INDEX

Explanations

components and systems

The neuron flags high-importance, domain-specific nouns (e.g. “company,” “shareholders,” “data,” “computer,” “environment”) that serve as topic keywords in technical or financial 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

and

-1.59

 this

-1.40

 then

-1.24

 again

-1.16

ことも

-1.15

ところは

-1.14

 dabar

-1.05

 these

-1.01

 asemenea

-1.00

際には

-0.98

POSITIVE LOGITS

的一种

1.70

WillAppear

1.37

or

1.27

usually

1.27

 aikana

1.26

is

1.26

段时间

1.24

who

1.21

において

1.16

欧式

1.14

Activations Density 0.140%