INDEX

Explanations

announcements and events

The neuron specializes in picking out words that name or announce formal events or exhibits (e.g. “exhibit,” “service,” “workshops,” “awards”) in press‐release–style text.

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

 basically

-1.68

 причем

-1.30

 organized

-1.27

 grayish

-1.24

but

-1.23

 Basically

-1.23

 merely

-1.22

 suele

-1.21

 organizers

-1.21

 probably

-1.21

POSITIVE LOGITS

Whilst

1.33

zension

1.13

égio

1.10

orsing

1.09

 whilst

1.06

 Whilst

1.05

Commenting

1.05

佈

1.04

 rafra

1.04

müş

1.01

Activations Density 0.183%