INDEX

Explanations

grammatical punctuation and structure

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

fir

-0.08

 Chili

-0.08

 legitimately

-0.08

SSIP

-0.08

MJ

-0.08

 unheard

-0.07

Bir

-0.07

 eagerly

-0.07

 recognizable

-0.07

fac

-0.07

POSITIVE LOGITS

 forgiving

0.13

 merc

0.11

 amused

0.11

tl

0.11

lyn

0.11

 silently

0.11

omn

0.10

rug

0.10

 dangling

0.10

anness

0.10

Activations Density 0.032%