arXiv subjects

Chao Yu

Publications and source records attributed to Chao Yu.

At least 73 records · Page 4Linked to original sources

arXiv2025

FUSCO: High-Performance Distributed Data Shuffling via Transformation-Communication Fusion

Large-scale Mixture-of-Experts (MoE) models rely on \emph{expert parallelism} for efficient training and inference, which splits experts across devices and necessitates distributed data shuffling to route each token to its assigned experts. However, existing communication libraries handle this shuffling poorly; its overhead can account for over half of end-to-end runtime. We present FUSCO, an MoE-friendly communication library that achieves efficient and lightweight data shuffling through fused data transformation and communication, based on the key observation that MoE's expert-major data layout conflicts with the device-major layout expected by communication operations. FUSCO captures the fine-grained data layout, which is then interpreted by a pipelined communication engine that performs the required shuffling efficiently along the communication path. Lightweight planning and load-balancing mechanisms complement the engine by eliminating redundant communication and dispersing traffic. Evaluations on representative benchmarks illustrate that FUSCO achieves up to 3.84$\times$ and 2.01$\times$ speedups over NCCL and DeepEP (the state-of-the-art MoE communication library), respectively. In end-to-end MoE tasks, compared to NCCL and DeepEP, FUSCO reduces the training latency by 1.17-1.39$\times$ and 1.10-1.19$\times$, and lowers the first-token generation latency in inference by 1.09-1.25$\times$ and 1.06-1.16$\times$.

cs.DC↗

arXiv2025

Automatic Reward Shaping from Multi-Objective Human Heuristics

Designing effective reward functions remains a central challenge in reinforcement learning, especially in multi-objective environments. In this work, we propose Multi-Objective Reward Shaping with Exploration (MORSE), a general framework that automatically combines multiple human-designed heuristic rewards into a unified reward function. MORSE formulates the shaping process as a bi-level optimization problem: the inner loop trains a policy to maximize the current shaped reward, while the outer loop updates the reward function to optimize task performance. To encourage exploration in the reward space and avoid suboptimal local minima, MORSE introduces stochasticity into the shaping process, injecting noise guided by task performance and the prediction error of a fixed, randomly initialized neural network. Experimental results in MuJoCo and Isaac Sim environments show that MORSE effectively balances multiple objectives across various robotic tasks, achieving task performance comparable to those obtained with manually tuned reward functions.

cs.LG↗

arXiv2025

RoboScape-R: Unified Reward-Observation World Models for Generalizable Robotics Training via RL

Achieving generalizable embodied policies remains a key challenge. Traditional policy learning paradigms, including both Imitation Learning (IL) and Reinforcement Learning (RL), struggle to cultivate generalizability across diverse scenarios. While IL policies often overfit to specific expert trajectories, RL suffers from the inherent lack of a unified and general reward signal necessary for effective multi-scene generalization. We posit that the world model is uniquely capable of serving as a universal environment proxy to address this limitation. However, current world models primarily focus on their ability to predict observations and still rely on task-specific, handcrafted reward functions, thereby failing to provide a truly general training environment. Toward this problem, we propose RoboScape-R, a framework leveraging the world model to serve as a versatile, general-purpose proxy for the embodied environment within the RL paradigm. We introduce a novel world model-based general reward mechanism that generates ''endogenous'' rewards derived from the model's intrinsic understanding of real-world state transition dynamics. Extensive experiments demonstrate that RoboScape-R effectively addresses the limitations of traditional RL methods by providing an efficient and general training environment that substantially enhances the generalization capability of embodied policies. Our approach offers critical insights into utilizing the world model as an online training strategy and achieves an average 37.5% performance improvement over baselines under out-of-domain scenarios.

cs.RO↗

arXiv2025

AED: Automatic Discovery of Effective and Diverse Vulnerabilities for Autonomous Driving Policy with Large Language Models

Assessing the safety of autonomous driving policy is of great importance, and reinforcement learning (RL) has emerged as a powerful method for discovering critical vulnerabilities in driving policies. However, existing RL-based approaches often struggle to identify vulnerabilities that are both effective-meaning the autonomous vehicle is genuinely responsible for the accidents-and diverse-meaning they span various failure types. To address these challenges, we propose AED, a framework that uses large language models (LLMs) to automatically discover effective and diverse vulnerabilities in autonomous driving policies. We first utilize an LLM to automatically design reward functions for RL training. Then we let the LLM consider a diverse set of accident types and train adversarial policies for different accident types in parallel. Finally, we use preference-based learning to filter ineffective accidents and enhance the effectiveness of each vulnerability. Experiments across multiple simulated traffic scenarios and tested policies show that AED uncovers a broader range of vulnerabilities and achieves higher attack success rates compared with expert-designed rewards, thereby reducing the need for manual reward engineering and improving the diversity and effectiveness of vulnerability discovery. The implementation can be found on: https://github.com/thu-nics/AED .

cs.CR↗

arXiv2025

Adapting Like Humans: A Metacognitive Agent with Test-time Reasoning

Recent Vision-Language Models (VLMs) exhibit strong perceptual reasoning abilities, yet they often struggle to adapt efficiently when encountering novel tasks at test time. In contrast, humans leverage the metacognitive model with memory, enabling continuous strategy refinement through metacognitive control when faced with new challenges. To bridge this gap, we propose metacognitive test-time reasoning (MCTR), a framework that equips models with the ability to learn, adapt, and improve during test time through metacognitive self-updating. Inspired by the dual structure of human metacognition, MCTR comprises meta-level and object-level VLM reasoning modules, each equipped with dedicated memory systems for hierarchical adaptive reasoning. Specifically, MCTR consists of (1) a meta-reasoning module which incrementally builds a structured memory by discovering and storing task-relevant rules, environmental patterns, and action-outcome relationships from test-time observations as natural language descriptions; and (2) an action-reasoning module that determines optimal actions through context-aware perception and strategic reasoning by dynamically retrieving and integrating knowledge from memory. The action-reasoning module continuously updates its policy through proposed metacognitive test-time reinforcement learning, adapting as knowledge memory evolves. We evaluate MCTR on 45 Atari games (33 seen, 12 unseen). MCTR demonstrates robust test-time adaptation, achieving 9/12 top-1 results on unseen games compared with baselines. Analyses through ablations, learning dynamics, and case studies reveal the complementary contributions of both components and show meta-reasoning evolving toward human-like adaptation strategies.

cs.AI↗

arXiv2025

Extending Test-Time Scaling: A 3D Perspective with Context, Batch, and Turn

Reasoning reinforcement learning (RL) has recently revealed a new scaling effect: test-time scaling. Thinking models such as R1 and o1 improve their reasoning accuracy at test time as the length of the reasoning context increases. However, compared with training-time scaling, test-time scaling is fundamentally limited by the limited context length of base models, which remains orders of magnitude smaller than the amount of tokens consumed during training. We revisit test-time enhancement techniques through the lens of scaling effect and introduce a unified framework of multi-dimensional test-time scaling to extend the capacity of test-time reasoning. Beyond conventional context-length scaling, we consider two additional dimensions: batch scaling, where accuracy improves with parallel sampling, and turn scaling, where iterative self-refinement enhances reasoning quality. Building on this perspective, we propose 3D test-time scaling, which integrates context, batch, and turn scaling. We show that: (1) each dimension demonstrates a test-time scaling effect, but with a bounded capacity; (2) combining all three dimensions substantially improves the reasoning performance of challenging testbeds, including IOI, IMO, and CPHO, and further benefits from human preference feedback; and (3) the human-in-the-loop framework naturally extends to a more open-ended domain, i.e., embodied learning, which enables the design of humanoid control behaviors.

cs.LG↗

arXiv2025

A Survey on Self-play Methods in Reinforcement Learning

Self-play, a learning paradigm where agents iteratively refine their policies by interacting with historical or concurrent versions of themselves or other evolving agents, has shown remarkable success in solving complex non-cooperative multi-agent tasks. Despite its growing prominence in multi-agent reinforcement learning (MARL), such as Go, poker, and video games, a comprehensive and structured understanding of self-play remains lacking. This survey fills this gap by offering a comprehensive roadmap to the diverse landscape of self-play methods. We begin by introducing the necessary preliminaries, including the MARL framework and basic game theory concepts. Then, it provides a unified framework and classifies existing self-play algorithms within this framework. Moreover, the paper bridges the gap between the algorithms and their practical implications by illustrating the role of self-play in different non-cooperative scenarios. Finally, the survey highlights open challenges and future research directions in self-play.

cs.AI↗

arXiv2025

EARL: Efficient Agentic Reinforcement Learning Systems for Large Language Models

Reinforcement learning (RL) has become a pivotal component of large language model (LLM) post-training, and agentic RL extends this paradigm to operate as agents through multi-turn interaction and tool use. Scaling such systems exposes two practical bottlenecks: (1) context length grows rapidly during training, inflating memory usage and latency, and triggering out-of-memory (OOM) failures; and (2) intermediate tensors accumulate with context length, making cross-device data movement a major system bottleneck. We present EARL, a scalable system for efficient agentic RL. EARL designs a parallelism selector that dynamically adapts model and training parallelism across RL stages based on sequence length and system load, and a data dispatcher that performs layout-aware, decentralized exchange of intermediate data batches. Together, these components increase throughput, reduce long-context failures, and enable stable large-scale training of agentic LLMs without relying on hard limits or penalties of context length.

cs.DC↗

arXiv2025

Doppler-Based Multistatic Drone Tracking via Cellular Downlink Signals

In this paper, a multistatic Doppler sensing system is proposed for the drone tracking via downlink Long-Term Evolution (LTE) signals. Specifically, the LTE base stations (BSs) are exploited as signal illuminators, and three passive sensing receivers are deployed at different locations to detect the bistatic Doppler frequencies of a target drone from received downlink signals. It is shown that even without the measurements of BS-drone-receiver range and angle, the Doppler measurements could provide sufficient information for trajectory tracking. Particularly, the trajectory of the target drone, consisting of the initial position and velocities of all the time slots, can be reconstructed by solving a minimum mean-squared error problem according to the above Doppler measurements. It is demonstrated by experiment that although the target drone and all the sensing receivers are around 200 meters away from the illuminating BSs, the complicated trajectories can be tracked with 90% errors below 90 centimeters. Since this accuracy is notably higher than the typical range resolution of LTE signals, the demonstration shows that drone trajectory tracking with a high accuracy could be feasible solely according to Doppler detection, as long as the deployment density of receivers is sufficiently high.

eess.SP↗

arXiv2025

Observation of non-Hermitian topology in cold Rydberg quantum gases

The pursuit of topological phenomena in non-Hermitian systems has unveiled new physics beyond the conventional Hermitian paradigm, yet their realization in interacting many-body platforms remains a critical challenge. Exploring this interplay is essential to understand how strong interactions and dissipation collectively shape topological phases in open quantum systems. Here, we experimentally demonstrate non-Hermitian spectra topology in a dissipative Rydberg atomic gas and characterize parameters-dependent winding numbers. By increasing the interaction strength, the system evolves from Hermitian to non-Hermitian regime, accompanying emergence of trajectory loop in the complex energy plane. As the scanning time is varied, the spectra topology becomes twisted in the complex energy plane manifesting as a topology phase transition with the sign winding number changed. When preparing the system in different initial states, we can access a nontrivial fractional phase within a parameter space that globally possesses an integer winding. Furthermore, by changing the scanning direction, we observe the differentiated loops, revealing the breaking of chirality symmetry. This work establishes cold Rydberg gases as a versatile platform for exploring the rich interplay between non-Hermitian topology, strong interactions, and dissipative quantum dynamics.

cond-mat.quant-gas↗

arXiv2025

Fine-tuning Diffusion Policies with Backpropagation Through Diffusion Timesteps

Diffusion policies, widely adopted in decision-making scenarios such as robotics, gaming and autonomous driving, are capable of learning diverse skills from demonstration data due to their high representation power. However, the sub-optimal and limited coverage of demonstration data could lead to diffusion policies that generate sub-optimal trajectories and even catastrophic failures. While reinforcement learning (RL)-based fine-tuning has emerged as a promising solution to address these limitations, existing approaches struggle to effectively adapt Proximal Policy Optimization (PPO) to diffusion models. This challenge stems from the computational intractability of action likelihood estimation during the denoising process, which leads to complicated optimization objectives. In our experiments starting from randomly initialized policies, we find that online tuning of Diffusion Policies demonstrates much lower sample efficiency compared to directly applying PPO on MLP policies (MLP+PPO). To address these challenges, we introduce NCDPO, a novel framework that reformulates Diffusion Policy as a noise-conditioned deterministic policy. By treating each denoising step as a differentiable transformation conditioned on pre-sampled noise, NCDPO enables tractable likelihood evaluation and gradient backpropagation through all diffusion timesteps. Our experiments demonstrate that NCDPO achieves sample efficiency comparable to MLP+PPO when training from scratch, outperforming existing methods in both sample efficiency and final performance across diverse benchmarks, including continuous robot control and multi-agent game scenarios. Furthermore, our experimental results show that our method is robust to the number denoising timesteps in the Diffusion Policy.

cs.LG↗

arXiv2025

Observation of Discrete Time Quasicrystal in Rydberg Atomic Gases

Discrete time quasicrystals (DTQC) constitute a class of non-equilibrium matter characterized by temporal order without strict periodicity, in contrast to conventional time crystals. Investigating these phenomena is essential for expanding our fundamental understanding of far-from-equilibrium quantum matter and spontaneous symmetry breaking beyond periodic regimes. Here, we experimentally observe a DTQC in a driven-dissipative ensemble of strongly interacting Rydberg atoms, displaying non-equilibrium dynamical response with a different finite Abelian group symmetry $\mathbb{Z}{_m} \times \mathbb{Z}{_n}$. By applying a quasi-periodic drive using a dual-frequency drive with incommensurate frequencies, we demonstrate that the system exhibits a robust subharmonic response at multiple incommensurate frequencies, signifying the emergence of a DTQC phase. We map the full phase diagram of the system, which includes the DTQC phase, and demonstrated its rigidity against perturbations in both RF field intensity and laser detuning. Moreover, we observe a cyclic group symmetry effect that constrains the construction of $\mathbb{Z}{_2} \times \mathbb{Z}{_3}$-symmetric DTQC. This work establishes a versatile platform for studying non-equilibrium phases of matter and provides insights into the dynamics of time-translation symmetry breaking in quantum many-body systems.

cond-mat.quant-gas↗

arXiv2025

Spec-VLA: Speculative Decoding for Vision-Language-Action Models with Relaxed Acceptance

Vision-Language-Action (VLA) models have made substantial progress by leveraging the robust capabilities of Visual Language Models (VLMs). However, VLMs' significant parameter size and autoregressive (AR) decoding nature impose considerable computational demands on VLA models. While Speculative Decoding (SD) has shown efficacy in accelerating Large Language Models (LLMs) by incorporating efficient drafting and parallel verification, allowing multiple tokens to be generated in one forward pass, its application to VLA models remains unexplored. This work introduces Spec-VLA, an SD framework designed to accelerate VLA models. Due to the difficulty of the action prediction task and the greedy decoding mechanism of the VLA models, the direct application of the advanced SD framework to the VLA prediction task yields a minor speed improvement. To boost the generation speed, we propose an effective mechanism to relax acceptance utilizing the relative distances represented by the action tokens of the VLA model. Empirical results across diverse test scenarios affirm the effectiveness of the Spec-VLA framework, and further analysis substantiates the impact of our proposed strategies, which enhance the acceptance length by 44%, achieving 1.42 times speedup compared with the OpenVLA baseline, without compromising the success rate. The success of the Spec-VLA framework highlights the potential for broader application of speculative execution in VLA prediction scenarios.

cs.LG↗

arXiv2025

H$^2$R: Hierarchical Hindsight Reflection for Multi-Task LLM Agents

Large language model (LLM)-based agents have shown strong potential in multi-task scenarios, owing to their ability to transfer knowledge across diverse tasks. However, existing approaches often treat prior experiences and knowledge as monolithic units, leading to inefficient and coarse-grained knowledge transfer. In this work, we propose a novel hierarchical memory architecture that enables fine-grained knowledge transfer by decoupling high-level planning memory from low-level execution memory. To construct and refine these hierarchical memories, we introduce Hierarchical Hindsight Reflection (H$^2$R), a mechanism that distills reusable and hierarchical knowledge from past agent-environment interactions. At test time, H$^2$R performs retrievals of high-level and low-level memories separately, allowing LLM-based agents to efficiently access and utilize task-relevant knowledge for new tasks.Experimental results across two benchmarks demonstrate that H$^2$R can improve generalization and decision-making performance, outperforming prior baselines such as Expel.

cs.AI↗

arXiv2025

Toward Real-World Cooperative and Competitive Soccer with Quadrupedal Robot Teams

Achieving coordinated teamwork among legged robots requires both fine-grained locomotion control and long-horizon strategic decision-making. Robot soccer offers a compelling testbed for this challenge, combining dynamic, competitive, and multi-agent interactions. In this work, we present a hierarchical multi-agent reinforcement learning (MARL) framework that enables fully autonomous and decentralized quadruped robot soccer. First, a set of highly dynamic low-level skills is trained for legged locomotion and ball manipulation, such as walking, dribbling, and kicking. On top of these, a high-level strategic planning policy is trained with Multi-Agent Proximal Policy Optimization (MAPPO) via Fictitious Self-Play (FSP). This learning framework allows agents to adapt to diverse opponent strategies and gives rise to sophisticated team behaviors, including coordinated passing, interception, and dynamic role allocation. With an extensive ablation study, the proposed learning method shows significant advantages in the cooperative and competitive multi-agent soccer game. We deploy the learned policies to real quadruped robots relying solely on onboard proprioception and decentralized localization, with the resulting system supporting autonomous robot-robot and robot-human soccer matches on indoor and outdoor soccer courts.

cs.RO↗

arXiv2025

D3P: Dynamic Denoising Diffusion Policy via Reinforcement Learning

Diffusion policies excel at learning complex action distributions for robotic visuomotor tasks, yet their iterative denoising process poses a major bottleneck for real-time deployment. Existing acceleration methods apply a fixed number of denoising steps per action, implicitly treating all actions as equally important. However, our experiments reveal that robotic tasks often contain a mix of \emph{crucial} and \emph{routine} actions, which differ in their impact on task success. Motivated by this finding, we propose \textbf{D}ynamic \textbf{D}enoising \textbf{D}iffusion \textbf{P}olicy \textbf{(D3P)}, a diffusion-based policy that adaptively allocates denoising steps across actions at test time. D3P uses a lightweight, state-aware adaptor to allocate the optimal number of denoising steps for each action. We jointly optimize the adaptor and base diffusion policy via reinforcement learning to balance task performance and inference efficiency. On simulated tasks, D3P achieves an averaged 2.2$\times$ inference speed-up over baselines without degrading success. Furthermore, we demonstrate D3P's effectiveness on a physical robot, achieving a 1.9$\times$ acceleration over the baseline.

cs.RO↗

arXiv2025

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd overall among all evaluated models and 2nd on agentic benchmarks. We release both GLM-4.5 (355B parameters) and a compact version, GLM-4.5-Air (106B parameters), to advance research in reasoning and agentic AI systems. Code, models, and more information are available at https://github.com/zai-org/GLM-4.5.

cs.CL↗

arXiv2025

CityLight: A Neighborhood-inclusive Universal Model for Coordinated City-scale Traffic Signal Control

City-scale traffic signal control (TSC) involves thousands of heterogeneous intersections with varying topologies, making cooperative decision-making across intersections particularly challenging. Given the prohibitive computational cost of learning individual policies for each intersection, some researchers explore learning a universal policy to control each intersection in a decentralized manner, where the key challenge is to construct a universal representation method for heterogeneous intersections. However, existing methods are limited to universally representing information of heterogeneous ego intersections, neglecting the essential representation of influence from their heterogeneous neighbors. Universally incorporating neighborhood information is nontrivial due to the intrinsic complexity of traffic flow interactions, as well as the challenge of modeling collective influences from neighbor intersections. To address these challenges, we propose CityLight, which learns a universal policy based on representations obtained with two major modules: a Neighbor Influence Encoder to explicitly model neighbor's influence with specified traffic flow relation and connectivity to the ego intersection; a Neighbor Influence Aggregator to attentively aggregate the influence of neighbors based on their mutual competitive relations. Extensive experiments on five city-scale datasets, ranging from 97 to 13,952 intersections, confirm the efficacy of CityLight, with an average throughput improvement of 11.68% and a lift of 22.59% for generalization.

eess.SY↗