10 papers
A Few Teacher Steps Go a Long Way: Cost-Efficient On-Policy Data Augmentation for Agent Post-Training
Junze Ye, Jiayi Cheng, Miao Lu +3
For LLM agents, supervised fine-tuning is not only about teacher labels' quality, but also about which interaction contexts those labels condition on. Pure behavioral cloning uses…
Calibrating Conservatism for Scalable Oversight
William Overman, Mohsen Bayati
Agentic AI systems capable of autonomous planning and extended environmental interaction pose a fundamental control problem: how can humans maintain meaningful oversight of systems…
Annealed Softmax Greedy in Many-Armed Bayesian Bandits
William Overman, Mohsen Bayati
Reinforcement learning with verifiable rewards and group-based policy optimization methods update a stochastic policy by sampling multiple completions per prompt and increasing the…
Text-to-Distribution Prediction with Quantile Tokens and Neighbor Context
Yilun Zhu, Yuan Zhuang, Nikhita Vedula +6
Many applications of LLM-based text regression require predicting a full conditional distribution rather than a single point value. We study distributional regression under empiric…
The Oversight Game: Learning to Cooperatively Balance an AI Agent's Safety and Autonomy
William Overman, Mohsen Bayati
As increasingly capable agents are deployed, a central safety challenge is how to retain meaningful human control without modifying the underlying system. We study a minimal contro…
On Aligning Prediction Models with Clinical Experiential Learning: A Prostate Cancer Case Study
Jacqueline J. Vallon, William Overman, Wanqiao Xu +11
Over the past decade, the use of machine learning (ML) models in healthcare applications has rapidly increased. Despite high performance, modern ML models do not always capture pat…
Scaling Clinician-Grade Feature Generation from Clinical Notes with Multi-Agent Language Models
Jiayi Wang, Jacqueline Jil Vallon, Nikhil V. Kotha +8
Developing accurate clinical prediction models is often bottlenecked by the difficulty of deriving meaningful structured features from unstructured EHR notes, a process that tradit…
Conformal Arbitrage: Risk-Controlled Balancing of Competing Objectives in Language Models
William Overman, Mohsen Bayati
Modern language model deployments must often balance competing objectives, for example, helpfulness versus harmlessness, cost versus accuracy, and reward versus safety. We introduc…
Aligning Model Properties via Conformal Risk Control
William Overman, Jacqueline Jil Vallon, Mohsen Bayati
AI model alignment is crucial due to inadvertent biases in training data and the underspecified machine learning pipeline, where models with excellent test metrics may not meet end…
A Probabilistic Approach for Model Alignment with Human Comparisons
Junyu Cao, Mohsen Bayati
A growing trend involves integrating human knowledge into learning frameworks, leveraging subtle human feedback to refine AI models. While these approaches have shown promising res…