INDEX

Explanations

be followed by state or action

The neuron strongly activates on instances of the verb “to be” (and related modals like “should be,” “able to be,” etc.).

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

but

-1.52

 There

-1.52

 Your

-1.39

we

-1.38

ronique

-1.33

 špat

-1.31

 Often

-1.29

 With

-1.27

NotEqualTo

-1.27

pfung

-1.26

POSITIVE LOGITS

1.74

1.63

1.54

be

1.51

1.48

1.26

Do

1.25

 minum

1.23

 ſta

1.23

–

1.21

Activations Density 0.068%