INDEX

Explanations

line styles and borders

The neuron activates on tokens naming line‐style keywords (e.g. “dashed,” “dotted,” “solid”).

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

geable

-1.29

 bizarre

-1.21

 บาง

-1.20

-1.19

oted

-1.18



-1.18

 crea

-1.14

în

-1.14

还不错

-1.14

altitude

-1.13

POSITIVE LOGITS

our

1.37

1.26

 Figures

1.20

Lain

1.19

</i>

1.18

woordig

1.13

 tarko

1.13

[]={

1.13

}`);

1.11

コチラ

1.10

Activations Density 0.002%