diffusers

[docs] Add a note on torchao/quanto benchmarks for CogVideoX and memory-efficient inference

#9296

Merged