Generative Representation Learning for Scalable Latent Tokenization
A generative tokenizer avoids representation collapse and reaches FID-50K 2.18 with JiT-L.
9 stories, newest first · one card per story
A generative tokenizer avoids representation collapse and reaches FID-50K 2.18 with JiT-L.
BudgetPix adapts image-diffusion token budgets to visual complexity, matching larger models at a fraction of compute.
Joint path-flow training with a stochastic regularizer improves ImageNet-256 FID without changing inference.
LiFT loops a shared Diffusion Transformer and lowers ImageNet FID while cutting parameters and FLOPs.
Persistent representation learning improves one-step Drifting Models, cutting pixel-space FID by about 82-95%.
DMAD distills few-step image and video generators via adversarial distribution matching, reporting top compared scores.
Improved distributional diffusion models reach 2.38 FID on ImageNet-256 with one from-scratch DiT.
Persistence Forcing uses specialized diffusion features to reach FID 1.63 on ImageNet 256.
Structured residual paths in diffusion transformers cut FID and training iterations with almost no extra parameters.
Today 1.2K readers · 1.4K views · 193 reads · all time 203K views