INDEX

Explanations

communication

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 merry

-0.10

EK

-0.09

 recounted

-0.09

 scenery

-0.09

abal

-0.09

agini

-0.09

($)

-0.08

 Kale

-0.08

 berg

-0.08

jb

-0.08

POSITIVE LOGITS

 communication

0.13

 Communic

0.13

Communication

0.12

.Comm

0.12

 communicating

0.12

communication

0.11

 misunderstand

0.11

communic

0.11

 effectively

0.11

 Message

0.11

Activations Density 0.141%

No Known Activations