INDEX

Explanations

words from various technical fields

The neuron strongly activates on the substring “of” within tokens (e.g. “of”, “offset,” and the “us” piece when “vagus” is split), i.e. it detects any token containing “of.”

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

 because

-1.67

 their

-1.62

 récord

-1.60

 всички

-1.59

 çünkü

-1.43

 boxeo

-1.34

 човек

-1.33

 mismos

-1.32

 cucharadas

-1.31

 gastronomía

-1.31

POSITIVE LOGITS

nell

1.34

完成

1.30

 完成

1.29

ൺ

1.28

 неприят

1.27

 向け

1.27

७

1.25

 урок

1.23

 サロン

1.23

的笑

1.22

Activations Density 0.000%