INDEX

Explanations

press release identifiers

This neuron detects the slash characters used in press‐release dateline headers (e.g. the “/” separators and accompanying numeric codes in PRNewswire-style metadata).

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

 vack

-0.86

 σύν

-0.81

 болезни

-0.81

()?;

-0.79

 krä

-0.79

 kalla

-0.78

)।

-0.76

 Mala

-0.76

 pada

-0.75

úcar

-0.75

POSITIVE LOGITS

PRNewswire

1.03

єра

0.82

{/

0.80

CCESS

0.80

 awning

0.77

mə

0.76

ників

0.75

ENSUS

0.74

おります

0.74

yentes

0.73

Activations Density 0.011%