INDEX

Explanations

forms of to be

The neuron is primarily detecting occurrences of the verb “be” (especially the form “be,” often in constructions like “could be,” “might be,” or “that’s a”).

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

 santiago

-1.56

 would

-1.52

 diego

-1.46

 castro

-1.34

 colombia

-1.33

 pablo

-1.33

 alexander

-1.32

 gabriel

-1.32

 miguel

-1.31

 valencia

-1.31

POSITIVE LOGITS

på

1.48

ୌ

1.41

return

1.39

 estos

1.35

 для

1.34

留意

1.34

 novin

1.34

(\

1.30

 voor

1.29

Activations Density 0.026%