INDEX

Explanations

prejudice and othering

The neuron fires on words that denote labeling or dehumanizing of entire groups (e.g. categorize, dehumanize, “less-than,” criminals, misogynistic).

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

亵

-1.15

なら

-1.14

 канала

-1.11

いれば

-1.10

 kanske

-1.09

严肃

-1.04

ßung

-0.99

ylla

-0.98

とお

-0.97

 wollte

-0.97

POSITIVE LOGITS

 their

1.38

 they

1.23

elegante

1.22

nosotros

1.20

 capacidades

1.20

karakter

1.19

kawaii

1.16

떳

1.13

cribes

1.13

tapete

1.12

Activations Density 0.025%