INDEX

Explanations

describing origins and creation

The neuron fires on section-heading keywords that are verbs or participles (e.g. “Born,” “Written,” “Consisting,” “Designed,” “Dating”) marking metadata or introductory lines.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 those

-1.84

 does

-1.63

can

-1.61

use

-1.59

because

-1.47

 their

-1.46

 other

-1.43

lo

-1.41

</i>

-1.41

 more

-1.40

POSITIVE LOGITS

 ſou

1.57

adalah

1.50

梠

1.48

澌

1.47

1.41

 moelle

1.39

 帅气

1.37

 stratég

1.36

瓛

1.35

 heeft

1.34

Activations Density 0.063%