INDEX

Explanations

method for procedures

The neuron is strongly triggered by the word “Method.”

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

 alrededores

-1.50

луйста

-1.47

作出

-1.45

pulseira

-1.42

IQUES

-1.36

estructuras

-1.36

 hábit

-1.35

齜

-1.34

står

-1.34

ted

-1.33

POSITIVE LOGITS

すでに

1.60

 gerekir

1.44

 each

1.39

 diinginkan

1.38

—

1.34

asantry

1.34

 escolas

1.33

 最後

1.33

 وما

1.33

 choose

1.33

Activations Density 0.016%