INDEX

Explanations

at the end of

This neuron responds to occurrences of the word “end” when used to mark the conclusion point of something (e.g. “end of the lending period,” “end of the study,” etc.).

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

 acoged

-1.68

and

-1.65

different

-1.56

ありますか

-1.53

 Erwachsene

-1.52

 ganchillo

-1.50

Different

-1.50

してます

-1.48

一时间

-1.45

 informace

-1.43

POSITIVE LOGITS

of

5.75

 when

1.81

 enge

1.61

taler

1.52

ของ

1.49

 sagt

1.40

 étoile

1.39

banken

1.38

 llamado

1.38

 herv

1.37

Activations Density 0.022%