INDEX

Explanations

should be or only

This neuron detects prescriptive modal verbs (especially “should”) that signal recommendations or guidance.

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

 that

-1.47

</strong>

-1.40

stel

-1.38

 reconoci

-1.31

 absence

-1.31

 listos

-1.30

 он

-1.27

immagine

-1.26

わずか

-1.26

 internetowa

-1.24

POSITIVE LOGITS

the

1.50

 rafael

1.48

nih

1.48

while

1.47

unless

1.46

 garcia

1.45

 antonio

1.43

already

1.42

 loto

1.41

ve

1.40

Activations Density 0.130%