INDEX

Explanations

code block openers

The neuron fires on embedded code delimiters (e.g. “<?” and “@”) indicating server‐side/script 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

-1.93

empio

-1.63

of

-1.52

its

-1.52

</b>

-1.52

);

-1.50

 that

-1.45

We

-1.39

 日期

-1.38

및

-1.30

POSITIVE LOGITS

 only

1.70

 first

1.59

ڧ

1.49

 stesse

1.47

Tecnologia

1.47

 शुर

1.44

はこんな感じ

1.42

 scoprire

1.42

Cuidado

1.37

sivamente

1.36

Activations Density 0.008%