INDEX

Explanations

approach to

The neuron fires strongly on key abstract nouns that serve as the objects of “of” or “to” in descriptive noun phrases (e.g. the noun in “approach to X” or “X of Y”).

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

确实是

-1.40

or

-1.36

可不是

-1.34

 jugando

-1.32

 Cualquier

-1.30

 verschillen

-1.28

 ИЗ

-1.28

 hubo

-1.28

 viendo

-1.27

已经有

-1.26

POSITIVE LOGITS

how

1.35

 what

1.33

 bantu

1.26

 every

1.23

 makna

1.23

呖

1.20

臯

1.20

 each

1.18

 morena

1.18

紲

1.17

Activations Density 0.042%