INDEX

Explanations

names and nationalities

The neuron strongly activates on numerical and quantitative tokens—numbers, measures, dates, and other figures—i.e. it’s looking for quantitative data in the text.

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

 Allegheny

0.84

 Mojave

0.80

 Wilmington

0.77

 Donegal

0.77

 Scot

0.74

沖縄

0.72

 NASCAR

0.71

0.70

 Tampa

0.70

 ज्वेल

0.70

POSITIVE LOGITS

 handicrafts

1.17

 paysans

1.05

 социа

1.03

 motorbike

1.02

 supaya

0.98

 адміністра

0.97

 cadast

0.97

 peasants

0.96

 magistrates

0.96

 railway

0.96

Activations Density 2.250%