INDEX

Explanations

directions or orientation-related tokens, specifically recognizing mentions of "left" and "right."

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

peicher

-0.81

Autoritní

-0.79

":[{

-0.68

 Tabor

-0.67

LookAnd

-0.67

WireFormatLite

-0.67

illaries

-0.66

그러나

-0.65

 Bradbury

-0.65

 Diener

-0.65

POSITIVE LOGITS

right

3.31

 right

2.92

Right

2.73

 Right

2.69

RIGHT

2.58

 RIGHT

2.52

 righ

2.08

left

2.02

右

1.87

 left

1.86

Activations Density 2.643%