INDEX

Explanations

weekly or monthly series

The neuron fires on words that are part of names or headings of series, programs, or article collections (e.g. “series,” “program,” “articles,” “focused on,” “of the Week,” 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

 when

-1.32

any

-1.28

 most

-1.21

 just

-1.17

 after

-1.16

if

-1.12

can

-1.10

an

-1.09

 before

-1.08

 both

-1.07

POSITIVE LOGITS

 each

1.81

 series

1.49

Each

1.34

陆续

1.33

 každý

1.32

 succession

1.28

 seront

1.27

each

1.27

 jeweils

1.27

 mỗi

1.20

Activations Density 0.084%