26 papers
Pin Once, Swap Light: Subspace-Aligned Centroid-Residual Training for Efficient Ultra-LoRA Serving
Xiang Li, Pengcheng Wang, Huazheng Wang +1
Modern multi-tenant Low-Rank Adapters (LoRAs) serving systems concurrently host tens to hundreds of LoRA adapters. Though powerful, this introduces a critical system dilemma betwee…
From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement
Qinsi Wang, Jing Shi, Huazheng Wang +8
Reinforcement Learning with Verifiable Rewards (RLVR) has driven recent progress in reasoning-oriented large language models (LLMs) by enabling large-scale optimization. However, i…
Online KL-Regularized Reinforcement Learning with Function Approximation under Misspecification
Haoyang Hong, Zichen Wang, Quanquan Gu +1
We study KL-regularized contextual bandits and episodic reinforcement learning (RL) under general function approximation with model misspecification. Existing guarantees rely on re…
Who&When Pro: Can LLMs Really Attribute Failures in AI Agents?
Jiale Liu, Huajun Xi, Shaokun Zhang +6
Automated failure attribution uses LLMs to identify where and why agentic systems fail. As agents become more capable, their failures become subtler, making automated attribution i…
When Can You Poison Rewards? A Tight Characterization of Reward Poisoning in Linear MDPs
Jose Efraim Aguilar Escamilla, Haoyang Hong, Jiawei Li +4
We study reward poisoning attacks in reinforcement learning (RL), where an adversary manipulates rewards within constrained budgets to force the target RL agent to adopt a policy t…
Speculative Pipeline Decoding: Higher-Accuracy Drafting with Hidden Latency via Pipeline Parallelism
Yijiong Yu, Huazheng Wang, Shuai Yuan +2
Speculative Decoding (SD) accelerates low-concurrency LLM inference with a draft-then-verify paradigm. Mainstream methods, however, rely on multi-token prediction, which incurs com…