INDEX

Explanations

numbers

The neuron is primarily activated by numeric tokens (digits and multi‐digit 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

Hundreds

-0.98

hundreds

-0.92

 centaines

-0.90

五百

-0.89

hundred

-0.87

 centenary

-0.87

undred

-0.86

 hundred

-0.85

рый

-0.83

 centuries

-0.79

POSITIVE LOGITS

 twenty

1.27

 thirty

1.15

ten

1.10

 Twenty

1.02

二十

1.02

 fifteen

0.95

 ۲۰

0.94

Twenty

0.90

 forty

0.88

 sixteen

0.82

Activations Density 0.090%