INDEX

Explanations

proton gradient

counting specific numbers

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

Türk

0.59

вича

0.55

üre

0.54

çuk

0.54

encija

0.54

dą

0.53

⥤

0.53

ancien

0.52

༠

0.52

仵

0.52

POSITIVE LOGITS

0.58

 population

0.51

 subgroups

0.50

 wine

0.50

 wines

0.49

ail

0.49

 triad

0.49

↵↵

0.48

 teamwork

0.48

 three

0.48

Activations Density 0.001%