INDEX

Explanations

package names and code elements

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

を選択します

-1.42

 verhind

-1.32

Več

-1.30

 nimmer

-1.30

を選ぶ

-1.30

 inond

-1.27

 investir

-1.23

を選び

-1.22

 kapture

-1.20

 erhi

-1.17

POSITIVE LOGITS

 with

1.27

 over

1.23

for

1.21

 while

1.16

or

1.09

 other

1.08

all

1.06

has

1.05

not

1.05

if

1.04

Activations Density 0.002%