DeepUnfolding Agent is a schema‑driven FastAPI platform that exposes deep‑unfolded adaptive control as an external software service, separating controller logic from domain‑specific execution. It allows external executors to register strategy schemas, submit diagnostic variables, receive bounded control actions, execute optimization rounds, and return feedback for controller updates. The framework includes a strategy registry, diagnostic‑to‑state adapter, controller factory, safety guard, executor abstraction, feedback update path, SQLite persistence, Docker deployment, and reproducible benchmark scripts, supporting classical, quantum‑aware, hybrid, Qiskit Aer, and IBM quantum processing pathways.
By Shanika Iroshi Nanayakkara, Shiva Raj Pokhrel
The paper introduces an iterative neural variational solver for the committor function in under‑damped Langevin dynamics. By splitting the infinitesimal generator into self‑adjoint and skew‑adjoint parts, each subproblem gains a variational form that can be efficiently solved with neural networks, resembling a generalized Concus‑Golub‑Widlund splitting. The method is validated on benchmark systems such as a double‑well potential, the Muller‑Brown potential, and alanine dipeptide, demonstrating practical accuracy and extensibility for studying rare transition events across different damping coefficients.
By Tiejun Li, Zhaoqi Liu, Xiang Zhou
The paper introduces Latent Annealed Posterior Sampling (LAPS), a method that maps sparsely observed random fields into a lower‑dimensional latent space and performs posterior sampling there. A variance‑preserving bridge controls local covariance inflation, while flow matching learns the latent prior. Experiments on Gaussian benchmarks, a bimodal posterior, and a Navier‑Stokes field show that LAPS accurately approximates posterior distributions, preserves multiple modes, and keeps conditional sampling costs nearly constant as spatial resolution increases.
By Jiankun Lei, Tian Qi
The paper introduces Modality Self-Play (MSP), a training framework that enables multimodal large language models to learn cross‑modal interactions without requiring paired non‑text data. MSP uses proxy‑mediated textual descriptions of unseen modalities as semantic bridges, allowing a model to generate and align these proxies with observed modalities. Experiments show that MSP achieves state‑of‑the‑art or competitive performance on several multimodal benchmarks and demonstrates effective zero‑shot modality composition through learned bridges and anchor‑supporting proxy descriptions.
By Robert Moseley, Md Kaykobad Reza, Ameya Patil, Edward Ayrapetian, Salman Asif
AgentEval is a system that tracks a contract-bound assessment record throughout a stateful agent’s tool execution, capturing state, events, and evidence to determine whether obligations are satisfied, violated, or unresolved. In 32 tool‑world administrations, independent checks confirmed all 30 complete episodes, while controlled interventions on 48 evidence views uncovered 36 binding faults that hash‑only checks missed. A public‑benchmark study on 192 held‑out executions identified ten completed subgoals within broader task failures and highlighted the impact of missing evidence on overall task success.
By Ruiqi Gao (Alibaba Cloud, Alibaba Group)
All-in-One Memory (AiO-Mem) is a reinforcement learning framework that unifies memory retrieval, writing, and stopping decisions into a single reward space. It compiles diverse histories into traceable multi‑view events, uses a unified policy actor to generate interaction trajectories, and employs a trajectory‑conditioned process reward model trained with supervised fine‑tuning and DPO. Experiments on long‑context memory benchmarks show AiO‑Mem consistently improves answer accuracy compared to static memory systems and other RL‑trained agents.
By Bingbing Wang, Jingjie Lin, Weiming Qiao, Jing Li, Ruifeng Xu
The paper introduces a test-time aggregation framework for multi-agent reasoning with large language models. It mitigates non-semantic biases by normalizing reasoning traces and injects bias to create decision instability, which signals reliability. The method improves accuracy and token efficiency on several mathematical reasoning benchmarks without requiring validation data or prior model strength knowledge.
By Sungjae Lee, Soheil Abbasloo, Jabin Koo, Jungseul Ok
The study compared six families of quantum machine learning models with classical counterparts using identical correlation-derived features across four large neuroimaging cohorts. Across all cohorts, quantum kernels performed statistically indistinguishable from matched classical kernels, with average accuracy differences within 1.4 percentage points. While a hybrid quantum convolutional network matched classical accuracy and a trainable quantum kernel slightly improved AUC on ABIDE I, overall quantum models did not consistently outperform classical ones, providing a feature‑matched benchmark of 284 comparisons.
By Utsav Poudel, Anuj Nepal, Vyas Viswanath, Subramaniyaswamy Vairavasundaram
Jev-LiteMAS is a multi‑agent framework that separates decision‑making from generative execution in large‑language‑model workflows. By using a lightweight decision model (Jev) to dynamically select experts and screen artifacts, it closes the planning and review gaps left by pure generative planners and reviewers. Experiments on four benchmarks show that Jev‑LiteMAS cuts generative‑LLM token usage by 12.7–54.6% while keeping accuracy within 1.34 percentage points and reducing latency in most settings.
By Kaixiang Wang, Zhaijiacheng Zhou, Jiong Lou, Jie Li
The paper introduces FlashAttention with Statistics, a method that computes attention‑based detector features without materializing the full attention matrix, thereby reducing memory usage and latency. By accumulating statistics during the FlashAttention pass or via sparse lookups, the approach adds only a small overhead—16% at 2K tokens and 1.4% at 35K tokens on an H100 GPU—and stores just 22.5 KiB per response token. Experiments show that frozen Qwen3 proxy models, even as small as 0.6 B parameters, can serve as strong hallucination detectors, outperforming larger zero‑shot judges and specialized grounding checkers on several benchmarks, with attention features further boosting performance.
By Andrei Volodichev, Maksim Savichev, Nikita Sviridov, Artem Yuliy, Dmitry Simakov, Alexey Zaytsev
QREF is a plug‑and‑play system that improves DeepSearch agents by refining search queries to match the current reasoning intent and filtering evidence to keep only what is relevant. The Query Refiner rewrites queries for complementary coverage across each search batch, while the Evidence Filter removes distracting retrieval results. Experiments show that QREF boosts DeepSearch performance on four benchmarks, achieving significant Pass@1 gains and dramatically reducing ineffective query batches without changing the agents’ architectures or retraining their reasoning policies.
By Yujia Chen, Yuhui Wang, Junzhe Wang, Shihan Dou, Di Liang, Yujiong Shen, Yuhao Zhou, Zhiheng Xi, Ming Zhang, Senjie Jin, Jiayi Chen, Dingwei Zhu, Yajie Yang, Yuan Hua, Chenyao Lu, Tao Gui, Qi Zhang, Xuanjing Huang
The paper identifies functionally specialized subnetworks in large language models using gradient‑attribution profiling, revealing that these subnetworks have minimal overlap. It introduces CiMEO, a continual learning framework that routes each training example to its corresponding subnetwork, limiting updates to selected parameters and optimizer states. On 1.4 million sequential instructions across five generative benchmarks, CiMEO outperforms other methods while reducing catastrophic forgetting and preserving pre‑trained representations.
By Shuoran Jiang, Wentao Zhu, Yang Xiang, Dongfeng Li, Yunan Zhang, Mengchen Zhao, Yuhui Guo, Kai Chen, Youcheng Pan, Xiangping Wu, Qingcai Chen
DeciCAD introduces a domain‑native harness for CAD agents that defines semantic CAD actions, decision‑oriented observations, environment feedback, and persistent state transitions. Using this harness, the authors develop a learning pipeline that extracts causal decision supervision from interaction trajectories and refines the policy through supervised fine‑tuning and geometry‑grounded online reinforcement learning. Across four CAD benchmarks, DeciCAD outperforms a general‑purpose coding harness, achieving higher F1 and IoU scores while reducing model‑output tokens by 45–88% and improving geometry availability from 66.5% to 73.5%.
By Yufeng Peng, Yixin Zhang, Ziyi Liu, Zhikun Xu, Jiarong Yang, Sijie Wang, Wenjie Feng, Qiyu Kang
Prism-1 introduces Position- and Rival-Independent Scoring for System One models, ensuring that option ordering and set changes do not affect decision probabilities. It achieves this by reading each option in its own block at shared positions, preserving softmax independence across options. Experiments on Qwen3-8B and Gemma 4 12B show Prism-1 maintains baseline accuracy while eliminating sensitivity to shuffling and irrelevant options.
By Yao Long Teng, Jianguo Huang, Yanchen Deng, Jiayi Cai, Bo An
HTN‑Orchestrate is a two‑layer framework that separates global task planning from local subtask execution using a hierarchical task network. The Planner builds and updates subtask hierarchies only when prerequisite results are available, while local Orchestrators schedule workers to carry out leaf subtasks. On five benchmarks—including web research, scientific reasoning, and research workflows—the framework outperforms ReAct, Agent Team, and DeepSeek Harness, achieving 89.73% accuracy on BrowseComp and 73.44% on SUPERChem.
By Yanchen Deng, Jianguo Huang, Yao Long Teng, Yixuan Li, Handuo Zhang, Ning Wang, Liangcai Su, Xinyu Wang, Bo An
The paper presents a new conflict graph model for layout decomposition in DSA-MP, addressing overlooked edge-to-edge spacing constraints. It introduces a graph neural network–powered local refinement strategy that improves solution quality and convergence speed. Experiments show the method outperforms state‑of‑the‑art techniques across various scales, styles, nodes, and mask counts.
By Yu Feng, Liyi Zhou, Haoyu Wang, Yiming Xu, Hao Geng, Shaozhi Deng, Ningsheng Xu, Shisheng Xiong
The paper introduces ImpactBench, an open, expert-guided platform designed to measure the multi-dimensional impact of generative AI on users. It evaluates AI models across psychological, physical, and social wellbeing metrics derived from domain experts and existing benchmarks, using simulated multi-turn conversations with teen and adult scenarios. Ten leading LLMs were tested on 48,540 conversations, showing 68.7% promotion of desirable behaviors but only 53.1% avoidance of harms, with an 83.3% agreement rate between automated judges and expert reviewers.
By Chayapatr Archiwaranguprok, Krystal Kirkland, Rachel Poonsiriwong, Stanley Huang, Constanze Albrecht, Jackson Trager, Emillie de Keulenaar, Sheer Karny, Peter Baxter, Nikos Bellos, Andre Luiz Freire de Carvalho Kato, Yuning Liu, Valdemar Danry, Auren Liu, Jennifer Pfister, Pattie Maes, Nathanael J. Fast, Ravi Iyer, Juliana Schroeder, Pat Pataranutaporn
Verdict-MQ is an open encoder that processes a single piece of state once and then answers multiple typed questions about that state, avoiding separate passes for each question. By using a structured attention mask, it ensures the state encoding is independent of question order and prevents questions from influencing one another. On the typed‑decisions benchmark, Verdict‑MQ outperforms a larger specialist model, achieving higher accuracy and better calibration, while on other benchmarks it shows mixed results compared to fine‑tuned baselines.
By Avdhesh Singh Chouhan, Jack Perry
The paper reviews how reinforcement learning (RL) can refine diffusion models (DMs) after pretraining, addressing limitations in reasoning, exploration, and alignment. It surveys RL-driven DMs across language, multimodal, robotic, and scientific domains, focusing on trajectory optimization and reward engineering. Experiments highlight the benefits of various RL post‑training strategies and identify key bottlenecks—reward fragility, exploration‑exploitation trade‑offs, and training instability—while proposing a roadmap for future research and a public repository for updates.
By Zhiyuan Ma, Jiabao Wei, Tianyu Wang, Hanqing Wang, Jiadong Pan, Haowen Xu, Liangliang Zhao, Ruixun Liu, Yuhan Wang, Yuwei Zhang, Liang Feng, Junhao Xiao, Kaiyan Zhang, Xuekai Zhu, Daizong Liu, Biqing Qi, Yihao Liu, Ganqu Cui, Ning Ding, Linfeng Zhang, Jianjun Li, Kun He, Xiang Bai, Bowen Zhou
The paper introduces a Bellman-consistent trajectory reinforcement learning approach for large language model (LLM) arithmetic reasoning, called Trajectory Bellman Residual Minimization (TBRM). By converting sparse terminal rewards into dense token-level signals and employing a Reinforce Leave-One-Out (RLOO) strategy to reduce gradient variance, the method improves learning stability. On the Countdown task, TBRM-RLOO achieves Pass@1 scores of 0.55–0.64 and a state-of-the-art Pass@16 accuracy of 76.0–84.0%, surpassing several baseline methods.
By Lianfa Li