INDEX

Explanations

reveal secret information

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

imos

-0.10

-0.09

 Quincy

-0.09

ï»¿↵

-0.08

 Marvin

-0.08

pector

-0.08

rop

-0.08

 Wald

-0.08

ior

-0.08

yles

-0.08

POSITIVE LOGITS

 revealing

0.49

 reveal

0.45

 betray

0.42

 reveals

0.39

 Reve

0.38

reve

0.37

 divul

0.36

 revealed

0.34

 disclosing

0.34

 giveaway

0.34

Activations Density 0.194%