INDEX

Explanations

approximations of degree or certainty

The neuron activates on vague or qualifying adverbs (e.g. “somehow,” “somewhat,” “extent”) that hedge or soften 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

 helft

-1.34

teki

-1.32

même

-1.28

คือ

-1.27

针

-1.24

pruch

-1.23

 façons

-1.23

ต

-1.23

 thèmes

-1.23

咣

-1.22

POSITIVE LOGITS

 have

1.68

émoc

1.60

 Página

1.47

metido

1.47

 consistently

1.45

at

1.44

橚

1.41

1.38

of

1.36

 Referenzen

1.36

Activations Density 0.012%