INDEX

Explanations

titles and descriptions

The neuron fires strongly on named titles or headings—e.g. program or project names and document or session titles (often in title case or in quotes).

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

 billions

0.75

Also

0.75

groups

0.73

ardır

0.71

charts

0.71

illions

0.70

 constituencies

0.70

 pundits

0.69

ratings

0.69

liners

0.68

POSITIVE LOGITS

 Urinary

0.86

 Fertility

0.85

 Floral

0.84

 Dissertation

0.84

 Mitochondrial

0.84

 Fire

0.83

 Insect

0.82

 Oste

0.81

 Rheumat

0.81

 Tuberculosis

0.81

Activations Density 2.106%