INDEX

Explanations

;

The neuron strongly activates on tokens that are capitalized identifiers (e.g. class names, frameworks, brand names, or other proper nouns) in code.

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

 only

-1.56

 both

-1.30

 from

-1.30

-1.28

-1.27

-1.24

all

-1.22

-1.20

providedIn

-1.20

POSITIVE LOGITS

↵

2.06

Groet

1.44

 ドア

1.41

NOTAS

1.38

Bibliografie

1.33

XDD

1.33

Spes

1.29

Kleur

1.24

Gek

1.23

SONY

1.22

Activations Density 0.204%