INDEX

Explanations

scripting and configuration details

The neuron fires on code identifiers (especially CamelCase/PascalCase names)—i.e. tokens that are class‐ or type‐style names.

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

of

-1.40

 maler

-1.34

 herren

-1.31

 never

-1.27

が

-1.23

他

-1.23

uan

-1.20

 with

-1.20

 tages

-1.20

 into

-1.18

POSITIVE LOGITS

When

1.51

 pudieran

1.47

了几

1.45

 ""))

1.40

expédition

1.39

 offensichtlich

1.33

During

1.32

 ',',

1.32

Usually

1.31

ométrique

1.31

Activations Density 0.018%