INDEX

Explanations

pale and light colors

The neuron fires on adjectives that modify or intensify colors or degrees—e.g. “dark,” “light,” “pale,” “extreme,” etc.

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

を

-1.69

がありますが

-1.55

 caricatura

-1.52

が

-1.50

骉

-1.39

mainly

-1.34

 delightfully

-1.33

⾞

-1.33

 และ

-1.32

 extravag

-1.32

POSITIVE LOGITS

 what

1.75

 when

1.73

if

1.66

 there

1.57

or

1.55

 something

1.55

There

1.55

 while

1.53

 because

1.52

 During

1.46

Activations Density 0.036%