INDEX

Explanations

case reports and documents

The neuron strongly activates on standalone title or section‐heading words (often in uppercase or as document labels like “Guide,” “Release Notes,” “Communiqué,” “Case Report,” “Update,” 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

-1.79

If

-1.59

and

-1.59

The

-1.46

-1.44

-1.39

 kupić

-1.35

or

-1.34

 obicei

-1.34

an

-1.33

POSITIVE LOGITS

 kän

1.64

 作り方

1.63

 kalori

1.60

 ekos

1.60

 maksi

1.55

骉

1.55

 katk

1.53

Spesifikasi

1.50

 chè

1.50

Makanan

1.46

Activations Density 0.056%