INDEX

Explanations

source and origin

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

otas

-0.09

awa

-0.09

rze

-0.09

.zoom

-0.08

 sext

-0.08

owy

-0.08

-alert

-0.08

atti

-0.08

 Buffer

-0.08

/configuration

-0.08

POSITIVE LOGITS

 source

0.69

 sources

0.69

 origin

0.62

 Sources

0.56

 origins

0.56

æĿ¥æºĲ

0.55

source

0.53

sources

0.51

 Source

0.51

-source

0.49

Activations Density 2.400%

No Known Activations