INDEX

Explanations

compound terms and phrases

This neuron detects specific product or model names—proper nouns often including brand identifiers and model numbers.

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

⽼

-1.40

of

-1.30

 commenti

-1.30

 estudante

-1.24

 chefe

-1.23

savevideo

-1.23

未必

-1.22

 następnie

-1.16

 Kommentare

-1.16

 fragt

-1.14

POSITIVE LOGITS

 betrek

1.34

 särsk

1.33

 костюм

1.27

печат

1.25

一丝

1.23

诸多

1.22

вобо

1.20

قرأ

1.20

=)

1.20

 which

1.18

Activations Density 0.043%