INDEX

Explanations

community

The neuron reliably fires on the word “community” (and closely related locality labels) in geographic or administrative descriptions.

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

 поча

-1.36

 Milliarden

-1.34

who

-1.32

 związ

-1.26

 wzgl

-1.24

Through

-1.24

 time

-1.23

鑾

-1.23

 детям

-1.21

各种

-1.20

POSITIVE LOGITS

alians

1.28

 ruines

1.24

 tão

1.24

﹣

1.23

футболка

1.23

ここは

1.23

⁻

1.22

--.

1.20

↵↵

1.17

hlten

1.17

Activations Density 0.075%