INDEX

Explanations

the

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

 accurate

-0.76

 accurately

-0.71

accurate

-0.69

Accurate

-0.69

 Accurate

-0.66

 correct

-0.59

 correctly

-0.57

 precise

-0.56

 inaccurate

-0.51

正確

-0.51

POSITIVE LOGITS

AnimationsModule

1.07



0.95

 ComVisible

0.85

AndEndTag

0.84

TagMode

0.79

findpost

0.78

IndentedString

0.76

出版年

0.74

躇

0.71

 GenerationType

0.70

Activations Density 0.008%