arXiv subjects

Chao Yu

Publications and source records attributed to Chao Yu.

At least 73 records · Page 4Linked to original sources

arXiv2026

GLM-5: from Vibe Coding to Agentic Engineering

We present GLM-5, a next-generation foundation model designed to transition the paradigm of vibe coding to agentic engineering. Building upon the agentic, reasoning, and coding (ARC) capabilities of its predecessor, GLM-5 adopts DSA to significantly reduce training and inference costs while maintaining long-context fidelity. To advance model alignment and autonomy, we implement a new asynchronous reinforcement learning infrastructure that drastically improves post-training efficiency by decoupling generation from training. Furthermore, we propose novel asynchronous agent RL algorithms that further improve RL quality, enabling the model to learn from complex, long-horizon interactions more effectively. Through these innovations, GLM-5 achieves state-of-the-art performance on major open benchmarks. Most critically, GLM-5 demonstrates unprecedented capability in real-world coding tasks, surpassing previous baselines in handling end-to-end software engineering challenges. Code, models, and more information are available at https://github.com/zai-org/GLM-5.

cs.LG↗

arXiv2026

Translate Policy to Language: Flow Matching Generated Rewards for LLM Explanations

As humans increasingly share environments with diverse agents powered by RL, LLMs, and beyond, the ability to explain agent policies in natural language is vital for reliable coexistence. We introduce a general-purpose framework that trains explanation-generating LLMs via reinforcement learning from AI feedback, with distributional rewards generated by generative continuous normalizing flows (CNFs). CNFs capture the pluralistic and probabilistic nature of human judgments about explanations. Moreover, under mild assumptions, CNFs provably bound deviations from true human reward distributions when trained on noisy proxy rewards from LLMs. We design a specialized CNF architecture that selectively attends to linguistic cues in the decision context and explanations when generating rewards. Human and LLM evaluators find that our method delivers explanations that enable more accurate predictions of true agent decisions, exhibit greater logical soundness and actionability, and impose lower cognitive load than explanations trained with proxy LLM rewards or state-of-the-art RLHF and RLAIF baselines.

cs.CL↗

arXiv2026

MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMs

Developing Large Language Models (LLMs) to cooperate and compete effectively within multi-agent systems (MASs) is a critical step towards more advanced intelligence. While reinforcement learning (RL) has proven effective for enhancing reasoning in single-agent tasks, its extension to multi-turn, multi-agent scenarios remains underexplored due to the challenges of long-horizon credit assignment and agent-specific advantage estimation. To address these challenges, we introduce MARSHAL, an end-to-end RL framework that incentivizes Multi-Agent Reasoning through Self-play witH strAtegic LLMs in both cooperative and competitive games. MARSHAL features a turn-level advantage estimator that aligns learning signals with each interaction for credit assignment, and an agent-specific advantage normalization to stabilize multi-agent training. By learning with self-play across cooperative and competitive games, MARSHAL agents trained from Qwen3-4B develop strong strategic abilities, with up to 28.7% performance improvements in held-out games. More importantly, the capability acquired through self-play generalizes beyond games, yielding consistent performance gains of MASs in reasoning benchmarks. When integrated into leading MASs, our MARSHAL agent achieves significant zero-shot performance gains of up to 10.0% on AIME, 7.6% on GPQA-Diamond, and 3.5% on average across all benchmarks. These results establish self-play in strategic games as a powerful approach for developing generalizable multi-agent reasoning capabilities in LLMs.

cs.AI↗

arXiv2026

An Experimental Study on Fine-Grained Bistatic Sensing of UAV Trajectory via Cellular Downlink Signals

In this letter, a dual-bistatic unmanned aerial vehicles (UAVs) tracking system utilizing downlink Long-Term Evolution (LTE) signals is proposed and demonstrated. Particularly, two LTE base stations (BSs) are exploited as illumination sources. Two passive sensing receivers are deployed at different locations to detect the bistatic Doppler frequencies of the target UAV at different directions according to downlink signals transmitted from their corresponding BSs, such that the velocities of the UAV versus time can be estimated. Hence, the trajectories of the target UAV can be reconstructed. Although both the target UAV and the sensing receivers are around 200 meters away from the illuminating BSs, it is demonstrated by experiments that the tracking errors are below 50 centimeters for 90% of the complicated trajectories, when the distances between the UAV and sensing receivers are less than 30 meters. Note this accuracy is significantly better than the ranging resolution of LTE signals, high-accuracy trajectory tracking for UAV might be feasible via multi-angle bistatic Doppler measurements if the receivers are deployed with a sufficient density.

eess.SP↗

arXiv2026

$π_\texttt{RL}$: Online RL Fine-tuning for Flow-based Vision-Language-Action Models

Vision-Language-Action (VLA) models enable robots to understand and perform complex tasks from multimodal input. Although recent work explores using reinforcement learning (RL) to automate the laborious data collection process in scaling supervised fine-tuning (SFT), applying RL to large-scale flow-based VLAs (\eg, $π_0$, $π_{0.5}$) remains challenging due to intractable action log-likelihoods raised from flow matching. We address this challenge with $π_{\texttt{RL}}$, featuring two technical approaches: (1) \textbf{Flow-Noise} models the denoising process as a discrete-time MDP with a learnable noise network for exact log-likelihood computation. (2) \textbf{Flow-SDE} integrates denoising with agent-environment interaction, formulating a two-layer MDP that employs ODE-to-SDE conversion for efficient RL exploration. We evaluate $π_{\texttt{RL}}$ across various benchmarks, with experiments demonstrating that RL yields significant performance improvements in both in-distribution and out-of-distribution settings.

cs.LG↗

arXiv2026

Context-Picker: Dynamic context selection using multi-stage reinforcement learning

In long-context question answering, selecting the appropriate scope of context for a query remains a key and unresolved challenge. Insufficient context can lead to missing essential information, whereas excessive context often introduces noise and degrades answer quality. Conventional methods, such as retrieving a fixed number of passages or applying reranking, struggle to dynamically determine which context to include. This is especially problematic for factoid questions, which typically depend only on a few precise pieces of evidence. To overcome this limitation, we propose Context-Picker, a reasoning-aware framework that reframes context selection as the task of identifying a minimal sufficient evidence subset, moving beyond conventional similarity-based ranking. Context-Picker uses a human-inspired two-stage reinforcement learning schedule: stage 1 focuses on improving the recall rate of critical passages, and stage 2 prioritizes pruning redundancy to distill a compact evidence set. To resolve reward sparsity, we propose an offline evidence distillation pipeline that mines ``minimal sufficient sets" via a Leave-One-Out (LOO) procedure, providing dense and task-aligned supervision. Experiments on five long-context and multi-hop QA datasets demonstrate that our method outperforms strong RAG baselines and achieved higher answer accuracy. Ablation studies also indicate that our coarse-to-fine optimization schedule, the redundancy-aware reward shaping, along with the rationale generated by the policy, all contribute substantially to these gains.

cs.AI↗

arXiv2026

What Can RL Bring to VLA Generalization? An Empirical Study

Large Vision-Language Action (VLA) models have shown significant potential for embodied AI. However, their predominant training via supervised fine-tuning (SFT) limits generalization due to susceptibility to compounding errors under distribution shifts. Reinforcement learning (RL) offers a path to overcome these limitations by optimizing for task objectives via trial-and-error, yet a systematic understanding of its specific generalization benefits for VLAs compared to SFT is lacking. To address this, our study introduces a comprehensive benchmark for evaluating VLA generalization and systematically investigates the impact of RL fine-tuning across diverse visual, semantic, and execution dimensions. Our extensive experiments reveal that RL fine-tuning, particularly with PPO, significantly enhances generalization in semantic understanding and execution robustness over SFT, while maintaining comparable visual robustness. We identify PPO as a more effective RL algorithm for VLAs than LLM-derived methods like DPO and GRPO. We also develop a simple recipe for efficient PPO training on VLAs, and demonstrate its practical utility for improving VLA generalization. The project page is at https://rlvla.github.io

cs.LG↗

arXiv2026

JuggleRL: Mastering Ball Juggling with a Quadrotor via Deep Reinforcement Learning

Aerial robots interacting with objects must perform precise, contact-rich maneuvers under uncertainty. In this paper, we study the problem of aerial ball juggling using a quadrotor equipped with a racket, a task that demands accurate timing, stable control, and continuous adaptation. We propose JuggleRL, the first reinforcement learning-based system for aerial juggling. It learns closed-loop policies in large-scale simulation using systematic calibration of quadrotor and ball dynamics to reduce the sim-to-real gap. The training incorporates reward shaping to encourage racket-centered hits and sustained juggling, as well as domain randomization over ball position and coefficient of restitution to enhance robustness and transferability. The learned policy outputs mid-level commands executed by a low-level controller and is deployed zero-shot on real hardware, where an enhanced perception module with a lightweight communication protocol reduces delays in high-frequency state estimation and ensures real-time control. Experiments show that JuggleRL achieves an average of $311$ hits over $10$ consecutive trials in the real world, with a maximum of $462$ hits observed, far exceeding a model-based baseline that reaches at most $14$ hits with an average of $3.1$. Moreover, the policy generalizes to unseen conditions, successfully juggling a lighter $5$ g ball with an average of $145.9$ hits. This work demonstrates that reinforcement learning can empower aerial robots with robust and stable control in dynamic interaction tasks.

cs.RO↗

arXiv2026

SAC Flow: Sample-Efficient Reinforcement Learning of Flow-Based Policies via Velocity-Reparameterized Sequential Modeling

Training expressive flow-based policies with off-policy reinforcement learning is notoriously unstable due to gradient pathologies in the multi-step action sampling process. We trace this instability to a fundamental connection: the flow rollout is algebraically equivalent to a residual recurrent computation, making it susceptible to the same vanishing and exploding gradients as RNNs. To address this, we reparameterize the velocity network using principles from modern sequential models, introducing two stable architectures: Flow-G, which incorporates a gated velocity, and Flow-T, which utilizes a decoded velocity. We then develop a practical SAC-based algorithm, enabled by a noise-augmented rollout, that facilitates direct end-to-end training of these policies. Our approach supports both from-scratch and offline-to-online learning and achieves state-of-the-art performance on continuous control and robotic manipulation benchmarks, eliminating the need for common workarounds like policy distillation or surrogate objectives.

cs.RO↗

arXiv2026

ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning

We propose ReinFlow, a simple yet effective online reinforcement learning (RL) framework that fine-tunes a family of flow matching policies for continuous robotic control. Derived from rigorous RL theory, ReinFlow injects learnable noise into a flow policy's deterministic path, converting the flow into a discrete-time Markov Process for exact and straightforward likelihood computation. This conversion facilitates exploration and ensures training stability, enabling ReinFlow to fine-tune diverse flow model variants, including Rectified Flow [35] and Shortcut Models [19], particularly at very few or even one denoising step. We benchmark ReinFlow in representative locomotion and manipulation tasks, including long-horizon planning with visual input and sparse reward. The episode reward of Rectified Flow policies obtained an average net growth of 135.36% after fine-tuning in challenging legged locomotion tasks while saving denoising steps and 82.63% of wall time compared to state-of-the-art diffusion RL fine-tuning method DPPO [43]. The success rate of the Shortcut Model policies in state and visual manipulation tasks achieved an average net increase of 40.34% after fine-tuning with ReinFlow at four or even one denoising step, whose performance is comparable to fine-tuned DDIM policies while saving computation time for an average of 23.20%. Project webpage: https://reinflow.github.io/

cs.RO↗

arXiv2025

RLinf: Flexible and Efficient Large-scale Reinforcement Learning via Macro-to-Micro Flow Transformation

Reinforcement learning (RL) has demonstrated immense potential in advancing artificial general intelligence, agentic intelligence, and embodied intelligence. However, the inherent heterogeneity and dynamicity of RL workflows often lead to low hardware utilization and slow training on existing systems. In this paper, we present RLinf, a high-performance RL training system based on our key observation that the major roadblock to efficient RL training lies in system flexibility. To maximize flexibility and efficiency, RLinf is built atop a novel RL system design paradigm called macro-to-micro flow transformation (M2Flow), which automatically breaks down high-level, easy-to-compose RL workflows at both the temporal and spatial dimensions, and recomposes them into optimized execution flows. Supported by RLinf worker's adaptive communication capability, we devise context switching and elastic pipelining to realize M2Flow transformation, and a profiling-guided scheduling policy to generate optimal execution plans. Extensive evaluations on both reasoning RL and embodied RL tasks demonstrate that RLinf consistently outperforms state-of-the-art systems, achieving $1.07\times-2.43\times$ speedup in end-to-end training throughput.

cs.LG↗

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↗