INDEX

Explanations

equal to

The neuron fires strongly on isolated high-register, technical or formal vocabulary items—especially verbs and adjectives (“exported,” “normal,” “emphasizes,” etc.) that you’d see in legal, regulatory, or scientific prose.

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

ﺭ

0.91

er

0.87

AndTime

0.81

 ভিতরে

0.80

 νό

0.80

<0x8C>

0.79

ber

0.77

 Эд

0.77

atuan

0.76

ffel

0.75

POSITIVE LOGITS

根据

0.92

opausal

0.77

讲解

0.75

随着

0.74

单纯

0.73

去

0.71

传统的

0.70

})^{

0.70

念

0.70

畴

0.70

Activations Density 0.000%