Xiao Liu

Zhenghao Lin, Yeyun Gong, Xiao Liu et al. · microsoft-research

Knowledge distillation is an effective way to transfer knowledge from a strong teacher to an efficient student model. Ideally, we expect the better the teacher is, the better the student. However, this expectation does not always come true. It is common that a better teacher model results in a bad student via distillation due to the nonnegligible gap between teacher and student. To bridge the gap, we propose PROD, a PROgressive Distillation method, for dense retrieval. PROD consists of a teacher progressive distillation and a data progressive distillation to gradually improve the student. We conduct extensive experiments on five widely-used benchmarks, MS MARCO Passage, TREC Passage 19, TREC Document 19, MS MARCO Document and Natural Questions, where PROD achieves the state-of-the-art within the distillation methods for dense retrieval. The code and models will be released.

Wubing Chen, Wenbin Li, Xiao Liu et al.

Cooperative multi-agent policy gradient (MAPG) algorithms have recently attracted wide attention and are regarded as a general scheme for the multi-agent system. Credit assignment plays an important role in MAPG and can induce cooperation among multiple agents. However, most MAPG algorithms cannot achieve good credit assignment because of the game-theoretic pathology known as \textit{centralized-decentralized mismatch}. To address this issue, this paper presents a novel method, \textit{\underline{M}ulti-\underline{A}gent \underline{P}olarization \underline{P}olicy \underline{G}radient} (MAPPG). MAPPG takes a simple but efficient polarization function to transform the optimal consistency of joint and individual actions into easily realized constraints, thus enabling efficient credit assignment in MAPG. Theoretically, we prove that individual policies of MAPPG can converge to the global optimum. Empirically, we evaluate MAPPG on the well-known matrix game and differential game, and verify that MAPPG can converge to the global optimum for both discrete and continuous action spaces. We also evaluate MAPPG on a set of StarCraft II micromanagement tasks and demonstrate that MAPPG outperforms the state-of-the-art MAPG algorithms.

Xiao Liu, Shuyang Liu, Wenbin Li et al.

Although deep reinforcement learning has become a universal solution for complex control tasks, its real-world applicability is still limited because lacking security guarantees for policies. To address this problem, we propose Boundary Characterization via the Minimum Experience Retention (BCMER), an end-to-end Interpretable Policy Distillation (IPD) framework. Unlike previous IPD approaches, BCMER distinguishes the importance of experiences and keeps a minimal but critical experience pool with almost no loss of policy similarity. Specifically, the proposed BCMER contains two basic steps. Firstly, we propose a novel multidimensional hyperspheres intersection (MHI) approach to divide experience points into boundary points and internal points, and reserve the crucial boundary points. Secondly, we develop a nearest-neighbor-based model to generate robust and interpretable decision rules based on the boundary points. Extensive experiments show that the proposed BCMER is able to reduce the amount of experience to 1.4%~19.1% (when the count of the naive experiences is 10k) and maintain high IPD performance. In general, the proposed BCMER is more suitable for the experience storage limited regime because it discovers the critical experience and eliminates redundant experience.

Henry Zheng, Chenyue Fang, Rui Huang et al.

Vision-language models (VLMs) have achieved strong performance in multimodal understanding and reasoning, yet grounded reasoning in 3D scenes remains underexplored. Effective 3D reasoning hinges on accurate grounding: to answer open-ended queries, a model must first identify query-relevant objects and regions in a complex scene, and then reason about their spatial and geometric relationships. Recent approaches have demonstrated strong potential for grounded 3D reasoning. However, they often rely on in-domain tuning or hand-crafted reasoning pipelines, which limit their flexibility and zero-shot generalization to novel environments. In this work, we present MAG-3D, a training-free multi-agent framework for grounded 3D reasoning with off-the-shelf VLMs. Instead of relying on task-specific training or fixed reasoning procedures, MAG-3D dynamically coordinates expert agents to address the key challenges of 3D reasoning. Specifically, we propose a planning agent that decomposes the task and orchestrates the overall reasoning process, a grounding agent that performs free-form 3D grounding and relevant frame retrieval from extensive 3D scene observations, and a coding agent that conducts flexible geometric reasoning and explicit verification through executable programs. This multi-agent collaborative design enables flexible training-free 3D grounded reasoning across diverse scenes and achieves state-of-the-art performance on challenging benchmarks.

Xiao Liu, Wubing Chen, Mao Tan

Deep Reinforcement Learning (DRL) has achieved remarkable success in sequential decision-making problems. However, existing DRL agents make decisions in an opaque fashion, hindering the user from establishing trust and scrutinizing weaknesses of the agents. While recent research has developed Interpretable Policy Extraction (IPE) methods for explaining how an agent takes actions, their explanations are often inconsistent with the agent's behavior and thus, frequently fail to explain. To tackle this issue, we propose a novel method, Fidelity-Induced Policy Extraction (FIPE). Specifically, we start by analyzing the optimization mechanism of existing IPE methods, elaborating on the issue of ignoring consistency while increasing cumulative rewards. We then design a fidelity-induced mechanism by integrate a fidelity measurement into the reinforcement learning feedback. We conduct experiments in the complex control environment of StarCraft II, an arena typically avoided by current IPE methods. The experiment results demonstrate that FIPE outperforms the baselines in terms of interaction performance and consistency, meanwhile easy to understand.

Yu Gu, Yiheng Shu, Hao Yu et al. · microsoft-research, tsinghua

The applications of large language models (LLMs) have expanded well beyond the confines of text processing, signaling a new era where LLMs are envisioned as generalist agents capable of operating within complex environments. These environments are often highly expansive, making it impossible for the LLM to process them within its short-term memory. Motivated by recent research on extending the capabilities of LLMs with tools, we seek to investigate the intriguing potential of tools to augment LLMs in handling such complexity by introducing a novel class of tools, termed middleware, to aid in the proactive exploration within these massive environments. Such specialized tools can serve as a middleware layer shielding the LLM from environmental complexity. In two representative complex environments -- knowledge bases (KBs) and databases -- we demonstrate the significant potential of augmenting language agents with tools in complex environments. Notably, equipped with the middleware, GPT-4 achieves 2.8X the performance of the best baseline in tasks requiring access to database content and 2.2X in KB tasks. Our findings illuminate the path for advancing language agents in real-world applications.

Wenjie Huang, Jing Jiang, Xiao Liu

In this paper, novel gradient-based online learning algorithms are developed to investigate an important environmental application: real-time river pollution source identification, which aims at estimating the released mass, location, and time of a river pollution source based on downstream sensor data monitoring the pollution concentration. The pollution is assumed to be instantaneously released once. The problem can be formulated as a non-convex loss minimization problem in statistical learning, and our online algorithms have vectorized and adaptive step sizes to ensure high estimation accuracy in three dimensions which have different magnitudes. In order to keep the algorithm from stucking to the saddle points of non-convex loss, the escaping from saddle points module and multi-start setting are derived to further improve the estimation accuracy by searching for the global minimizer of the loss functions. This can be shown theoretically and experimentally as the $O(N)$ local regret of the algorithms and the high probability cumulative regret bound $O(N)$ under a particular error bound condition in loss functions. A real-life river pollution source identification example shows the superior performance of our algorithms compared to existing methods in terms of estimation accuracy. Managerial insights for the decision maker to use the algorithms are also provided.