INDEX

Explanations

future or past states

The neuron fires on the principal topic noun that heads a paragraph or section—i.e. the key subject term introducing each passage.

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

 verranno

-1.48

 公园

-1.44

 applau

-1.42

 vanille

-1.39

青木

-1.38

 XXIII

-1.37

榎

-1.36

 gosta

-1.35

-1.34

 עוד

-1.34

POSITIVE LOGITS

 that

2.25

if

2.13

by

1.81

だったら

1.70

or

1.69

 geweldig

1.66

 prachtig

1.66

 about

1.66

 more

1.60

);

1.59

Activations Density 0.132%