INDEX

Explanations

multi-line string delimiters and terminators

The neuron fires strongly on here-document or multi-line string opening delimiters (e.g. “<<<” and “<<-”).

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

 Mondial

-1.66

and

-1.48

 billi

-1.48

 mardi

-1.46

过程中

-1.45

 nagu

-1.45

 هناك

-1.44

 võib

-1.38

衚

-1.36

uña

-1.35

POSITIVE LOGITS

 others

2.09

 other

1.93

1.92

</h4>

1.77

 innych

1.70

</sup>

1.63

 überzeugt

1.57

 otras

1.57

is

1.55

of

1.54

Activations Density 0.009%