INDEX

Explanations

here

This neuron primarily detects the deictic location term “here.”

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

 Näch

-1.60

になるので

-1.43

 تفاوت

-1.38

of

-1.34

 zawsze

-1.33

衚

-1.30

が必要です

-1.27

總是

-1.27

可能會

-1.27

 italie

-1.25

POSITIVE LOGITS

are

1.62

了一会儿

1.59

is

1.58

will

1.42

before

1.41

they

1.38

„

1.37

 карт

1.33

because

1.33

1.30

Activations Density 0.018%