INDEX

Explanations

Italian names and terms

The neuron fires on tokens that are capitalized names or acronyms (i.e. proper nouns and all-caps abbreviations).

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

</td>

-3.06

</b>

-2.86

-2.61

-2.39

 also

-2.31

 های

-2.17

”。

-2.14

𐄁

-2.08

$};

-2.03

правление

-2.02

POSITIVE LOGITS

࿚

2.98

 Then

2.50

ments

2.50

able

2.48

Ᏺ

2.45

ν

2.39

 réessayer

2.30

over

2.28

dzić

2.27

."

2.25

Activations Density 0.008%