INDEX

Explanations

diagnose OS issues

This neuron flags unusually rare or specific tokens—especially long, specialized terms, proper names, or URL‐style strings.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 locus

0.75

 nest

0.71

 opción

0.70

 Từ

0.69

 option

0.68

న్

0.66

 twee

0.66

的主

0.66

 centrifuge

0.66

 urbanization

0.66

POSITIVE LOGITS

}))

0.82

tober

0.79

㔹

0.79

بے

0.77

워

0.77

봐

0.77

})}

0.77

=\{

0.76

ppi

0.73

blox

0.73

Activations Density 0.000%