INDEX

Explanations

code structure

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

 великолеп

-1.40

Pře

-1.30

Different

-1.24

 pengaturan

-1.21

 искрен

-1.16

Numerous

-1.16

 Hinsicht

-1.13

 Different

-1.13

 двумя

-1.12

Obsah

-1.12

POSITIVE LOGITS

)}$.

1.25

ことも

1.18

ことを

1.18

 даже

1.13

也

1.13

item

1.10

}));

1.10

，「

1.09

INSTRUCTIONS

1.08

珐

1.08

Activations Density 0.081%