INDEX

Explanations

the city of Paris

This attention head attends to the token "Paris" from tokens related to Paris.

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

Negative Logits

ChildScrollView

-0.66

usten

-0.63

DebuggerNonUser

-0.61

Myr

-0.60

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

-0.60

imag

-0.60

<?>>

-0.60

 demas

-0.60

Datuak

-0.59

 arro

-0.59

POSITIVE LOGITS

 Paris

1.25

Paris

1.16

paris

1.06

 PARIS

1.05

 paris

1.04

kegaard

0.96

 Parisi

0.95

 quilt

0.88

 unknownFields

0.87

 quilts

0.84

Activations Density 0.002%