INDEX

Explanations

at least, it's better to

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 inhomogeneities

1.67

integrity

1.60

وصل

1.57

 dishonest

1.53

 inhomogeneity

1.50

Insufficient

1.49

<unused758>

1.47

<unused302>

1.44

sufficient

1.40

ၺ

1.38

POSITIVE LOGITS

스로

1.55

 посвящен

1.32

Suk

1.29

なった

1.28

osci

1.26

チェ

1.24

Cy

1.24

 powered

1.23

ské

1.22

 laquelle

1.20

Activations Density 0.123%