Lin Xiao

Research Scientist@Facebook·美国·OpenReview

研究方向

deep learning algorithms · generative adversary networks · reinforcement learning · nonconvex optimization · variance reduction · convex optimization · stochastic gradient methods · randomized algorithms

3.5

Smoothing DiLoCo with Primal Averaging for Faster Training of LLMs

ICLR 2026Rejected

通讯

7.8

Exploration from a Primal-Dual Lens: Value-Incentivized Actor-Critic Methods for Sample-Efficient Online RL

NeurIPS 2025Poster

三作

7.3

Incentivize without Bonus: Provably Efficient Model-based Online Multi-agent RL for Markov Games

合作者 (20)

Lin Xiao

Smoothing DiLoCo with Primal Averaging for Faster Training of LLMs

Exploration from a Primal-Dual Lens: Value-Incentivized Actor-Critic Methods for Sample-Efficient Online RL

ParetoQ: Improving Scaling Laws in Extremely Low-bit LLM Quantization

LoRe: Personalizing LLMs via Low-Rank Reward Modeling

PARQ: Piecewise-Affine Regularized Quantization

BCOS: A Method for Stochastic Approximation

Incentivize without Bonus: Provably Efficient Model-based Online Multi-agent RL for Markov Games