INDEX

Explanations

how/however + magnitude/degree

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

❉

-1.63

除此之外

-1.45

 maji

-1.44

曏

-1.44

 trimestre

-1.42

 विश्वसनीयता

-1.41

in

-1.35

 использова

-1.35

yyb

-1.34

 Lumpur

-1.34

POSITIVE LOGITS

or

2.41

its

1.88

 certaines

1.69

 czasem

1.66

 this

1.59

 wcześniej

1.54

 esperamos

1.54

 berücksich

1.51

 bouw

1.51

we

1.50

Activations Density 0.007%