INDEX

Explanations

numbers around nine

This neuron activates on tokens containing digits (numeric values, years, amounts, etc.).

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

二十

0.63

 Sixteen

0.61

 trente

0.61

 sixteen

0.60

٢

0.59

 seventeenth

0.58

 २५

0.57

 ১৫০

0.57

 thirty

0.57

十七

0.57

POSITIVE LOGITS

2.06

９

2.03

۹

1.77

 ninth

1.74

 Ninth

1.73

۹

1.73

 nueve

1.70

 nine

1.61

⁹

1.59

৯

1.53

Activations Density 0.338%