INDEX

Explanations

coding and testing constructs

The neuron fires on the character sequence “Dup” (as in parts of words like “Duplicator”).

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

or

-1.97

 because

-1.80

 there

-1.60

as

-1.57

 Without

-1.52

 Numerous

-1.50

 When

-1.47

for

-1.46

 Often

-1.45

 Those

-1.44

POSITIVE LOGITS

getMock

1.48

龇

1.48

 EXPECT

1.46

 {}));

1.43

cestry

1.39

1.38

 fluide

1.38

 aktiv

1.38

都想

1.35

 neve

1.34

Activations Density 0.002%