INDEX

Explanations

et al

The neuron is highly sensitive to standalone numeric tokens—particularly years, volume, and page numbers in reference citations.

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

":"+

-2.22

 ($('#

-2.19

usschuss

-2.00

 ($("#

-1.91

硖

-1.84

 }));

-1.80

ONESIA

-1.72

/\.

-1.68

 которыми

-1.67

 selera

-1.66

POSITIVE LOGITS

1.70

 Moreover

1.55

1.54

夘

1.48

却是

1.46

 Also

1.45

the

1.44

増えて

1.43

 概念

1.42

 介绍

1.41

Activations Density 0.003%