INDEX

Explanations

hexadecimal bytes and colors

The neuron activates on hexadecimal byte literals (e.g. “0xFF”, “FA”, “DE”, etc.) in source code.

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

-1.91

 from

-1.89

 have

-1.70

 takes

-1.68

 keeps

-1.65

 regarding

-1.63

 should

-1.63

 makes

-1.62

 gives

-1.62

 through

-1.59

POSITIVE LOGITS

</b>

1.73

,",

1.57

 bloquer

1.39

 flamboyant

1.38

мок

1.35

—-

1.35

 大きい

1.35

,;

1.34

Iam

1.33

 viejos

1.31

Activations Density 0.001%