OpenAI 在 2024年3月 发布的内容

Video generation models as world simulators

OpenAI has developed Sora, a large-scale generative model capable of producing high-fidelity videos up to a minute long. Sora is a text-conditional diffusion model trained on both videos and images, l…