INDEX

Explanations

lives claimed or cost

The neuron detects mentions of deaths or casualties—particularly words like “death,” “claimed,” and “lives” in the context of counting fatalities.

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

are

-2.22

 such

-1.45

,,

-1.42

 techniques

-1.40

–

-1.37

 mentre

-1.30

ють

-1.30

it

-1.30

ца

-1.29

}$

-1.27

POSITIVE LOGITS

 cafetería

1.37

帮你

1.36

 yaklaşık

1.28

⓿

1.27

峦

1.27

 femininas

1.27

Цент

1.26

 preciosa

1.24

魨

1.23

(*^^

1.22

Activations Density 0.058%