INDEX

Explanations

The neuron strongly activates on the principal content word at the head of a bibliographic entry or title (i.e. the main title term just before a subtitle or colon) in lists of publications.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

1.39

is

1.34

in

1.27

as

1.23

to

1.16

are

1.15

be

1.14

on

1.12

POSITIVE LOGITS

 വിവിധ

1.43

堈

1.06

<unused327>

1.06

羵

1.02

<unused336>

1.02

誢

1.00

<unused2031>

1.00

direccion

0.99

<unused2044>

0.99

龧

0.98

Activations Density 0.000%