INDEX

Explanations

checks, communication, data, resistance, advertising

The neuron fires on proper-noun tokens—that is, names of people, places, titles, brands, or other named entities.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

0.86

（

0.84

 amelyet

0.80

（

0.79

 เนื่องจาก

0.78

 אשר

0.75

0.74

 καθώς

0.73

−

0.73

(_

0.73

POSITIVE LOGITS

 doesn

1.91

 didn

1.77

isn

1.73

 hasn

1.68

 shouldn

1.66

 wasn

1.63

 deserves

1.62

 gotta

1.58

 aren

1.55

 wouldn

1.54

Activations Density 0.850%