INDEX

Explanations

forms of "to be"

The neuron fires on moderately common, mid-frequency words (e.g. content and function words that occur often but aren’t the highest-frequency tokens).

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

 होतात

0.75

 હંમે

0.72

 हमेशा

0.70

 alır

0.67

になります

0.66

 får

0.65

 ہمیشہ

0.64

 пройдет

0.64

可以让

0.63

 станет

0.63

POSITIVE LOGITS

 đang

2.52

 sedang

2.36

正在

2.28

กำลัง

1.98

 estamos

1.87

 estão

1.81

 estás

1.79

 están

1.72

 estoy

1.70

 está

1.67

Activations Density 1.264%