INDEX

Explanations

while introducing contrasts

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

 spricht

1.61

*)$

1.55

랭

1.52

 forbid

1.51

 classification

1.51

 berücksicht

1.51

 recirc

1.50

 cellpadding

1.50

排水

1.46

 boats

1.46

POSITIVE LOGITS

 நிறைவே

2.03

executor

1.77

müşt

1.76

তিপ

1.74

ঙ্খ

1.73

fak

1.73

ഷ്ട

1.70

ierungs

1.68

Fet

1.68

 साकार

1.64

Activations Density 0.214%