4 ms·Generative vision model trains with high efficiency and breaks text CAPTCHAs3 points by kensai 9y ago