INDEX

Explanations

date ranges

This neuron fires on numeric timeline markers—especially years, dates, and “Present”—i.e. date and duration tokens in experience or study periods.

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

 empêche

-1.61

 querían

-1.58

 kosme

-1.57

 simplesmente

-1.55

 spektak

-1.52

 deklar

-1.52

 tunik

-1.51

you

-1.49

 bakteri

-1.49

 komplet

-1.48

POSITIVE LOGITS

1.97

at

1.66

1.64

 Although

1.63

 Despite

1.56

 However

1.55

As

1.51

 with

1.50

 Unlike

1.45

1.41

Activations Density 0.013%