INDEX

Explanations

year markers

The neuron selectively activates on numeric tokens—especially four-digit year numbers, large figures, and other numerals (e.g., dates, statistics, prices).

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

 इलाकों

0.64

 departments

0.54

 नेहमी

0.54

枌

0.54

 ১৯৮

0.52

দিত

0.51

screwdriver

0.51

দার

0.50

 지난

0.50

מין

0.50

POSITIVE LOGITS

年

1.55

년

1.52

 yılında

1.37

年在

1.36

 সালের

1.34

년에

1.29

年的

1.25

年に

1.24

 yılı

1.22

년부터

1.22

Activations Density 0.345%