INDEX

Explanations

categories of things

The neuron is tuned to spot nouns and descriptors that immediately follow numeric mentions (i.e. words naming categories or items right after a number).

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

if

-1.63

two

-1.62

әрмәләр

-1.36

 podremos

-1.35

取締

-1.34

 девушка

-1.30

 segíts

-1.23

 only

-1.23

èt

-1.22

ukone

-1.22

POSITIVE LOGITS

 that

1.87

 различ

1.84

 including

1.79

 включая

1.62

 которые

1.60

 diferentes

1.54

who

1.48

including

1.47

 olika

1.46

 هذا

1.41

Activations Density 0.456%