A PyTorch-native and Flexible Inference Engine with Hybrid Cache Acceleration and Parallelism for 🤗DiTs.
