INDEX

Explanations

expressing confusion

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

 confounded

-0.68

原始内容存档于

-0.59

windowFixed

-0.58

 HasFactory

-0.57

 baffled

-0.57

Referanser

-0.56

 chofe

-0.55

Showing

-0.55

 surprised

-0.55

 doubtful

-0.54

POSITIVE LOGITS

kloped

0.67

AndEndTag

0.67

 where

0.60

̍t

0.59

why

0.54

 utafitiHapana

0.54

 which

0.52

WireFormat

0.50

THON

0.48

DllImport

0.48

Activations Density 0.008%