INDEX

Explanations

new to or relationship to

This neuron lights up on content‐bearing words (open‐class tokens like nouns, adjectives, and adverbs) and stays off for function 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

 znaleźć

-1.30

-1.29

を行いました

-1.29

 zostać

-1.23

肩膀

-1.23

 compleja

-1.21

 acoged

-1.20

在我的

-1.20

廁

-1.19

 restauración

-1.18

POSITIVE LOGITS

of

3.95

to

1.91

 when

1.48

Então

1.44

at

1.44

for

1.41

möglichen

1.37

gül

1.30

when

1.24

спасибо

1.22

Activations Density 0.109%