INDEX

Explanations

as if, like, as though

The neuron detects simile or hypothetical comparison phrases built with “were” (e.g. “as if they were…,” “like it was…”).

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

who

-1.49

habited

-1.41

 Início

-1.39

of

-1.38

Your

-1.38

讓

-1.32

 Aragón

-1.30

您好

-1.27

㌔

-1.27

in

-1.26

POSITIVE LOGITS

 another

2.39

 other

1.80

1.79

 autre

1.52

 either

1.51

 başka

1.50

 otras

1.49

 soluciones

1.48

an

1.48

one

1.47

Activations Density 0.009%