INDEX

Explanations

promise to release

sometimes good, sometimes bad

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 تلوار

0.63

砦

0.49

是你

0.47

اردوش

0.47

 rozpozn

0.47

 شە

0.45

۲

0.42

 espada

0.42

neutron

0.42

 کہاں

0.42

POSITIVE LOGITS

 with

0.54

it

0.50

to

0.50

in

0.46

get

0.46

 require

0.45

 British

0.45

 Standard

0.43

 usually

0.43

 needs

0.43

Activations Density 0.001%