INDEX

Explanations

places described by "is", "has", "lies", "was", "sits"

The neuron activates strongly on tokens naming types of places or facilities (e.g. “Field,” “Prison,” “Theatre,” “Airport,” “Hall,” “Place,” “School”).

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

 بودند

-1.20

 должны

-1.18

 môžu

-1.02

 szükség

-1.00

 weten

-1.00

 alapján

-1.00

 szüks

-0.98

 móds

-0.98

står

-0.97

 byen

-0.95

POSITIVE LOGITS

is

2.56

 lies

2.30

was

2.27

 sits

1.82

lie

1.80

 nằm

1.50

sit

1.47

has

1.47

had

1.44

 lying

1.42

Activations Density 0.149%