INDEX

Explanations

word endings like -istic, -atory, -ize, -ism

This neuron fires on long, multi-syllable words—especially those of Latin/Greek origin (e.g. “handicapping,” “laundering,” “ventriloquism,” “severability,” 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

 hủy

0.55

 mudah

0.55

 yang

0.54

tfrac

0.52

↻

0.50

px

0.49

 tambahan

0.49

 divis

0.49

 hilang

0.48

 tăng

0.48

POSITIVE LOGITS

ing

0.82

ed

0.72

揄

0.67

entially

0.66

itatively

0.61

edRight

0.61

함으로써

0.61

 રીતે

0.59

 کردن

0.58

了下来

0.58

Activations Density 0.000%