INDEX

Explanations

as

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 izdel

-0.56

 preuve

-0.53

-0.46

 enkelte

-0.45

 chiaramente

-0.45

 obviamente

-0.45

obviously

-0.43

 igång

-0.43

 simplemente

-0.42

POSITIVE LOGITS

UnusedPrivate

0.86

InjectAttribute

0.82

WarningLevel

0.77

 <=",

0.77

 Roskov

0.73

➟

0.71

EndProject

0.71

 createState

0.68

<bos>

0.67

tvguidetime

0.67

Activations Density 0.002%