Towards End-to-End Generative Modeling of Long Videos With Memory-Efficient Bidirectional Transformers | Read Paper on Bytez