INDEX

Explanations

fields of study and concepts

The neuron fires on uncommon, domain-specific multisyllabic terms—particularly scientific or technical jargon built from Greek/Latin roots.

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

 then

-2.03

 first

-1.85

 start

-1.75

of

-1.66

an

-1.63

 with

-1.60

 Even

-1.57

so

-1.56

 this

-1.51

its

-1.50

POSITIVE LOGITS

 accompagner

1.70

izedBox

1.56

…»

1.53



1.48

 jahre

1.39

igheid

1.38

 sosis

1.37

 Vitamina

1.37

ihnachts

1.35

 vivre

1.34

Activations Density 0.138%