INDEX

Explanations

names and specific phrases

The neuron is primarily detecting numeric tokens (numbers and numeric strings) in the 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

formatics

-0.83

ynchron

-0.80

randomUUID

-0.75

 fasting

-0.74

 Advocacy

-0.73

lecteurs

-0.73

ccf

-0.73

."\

-0.72

 Mindanao

-0.71

 eval

-0.70

POSITIVE LOGITS

 Tahoe

0.80

برا

0.79

 gallop

0.75

プリ

0.73

 besoins

0.72

 stomat

0.71

Ska

0.71

 khaki

0.70

前的

0.70

蛎

0.70

Activations Density 0.009%