INDEX

Explanations

websites and official documentation

The neuron detects calls-to-action or “click here”–style promotional link phrases.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 estábamos

0.58

何の

0.55

تقرير

0.54

小説

0.51

 podríamos

0.51

বিষ্ট

0.51

 Abbiamo

0.50

 그려

0.50

 Seorang

0.50

 empezamos

0.50

POSITIVE LOGITS

 HERE

1.51

 https

1.47

 http

1.44

www

1.27

https

1.26

 here

1.24

http

1.23

HERE

1.21

>>

1.04

here

1.02

Activations Density 0.789%