INDEX

Explanations

preventing dissipation

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

 culpa

1.45

 korišten

1.40

 bisogno

1.31

 Going

1.27

 يقول

1.27

 riba

1.26

Pues

1.24

 slecht

1.22

 pergunt

1.22

 возь

1.22

POSITIVE LOGITS

 maximizes

1.34

æld

1.30

 avoid

1.27

 maximized

1.26

 avoids

1.21

 maximize

1.17

 maximise

1.17

 \%)$.

1.16

 homogene

1.14

jabi

1.13

Activations Density 0.473%