INDEX

Explanations

car model and spec information

The neuron activates on the numeric (and adjoining letter) tokens that make up car model or performance variant codes (e.g. “M4,” “N,” “GT5,” 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

 that

-1.70

intégr

-1.34

＊＊＊

-1.30

 caneca

-1.29

 Deformation

-1.27

 比較

-1.26

 fundamenta

-1.23

indicates

-1.22

igreja

-1.22

 ENDS

-1.20

POSITIVE LOGITS

之后

1.49

1.41

 then

1.35

 ikkje

1.34

 специальные

1.30

 alors

1.29

1.27

cazione

1.27

ÇÃO

1.26

Activations Density 0.018%