INDEX

Explanations

preposition followed by identifier

This neuron activates on common English connective words (prepositions and similar tokens) used in code identifiers, such as “from,” “on,” “in,” “to,” “by,” and “with.”

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

 perfekte

-1.54

立马

-1.48

 gratuita

-1.48

-1.41

its

-1.38

 motive

-1.35

 citada

-1.35

 qualquer

-1.35

楽しめます

-1.33

及び

-1.32

POSITIVE LOGITS

苼

1.95

 something

1.57

]-\

1.55

ஂ

1.52

 also

1.52

Jawab

1.50

 воспользова

1.47

gestellt

1.45

 bukan

1.42

Something

1.41

Activations Density 0.035%