INDEX

Explanations

dark colors and materials

The neuron detects color-describing adjectives (especially darker hues like “dark,” “brown,” “black,” etc.) in the 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

 ,‬

-1.61

ाइल

-1.59

``.

-1.53

bited

-1.48

敉

-1.45

strado

-1.43

スリ

-1.40

 Potato

-1.39

asına

-1.39

ρας

-1.38

POSITIVE LOGITS

and

1.54

 sechs

1.31

 แต่

1.30

六年

1.30

 один

1.29

↵

1.29

 більше

1.23

 Certain

1.22

 texting

1.21

 drewna

1.21

Activations Density 0.051%