INDEX

Explanations

that Congress

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

 seems

-0.10

 apparently

-0.09

 seemingly

-0.09

 perhaps

-0.09

 probably

-0.09

 Bravo

-0.08

Wow

-0.08

Bac

-0.08

 seem

-0.08

POSITIVE LOGITS

 misunderstood

0.12

 “âĢ¦

0.12

 "...

0.12

“

0.11

because

0.11

 blah

0.10

 Constit

0.10

uest

0.10

 wegen

0.10

Activations Density 0.091%