INDEX

Explanations

what does

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

存知

-1.05

 Epilepsy

-0.96

 разные

-0.95

 CHINA

-0.95

 بيانات

-0.95

 différents

-0.92

 reci

-0.91

 VIOL

-0.90

 extraordinaire

-0.89

 قوات

-0.89

POSITIVE LOGITS

 what

1.64

how

1.45

 What

1.30

 does

1.22

How

1.12

 which

1.05

can

1.05

 best

1.02

 Does

1.00

low

0.98

Activations Density 0.005%