INDEX

Explanations

Miss followed by a name

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 milj

-1.34

 kazak

-1.32

 geene

-1.30

 eksemp

-1.27

 russa

-1.26

 lorsque

-1.24

 عندما

-1.22

 hvil

-1.22

 helg

-1.21

 egent

-1.20

POSITIVE LOGITS

1.30

↵↵↵↵

1.23

toThrow

1.19

↵↵↵↵↵↵

1.18

1.17

</td>

1.16

裴

1.15

（

1.15

 that

1.12

andReturn

1.12

Activations Density 0.080%