INDEX

Explanations

the followed by a noun

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

 länder

-1.33

 that

-1.20

middels

-1.15

が

-1.11

それが

-1.05

 knowing

-1.04

seende

-1.01

divers

-1.01

 것을

-0.99

を与

-0.98

POSITIVE LOGITS

 performs

1.20

poir

1.18

</h1>

1.15

 だっ

1.14

他已经

1.10

 minds

1.09

ところに

1.08

 jeżeli

1.07

meteries

1.06

Activations Density 0.037%