INDEX

Explanations

numbers or specific details that follow certain words

The neuron is highly activated by common high‐frequency function words (articles, prepositions, simple numerals) rather than content words.

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

了一個

-1.32

ండి

-1.31

ントン

-1.29

了个

-1.28

 hicimos

-1.27



-1.27

 destiné

-1.25

一阵

-1.23

娴

-1.23

 अन्य

-1.21

POSITIVE LOGITS

 which

1.93

are

1.91

 these

1.53

 доступ

1.43

 plastica

1.34

 maglia

1.34

 potranno

1.34

 doccia

1.33

 will

1.29

Zitat

1.27

Activations Density 1.663%