INDEX

Explanations

politics and social issues

The neuron strongly activates on words and tokens associated with geopolitical, military, and high‐level policy topics (e.g. “violence,” “ethnic,” “democracy,” “military,” “terrorism,” names of leaders, dates/centuries, counts).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

It

-1.81

 part

-1.50

 не

-1.47

 used

-1.29

As

-1.28

use

-1.27

of

-1.26

Not

-1.26

In

-1.24

 provide

-1.22

POSITIVE LOGITS

 Шаг

1.45

 Материал

1.44

 extremadamente

1.40

 descendre

1.38

گزاری

1.38

 numerosas

1.37

몀

1.35

🚍

1.35

 fenómenos

1.34

様々

1.34

Activations Density 0.152%