INDEX

Explanations

reducing score deficits

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Reina

-0.93

 séjours

-0.82

fondos

-0.81

蜱

-0.81

 mengembangkan

-0.80

を超

-0.80

stützung

-0.80

bahaya

-0.80

 conflictos

-0.80

věd

-0.79

POSITIVE LOGITS

 reduce

1.77

 reduction

1.66

 reducing

1.60

 reduced

1.52

 reduces

1.48

 narrowed

1.41

 narrowing

1.40

 Reduce

1.39

Reduce

1.30

 сокра

1.30

Activations Density 0.010%