INDEX

Explanations

imperceptible / критически / would

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

whe

0.78

 andre

0.75

[*

0.75

int

0.74

assing

0.73

 irrelevant

0.71

ail

0.71

aqu

0.70

 ments

0.69

ractions

0.67

POSITIVE LOGITS

ικής

1.28

ских

1.22

ской

1.18

ային

1.15

ных

1.15

적인

1.12

κής

1.11

ковых

1.08

ческих

1.08

들에

1.08

Activations Density 0.114%