INDEX

Explanations

job titles like director or principal

The neuron flags capitalized professional role or job‐title tokens (e.g. “Director,” “Principal,” “Associate,” “Manager”) in text.

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

 \%$\\

-1.37

 Algunas

-1.36

 üblichen

-1.32

 Algunos

-1.27

 Einige

-1.25

 ciertos

-1.23

 eigentlichen

-1.23

 razie

-1.18

 zrobić

-1.18

alyptus

-1.18

POSITIVE LOGITS

of

1.46

at

1.45

 danke

1.35

</h6>

1.28

and

1.25

 CONDU

1.22

</i>

1.20

 through

1.13

</h3>

1.12

ских

1.11

Activations Density 0.033%