INDEX

Explanations

foreign language names

The neuron is tuned to spot numeric expressions—particularly standalone numbers or measurements (e.g. “45,” “47.75,” “41.25,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

聼

-1.45

 creazione

-1.41

それとも

-1.40

 possibilità

-1.38

 only

-1.35

$.

-1.32

 entire

-1.30

prime

-1.30

 vittoria

-1.30

hline

-1.29

POSITIVE LOGITS

 Mereka

1.43

Fotografía

1.33

 Mulai

1.33

ГЛА

1.32

 that

1.32

 juſ

1.30

 metic

1.29

 outcry

1.28

 Atención

1.27

ዡ

1.26

Activations Density 0.093%