INDEX

Explanations

words followed by common endings

The neuron responds to capitalized words—especially proper names and other title‐cased tokens.

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

-2.84

-2.05

↵↵

-1.99

of

-1.87

-1.78

like

-1.73

all

-1.67

-1.64

 acudir

-1.59

ஂ

-1.55

POSITIVE LOGITS

蟜

2.03

хіво

1.92

]")]

1.90

翃

1.90

}$.

1.84

ANDUM

1.84

ppuden

1.84

yskland

1.83

antMatchers

1.82

>'+

1.81

Activations Density 0.064%