INDEX

Explanations

black followed by demographic groups

The neuron activates on capitalized racial/ethnic descriptors—particularly the word “Black” (and related terms like “African-American”) when referring to people’s race.

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

༷

-1.77

之为

-1.68

 möchten

-1.53

 camiset

-1.52

ING

-1.51

还未

-1.49

 bels

-1.46

 sumber

-1.45

 that

-1.45

 doble

-1.45

POSITIVE LOGITS

and

1.78

 فقط

1.62

 хора

1.46

zelfde

1.45

」、「

1.45

 tärke

1.43

 pantaloni

1.43

 illustrationer

1.41

 cire

1.37

umbersome

1.36

Activations Density 0.026%