INDEX

Explanations

car burning

The neuron fires strongly on multiword noun phrases—especially those introduced by a determiner (“the number on the screen,” “the piece of paper,” 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

 then

-1.84

you

-1.60

so

-1.50

 both

-1.46

 These

-1.45

lo

-1.40

 would

-1.34

 have

-1.30

 because

-1.27

-1.26

POSITIVE LOGITS

 clínicos

1.42

 الذي

1.40

houettes

1.34

boleh

1.33

 químicos

1.31

 delito

1.31

 možnosti

1.31

 dispuestos

1.31

 possibilité

1.29

 который

1.28

Activations Density 0.017%