INDEX

Explanations

```

The neuron fires on date‐stamp patterns (month day, year, and range separators) at the beginnings of articles.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

that

0.24

abundance

0.23

 that

0.23

cludes

0.23

大家都

0.22

 THAT

0.22

 facilitating

0.22

ேயே

0.22

 brooch

0.21

 That

0.21

POSITIVE LOGITS

 utenti

0.25

 joten

0.23

וק

0.22

อ

0.22

Pentru

0.21

คุณ

0.21

 คณิต

0.20

|।

0.20

 cikin

0.20

Activations Density 0.005%