INDEX

Explanations

translation, id tags

The neuron fires on metadata keys in localization/translation files—e.g. XML `<source>` tags and `msgid` identifiers.

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

 gewor

-2.02

 anhänger

-1.91

なども

-1.91

 zubehör

-1.89

 dando

-1.88

-1.85

 électriques

-1.80

 páns

-1.80

 elaboración

-1.78

 fantástica

-1.77

POSITIVE LOGITS

ဴ

1.98

}$.

1.84

 This

1.83

 appear

1.78

 при

1.77

 शुर

1.76

 receive

1.74

 indicate

1.73

icoot

1.73

 will

1.72

Activations Density 0.002%