CogVideo: Large-Scale Pretraining for Text-to-Video Generation via Transformersgithub.com/THUDM9 pointslnyan4 years ago