INDEX

Explanations

symbol-delimited keywords

The neuron is triggered by long runs of repeated punctuation characters (like sequences of hyphens or equals signs) that serve as section dividers or horizontal rules.

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

an

-1.77

as

-1.68

 then

-1.66

وحة

-1.50

<em>

-1.49

 sometime

-1.47

 FINDINGS

-1.45

ikbaar

-1.41

 principalement

-1.41

 from

-1.39

POSITIVE LOGITS

which

1.73

êtes

1.73

 anderen

1.68

ซึ่ง

1.66

 WHICH

1.66

 Efek

1.59

 или

1.57

},

1.56

Ivoire

1.55

 VERY

1.52

Activations Density 0.094%