Fetching the paper…

Lformer: Text-to-Image Generation with L-shape Block Parallel Decoding · Around