INDEX

Explanations

when followed by a word

The neuron strongly activates on very common, high-frequency function words (e.g. “the,” “we,” “have,” “a”) and related punctuation or citation markers.

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

 that

-2.33

 this

-2.03

one

-1.60

 your

-1.39

of

-1.31

</em>

-1.23

ümüz

-1.20

 ďal

-1.20

 mūsų

-1.18

 innumerable

-1.17

POSITIVE LOGITS

They

1.35

 rolex

1.27

vités

1.18

Which

1.16

 they

1.13

 принтом

1.13

猢

1.12

 لهم

1.11

oudoune

1.11

 denen

1.10

Activations Density 0.017%