INDEX

Explanations

in fact

The neuron detects high-level discourse markers or sentence‐opening adverbs (e.g. “In,” “Here,” “We”) that introduce or transition between statements.

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

 will

-1.72

 then

-1.55

–

-1.50

ke

-1.37

te

-1.31

 Hauses

-1.30

-1.28

-1.27

-1.23

may

-1.21

POSITIVE LOGITS

 этого

1.48

achuca

1.38

 հղումներ

1.36

ellä

1.29

Ꮥ

1.27

катери

1.25

klere

1.25

 verschiedenen

1.25

 когда

1.24

Médaille

1.23

Activations Density 0.699%