INDEX

Explanations

type, category, or genre

This neuron flags document tokens that denote product‐spec metadata or quantitative specs—e.g. field labels (“Item type,” “Brand name,” “Pattern printed,” “Package weight,” “overnight delivery”) and unit‐of‐measure expressions (mg, mcg, cm, lb).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ﻈ

0.67

aría

0.64

小时

0.64

Mans

0.64

Marble

0.63

盟

0.63

érer

0.62

मिल

0.62

orah

0.62

Ві

0.62

POSITIVE LOGITS

 category

0.80

 genre

0.78

 type

0.76

 kategorie

0.71

 kategori

0.70

 shank

0.69

 categorie

0.69

 typology

0.69

 Kategorie

0.68

 genus

0.68

Activations Density 0.003%