INDEX

Explanations

rich descriptions

The neuron activates on constructions that signal abundance or richness—e.g. “rich theory,” “rich metadata,” “rich diversity,” “rich variety,” “rich history,” or “rich source.”

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

or

-2.13

 such

-1.85

 what

-1.84

braio

-1.51

 verdades

-1.45

...”

-1.45

しないで

-1.45

 dieną

-1.43

”,

-1.41

しないと

-1.41

POSITIVE LOGITS

rich

1.95

丰富的

1.57



1.55

1.52

sender

1.47

 richness

1.46

1.45



1.43

Rich

1.41

 mereka

1.39

Activations Density 0.023%