INDEX

Explanations

let someone/something do/go

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

 garantizar

-1.07

artigen

-1.07

 možná

-1.05

foundland

-1.02

 garantire

-1.00

 Klä

-0.98

 mindig

-0.97

 ensuring

-0.97

 alimentare

-0.97

 potrebbero

-0.94

POSITIVE LOGITS

 into

1.80

go

1.66

 know

1.54

be

1.47

 take

1.31

 have

1.30

 become

1.30

 continue

1.27

 slip

1.25

get

1.24

Activations Density 0.031%