INDEX

Explanations

java code keywords

The neuron activates on isolated capitalized identifiers—i.e. the simple names of classes or types (the parts after the last “.”) in import statements.

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

 permitirá

-1.70

-1.65

 generó

-1.64

-1.63

 conseguido

-1.59

 provocado

-1.58

 esteemed

-1.57

 funcionam

-1.56

 both

-1.55

POSITIVE LOGITS

of

1.46

苼

1.43

加拿大

1.35

鐽

1.30

ТОВ

1.30

 hiver

1.30

 denna

1.30

 違い

1.30

 этого

1.29

את

1.29

Activations Density 0.027%