INDEX

Explanations

patent dates and numbers

The neuron detects numeric sequences and classification numbers commonly found in patent citations and prior‐art listings.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 this

-2.28

our

-1.70

 junger

-1.66

 cuillère

-1.61

pullover

-1.47

 этой

-1.46

 שלנו

-1.39

gemaakt

-1.38

 these

-1.38

it

-1.34

POSITIVE LOGITS

єра

1.49

无论是

1.49

诟

1.48

1.45

 државе

1.42

 אך

1.40

两种

1.36

waf

1.34

 времена

1.34

 แต่

1.33

Activations Density 0.158%