INDEX

Explanations

floors, relief, silhouette, policy, information

This neuron detects promotional or event–related terms, especially those tied to film/video showcases, streaming services, and related marketing language.

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

 inasmuch

0.96

 etcétera

0.91

 agak

0.85

 “‘

0.85

 తాను

0.85

之类的

0.83

 తనకు

0.83



0.81



0.80

 انھوں

0.77

POSITIVE LOGITS

 YOUR

1.48

YOU

1.46

1.40

1.31

 ONLY

1.30

!</

1.26

!\

1.24

 your

1.20

！

1.19

让你

1.16

Activations Density 0.997%