INDEX

Explanations

removing parts

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 holes

-0.09

holes

-0.09

 bare

-0.08

 hole

-0.08

 Bare

-0.08

 naked

-0.08

 Serg

-0.08

onth

-0.07

 faces

-0.07

 Faces

-0.07

POSITIVE LOGITS

 cumbersome

0.18

 offending

0.18

 unnecessary

0.17

 barriers

0.17

 protections

0.16

 loos

0.16

pes

0.15

 unwanted

0.15

 redundant

0.15

 conce

0.15

Activations Density 0.110%

No Known Activations