INDEX

Explanations

prime factors, gel electrophoresis, nationwide package

The neuron fires on the first high‐level content word (or leading marker) at the start of a new sentence or comment line—i.e. the key “headline” token opening each statement.

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

infeld

-0.73

 estuary

-0.73

 dimensionality

-0.73

 Sectors

-0.72

 Karls

-0.71

ømme

-0.70

顿

-0.69

ഋ

-0.69

フリー

-0.69

 aeration

-0.68

POSITIVE LOGITS

polazione

0.80

 fuel

0.80

 Fuel

0.78

 ограни

0.75

limiting

0.74

 necesit

0.73

nyttet

0.73

odm

0.72

vea

0.72

spection

0.72

Activations Density 0.003%