INDEX

Explanations

relates to invention

This neuron fires on high-salience discourse cues that introduce or spotlight key content—e.g. section headings (“the present invention relates…”) and reporting/attention markers (“said,” “hot topic,” “forthcoming”).

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

﹍﹍

-1.65

 lebte

-1.63

 bedeckt

-1.60

 markiert

-1.59

 beeind

-1.58

</h5>

-1.55

 faktiskt

-1.54

 überrascht

-1.54

 bewegt

-1.52

 kupa

-1.52

POSITIVE LOGITS

of

2.63

by

1.66

 This

1.63

at

1.51

Not

1.38

for

1.35

 while

1.35

pagesize

1.32

нт

1.32

あるので

1.31

Activations Density 0.051%