INDEX

Explanations

Identifier, subdivision, Blue, Prime, id, email

The neuron detects tokens that serve as structured field or parameter names in markup—e.g. XML/infobox keys like “phylum,” “class,” “species,” “binomial,” “CFBundleIdentifier,” “Color Space,” “Module,” “Date,” 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

 have

-1.23

all

-1.21

 their

-1.18

 everything

-1.13

 Automó

-1.13

время

-1.10

 bereits

-1.10

第一天

-1.08

 nothing

-1.07

 Hitze

-1.07

POSITIVE LOGITS

</strong>

2.02

</b>

1.32

</i>

1.31

</em>

1.29

 котором

1.26

Allgemeine

1.22

 بيها

1.20

 ailing

1.19

 mortos

1.19

rfloor

1.16

Activations Density 0.003%