INDEX

Explanations

explaining purpose or discussion

This neuron fires strongly on UI element or section‐heading labels (e.g. “popup,” “Tools,” “consultation,” “roundtable forum”), i.e. names of interface pop-ups, tools, or menu headings.

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

ิ

0.75

所在的

0.74

 enthusiast

0.71

 hails

0.70

 bằng

0.69

otte

0.68

 criterion

0.68

from

0.67

ismatic

0.67

ibald

0.66

POSITIVE LOGITS

เพื่อ

1.63

 để

1.41

 чтобы

1.38

 membahas

1.34

 чтоб

1.32

 щоб

1.31

 upang

1.29

เพื่อให้

1.29

 เพื่อ

1.27

ເພື່ອ

1.23

Activations Density 2.704%