INDEX

Explanations

Sharpness/ultrasharp

The neuron strongly activates on capitalized product‐model names or alphanumeric model identifiers (e.g. PowerEdge, OptiPlex, UltraSharp, Cobas) rather than ordinary words.

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

 اگر

-1.44

 може

-1.43

quierda

-1.40

 when

-1.32

（※

-1.28

źni

-1.27

 filmes

-1.26

can

-1.25

rådet

-1.25

 with

-1.23

POSITIVE LOGITS

to

1.75

ということ

1.42

さまざま

1.33

 kiezen

1.27

 about

1.26

様々

1.25

 itinerario

1.20

Choose

1.13

 April

1.13

𐬹

1.13

Activations Density 0.152%