Sora: first impressions

We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios. We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of…

We are launching a new generation of embedding models, new GPT-4 Turbo and moderation models, new API usage management tools, and soon, lower pricing on GPT-3.5 Turbo. Post Views: 44

In many areas of natural language processing, including language interpretation and natural language synthesis, large-scale training of machine learning models utilizing transformer topologies has produced ground-breaking advances. The widely acknowledged behavior of these systems is their ability to stably scale or to continue to perform better as the number of model parameters and the volume…

Post Content Post Views: 63

We’re developing a blueprint for evaluating the risk that a large language model (LLM) could aid someone in creating a biological threat. In an evaluation involving both biology experts and students, we found that GPT-4 provides at most a mild uplift in biological threat creation accuracy. While this uplift is not large enough to be conclusive,…

We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios. We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of…