INDEX

Explanations

construction dates

The neuron fires on numeric tokens that form dates or years (e.g. “Built in 1986,” “first registration 2021,” “year of manufacture 2017”).

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

 pihaknya

-1.18

引用元

-1.16

いま

-1.12

 llanto

-1.12

 ロン

-1.10

lokal

-1.09

ίων

-1.09

 karena

-1.09

Πηγές

-1.09

 ライブ

-1.09

POSITIVE LOGITS

1.28

 there

1.20

歼

1.12

[{

1.11

 kindred

1.05

 mustn

1.02

 quint

1.02

may

1.02

calfe

1.00

Ships

0.98

Activations Density 0.083%