INDEX

Explanations

5 thru 8

This neuron detects tokens that are identifiers or literals containing numeric characters (e.g. type sizes like “u32,” hex/escape sequences like “\xE2,” constants, version numbers).

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

ちなみに

-1.95

Despite

-1.90

and

-1.78

 Despite

-1.73

While

-1.64

-1.59

 While

-1.59

Though

-1.58

For

-1.57

After

-1.56

POSITIVE LOGITS

婠

1.84

 wonderfully

1.59

做什么

1.58

 χρησιμοποι

1.56

probably

1.55

 nowy

1.54

 amazingly

1.53

 stary

1.50

unfortunately

1.49

primarily

1.48

Activations Density 0.002%