Author

Jonathan Light

Observed name: Light, Jonathan

Institutions: Microsoft (United States), University of Illinois Chicago, University of Chicago, Microsoft Research Montréal (Canada), Rush University

Observed institutions: Microsoft (United States), Rensselaer Polytechnic Institute, University of Illinois Chicago, University of Chicago, Microsoft Research Montréal (Canada), Rush University

OpenAlex ID: A5107672050

DISC: Dynamic Decomposition Improves LLM Inference Scaling

2025 · Light, Jonathan, Cheng, Wei, et al. · arXiv (Cornell University) · 7

On the Effect of Sampling Diversity in Scaling LLM Inference

2025 · Wang, Tianchun, Liu, Zichuan, et al. · arXiv (Cornell University) · 7

TDRM: Smooth Reward Models with Temporal Difference for LLM RL and Inference

2025 · Zhang, Dan, Cai, Min, et al. · arXiv (Cornell University) · 3

Scattered Forest Search: Smarter Code Space Exploration with LLMs

2024 · Light, Jonathan, Wu, Yue, et al. · arXiv (Cornell University) · 2

Strategist: Self-improvement of LLM Decision Making via Bi-Level Tree Search

2024 · Light, Jonathan, Cai, Min, et al. · arXiv (Cornell University) · 2

Shockingly Simple Self-retrospection Improves Agentic Models Without RL

2026 · Light, Jonathan, Cui, Christopher Zhang, et al. · arXiv (Cornell University)

KV-streams for Efficient Compaction in Agentic Reinforcement Learning

2026 · Penaloza, Emiliano, Malenfant, Dane, et al. · arXiv (Cornell University)

FrogNano: Training a 4B Coding Agent via Online Task Synthesis

2026 · Kim, Minseon, Shi, Zhengyan, et al. · arXiv (Cornell University)

Actor-Curator: Co-adaptive Curriculum Learning via Policy-Improvement Bandits for RL Post-Training

2026 · Gu, Zhengyao, Light, Jonathan, et al. · arXiv (Cornell University)

PIANIST: Learning Partially Observable World Models with LLMs for Multi-Agent Decision Making

2024 · Light, Jonathan, Xing, Sixue, et al. · arXiv (Cornell University)

Works count: 10

Citation count: 3

H-index: 1

I10-index: