INDEX

Explanations

check out

The neuron flags promotional calls to action—phrases urging the reader to “check out,” “download,” “watch,” “love,” etc.

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

</h2>

-1.35

 everything

-1.34

 если

-1.29

ertos

-1.27

訁

-1.27

titial

-1.24

ladelphia

-1.19

 $^{\

-1.18

Everything

-1.16

if

-1.15

POSITIVE LOGITS

or

1.77

in

1.60

one

1.45

of

1.37

 several

1.23

或者

1.22

1.21

 дописа

1.20

our

1.20

 this

1.20

Activations Density 0.053%