Guozheng Li

Pretrained 3D Gaussian scenes can be compressed 100-fold post-hoc without any per-scene optimization, beating existing simplification baselines by 1.3 dB PSNR at 12x the speed.

Leveraging hidden states from reward models can boost RLHF performance by over 6% on challenging benchmarks, transforming how we utilize reward signals.

Tactile feedback can transform robotic manipulation, enabling real-time adaptation that boosts success rates in challenging contact-rich tasks.

Forget tedious poster design – iPoster lets you sketch your vision and then uses a smart diffusion model to instantly generate polished, content-aware layouts that respect your constraints.