INDEX

Explanations

first, comprises, includes

The neuron strongly responds to formal patent-claim style wording—terms like “comprising,” “includes,” “system,” “method,” “at least one,” etc.

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

Such

-0.93

だったり

-0.91

 उदाहरण

-0.91

corrhi

-0.91

ଂ

-0.91

必要がある

-0.89

Szer

-0.89

 häng

-0.88

工程师

-0.87

ieš

-0.85

POSITIVE LOGITS

 first

3.02

first

2.11

第一

1.87

 second

1.65

 первый

1.59

 první

1.58

 primeiro

1.52

 первым

1.50

 primeira

1.49

的第一

1.49

Activations Density 0.430%