Zhen Li

Z-Image Turbo++ achieves high-quality 2-step image generation that rivals 8-step models, transforming the landscape of diffusion distillation.

Z-Reward achieves 41.3% better human preference alignment in text-to-image generation by transforming complex reasoning into efficient score distributions.

Resource-constrained edge devices can achieve Pareto-optimal trade-offs between inference accuracy, latency, and energy consumption in federated learning by using a constrained multi-objective reinforcement learning approach.

Fine-tuning efficient few-step diffusion models no longer requires sacrificing their speed, thanks to a self-distillation approach that preserves inference capabilities.

Forget reward function dependencies – this new approach to contextual bandits with latent state dynamics achieves stronger regret bounds by directly modeling hidden state dependencies and adaptively estimating HMM parameters.

Pre-training with Dual Latent World Models unlocks significant performance gains in autonomous driving tasks by learning holistic Gaussian-centric representations.

You don't need 80B parameters to rival top-tier commercial image generators: Z-Image proves that a carefully optimized 6B model can deliver comparable performance with dramatically lower computational cost.