INDEX

Explanations

UK official communications and advice

The neuron fires on words naming trades or professional services (e.g. builder, bricklayer, quotes) and on terms referring to legal or violent actions/proceedings (e.g. attack, offence, courts).

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

 railroad

1.06

 regimens

1.06

 outfitted

1.06

 yalnızca

1.05

 marijuana

1.03

 arugula

1.03

 nighttime

1.02

 amerik

1.02

 미국

1.01

 caloric

1.00

POSITIVE LOGITS

Whilst

1.33

 Doncaster

1.27

 Whilst

1.24

 postcode

1.21

UK

1.20

 enquiries

1.20

 HMRC

1.17

 underpinned

1.14

 Sainsbury

1.14

£

1.12

Activations Density 0.033%