INDEX

Explanations

brand names and product models

This neuron detects technical product or equipment names, especially branded model identifiers (brand + 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

成績

-0.90

 kostenlosen

-0.81

职工

-0.78

韓

-0.78

ふるさと

-0.76

滸

-0.75

ข่าว

-0.74

基金

-0.73

愛知

-0.73

 adoles

-0.72

POSITIVE LOGITS

凉

0.83

 rekord

0.80

存于互联网档案馆

0.79

 swallowed

0.79

BigDecimal

0.77

Aihe

0.77

 Meister

0.75

$<$

0.75

¥

0.74

 Lavender

0.74

Activations Density 0.040%