Embedding Prediction Helps Image Generation
NEPA-DiT-XL predicts embeddings each denoising step and reaches ImageNet FID 1.32 with about one-third of REPA's compute.
3 stories, newest first · one card per story
NEPA-DiT-XL predicts embeddings each denoising step and reaches ImageNet FID 1.32 with about one-third of REPA's compute.
PixelDense separates semantic and geometric alignment teachers, improving pixel diffusion quality and training speed.
Structured residual paths in diffusion transformers cut FID and training iterations with almost no extra parameters.
Today 1.2K readers · 1.4K views · 189 reads · all time 203K views