INDEX

Explanations

after a particular event

The neuron is strongly triggered by large spans of whitespace or layout characters—i.e. long runs of spaces, tabs, or blank‐line separators.

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

-1.53

 OGSÅ

-1.52

}),

-1.42

當中

-1.42

and

-1.38

 crossorigin

-1.36

Namara

-1.34

 Italij

-1.33

俱乐

-1.31

POSITIVE LOGITS

 they

1.81

1.55

not

1.48

 maravill

1.46

his

1.39

 that

1.38

 them

1.34

 rigue

1.33

 verbre

1.28

 natür

1.27

Activations Density 0.025%