INDEX

Explanations

final answer is boxed

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

 четы

1.82

 вкус

1.81

اکي

1.72

 отре

1.62

 подня

1.60

 удоволь

1.58

طل

1.56

 بالس

1.55

 ترك

1.54

 ярко

1.52

POSITIVE LOGITS

ijnen

1.83

ichting

1.82

 gebruik

1.82

zijde

1.80

 buiten

1.73

wijs

1.73

waarden

1.72

vaart

1.72

 Nederlandse

1.71

ouw

1.68

Activations Density 0.177%