INDEX

Explanations

difficulty or learning curve

The neuron detects promotional or marketing language—positive, attention-grabbing adjectives and venue/product terms used in ad-style copy.

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

 Yale

0.80

 coales

0.79

 Gasoline

0.75

깡

0.75

 caffe

0.75

 Chipotle

0.75

ud

0.73

 Benz

0.73

 shunt

0.72

Yale

0.72

POSITIVE LOGITS

ور

0.79

🚻

0.77

 শ্রীযুক্ত

0.73

दक्षिण

0.71

 awaits

0.70

 stärker

0.70

 Annotated

0.68

ఏ

0.68

می

0.68

مه

0.67

Activations Density 0.001%