INDEX

Explanations

agent roles

This neuron detects long, contiguous uppercase or prominently capitalized name-and-title fragments (e.g. EMANCIPATOR, Devourer) marking proper nouns or epic titles.

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

-1.50

Joanne

-1.36

pherds

-1.34

”？

-1.34

 dotyczące

-1.34

yaret

-1.30

,\\

-1.29

Bew

-1.27

 extranjeros

-1.27

 incu

-1.27

POSITIVE LOGITS

of

1.66

 beurre

1.53

Then

1.45

if

1.45

for

1.39

 Potosí

1.39

UCIÓN

1.38

還可以

1.37

 calculées

1.35

Most

1.33

Activations Density 0.030%