INDEX

Explanations

years followed by punctuation

The neuron fires on the substring “201” in four-digit year numbers (e.g. in “2013,” “2015,” etc.).

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

dersfield

-1.76

ICOLON

-1.55

ありました

-1.54

 traffico

-1.49

舠

-1.48

 initiate

-1.47

 schönes

-1.46

tiqueta

-1.46

 хочется

-1.45

 bisschen

-1.42

POSITIVE LOGITS

1.73

1.67

{~

1.53

1.50

 uren

1.48

：

1.47

1.45

("

1.38

...

1.37

 både

1.34

Activations Density 0.043%