arXiv subjects

Yuyang Song

Publications and source records attributed to Yuyang Song.

12 recordsLinked to original sources

arXiv2026

PRAXIS: Case-distilled and code-verified AI agents for biological research

Large language models are moving scientific research from text assistance toward agentic workflows, yet biological research requires strong object validation, methodological suitability, reproducibility, and auditability. Prompt engineering, general RAG, or tool use alone cannot reliably produce domain-specific scientific judgment. Here, we present PRAXIS, a verifiable biological research agent framework driven by literature learning and case distillation. PRAXIS converts research experience, failure boundaries, domain rules, and executable procedures into structured long-term memory. By coordinating successful cases, negative cases, rules, and skills, PRAXIS supports problem definition, object validation, method selection, workflow execution, result interpretation, and review feedback across diverse biocomputational tasks. We instantiated PRAXIS as an agent suite for biomedical computing and evaluated it through object validation, case retrieval, memory ablation, public benchmarks, and cross-agent workflows. The results show that case-based learning improves method selection, error suppression, and workflow organization in complex biological research tasks. Rather than replacing scientists, PRAXIS provides a general pathway for transforming research experience into executable, auditable, and transferable agent capabilities.

q-bio.QM↗

arXiv2026

MDAgent: A Multi-Agent Framework for End-to-End Molecular Dynamics Research

Molecular dynamics (MD) simulation is a powerful tool for studying biomolecular structural changes, molecular recognition, transmembrane transport, and functional mechanisms. However, its practical bottleneck lies not only in software operation or parameter setup, but in translating experimental questions into executable, interpretable, and reviewable computational workflows. Here, we present MDAgent, a multi-agent system for end-to-end molecular dynamics research. The system integrates problem understanding, literature-guided strategy design, simulation execution, trajectory analysis, mechanistic interpretation, and quality supervision into a unified workflow, enabling agents not only to run simulations but also to generate research-oriented computational plans and analytical reports. We further introduce a case-based learning mechanism based on Skill and Memory, which stores reusable knowledge from prior tasks, including parameter choices, operational rules, analytical logic, and problem-solving pathways, thereby supporting cross-task transfer without retraining the underlying model. Across multiple representative molecular simulation tasks, MDAgent achieved stable end-to-end performance with improved strategic adaptability, interpretability, and generalization. In an independent complex task involving conformational transitions of TMEM16F and XKR8, the system successfully completed system design, simulation, and mechanistic analysis for large membrane proteins. These results show that combining multi-agent collaboration with case-based learning can transform MD agents from workflow automation tools into scientific question-oriented computational research systems, providing a scalable framework for AI-driven automated research.

q-bio.QM↗

arXiv2026

Transferable Expertise for Autonomous Agents via Real-World Case-Based Learning

LLM-based autonomous agents perform well on general reasoning tasks but still struggle to reliably use task structure, key constraints, and prior experience in complex real-world settings. We propose a case-based learning framework that converts experience from past tasks into reusable knowledge assets, allowing agents to transfer prior case experience to new tasks and perform more structured analysis. Unlike methods based mainly on pretrained knowledge or static prompts, our framework emphasizes extracting and reusing task-relevant knowledge, analytical prompts, and operational skills from real cases. We evaluate the method on a unified benchmark of six complex task categories and compare it with Zero-Shot, Few-Shot, Checklist Prompt, and Rule Memory baselines. Results show that our method achieves consistently strong performance across all tasks and matches or outperforms the best baseline in every case, with especially clear gains on more complex tasks. Further analysis shows that the advantage of case-based learning increases with task complexity, and that practical knowledge acquired by one agent can be reused by others. These findings suggest that case-based learning offers a promising path for building professional agents for real-world work.

cs.AI↗

arXiv2026

IQuest-Coder-V1 Technical Report

In this report, we introduce the IQuest-Coder-V1 series-(7B/14B/40B/40B-Loop), a new family of code large language models (LLMs). Moving beyond static code representations, we propose the code-flow multi-stage training paradigm, which captures the dynamic evolution of software logic through different phases of the pipeline. Our models are developed through the evolutionary pipeline, starting with the initial pre-training consisting of code facts, repository, and completion data. Following that, we implement a specialized mid-training stage that integrates reasoning and agentic trajectories in 32k-context and repository-scale in 128k-context to forge deep logical foundations. The models are then finalized with post-training of specialized coding capabilities, which is bifurcated into two specialized paths: the thinking path (utilizing reasoning-driven RL) and the instruct path (optimized for general assistance). IQuest-Coder-V1 achieves state-of-the-art performance among competitive models across critical dimensions of code intelligence: agentic software engineering, competitive programming, and complex tool use. To address deployment constraints, the IQuest-Coder-V1-Loop variant introduces a recurrent mechanism designed to optimize the trade-off between model capacity and deployment footprint, offering an architecturally enhanced path for efficacy-efficiency trade-off. We believe the release of the IQuest-Coder-V1 series, including the complete white-box chain of checkpoints from pre-training bases to the final thinking and instruction models, will advance research in autonomous code intelligence and real-world agentic systems.

cs.AI↗

arXiv2026

Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments

Training agentic models for terminal-based tasks critically depends on high-quality terminal trajectories that capture realistic long-horizon interactions across diverse domains. However, constructing such data at scale remains challenging due to two key requirements: \textbf{\emph{Executability}}, since each instance requires a suitable and often distinct Docker environment; and \textbf{\emph{Verifiability}}, because heterogeneous task outputs preclude unified, standardized verification. To address these challenges, we propose \textbf{TerminalTraj}, a scalable pipeline that (i) filters high-quality repositories to construct Dockerized execution environments, (ii) generates Docker-aligned task instances, and (iii) synthesizes agent trajectories with executable validation code. Using TerminalTraj, we curate 32K Docker images and generate 50,733 verified terminal trajectories across eight domains. Models trained on this data with the Qwen2.5-Coder backbone achieve consistent performance improvements on TerminalBench (TB), with gains of up to 20\% on TB~1.0 and 10\% on TB~2.0 over their respective backbones. Notably, \textbf{TerminalTraj-32B} achieves strong performance among models with fewer than 100B parameters, reaching 35.30\% on TB~1.0 and 22.00\% on TB~2.0, and demonstrates improved test-time scaling behavior. All code and data are available at https://github.com/Wusiwei0410/TerminalTraj.

cs.CL↗

arXiv2025

AGRO-SQL: Agentic Group-Relative Optimization with High-Fidelity Data Synthesis

The advancement of Text-to-SQL systems is currently hindered by the scarcity of high-quality training data and the limited reasoning capabilities of models in complex scenarios. In this paper, we propose a holistic framework that addresses these issues through a dual-centric approach. From a Data-Centric perspective, we construct an iterative data factory that synthesizes RL-ready data characterized by high correctness and precise semantic-logic alignment, ensured by strict verification. From a Model-Centric perspective, we introduce a novel Agentic Reinforcement Learning framework. This framework employs a Diversity-Aware Cold Start stage to initialize a robust policy, followed by Group Relative Policy Optimization (GRPO) to refine the agent's reasoning via environmental feedback. Extensive experiments on BIRD and Spider benchmarks demonstrate that our synergistic approach achieves state-of-the-art performance among single-model methods.

cs.DB↗

arXiv2025

P-Aligner: Enabling Pre-Alignment of Language Models via Principled Instruction Synthesis

Large Language Models (LLMs) are expected to produce safe, helpful, and honest content during interaction with human users, but they frequently fail to align with such values when given flawed instructions, e.g., missing context, ambiguous directives, or inappropriate tone, leaving substantial room for improvement along multiple dimensions. A cost-effective yet high-impact way is to pre-align instructions before the model begins decoding. Existing approaches either rely on prohibitive test-time search costs or end-to-end model rewrite, which is powered by a customized training corpus with unclear objectives. In this work, we demonstrate that the goal of efficient and effective preference alignment can be achieved by P-Aligner, a lightweight module generating instructions that preserve the original intents while being expressed in a more human-preferred form. P-Aligner is trained on UltraPrompt, a new dataset synthesized via a proposed principle-guided pipeline using Monte-Carlo Tree Search, which systematically explores the space of candidate instructions that are closely tied to human preference. Experiments across different methods show that P-Aligner generally outperforms strong baselines across various models and benchmarks, including average win-rate gains of 28.35% and 8.69% on GPT-4-turbo and Gemma-2-SimPO, respectively. Further analyses validate its effectiveness and efficiency through multiple perspectives, including data quality, search strategies, iterative deployment, and time overhead.

cs.CL↗

arXiv2025

QUITE: A Query Rewrite System Beyond Rules with LLM Agents

Query rewrite transforms SQL queries into semantically equivalent forms that run more efficiently. Existing approaches mainly rely on predefined rewrite rules, but they handle a limited subset of queries and can cause performance regressions. This limitation stems from three challenges of rule-based query rewrite: (1) it is hard to discover and verify new rules, (2) fixed rewrite rules do not generalize to new query patterns, and (3) some rewrite techniques cannot be expressed as fixed rules. Motivated by the fact that human experts exhibit significantly better rewrite ability but suffer from scalability, and Large Language Models (LLMs) have demonstrated nearly human-level semantic and reasoning abilities, we propose a new approach of using LLMs to rewrite SQL queries beyond rules. Due to the hallucination problems in LLMs, directly applying LLMs often leads to nonequivalent and suboptimal queries. To address this issue, we propose QUITE (query rewrite), a training-free and feedback-aware system based on LLM agents that rewrites SQL queries into semantically equivalent forms with significantly better performance, covering a broader range of query patterns and rewrite strategies compared to rule-based methods. Firstly, we design a multi-agent framework controlled by a finite state machine (FSM) to equip LLMs with the ability to use external tools and enhance the rewrite process with real-time database feedback. Secondly, we develop a rewrite middleware to enhance the ability of LLMs to generate optimized query equivalents. Finally, we employ a novel hint injection technique to improve execution plans for rewritten queries. Extensive experiments show that QUITE reduces query execution time by up to 35.8% over state-of-the-art approaches and produces 24.1% more rewrites than prior methods, covering query cases that earlier systems did not handle.

cs.DB↗

arXiv2025

Deployable 3D mesoscale structures through wafer fabrication, geometric frustration and bistable auxeticity

Transforming planar mesoscale devices into precise 3-D architectures is vital for next-generation flexible electronics, implants, and adaptive optics, yet wafer-based manufacturing to free-standing 3-D structures remain elusive. We fabricate polyimide architected 2-D precursors whose bistable unit cells deploy into stable 3-D mesoscale structures. Target Gaussian curvature is encoded by conformally flattening the desired mesh and locally tuning each cell so its second equilibrium matches the required scaling factor, aided by a computed library of isotropically expanding, bistable microstructures. The resulting heterogeneous tessellations uniquely morph into complex shapes. A flat disk deploys into a hemispherical dome with sub-millimeter accuracy and retains its shape after indentation. The same process yields positive- and negative-curvature geometries and tunable-focus paraboloidal mirrors whose reflected laser patterns coincide with geometric optics calculations. Our wafer-compatible, generative algorithm extends far beyond flexible substrates, enabling truly deployable, high-performance electronics and optical devices.

physics.app-ph↗

arXiv2024

Enhancing Code LLMs with Reinforcement Learning in Code Generation: A Survey

With the rapid evolution of large language models (LLM), reinforcement learning (RL) has emerged as a pivotal technique for code generation and optimization in various domains. This paper presents a systematic survey of the application of RL in code optimization and generation, highlighting its role in enhancing compiler optimization, resource allocation, and the development of frameworks and tools. Subsequent sections first delve into the intricate processes of compiler optimization, where RL algorithms are leveraged to improve efficiency and resource utilization. The discussion then progresses to the function of RL in resource allocation, emphasizing register allocation and system optimization. We also explore the burgeoning role of frameworks and tools in code generation, examining how RL can be integrated to bolster their capabilities. This survey aims to serve as a comprehensive resource for researchers and practitioners interested in harnessing the power of RL to advance code generation and optimization techniques.

cs.SE↗

arXiv2024

Unveiling dynamic bifurcation of Resch-patterned origami for self-adaptive impact mitigation structure

A long-standing challenge in impact mitigation is the development of versatile and omnifarious protective structures capable of encompassing a wide spectrum of scenarios, for example, ranging from low-speed pedestrian impacts to high-speed vehicle collisions. However, most existing impact mitigation strategies rely on fixed geometries or pre-tuned material properties targeting specific impact speed, lacking the ability to adapt in real time. Here, we draw inspiration from origami to design impact mitigation structures that exhibit multi-modal and self-adaptive behavior. We introduce a Resch-patterned origami structure that hosts two distinctive deformation modes: a monostable folding mode and a bistable unfolding mode featuring snap-through. Impact experiments reveal a speed-dependent dynamic bifurcation, wherein the structure autonomously switches between folding and unfolding in response to the applied impact velocity. This dynamic bifurcation, intrinsically distinct from kinematic or static origami bifurcations, enables real-time selection of deformation pathways that enhance energy dissipation across a broad range of impact conditions. We further demonstrate the scalability and practical relevance of this mechanism by fabricating tessellations in a bumper-like configuration and evaluating their performance using a pendulum-based mannequin impact test. Together, these results establish dynamic bifurcation in origami-based structures as an adaptive impact mitigation strategy. This approach enables scalable and programmable protective systems that autonomously select deformation modes in real time, with broad relevance to adaptive robotics, smart protective armor, and aerospace damping technologies.

cond-mat.mtrl-sci↗

arXiv2022

Electroadhesive Clutches for Programmable Shape Morphing of Soft Actuators

Soft robotic actuators are safe and adaptable devices with inherent compliance, which makes them attractive for manipulating delicate and complex objects. Researchers have integrated stiff materials into soft actuators to increase their force capacity and direct their deformation. However, these embedded materials have largely been pre-prescribed and static, which constrains the actuators to a predetermined range of motion. In this work, electroadhesive (EA) clutches integrated on a single-chamber soft pneumatic actuator (SPA) provide local programmable stiffness modulation to control the actuator deformation. We show that activating different clutch patterns inflates a silicone membrane into pyramidal, round, and plateau shapes. Curvatures from these shapes are combined during actuation to apply forces on both a 3.7 g and 820 g object along five different degrees of freedom (DoF). The actuator workspace is up to 12 mm for light objects. Clutch deactivation, which results in local elastomeric expansion, rapidly applies forces up to 3.2 N to an object resting on the surface and launches a 3.7 g object in controlled directions. The actuator also rotates a heavier, 820 g, object by 5 degrees and rapidly restores it to horizontal alignment after clutch deactivation. This actuator is fully powered by a 5 V battery, AA battery, DC-DC transformer, and 4.5 V (63 g) DC air pump. These results demonstrate a first step towards realizing a soft actuator with high DoF shape change that preserves the inherent benefits of pneumatic actuation while gaining the electrical controllability and strength of EA clutches. We envision such a system supplying human contact forces in the form of a low-profile sit-to-stand assistance device, bed-ridden patient manipulator, or other ergonomic mechanism. This technology was also demonstrated at ICRA 2022: https://www.youtube.com/watch?v=6Y6-iHWNi6s

cs.RO↗