INDEX

Explanations

numbers and hyphens followed by codes

The neuron fires on numeric characters (digits) within document IDs, case numbers, URLs, etc.

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

😂😂😂

-0.86

 🙃

-0.78

 😂😂😂

-0.77

 equity

-0.77

 😢

-0.76

HAHAHA

-0.75

Ikr

-0.75

🤦

-0.75

😂😂😂😂

-0.74

 😆

-0.73

POSITIVE LOGITS

iments

0.79

 Fact

0.79

 commerciales

0.77

arreglo

0.75

(-

0.75

xla

0.73

 неделю

0.72

 lender

0.71

हां

0.71

τε

0.70

Activations Density 0.197%