INDEX

Explanations

chill, hanging out

The neuron selectively activates on numerical measurements, especially decimal-formatted numbers.

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

 clouded

0.88

MNR

0.88

 Bhojpuri

0.77

ALJ

0.77

 ",")

0.76

JSC

0.76

 spanned

0.76

 mistrust

0.75

 violated

0.74

 Masjid

0.74

POSITIVE LOGITS

記憶

0.77

はじめ

0.77

図

0.74

 règles

0.73

sections

0.72

channels

0.68

 ఉండే

0.68

圖

0.68

 వు

0.67

раз

0.66

Activations Density 0.001%