INDEX

Explanations

questions starting with can or does

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

 scolded

1.53

çok

1.42

 berühm

1.40

 пришлось

1.37

 molt

1.35

<unused6>

1.34

বাবুর

1.33

 arbeitete

1.33

첪

1.33

 heartbreaking

1.33

POSITIVE LOGITS

1.34

1.31

1.23

to

1.17

 удовлетво

1.12

 facilitate

1.10

1.05

 satisf

1.04

 создать

1.04

 shall

1.02

Activations Density 0.188%