INDEX

Explanations

hold positions or continue states

holding

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

hedron

0.43

ás

0.39

 hängt

0.39

 μέσω

0.39

 dịp

0.39

 piled

0.38

ктор

0.38

 பயணம்

0.37

之后

0.37

之旅

0.37

POSITIVE LOGITS

 sway

1.16

 aloft

1.04

 onto

1.01

 accountable

0.95

 hostage

0.91

 grud

0.79

 captive

0.78

onto

0.77

 hold

0.76

 holds

0.74

Activations Density 0.070%