INDEX

Explanations

reddit comments, quotes, or titles

The neuron strongly activates on floating‐point numerals (tokens containing digits with a decimal point).

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

 现在

0.75

 experiences

0.74

NYC

0.73

 agri

0.72

Ahm

0.71

我相信

0.71

 woes

0.70

 razvoja

0.70

 Lond

0.69

 pengembangan

0.68

POSITIVE LOGITS

Footer

0.73

📹

0.68

="<?=

0.67

Durch

0.67

Video

0.65

Viewer

0.64

Hand

0.64

Skip

0.64



0.63

Downloads

0.62

Activations Density 0.007%