INDEX

Explanations

willpower and resolve

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

玙

-1.99

阋

-1.89

vaných

-1.87

the

-1.77

 adopting

-1.73

-1.72

一个小

-1.68

aliśmy

-1.68

çou

-1.66

jącego

-1.64

POSITIVE LOGITS

1.87

↵↵↵↵↵↵↵

1.72

↵↵↵↵

1.67

↵↵↵↵↵↵

1.63

↵↵↵↵↵↵↵↵↵↵↵↵↵

1.59

↵↵↵↵↵↵↵↵↵↵

1.58

↵↵↵↵↵↵↵↵↵↵↵

1.56

↵↵↵↵↵↵↵↵↵↵↵↵

1.55

↵↵↵↵↵↵↵↵↵

1.52

1.43

Activations Density 0.008%