INDEX

Explanations

place names and publishers

The neuron fires on bibliographic and citation metadata—publication details such as dates, page or volume numbers, journal abbreviations, and related reference markers.

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

Bibli

0.88

Museum

0.86

Thr

0.79

Conclusions

0.78

Bench

0.75

Bo

0.74

Paste

0.73

Kak

0.73

Pepper

0.73

Cruise

0.72

POSITIVE LOGITS

 playmaker

0.75

 carnage

0.72

är

0.71

då

0.71

 deluxe

0.71

 opportune

0.71

 ombre

0.70

 badass

0.70

с

0.70

Activations Density 0.044%