INDEX

Explanations

too or either

The neuron strongly activates on the word immediately preceding major punctuation (i.e. clause‐ or sentence‐final words).

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

‘

-1.39

 justru

-1.38

 tuturor

-1.25

 daž

-1.20

-1.16

-1.15

 població

-1.14

这几

-1.13

 Horm

-1.13

POSITIVE LOGITS

They

1.66

but

1.38

 现场

1.36

lapsingToolbar

1.34

ckles

1.32

"-

1.32

หว่าง

1.32

 einschließlich

1.30

有無

1.29

紗

1.28

Activations Density 0.009%