INDEX

Explanations

names of people

The neuron activates on fully qualified Java I/O package references (e.g., “java.io”).

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

and

-1.54

by

-1.54

 there

-1.53

 based

-1.43

of

-1.33

 again

-1.33

以及

-1.32

 língua

-1.25

 vitória

-1.24

you

-1.20

POSITIVE LOGITS

gesehen

1.48

 interrump

1.41

σία

1.38

){

1.26

 octobre

1.25

щений

1.23

%).

1.22

钾

1.21

 hawk

1.21

ⴱ

1.20

Activations Density 0.001%