INDEX

Explanations

code assignments and declarations

The neuron fires on numeric literal tokens (integers or floats) in code.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 such

-1.45

 each

-1.33

ようになります

-1.28

of

-1.27

我们将

-1.23

 vienas

-1.20

 jednej

-1.19

one

-1.16

 gleiche

-1.14

 koła

-1.14

POSITIVE LOGITS

\_

1.55

They

1.53

they

1.37

nitř

1.34

</h2>

1.34

these

1.32

for

1.30

 yPos

1.28

ázquez

1.28

子を

1.27

Activations Density 0.059%