INDEX

Explanations

environment based

The neuron fires on named entities and other capitalized, topic‐specific keywords (proper nouns and titles).

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

由于

1.05

においては

1.02

】，

0.98

Because

0.96

においても

0.95

虽然

0.93

 ponieważ

0.92

 เนื่องจาก

0.92

जबकि

0.91

因為

0.91

POSITIVE LOGITS

 まとめ

0.79

 sowas

0.73

 utiles

0.70

XPS

0.70

 সবই

0.70

 கொண்டனர்

0.69

 vorgestellt

0.67

 umana

0.66

 нужна

0.66

Ils

0.65

Activations Density 3.206%