INDEX

Explanations

nuclear envelope, hundreds of, km, colored, Evidence

non-English text or technical notation in various languages and scripts.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

and

-1.23

 first

-1.23

 went

-1.21

 might

-1.20

 which

-1.20

 supremely

-1.16

jaj

-1.16

 this

-1.12

 both

-1.10

 could

-1.06

POSITIVE LOGITS

С

1.30

epy

1.22

THERE

1.20

 tež

1.20

 EVEN

1.19

 Provides

1.18

 povez

1.15

В

1.15

zhan

1.14

chao

1.12

Activations Density 0.000%