horseee / learning-to-cache

[NeurIPS 2024] Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching
71 stars 5 forks source link