53 citations · 58 across the 14 of their papers we have counts for
11 papers · 1 filter
Demystifying On-Policy Distillation: Where Does the Teacher's Advantage Live?
Yiping Li, Fan Zhao, Zhibo Hou +2
On-policy distillation (OPD) is increasingly studied as a cheaper alternative to reinforcement learning, yet it remains a form of knowledge distillation (KD) that relies on teacher…
Vision: "What Don't I Know About User's Preferences?" How Uncertainty Estimation Improves Proactive Assistant Alignment
Zhiyu An, Ziyi Xuan, Yu Yang +1
Proactive Smart Assistants (PSAs) increasingly use general knowledge, learned patterns from group-level data, patterns from individual user's data, and recent sensor observations t…
Sanzaru: Benchmarking Protocol Formation under Epistemic & Communication Asymmetry
Yiping Li, Zhiyu An, Wan Du
A language model's ability to solve a task individually does not ensure that a team can solve it under unequal access to information and communication. SANZARU examines this gap th…
Neither True nor False: Do LLMs Represent Truth-Value Gaps in Plural Predication?
Duaa Nakshbandi, Zhiyu An, Yiping Li +2
Large Language Models (LLMs) are increasingly being used to summarize and verify claims in critical sources that inform consequential decisions. Such claims are full of plural defi…
Sparse-TD-JEPA: Improving Zero-Shot Reinforcement Learning With Sparse Latent Task Representations
Duaa Nakshbandi, Zhiyu An, Wan Du
Successor-feature-based zero-shot reinforcement learning methods pre-train a task encoder and successor-feature predictor on offline, reward-free data, then recover a policy at tes…
From Memorization to Absorption: Mixed-Policy RL for Continual Knowledge Injection
Zhibo Hou, Fan Zhao, Zhiyu An +1
Continual knowledge injection is essential for keeping large language models up-to-date in a fast-evolving world. Existing methods rely on supervised fine-tuning (SFT), which memor…
When Less Latent Leads to Better Relay: Information-Preserving Compression for Latent Multi-Agent LLM Collaboration
Yiping Li, Zhiyu An, Wan Du
Multi-agent LLM systems are moving beyond discrete-token messages toward richer relays that preserve internal state. Recent work such as LatentMAS transmits full key-value (KV) cac…
Open Problems in Differentiable Social Choice: Learning Mechanisms, Decisions, and Alignment
Zhiyu An, Wan Du
Social choice has become a foundational component of modern machine learning systems. From auctions and resource allocation to the alignment of large generative models, machine lea…
Differential Voting: Loss Functions For Axiomatically Diverse Aggregation of Heterogeneous Preferences
Zhiyu An, Duaa Nakshbandi, Wan Du
Reinforcement learning from human feedback (RLHF) implicitly aggregates heterogeneous human preferences into a single utility function, even though the underlying utilities of the…
DIML: Differentiable Inverse Mechanism Learning from Behaviors of Multi-Agent Learning Trajectories
Zhiyu An, Wan Du
We study inverse mechanism learning: recovering an unknown incentive-generating mechanism from observed strategic interaction traces of self-interested learning agents. Unlike inve…
Representational Homomorphism Predicts and Improves Compositional Generalization In Transformer Language Model
Zhiyu An, Wan Du
Compositional generalization-the ability to interpret novel combinations of familiar components-remains a persistent challenge for neural networks. Behavioral evaluations reveal \e…