INDEX

Explanations

foreign, lower, treatment, autumn

The neuron fires on category labels and classificatory descriptors—e.g. geographical qualifiers like “foreign”/“abroad” and positional labels like “upper”/“lower.”

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

-1.59

antry

-1.58

-1.48

nyard

-1.46

by

-1.46

в

-1.44

氏

-1.41

urably

-1.37

onians

-1.35

and

-1.34

POSITIVE LOGITS

 вони

1.79

こともあり

1.76

 kommunen

1.56

 насеља

1.55

Життє

1.54

,“

1.53

 außerord

1.49

─

1.45

Рей

1.45

 .”

1.44

Activations Density 0.017%