INDEX

Explanations

multiple items or variety

The main thing this neuron does is detect general quantifier words expressing large or multiple amounts (e.g. “many,” “multiple,” “reasons”).

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

tty

1.12

𝗮

1.07

tt

1.04

)");

1.04

ೃಹ

1.02

tting

1.02

wijs

1.02

 annulus

1.01

 выражение

1.00

ất

0.99

POSITIVE LOGITS

 sclerosis

1.48

 centaines

1.33

 variously

1.27

 milliers

1.23

 kinds

1.23

 MANY

1.23

 dozens

1.19

 hundert

1.16

 verschiedene

1.14

 forskellige

1.10

Activations Density 0.866%