INDEX

Explanations

kind of thing

The neuron tends to light up on the single head word (usually a noun or gerund) that names the main topic or object of a phrase.

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

habited

-1.55

”;

-1.42

];

-1.38

 loca

-1.34

hysema

-1.32

rália

-1.31

 "..\..\

-1.28

herst

-1.27

neuf

-1.27

|;

-1.26

POSITIVE LOGITS

 such

1.63

 those

1.30

 rígido

1.30

 eigentliche

1.29

or

1.28

 também

1.23

ութ

1.23

 being

1.22

 deiner

1.21

 también

1.20

Activations Density 0.143%