INDEX

Explanations

locations and adjacent street descriptions

The neuron activates on words and phrases that denote physical locations, landmarks, or spatial orientations (e.g. “seafront,” “promenade,” “heart of town,” “harbour,” “downtown”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

Lorem

0.91

Regional

0.86

CONFLICT

0.78

❖

0.77

REGISTER

0.76

TRI

0.76

UserModel

0.75

UserID

0.74

Тех

0.73

Haryana

0.73

POSITIVE LOGITS

 restaurant

1.22

 hotel

1.20

 restaurants

1.15

but

1.12

 street

1.10

 pero

1.07

(?)

1.07

 tourist

1.05

 piazza

1.03

 supermarket

1.02

Activations Density 0.452%