INDEX

Explanations

one of

This neuron detects cardinal number words (like “one,” “two,” etc.) used as list or item enumerators.

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

 difficult

-1.63

 time

-1.35

 people

-1.34

 chance

-1.34

</h1>

-1.34

and

-1.31

use

-1.31

 chances

-1.30

 schließlich

-1.29

 genuinely

-1.28

POSITIVE LOGITS

 fameux

1.36

 principaux

1.35

deja

1.34

 가장

1.33

 herhangi

1.30

áculo

1.27

tituto

1.27

презента

1.25

cabo

1.24

fifa

1.23

Activations Density 0.034%