INDEX

Explanations

Emotional Freedom techniques

The neuron primarily activates on capitalized names, acronyms, and title‐style terms (e.g. proper nouns, all‐caps abbreviations, and numbered sequel markers).

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

in

-1.70

 this

-1.63

and

-1.57

 both

-1.51

 provides

-1.44

as

-1.42

 only

-1.41

 requires

-1.41

 should

-1.39

at

-1.38

POSITIVE LOGITS

 majeurs

1.41

ビーフ

1.28

 ウール

1.28

 ストライプ

1.28

 刺繍

1.24

 actuels

1.23

 インナー

1.22

たまに

1.22

зывы

1.19

 naturels

1.17

Activations Density 0.094%