LOGITS

氯

0.56

ifference

0.48

 abnormalities

0.47

 хороший

0.46

 dioxide

0.46

 biopsies

0.46

 creatinine

0.46

ioxid

0.46

 absences

0.46

oh

0.45

Activations Density 0.002%