INDEX

Explanations

separating descriptive phrases

The neuron activates on terms that appear in enumerations or section‐heading lists (e.g. items in “Phase 1: Discovery, Phase 2: Exploration…” or “disputes and investigations, corporate finance, strategy and operations”).

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

 العديد

0.74

 الكثير

0.68

 ಅನೇಕ

0.67

zens

0.67

 многих

0.64

 అనేక

0.63

 وغيرها

0.63

许多

0.63

 పలు

0.62

許多

0.60

POSITIVE LOGITS

 which

0.75

which

0.71

 जिसे

0.70

 kterou

0.68

 iaitu

0.68

 которая

0.68

 Which

0.66

 যেটি

0.65

 która

0.65

 která

0.65

Activations Density 0.222%