INDEX

Explanations

events and announcements

The neuron lights up on prominent topic-setting nouns that signal the subject or headline of a news item—words like “deal,” “case,” “agreement,” “race,” “show,” or “proposal.”

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

to

-1.63

 with

-1.47

 różnych

-1.29

or

-1.29

 some

-1.28

 bilim

-1.26

できるので

-1.25

 daž

-1.23

 berbeda

-1.19

 různých

-1.18

POSITIVE LOGITS

↵↵

1.27

…

1.27

 author

1.23

 itself

1.23

 assures

1.21

1.20

 aguja

1.20

 adhesivo

1.20

 последних

1.20

 apoy

1.18

Activations Density 0.215%