INDEX

Explanations

titles and professions

The neuron detects copular constructions (“is,” “was,” etc.) used to link a subject to a description (especially patterns like “is a…” or “was…”).

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

 both

-1.38

 each

-1.35

-1.34

 inscripción

-1.30

 their

-1.30

 bigger

-1.30

 such

-1.28

랐

-1.28

 itself

-1.27

its

-1.27

POSITIVE LOGITS

if

1.88

 there

1.80

朱印

1.34

他是

1.33

出现在

1.23

 beliau

1.21

Mr

1.20

 affez

1.20

 hadapan

1.16

艷

1.16

Activations Density 0.035%