INDEX

Explanations

appears in publications

explaining parts and versions

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 milyen

0.48

 discontinuities

0.47

 Spectroscopy

0.44

 anastom

0.44

辙

0.44

Prothorax

0.43

 stockno

0.43

 উদ্যোগে

0.42

緑

0.42

 Polaribacter

0.42

POSITIVE LOGITS

くだ

0.47

 servant

0.47

de

0.46

 briefcase

0.45

de

0.44

 લાઇ

0.44

simo

0.44

 suff

0.43

 prive

0.43

Activations Density 0.000%