INDEX

Explanations

Arab, Mecca, Arabian, Arabic

This neuron fires on Arabic‐derived terms and names (e.g. the “al-” prefix, Arabic place names, and words like “Arab” or “Arabic”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

澌

-1.85

鹇

-1.80

婠

-1.52

ılığı

-1.51

-1.50

criminate

-1.48

鱘

-1.43

for

-1.42

謦

-1.40

hibits

-1.36

POSITIVE LOGITS

嵛

1.66

 facilidad

1.56

↵

1.48

 потому

1.43

鴣

1.39

 protože

1.33

 kteří

1.30

 kvůli

1.29

dateTimePicker

1.29

емного

1.28

Activations Density 0.030%