INDEX

Explanations

studies suggest or identify

The neuron fires on specialized scientific–technical jargon—especially method and result words (e.g. “development,” “structure-based,” “design,” “evaluation,” “glycodendrimers,” “yielded,” etc.) in research‐style writing.

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

蟊

-1.27

กรณ์

-1.13

 fără

-1.11

默认

-1.09

さまで

-1.07

 تساعد

-1.05

Ô

-1.03

Ă

-1.03

фы

-1.02

POSITIVE LOGITS

 recently

1.48

 studies

1.47

Recently

1.38

dreamstime

1.31

by

1.27

 Recently

1.27

 such

1.16

can

1.16

 using

1.13

has

1.11

Activations Density 0.063%