INDEX

Explanations

first opening, type Snow

can be "strong", "weak"

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

лна

0.46

 purchasing

0.43

čené

0.43

 neke

0.43

"${

0.42

phy

0.42

 Ordin

0.41

ኅ

0.41

 নিঃ

0.41

加工

0.40

POSITIVE LOGITS

ibert

0.48

zag

0.47

āja

0.46

ienes

0.46

 reverber

0.46

agers

0.46

 जूनियर

0.45

abri

0.45

onde

0.45

raum

0.45

Activations Density 0.001%