INDEX

Explanations

introduces core explanation

The neuron detects discourse-level qualifier adverbs (e.g. “essentially,” “basically,” “fundamentally,” “characteristically”) that signal a summary or emphasis.

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

for

-1.94

on

-1.52

or

-1.37

-1.32

niosek

-1.32

to

-1.31

&#

-1.29

 Deze

-1.28

?"

-1.27

 particolar

-1.25

POSITIVE LOGITS

 confuso

1.59

mathrm

1.57

 doloroso

1.50

である

1.46

 cerrada

1.45

 decorativa

1.41

menade

1.41

ABLO

1.40

 Inns

1.39

 sorgfäl

1.39

Activations Density 0.030%