INDEX

Explanations

numbers followed by ordinals or fractions

The neuron strongly activates on inline mathematical notation and numeric tokens—e.g. Greek letters (ε, τ), dollar‐sign delimiters, digits, and Roman‐numeral identifiers.

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

違います

-1.76

 querem

-1.49

 different

-1.43

歘

-1.38

 különböz

-1.37

 različ

-1.34

となっています

-1.32

 різних

-1.30

 őket

-1.28

-1.27

POSITIVE LOGITS

 "..\..\

1.88

rrrr

1.63

1.62

MAXN

1.52

 Affaires

1.44

byl

1.43

//{

1.43

 "..\..\..\

1.43

 índ

1.43

//}

1.42

Activations Density 0.069%