INDEX

Explanations

judging from or by

The neuron fires on words and short phrases that introduce hedges or qualifications—i.e. linguistic markers of uncertainty or estimation (like “judging from” or “grain of salt”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 konfigura

-1.70

 konkret

-1.56

 lokale

-1.54

 defekt

-1.53

 infrastruktur

-1.52

but

-1.51

 sanno

-1.48

 eksklu

-1.48

 katalog

-1.46

 kompet

-1.45

POSITIVE LOGITS

of

1.98

on

1.74

in

1.63

 other

1.48

IN

1.46

↵↵↵

1.41

by

1.41

 nesta

1.40

 often

1.40

1.36

Activations Density 0.165%