INDEX

Explanations

member of or named

This neuron picks out references to missing or undefined properties/methods (e.g. “has no attribute,” “undefined method,” “not a member of”).

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

 bordados

-1.39

)=\

-1.36

 icône

-1.34

moiselle

-1.31

 būti

-1.30

你是不是

-1.29

釵

-1.28

.;

-1.27

POSITIVE LOGITS

who

1.58

ishwa

1.41

 desired

1.38

EACH

1.38

 Kudos

1.37

suo

1.37

to

1.34

わい

1.31

 tampa

1.31

be

1.31

Activations Density 0.029%