INDEX

Explanations

gentle, safe, careful actions

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

dam

-0.12

 Buddy

-0.10

ovel

-0.10

_pb

-0.10

 Ernest

-0.10

aad

-0.10

dy

-0.10

GRA

-0.10

 Guth

-0.10

Mam

-0.10

POSITIVE LOGITS

 gently

0.16

 gentle

0.13

 safe

0.13

 safely

0.12

 care

0.12

 Safe

0.11

 carefully

0.11

 Gentle

0.10

 softly

0.10

-safe

0.10

Activations Density 0.021%