INDEX

Explanations

proper nouns followed by identifiers

The neuron detects capitalized named‐entity tokens—proper nouns such as people’s names, organizations, locations, products, and other multiword entity components.

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

ícias

-1.45

 огром

-1.40

zensionen

-1.38

₀

-1.38

isn

-1.38

 etwa

-1.38

 either

-1.37

("/:

-1.36

 étant

-1.35

 innymi

-1.34

POSITIVE LOGITS

 хотел

1.45

 iyong

1.42

 with

1.42

 valde

1.36

 lópez

1.35

esist

1.32

 باشید

1.31

 then

1.29

 χρησιμοποι

1.29

 händ

1.26

Activations Density 0.247%