INDEX

Explanations

academic publication citations

The neuron activates on numeric substrings—particularly digits and numbers (e.g. volume, page, year, or issue numbers) as found in citations and references.

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

</h3>

-1.55

Conclusie

-1.45

。"

-1.44

quarelle

-1.38

šie

-1.30

 الخارجية

-1.24

 occuper

-1.23

Terug

-1.21

そうですが

-1.18

たび

-1.17

POSITIVE LOGITS

 that

2.05

to

1.76

 those

1.59

not

1.46

1.38

1.37

in

1.34

\&

1.31

 início

1.23

 जिन

1.23

Activations Density 0.018%