INDEX

Explanations

case numbers and dates

The main thing this neuron does is detect sequences of digits and related punctuation used as case or file identification numbers (e.g., docket 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

śmy

-0.74

รา

-0.74

tab

-0.73

 Rache

-0.71

resses

-0.69

LOCAL

-0.68

lastic

-0.68

FIT

-0.66

 Behind

-0.65

quinho

-0.65

POSITIVE LOGITS

州

0.75

 minst

0.70

Schedules

0.63

र्शन

0.63

 Waik

0.63

чень

0.63

 كمان

0.62

UIP

0.62

端口

0.62

ंदी

0.62

Activations Density 0.103%