INDEX

Explanations

can or could followed by action

The neuron activates on numeric tokens, especially integers and decimal numbers.

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

-1.50

__

-1.39

 another

-1.30

 другую

-1.24

化した

-1.24

and

-1.23

 would

-1.23

=}

-1.23

 andere

-1.21

 maar

-1.21

POSITIVE LOGITS

Also

1.67

 también

1.63

 also

1.56

 همچنین

1.55

 также

1.52

Ahol

1.48

 їх

1.48

 Pristup

1.45

这些

1.43

 myös

1.42

Activations Density 0.045%