INDEX

Explanations

global namespace pollution

This neuron fires on imperative action words (commands) in technical or code‐instruction contexts.

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

 revolutionized

0.54

 Barnett

0.53

 kidding

0.52

 minimalistic

0.52

verbose

0.52

 redefined

0.50

 sécrétion

0.50

 vile

0.50

 Borel

0.50

tokenize

0.49

POSITIVE LOGITS

 ViewBag

0.97

 ViewData

0.82

FLASH

0.75

Flash

0.71

 flash

0.70

上下文

0.70

flash

0.70

Redirect

0.69

 FLASH

0.68

 Redirect

0.67

Activations Density 0.017%