INDEX

Explanations

The neuron fires strongly on words denoting formal titles, ranks, positions, or honors (e.g. “title,” “Commander,” “President,” “FIDE,” “honor,” etc.).

appointed to the position of

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

škin

-1.13

嵓

-1.05

それぞれの

-1.00

ジャンル

-0.99

われて

-0.98

 vodou

-0.96

苙

-0.95

袛

-0.94

アプローチ

-0.94

σεων

-0.94

POSITIVE LOGITS

 being

1.49

in

1.18

of

1.18

 from

1.16

 быть

1.07

at

1.05

 having

1.05

+"_

1.01

 president

1.00

as

0.98

Activations Density 0.047%