INDEX

Explanations

look, read, watch

The neuron detects imperative, call-to-action phrases that urge the reader to look, check out, watch, or read something.

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

 biasanya

-1.38

 erneut

-1.34

 ponownie

-1.33

kaos

-1.27

 всегда

-1.27

 with

-1.26

caucasian

-1.26

 после

-1.23

𖥧

-1.22

lahir

-1.21

POSITIVE LOGITS

设计的

1.35

📧

1.34

taient

1.32

tront

1.29

 його

1.25

пература

1.22

 bambou

1.22

 these

1.20

 любы

1.18

 appétit

1.17

Activations Density 0.093%