INDEX

Explanations

words from different languages

The neuron detects Cyrillic word stems—especially common Russian technical or domain-specific roots (like “сред(ы)”, “способ(о)”, “микро”, “контро”, “фай(л)”, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

to

-2.16

 from

-1.51

for

-1.41

Only

-1.40

 when

-1.36

 have

-1.34

new

-1.28

 only

-1.27

-1.25

All

-1.25

POSITIVE LOGITS

)_/¯

1.50

 anún

1.43

 Calidad

1.41

olge

1.41

 enseignants

1.39

 krij

1.39

 aranha

1.39

itisation

1.38

 cachorro

1.37

warzys

1.37

Activations Density 0.010%