INDEX

Explanations

years and durations

This neuron detects numeric tokens—especially multi‐digit numbers representing dates, counts, percentages, monetary amounts, and similar numeric quantities.

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

几十

0.66

month

0.62

 месяц

0.58

Fifty

0.57

Month

0.53

antica

0.53

week

0.52

月

0.52

昭和

0.52

 mês

0.52

POSITIVE LOGITS

四年

2.30

三年

2.00

五年

2.00

六年

1.98

两年

1.77

 years

1.44

 साढ़े

1.40

二年

1.37

 Years

1.36

 YEARS

1.35

Activations Density 0.333%