INDEX

Explanations

return values

The neuron fires on closing curly‐brace tokens (“}”), i.e. it spots the ends of code blocks.

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

۶

-1.46

''

-1.41

was

-1.39

など

-1.35

"/",

-1.32

 ۱۵

-1.30

ядок

-1.26

 ۳۰

-1.24

を

-1.23

۸

-1.22

POSITIVE LOGITS

に行って

1.37

 verschillen

1.28

镜子

1.22

its

1.21

 absorción

1.20

it

1.19

 cálido

1.18

槼

1.13

gencias

1.12

 agujeros

1.12

Activations Density 0.001%