INDEX

Explanations

giveaways and prizes

The neuron specializes in spotting promotional contest and giveaway language—specifically prize mentions, cash amounts, and related numeric or “win”-style tokens.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

埼玉

-1.26

 grève

-1.24

 nôtre

-1.22



-1.21

艄

-1.20

says

-1.18

 vôtre

-1.18

 theyre

-1.16

۱۹

-1.16

tingham

-1.16

POSITIVE LOGITS

 from

2.22

for

1.94

by

1.86

1.50

 both

1.45

 what

1.38

an

1.31

 būt

1.24

 also

1.24

was

1.22

Activations Density 0.048%