INDEX

Explanations

place names with common endings

The neuron fires on tokens containing non‐ASCII or accented characters—i.e. foreign proper names and place names with diacritics.

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

etsk

-1.43

-1.42

 возможно

-1.42

 konnte

-1.34

נון

-1.24

⠁

-1.22

 einzige

-1.22

lij

-1.20

勝利

-1.20

mühle

-1.20

POSITIVE LOGITS

1.82

new

1.59

and

1.51

1.48

 MessageBoxIcon

1.45

...”

1.44

now

1.42

out

1.41

ur

1.36

out

1.34

Activations Density 0.761%