INDEX

Explanations

government offices and roles

The neuron detects mentions of formal organizational or governmental office titles (e.g., “Office,” “Directorate,” “Ministry of…,” “Presidency”).

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

-1.50

</em>

-1.42

!"

-1.37

"",

-1.37

ご注意

-1.34

 или

-1.27

Then

-1.26

Instead

-1.25

僅

-1.25

POSITIVE LOGITS

it

1.50

𝗞

1.34

Jährige

1.34

of

1.28

 sencillas

1.27

 inalámb

1.27

 таких

1.20

起来的

1.19

层的

1.16

Dagger

1.16

Activations Density 0.043%