INDEX

Explanations

list item punctuation and separators

The neuron strongly flags infrequent, content-specific words—i.e. rare vocabulary items.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

⚉

1.10

ૄ

1.01

 pouquinho

0.95

ákat

0.93

 తొలి

0.92

 войска

0.90

 Quadrupèdes

0.90

 betroffen

0.89

 ショ

0.89

 reglamento

0.89

POSITIVE LOGITS

1.16

0.89

0.86

0.84

non

0.83

0.81

 highly

0.80

an

0.78

Activations Density 0.000%