INDEX

Explanations

divided into

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

 probably

-0.90

 Through

-0.88

ForRow

-0.86

 Towards

-0.86

 posibilidades

-0.84

 directly

-0.81

 notas

-0.81

kelijke

-0.80

 ruft

-0.79

Through

-0.79

POSITIVE LOGITS

 into

1.48

in

0.96

 dialec

0.88

俩

0.87

 between

0.86

 according

0.86

 ثلاثة

0.86

Kurs

0.81

 HOLY

0.80

ifiques

0.78

Activations Density 0.034%