INDEX

Explanations

Indonesian, Spanish, or Hindi words

The neuron strongly activates on transliterated foreign or loanwords (e.g. non-English terms like “intifada,” “madrasahs,” “taiga,” “oblast,” “warungs,” etc.).

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

↑↑↑</

-1.51

 povezave

-1.47

这里

-1.44

Gön

-1.40

ّه

-1.39

-1.34

 nhưng

-1.34

 alebo

-1.33

at

-1.31

 zealand

-1.31

POSITIVE LOGITS

 recogida

1.52

 into

1.45

 Selecciona

1.44

 delantera

1.38

 Otras

1.38

牖

1.37

 Aprove

1.35

 rejoint

1.34

BeginContext

1.33

 dentro

1.30

Activations Density 0.143%