INDEX

Explanations

templating filters

The neuron is detecting code‐style tokens—particularly templating or variable‐access expressions marked by dots, underscores, and pipe/filter syntax.

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

for

-1.84

 therefore

-1.58

of

-1.51

 where

-1.50

with

-1.41

 while

-1.39

 without

-1.38

 would

-1.38

 nevertheless

-1.34

POSITIVE LOGITS

inerario

1.76

 desnuda

1.66

Perché

1.56

Ꮏ

1.54

 शुर

1.50

zzet

1.46

蜮

1.45

ähteet

1.45

ԥ

1.45

珦

1.44

Activations Density 0.041%