INDEX

Explanations

Jiu-Jitsu

The neuron strongly activates on isolated numeric or technical alphanumeric tokens (e.g. standalone digits, number-containing codes, or uppercase acronyms).

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

fuo

-0.91

นั้น

-0.87

Projeto

-0.86

谷

-0.84

SHIRE

-0.83

Dział

-0.79

 ravioli

-0.78

 sulfu

-0.78

 英語

-0.78

さん

-0.77

POSITIVE LOGITS

 $^{\

1.11

 einzigartige

0.94

$^{

0.89

 szükség

0.88

 megs

0.88

 reason

0.85

meg

0.83

}\,\

0.81

 breite

0.81

 листа

0.81

Activations Density 0.037%