INDEX

Explanations

scientific and technical descriptions

This neuron detects tokens that mark section or list headings—especially standalone titles or labels (often in uppercase or title case) and their trailing punctuation (e.g. periods after “Magic.”, “Disk.”, “Chapter”, 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

our

-1.45

at

-1.43

 this

-1.41

it

-1.37

-1.29

 nuestras

-1.27

Pá

-1.24

Sumber

-1.23

Kenapa

-1.22

 barco

-1.21

POSITIVE LOGITS

>−

1.55

و

1.55

%"),

1.50

ς

1.47

 και

1.46

]}$

1.43

1.41

cribing

1.39

},

1.38

}{

1.37

Activations Density 0.018%