INDEX

Explanations

Specific entities or events

The neuron selectively fires on high‐level headline or title phrases—i.e. the section or article headings rather than body 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

 deines

-1.20

Hundreds

-1.09

뿐

-1.07

 giud

-1.07

⾒

-1.06

他自己

-1.05

 efficacement

-1.03

 profondément

-1.01

 telles

-1.00

 bajo

-0.98

POSITIVE LOGITS

and

1.63

it

1.30

さんも

1.27

さんに

1.18

但却

1.15

さんを

1.13

さんが

1.09

ſi

1.08

 باقی

1.05

หรือ

0.99

Activations Density 21.173%