INDEX

Explanations

Thank you

This neuron doesn’t activate on any tokens—it effectively never fires and thus doesn’t detect any particular pattern.

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

 meil

-1.34

 medan

-1.31

 Tenis

-1.28

 المعيار

-1.27

 suscit

-1.27

 varier

-1.25

了一眼

-1.24

 berge

-1.24

 biens

-1.23

 kafe

-1.23

POSITIVE LOGITS

 still

1.29

not

1.27

 doesn

1.26

one

1.25

 only

1.23

 never

1.22

 more

1.16

 actually

1.16

 first

1.16

 even

1.15

Activations Density 0.037%