INDEX

Explanations

for

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

 betweenstory

-1.18

клопе

-0.98

 مشين

-0.97

StructEnd

-0.93

 صوتيه

-0.93

migrationBuilder

-0.92

__':

-0.90

 Meksiku

-0.89

AddTagHelper

-0.87

contentLoaded

-0.87

POSITIVE LOGITS

0.73

an

0.55

new

0.48

 surviving

0.41

 extra

0.41

 those

0.40

 giving

0.40

 nouvel

0.40

NEW

0.39

 receiving

0.39

Activations Density 0.003%