Is your feature request related to a problem? Please describe.
I want to make some experiments about DiT based flow-matching model, I need an implementation of the common DiT block, but did not found it in both huggingface/diffusers and huggingface/transformers. Is there any implementation about it with just some other file names?
Describe the solution you'd like.
A clear DiT implementation
Describe alternatives you've considered.
Additional context.