INDEX

Explanations

communication and control systems

The neuron is highly sensitive to multiword technical terms and titles—especially capitalized domain-specific phrases (e.g. component/section headings and product names) in patent-style 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

 dergleichen

-1.41

 merkwür

-1.29

 obiet

-1.26

 задний

-1.21

 christlichen

-1.21

 other

-1.20

TERS

-1.18

 dunklen

-1.18

ρει

-1.17

 ungewöhn

-1.16

POSITIVE LOGITS

 with

1.38

and

1.32

1.21

 from

1.10

médi

1.06

of

1.05

 zoll

1.04

 qualit

1.04

 tats

1.00

lacian

1.00

Activations Density 0.504%