INDEX

Explanations

developed, designed, crafted, chosen

The neuron detects words describing product creation or design processes (e.g., “developed,” “designed,” “crafted”).

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

 when

-1.66

or

-1.53

two

-1.48

vého

-1.34

 three

-1.34

 begins

-1.24

釐

-1.23

but

-1.23

 concursos

-1.21

ătă

-1.17

POSITIVE LOGITS

 специально

1.32

caref

1.14

You

1.14

 designed

1.13

 design

1.13

 spécialement

1.13

偷偷

1.12

 semelh

1.10

 تصميم

1.09

特意

1.08

Activations Density 0.047%