INDEX

Explanations

numbers followed by other numbers

The neuron fires on numeric characters and punctuation that form docket or reference numbers (i.e. sequences of digits and hyphens).

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

and

-2.56



-1.86

 مختلف

-1.68

垕

-1.68

antaranya

-1.66

𞤢

-1.63

DebuggerNonUser

-1.59

TagHelpers

-1.59

gebouw

-1.58

ální

-1.57

POSITIVE LOGITS

 this

1.88

</h2>

1.72

тори

1.70

翆

1.67

ו

1.64

 digitais

1.63

 naturais

1.63

 получила

1.63

The

1.62

𓋼

1.62

Activations Density 0.015%