INDEX

Explanations

imitating actions or sounds

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 grac

-0.11

 thanking

-0.11

 cryst

-0.11

 Evan

-0.11

 pickups

-0.11

 politely

-0.11

hell

-0.10

elige

-0.10

 pinned

-0.10

 upgraded

-0.10

POSITIVE LOGITS

mim

0.34

 mimic

0.32

 imitation

0.30

Mim

0.26

 mime

0.25

im

0.24

mime

0.21

 imperson

0.20

 pant

0.20

 gestures

0.19

Activations Density 0.153%

No Known Activations