INDEX

Explanations

process

The neuron fires on occurrences of the word “process” (in any case or compound form) and related command names involving “process.”

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

atação

-1.55

ték

-1.55

ktas

-1.53

</b>

-1.48

са

-1.47

 rheinland

-1.45

鲞

-1.43

 directora

-1.38

zación

-1.38

垵

-1.38

POSITIVE LOGITS

着脸

1.66

xvii

1.63

ならば

1.59

 then

1.59

 obyvateľov

1.56

});

1.55

 Then

1.53

Sx

1.50

 empfangen

1.49

Activations Density 0.045%