Xu

Forget expensive compression trials – a simple spectral statistic can accurately predict how much your LLM will degrade *before* you even compress it.

Fusing MPC with RL yields safer and more efficient autonomous driving at intersections, outperforming both standalone MPC and end-to-end RL, and surprisingly generalizing better to new scenarios.

Run code LLMs 10x faster and with 6x less memory on your laptop without sacrificing accuracy, thanks to a new quantization and compilation approach.

Dataset distillation can be sped up by 18x on ImageNet-1K without sacrificing accuracy by focusing optimization on high-loss regions.