INDEX

Explanations

This neuron detects named entities—especially multiword organization, project, or program titles and acronyms.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 hypnot

0.72

 extermin

0.72

電腦

0.68

电脑

0.67

 контроли

0.66

 handicapped

0.66

 blemishes

0.65

 homosexual

0.64

 recollections

0.64

 invariable

0.63

POSITIVE LOGITS

 Initiative

1.72

 Initiatives

1.45

 Innovation

1.45

 Partnerships

1.43

 initiative

1.40

 Collaborative

1.38

 Partnership

1.32

 Coalition

1.31

 Sustainability

1.30

 Roadmap

1.28

Activations Density 0.970%