Artificial Intelligence
The Agentic Scientist: Orchestrating Coding Agents for Automated Scientific (Co-)Discovery
Robert Tjarko Lange, Luca Grillotti, Yutaro Yamada et al. (+2 more)
Modern coding
agents are fundamentally reshaping the role of human input for computational research: they write code, launch in silico experiments, analyze their results, and autonomously operate with ever-increasing time horizons. Despite these developments, fully autonomous research
agents suffer from a limited ideation radius and myopic hill-climbing search dynamics. To bridge this gap, we introduce The Agentic Scientist, an end-to-end distributed pipeline for human researchers to conduct computational research alongside modern coding
agents. Building on its predecessor, our system decomposes research into task drafting, idea generation, planning, sandboxed experimentation, result aggregation, baseline and ablation construction, post-hoc analysis, and report writing. Importantly, each step in the pipeline is powered by the next generation of coding
agents and leverages agentic self-verification. The overall research process is driven by a meta-controller that manages parallel research
agents and enforces rubric verification at each stage. This design enables the same orchestration layer to be reused across heterogeneous domains with different execution environments, agentic coding harnesses, and evaluation metrics. Our system can be used across a spectrum of autonomy: either as a co-scientist tool or to perform end-to-end scientific investigation. Within the co-scientist setting, we leveraged a development version of our system to submit a single research paper, which got accepted for publication at ALIFE 2026. We further apply The Agentic Scientist fully autonomously to open problem-solving tasks in machine learning, and demonstrate the system's capability of scientific discovery across different domains of computational research. Our ablations suggest that reliable scientific autonomy depends not only on base model capabilities, but also on the cross-
agent orchestration layer.
Read the full abstract ▾
All authors Robert Tjarko Lange, Luca Grillotti, Yutaro Yamada, Rujikorn Charakorn, David Ha
Artificial Intelligence
Why Static Skill Inconsistencies Do Not Equate to Runtime Risks: A Behavioral Analysis of LLM Agent Execution
Fengrui Fan, Huahao Wang, Jingyi Fang et al. (+3 more)
LLM
agent frameworks heavily rely on external skill assets to standardize tool usage, encode operational workflows, and supply executable runtime resources. Despite their widespread adoption, multi-component skill packages frequently suffer from cross-artifact inconsistencies, where documentation, scripts, and configuration files specify conflicting behavioral constraints. Although static auditing techniques can uncover such contractual discrepancies, static defect volume cannot reliably reflect real-world
agent performance degradation. This gap leaves practitioners without rigorous evidence to assess the practical risk of flawed
agent skills. This paper presents a systematic empirical study to demystify the runtime implications of skill inconsistencies. We conduct large-scale auditing over 575 community and
benchmark skills, yielding 1,182 manually validated contract inconsistencies. We further build a controlled evaluation pipeline with 49 executable
agent tasks and three representative LLMs across NoSkill, OriginalSkill, and FixedSkill experimental settings. Through fine-grained behavioral analysis of 441 paired execution trajectories, we quantify the actual performance impacts of resolved defects.
Read the full abstract ▾
All authors Fengrui Fan, Huahao Wang, Jingyi Fang, Yutao Hu, Yueming Wu, Deqing Zou
Artificial Intelligence · Cross-listed from eess.IV
Personalising Deformable Image Registration with Clinical Context and Deformation History for Adaptive Radiotherapy
Caiwen Jiang, Songlin Zhao, Yuzhen Ding et al. (+23 more)
Adaptive radiotherapy must account for anatomical changes during treatment, but doing so depends on reliably mapping targets, organs at risk and dose across serial images. Existing registration methods treat each image pair independently, without using radiotherapy-specific context or the patient's preceding anatomical trajectory. We developed Clinical-context-Aware and Patient-adaptive Deformable Image Registration (CAP-DIR), integrating source contours, dose, clinical text and prior-registration features. We evaluated 16,361 CT examinations from 3,473 patients across four institutional cohorts and five public
benchmarks. CAP-DIR increased internal-test organ-at-risk-region normalized cross-correlation from 84.76% to 96.92% relative to the strongest learning-based comparator, with low deformation folding, and remained superior across seven external cohorts. In a blinded review of 20 longitudinal cases, clinically acceptable contours increased from 45.0% to 82.5% for clinical target volumes and from 22.5% to 62.5% for organs at risk, while estimated editing time decreased. In a controlled
benchmark, CAP-DIR reduced global dose-mapping error by 34.5% relative to the strongest comparator. These findings show that clinically informed, longitudinal registration can improve contour usability and dose mapping for adaptive radiotherapy.
Read the full abstract ▾
All authors Caiwen Jiang, Songlin Zhao, Yuzhen Ding, Jia Mi, Samir H. Patel, Alexandra Bennett, Keldon K. Lin, Terence T. Sio, Jonathan B. Ashman, Lisa A. McGee, Jean-Claude M. Rwigema, William G. Rule, Sameer R. Keole, Sujay A. Vora, William W. Wong, Nathan Y. Yu, Michele Y. Halyard, Steven E. Schild, Safia K. Ahmed, Daniel J. Ma, Robert L. Foote, Michelle A. Neben Wittich, Thomas J. Whitaker, Haibo Lin, Dinggang Shen, Wei Liu
Artificial Intelligence · Cross-listed from cs.RO
ChunkAct: Action Chunking for Agentic Robotics
Bingxin Xu, Wenrui Bao, Xinxin Liu et al. (+3 more)
Agentic robotics places a large
language model (LLM)
agent in the
robot control loop, where it composes perception, motion, and
vision-
language-action (VLA) tools under the guidance of a
memory and thereby generalizes to new tasks without retraining. This generality comes at a steep price in efficiency. Every primitive requires a round of LLM
reasoning over the full interaction history, so the cost of an episode grows with the number of primitives. We find this
reasoning largely redundant: once a sub-goal is chosen, its primitives follow a routine already prescribed by
memory. We present ChunkAct, which introduces action chunking to agentic robotics. Online, a single call launches a chunk of chained primitives, which executes autonomously and returns control only upon failure. Offline, ChunkAct compiles the
memory into these chunks, so execution is
memory-free at test time and the planner is reserved for decisions that require
reasoning. LLM computation therefore scales with the number of decisions rather than the number of primitives. On LIBERO-Pro, ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls, 4.4x fewer input
tokens, and 2.3x fewer output
tokens.
Read the full abstract ▾
All authors Bingxin Xu, Wenrui Bao, Xinxin Liu, Jiawei Zhou, Yuzhang Shang, Emilio Ferrara
Artificial Intelligence · Cross-listed from cs.RO
What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study
Mingxuan Yan, Litian Gong, Dihong Huang et al. (+19 more)
How should demonstration quality be used during
vision-
language-action (VLA) pretraining to improve performance after target-task fine-tuning? We compare weighting, filtering, and conditioning across reward definitions and label granularities using 5,000 naturally suboptimal human demonstrations and 24 target tasks. Within each VLA, we match demonstrations and training budgets. Additional comparisons examine pretraining data size and the roles of quality tags during fine-tuning and inference. Across three VLAs, we find that approach-aware episode conditioning improves mean success over plain behavioral cloning; conditioning and filtering favor episode and chunk labels, respectively, on average, with model-dependent reversals; and constant-tag fine-tuning improves mean success after conditioned pretraining with either tagged or untagged inference. We introduce AXIS-Bench to support these controlled comparisons. Built on crowdsourced simulation demonstrations, it provides recorded object and contact states for recomputing quality labels on shared trajectories. Replayable trajectories support workspace-aligned rendering and user-defined real-world target tasks. In our real-world experiment, simulation pretraining raises success from 9.6% to 31.7%. Project page: https://axis-bench.github.io/.
Read the full abstract ▾
All authors Mingxuan Yan, Litian Gong, Dihong Huang, Zhixu Li, Tianyu Zhang, Wenqian Zhang, Hefeifei Jiang, Peihao Li, Jian Zhang, Mengfei Zhao, Yikai Tang, Hai Zhai, Zehao Wang, Ruijian Liang, Yanjia Huang, Lin Shao, Changliu Liu, Kaiyu Hang, Zhiwen Fan, Masayoshi Tomizuka, Jianfei Yang, Jiachen Li
Artificial Intelligence · Cross-listed from cs.CL
Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards
Wenbo Hu, Shraman Pramanick, Krishna Kumar Singh et al. (+3 more)
Long-horizon
agents have drawn growing
attention for their ability to work autonomously – from hours to days – on economically valuable tasks, a capability essential to automated discovery and recursive self-improvement (RSI). Yet open-source models still struggle on long-horizon tasks, largely for lack of both training data and a working recipe, and scaling both is hard: long, complex tasks are difficult to curate and verify, and their rewards are sparse. We address both. First, we introduce Tev-Gym, the first long-horizon
agent training
dataset with verifiable, dense, progressive rewards – 1,170 verified task environments across 7 domains. It is built on a latent-world-first approach that first constructs a latent world: a complete and correct hidden specification of the task domain, workflow, and expected state after each step. Each task is then progressively rendered from this specification into an
agent-facing environment, enabling controllable complexity and a natural progress metric. Second, we turn this metric into a training recipe: via progress-potential reward shaping (PPRS), we convert progress into a dense per-step reward that improves training efficiency and makes otherwise-intractable tasks trainable. Training with this recipe yields Tev-27B, which achieves substantial gains on the
Agents' Last Exam computing and math sciences domains (ALE-CM), matching GPT-6 Astra and Claude Opus 5 at Pass@5. Despite training only on ALE-related data, it generalizes to the Terminal-Bench (TB) 3.0 and 4.0 subsets, surpassing GLM 5.2 and Gemini 3.7 Flash, and reaches Claude Opus 4.8-level performance on middle-horizon tasks such as TB 2.1.
Read the full abstract ▾
All authors Wenbo Hu, Shraman Pramanick, Krishna Kumar Singh, Nanyun Peng, Kai-Wei Chang, Yong Jae Lee
Artificial Intelligence · Cross-listed from cs.CL
DoubtLess: Training-Free Contrastive Decoding for Efficient Reasoning
Minghao Luo and Liang Chen
Reasoning models achieve strong performance through long chains of thought (CoT), but often spend more
tokens than needed to solve a problem. Existing ways to shorten the CoT require additional training for each model, add a fixed steering vector to the model's hidden states, or use decoding heuristics such as stopping early or banning words like "wait", which can reduce accuracy. We propose DoubtLess, a training-free decoding method in which a much smaller
reasoning model steers the large one towards brevity. The small model reads the large model's CoT so far twice, once prompted to be brief (the brief thinker) and once prompted to be verbose (the verbose thinker), and the difference between its two logit vectors, scaled by a strength, is added to the large model's logits at every step of the CoT. On seven
benchmarks of competition mathematics, graduate-level science and competitive programming, DoubtLess shortens the CoT of Qwen3-32B by 31% with unchanged accuracy, outperforming five inference-time baselines. We also show that DoubtLess reduces thinking
tokens per turn by 20% in a self-debugging coding
agent with unchanged task performance. It also generalizes to two other model families with unchanged accuracy. When combined with training-based compression, it further shortens the CoT of a model already trained for brevity. Ablations and behavioral analysis show that DoubtLess reduces repeated checking, with most
token savings coming after the correct answer first appears in the CoT. Our code is available at https://github.com/leoluolol/DoubtLess.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CL
Jiaru Zou, Yunzhe Qi, Sirui Chen et al. (+5 more)
How can we build an
agent that thinks beyond human
language, yet still acts through it in the external environment? Modern LLM
agents still reason and interact through text, while latent
reasoning offers an efficient internal medium but lacks native tool access and
reasoning transparency. We bridge these two directions with Hybrid
Reasoning Agents (HybridAgents), which dynamically navigate between latent
reasoning and text-based interaction throughout a unified generation process. Architecturally, a lightweight HybridLink enables seamless transitions between text and latent modes, while the
agent's continuous working
memory preserves lossless hybrid context across multiple turns. These designs preserve tool-integrated agentic capabilities while making latent thoughts directly inspectable. Building on this architecture, we further develop dedicated post-training
algorithms for hybrid
reasoning: HybridSFT stabilizes tool-integrated hybrid generation by teaching the backbone to coordinate with HybridLink, then HybridRL scales
reasoning by learning when to think in latent space and when to act through text. We evaluate HybridAgent at 3 model scales across 11
benchmarks covering mathematics, science, medicine, search, and agentic
reasoning. Compared with frontier LLM
Agents and agentic frameworks spanning both text-based and latent methods, HybridAgent consistently improves performance by 10.4%, while delivering 35.5%-54.7%
token reduction, 1.8x-2.6x faster inference, and 2.7x-3.7x cost savings. Code and data are available at https://github.com/togethercomputer/HybridAgent.
Read the full abstract ▾
All authors Jiaru Zou, Yunzhe Qi, Sirui Chen, Federico Bianchi, Shang Zhu, Pan Lu, Jingrui He, James Zou
Artificial Intelligence
AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewards
Xin Gao, Yifan Yang, Xin Li et al. (+13 more)
Generating information-rich visual artifacts such as slides, posters, and dashboards as code is appealing because the outputs are structured, editable, and verifiable. Yet code models cannot see how their choices of layout and color come together on the canvas, often limiting spatial planning and aesthetic quality. Image generators excel at visual design but can misrender requested content, making their outputs unreliable targets for direct reproduction. We therefore introduce AesCode, which generates code grounded in the content
prompt while using an image generated from the same
prompt as an aesthetic guidance. To do so, AesCode represents the target as a design graph that describes the whole canvas and makes each property attributable. From this graph, we derive complementary cross-modal rewards: deterministic verifiers score properties parsable from code and rendering, while a VLM judge assesses the non-parsable ones with a sample-specific Visual Graph Rubric (VGR) tied to graph elements and relations. We optimize these decoupled rewards through reinforcement learning with per-channel normalization to jointly improve spatial planning, aesthetic quality, and verifiable correctness. To evaluate this setting, we build a
benchmark that scores verifiable correctness and visual quality separately, with visual scores highly consistent with human judgments. We find that reference images substantially improve the visual quality of code generated by existing models, and ablations show that AesCode benefits from references both as model input and as a source of visual rewards. Relative to same-scale Qwen3-VL baselines given the same references, AesCode-8B and AesCode-32B improve visual quality by 31.1 and 22.4 percentage points, reaching scores comparable to those of closed-source frontier models such as GPT-5.5 and Claude Opus 4.8.
Read the full abstract ▾
All authors Xin Gao, Yifan Yang, Xin Li, Pan Wang, Yan Li, Muzhao Tian, Mingxi Cheng, Bei Liu, Kai Qiu, Qi Dai, Dong Chen, Yue Dong, Leandros Tassiulas, Rex Ying, Ji Li, Chong Luo
Artificial Intelligence · Cross-listed from cs.LG
Two Clocks: Why Larger Rollout Batches Learn Slower but Fail on Schedule in Reinforcement Learning with Verifiable Rewards
Yuante Li, Junwei Quan, Kai Qiu et al. (+2 more)
Reinforcement learning with verifiable rewards (RLVR) relies on costly model-generated responses, motivating larger rollout batches and square-root learning-rate scaling. Reported to approximately preserve progress per sample with one optimizer step per rollout, this rule is less understood in practical multi-step training. We develop a two-clock framework that separates the samples needed to make progress from those consumed before collapse. Under a local quadratic model, the square-root rule preserves first-order drift per sample but not progress: larger batches take fewer, longer updates and lose more to
curvature, yielding a learning-rate-dependent critical batch beyond which sample efficiency degrades. A drift-threshold hypothesis predicts a complementary failure clock whose sample count remains approximately invariant along the square-root rule. Controlled GRPO experiments on Qwen2.5-Math-1.5B across a $16\times$ batch range support both predictions. Ratio clipping delays failure, especially for larger batches, but the progress gap persists without it. Widening the clip trades stability for a transient speed gain. The progress gap reappears on another
dataset, a general-purpose checkpoint, and a 7B model. Finally, we use the two-clock framework to derive a simple multi-step RLVR schedule: start below the critical batch and switch to a larger batch before failure. Our code is available at: https://github.com/WinstonLiyt/two-clocks-rlvr.
Read the full abstract ▾
All authors Yuante Li, Junwei Quan, Kai Qiu, Kate Zhang, Yicheng Tao
Artificial Intelligence
MFG: Multi-turn Feedback Guided Molecular Optimization
Trieu Nguyen, Frazier N. Baker, Ruoxi Gao et al. (+2 more)
Multi-objective molecular optimization is a critical process in drug discovery that aims to modify a given molecule to simultaneously improve multiple properties while maintaining a similar overall structure. This process often requires iterative modifications guided by computational or experimental feedback. Recent large
language model (LLM)-based methods have begun to explore this iterative approach; however, these methods are not explicitly trained to leverage such feedback, while existing approaches primarily focus on single-turn molecular optimization. To address this gap, we introduce Multi-turn Feedback Guided (MFG) molecular optimization, a reinforcement learning framework that trains an LLM to progressively optimize molecules using feedback from previous turns. MFG introduces a feedback vector capturing both property improvements and source molecule similarity, alongside a feedback-aligned reward function that promotes progressive improvement across turns. We apply MFG over $\text{GeLLM}^{3}\text{O}$, a state-of-the-art LLM model for multi-objective molecular optimization, evaluate $\text{GeLLM}^{3}\text{O-MFG}$ on ten multi-objective molecular optimization tasks. The results demonstrate that $\text{GeLLM}^{3}\text{O-MFG}$ achieves the highest average similarity-constrained success rate ($81.50%$), outperforming the base model $\text{GeLLM}^{3}\text{O}$ ($70.81%$).
Read the full abstract ▾
All authors Trieu Nguyen, Frazier N. Baker, Ruoxi Gao, Daniel Adu-Ampratwum, Xia Ning
Artificial Intelligence · Cross-listed from cs.RO
LIBERO-RIG: A Hardware-Setup Robustness Benchmark for Vision-Language-Action Models
Andrea Yaoyun Cui, Hanyu Wang, Xinyi Wang et al. (+6 more)
Robot policies are expected to generalize across deployment environments, including variations in the
robot itself: cameras are bumped and remounted, calibrations drift, controllers run at different rates, and end-effector pads wear down. We present LIBERO-RIG, a
benchmark that perturbs only the
robot's hardware and setup while holding every LIBERO task fixed. The
benchmark spans eleven perturbation axes, including camera mounting and optics, camera-to-arm calibration, actuation gain, control rate, proprioceptive noise, base pose, table height, and gripper pads. Each axis has five severity levels, yielding 210 conditions across the four LIBERO suites. We evaluate nine recent
vision-
language-action models using 500 episodes per condition and substantiate each finding with rigorous statistical tests. Pooled success falls from 0.932 to 0.767 at the middle severity and 0.509 at the highest severity. Performance degradation varies substantially across perturbation axes and action interfaces, and training on visual perturbations does not mitigate faults that leave camera images unchanged. We release the
benchmark together with manifests specifying the fixed perturbation parameters for each episode.
Read the full abstract ▾
All authors Andrea Yaoyun Cui, Hanyu Wang, Xinyi Wang, Bangzheng Li, Lin Zhang, Zhenggang Tang, Shyamal Buch, Dejia Xu, Matthew Tancik
Artificial Intelligence
Who Improves the Robot? Toward Physical Recursive Self-Improvement
Hai Zhou, Fan Yang, Zeyi Chen et al. (+27 more)
Robots are increasingly autonomous in task execution, yet their improvement remains largely human-driven. This study investigates Physical Recursive Self-Improvement (Physical RSI), in which physical experience guides lasting updates to both a
robot's capabilities and its improvement procedure, making subsequent improvement more effective across tasks. Physical RSI faces challenges from noisy observations, continuous dynamics, costly trials, and irreversible consequences. We present an L0-L5 framework that differentiates capability-level from meta-level updates and within-task from across-task gains, reserving L5 for independently verified recursive gain. We study existing methods around harness adaptation and model training, alongside safety and control mechanisms that govern execution and successive updates. Our analysis reveals that, despite substantial progress in self-improvement, the surveyed literature does not yet demonstrate Physical RSI as defined by our framework. We identify research priorities in reliable feedback, causal attribution, safety preservation, and benchmarking, outlining a path toward
robots that autonomously improve how they learn while ensuring safety. The GitHub repository for Physical RSI is available at https://github.com/SAIL-Research-Lab/Physical-RSI.
Read the full abstract ▾
All authors Hai Zhou, Fan Yang, Zeyi Chen, Lejun Zhang, Chenyang Fu, Jianping Li, Jiayi Zhu, Xianyang Liu, Yingxuan Yang, Lili Meng, Lu Wang, Xinyang Han, Jiankai Sun, Weirui Ye, Guanchu Wang, Wanhua Li, Ming Zhou, Boyi Li, Ming Jin, Kun Shao, Muhao Chen, Guang Chen, Lei Ma, Joey Tianyi Zhou, Costas Spanos, Xia Hu, Mac Schwager, Weinan Zhang, Qian Feng, Shangding Gu
Artificial Intelligence · Cross-listed from cs.LG
Fast Rates for Offline Q-Learning with KL Regularization
Yize Hao, Guangxin Jiang and Shuang Qiu
This paper investigates offline Q-learning for finite-horizon Markov decision processes with Kullback-Leibler (KL) regularization. Recent studies have shown that KL regularization can yield fast rates in offline decision-making problems. However, it remains unclear whether Q-learning, which adopts incremental
stochastic approximation updates, can achieve such fast rates in the offline setting. We first propose KL-regularized Q-Learning (KLQ), a simple variant of classical Q-learning. By exploiting KL regularization, we prove that KLQ achieves a sample complexity of $\widetilde{O}(1/\epsilon)$ without explicit pessimism. This guarantee, however, has an exponential dependence on the KL regularization parameter and the horizon. To address this limitation, we further propose KL-regularized Pessimistic Q-Learning (KLPQ), which incorporates pessimism into KLQ. We prove that KLPQ retains the same \(\widetilde O(1/\epsilon)\) rate while removing the explicit exponential dependence, thereby further improving the sample complexity. To the best of our knowledge, these results establish the first fast-rate guarantees for offline Q-learning in tabular finite-horizon KL-regularized MDPs.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.RO
RoboSimStudio: An Interactive Toolkit for Human and Agent Rapid Real2Sim of Robot Manipulation Tasks
Kaiyuan Chen, Shuangyu Xie, Uksang Yoo et al. (+9 more)
It has been difficult to accurately model the physics of the interaction between a
robot and the objects it manipulates, including contact mechanics, coupled contact dynamics, deformations, stiffness and friction. We present RoboSimStudio (RSS), an interactive toolkit that makes choices of solver, contact model, and physical parameters explicit and editable, allowing both humans and coding
agents to quickly create and iteratively refine simulations based on observed physical behavior. We implement RSS with the open Newton physics simulation system and describe experiments with human-guided real2sim using (1) visual evaluation, (2) force-feedback and (3) agentic Real2Sim2Real. Code,
video and additional information can be found on the project website: https://robosimstudio.github.io
Read the full abstract ▾
All authors Kaiyuan Chen, Shuangyu Xie, Uksang Yoo, Xinyue Zhu, Cael Magner, Alvin Zhu, Daehwa Kim, Eric Heiden, Erwin Coumans, Miles Macklin, Spencer Huang, Ken Goldberg
Artificial Intelligence
Read the Room Before Your Model Generates
Muyao Kong and Zhihao Lin
Large
language models (LLMs) generally rely on sampling temperature to regulate the amount of randomness in their outputs. Mainstream methods for selecting temperature typically involve fixed, heuristic global settings and context or
token-level calibration, rather than a geometrically interpretable representation of the communicative scenario. In this work, we introduce a simple and interpretable temperature selection framework, the Scenario-Actor, which adapts decoding to the inferred situational characteristics. Scenario-Actor extracts scenario-relevant structures from
Transformer hidden states and determines the temperature from scenario
embedding, a structured
geometric representation of the communicative setting. Empirical results show that on Llama-3.1-8B-Instruct, Scenario-Actor achieves strong goal-completion performance in interactive social scenarios, significant gains on open-ended instruction following, and competitive performance on multi-turn conversation compared to multiple baselines.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CL
Back-Translation Still Matters: Leveraging Monolingual Data for Low-Resource Machine Translation with LLMs
Divyanshu Bathla, Parag Singla and Jitin Singla
Pre-trained Large
Language Model (LLMs) are effective on many tasks in zero-shot and few-shot settings. For machine translation of low-resource
languages, the obvious options are few-shot prompting and fine-tuning on a small parallel set. We ask: what if moderate amounts of low-resource monolingual data are also available? We argue that, rather than continued pretraining, a more effective use of this data is the pre-LLM era technique of back-translation (BT). The current fine-tuned LLM translates monolingual sentences in each
language, and the resulting pairs train it in the reverse direction. Updates on the parallel set are interleaved for stable learning. With 1K parallel pairs on Odia and Swahili, BT beats continued pretraining into the low-resource
language by 7.7 to 16.4 XCOMET-XXL points in domain. With beam-decoded intermediate translations, it also beats a model with 10 time more parallel data. In-context examples retrieved from the parallel data further improve BT. We also analyze the source, quality and amount of monolingual text. Across five low-resource
languages, with Hindi as a high-resource control, and two 3B backbones, BT improves translation into the low-resource
language by 16 to 173% XCOMET-XXL over parallel-only fine-tuning. Blind human evaluation on Odia and Sanskrit agrees with the automatic metrics.
Read the full abstract ▾
Artificial Intelligence
Self-Evolving Agent Harness
Chuyang Wei, Haoxiang Guan, Minghang Zhu et al. (+5 more)
Harness self-
evolution revises an
agent's software scaffolding, including
prompts,
memory, workflows, and executable components, while keeping model weights fixed. We formalize a feedback-driven process in which accepted revisions persist across runs. The framework separates update targets from proposal and admission mechanisms. It distinguishes changes to interpreted state from revisions to tools and code, repository-level control logic, and plugin composition. We characterize search through evolutionary optimization and examine its reliability through continual PAC-learning. Under explicit assumptions, the analysis yields an expected generalization bound, a retention bound conditional on historical regression budgets, and a high-probability admission rule using fresh, independent validation data. We examine how persistent harness revisions affect knowledge sharing and organizational workflows, with implications for scientific reliability, physical safety, and institutional accountability. Cases from prior work illustrate model–harness co-design, distributed execution, and plugin reuse. Reliable deployment requires continual validation as task distributions and evaluators change, with independent safeguards for learning, runtime privileges, and consequential actions. We provide a comprehensive 66-page discussion of related work, theory,
algorithms, evaluation, applications, and the broader impact of Harness RSI, together with continuously updated code and supplementary materials, at https://github.com/zgcagi/Harness-RSI.
Read the full abstract ▾
All authors Chuyang Wei, Haoxiang Guan, Minghang Zhu, Kexin Zhou, Jiyan He, Xiren Zhou, Huanhuan Chen, Yifei Shen
Artificial Intelligence · Cross-listed from cs.MA
Optimizing Team Behavior via Extremum-seeking Control in Multi-agent Reinforcement Learning
Svar Rajankumar Patel and Kshitij Jerath
In cooperative Multi-
Agent Reinforcement Learning (MARL),
agents must learn to coordinate their actions to achieve shared objectives. A key indicator of effective coordination is behavioral diversity, i.e., the
agents' potential to enact distinct yet complementary control policies given the same set of observations. Although behavioral diversity can enable role specialization and richer joint behaviors, its appropriate level is highly context- and task-dependent and often costly to tune. We propose Adaptive Diversity Control (ADiCo), a two-timescale framework that adapts behavioral diversity online during training. On a fast timescale, Diversity Control (DiCo) regulates system diversity toward a target value. On a slower timescale, an Extremum Seeking Controller (ESC) adjusts this target using only mean episodic returns, without requiring an explicit model or analytical gradient of the diversity-performance landscape. We evaluate ADiCo on six configurations across two cooperative MARL tasks, where it maintains task-appropriate diversity levels: low diversity when homogeneous policies are beneficial and higher diversity when heterogeneous policies are required. Through this adaptation, ADiCo matches or exceeds the mean performance of DiCo (best grid-searched diversity target) and Unconstrained DiCo (diversity allowed to emerge without target regulation). These results demonstrate that incorporating classical feedback control into MARL training enables effective coordination by providing a systematic alternative to task-specific manual selection of the diversity target.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.RO
SEAR: Self-Evolving Agentic Robotics
Xinyi Wang, Bangzheng Li, Hanyu Wang et al. (+5 more)
Frontier
language-model
agents can now improve
robots on their own: they write control code, test it, and keep what works as skills and
memories for the next task. We introduce SEAR (Self-Evolving Agentic Robotics), a framework and
benchmark built around two questions. Is the
agent really learning, or just trying more often? And does what it learns still help on new tasks, in a new simulator, and on a real
robot? The framework runs an
agent in three modes: it writes code that becomes the policy (code-as-policy), acts live in the control loop (
agent-as-policy), or works as a researcher that fine-tunes a
vision-
language-action model (auto-research). The
benchmark adds controls that separate retained experience from repeated attempts, familiar tasks, privileged simulator state, larger budgets, and luck. Across seven frontier
language models, 330 simulated and five physical tasks, and over four years of
agent runtime, we highlight three findings. First, the learning curve of a robotic
agent depends both on the overall time budget and the task difficulty relative to model capability. Harder tasks make longer sessions worthwhile; on easier tasks, re-rolling short attempts does just as well. Second, control code and shared routines transfer across tasks and simulators, while task-specific solutions do not. On a real
robot, we observe a performance gap between
agent-designed practice and novel human-designed scenes. Third, different models evolve more effectively in different ways. Some do best acting live, others writing code first. And a text-only model keeps pace with a comparable
multimodal one: perception helps with error assessment, but agentic
reasoning and tool use can matter more. We release the
benchmark and codebase to support future research.
Read the full abstract ▾
All authors Xinyi Wang, Bangzheng Li, Hanyu Wang, Lin Zhang, Andrea Yaoyun Cui, Zhenggang Tang, Shyamal Buch, Dejia Xu
Artificial Intelligence · Cross-listed from cs.CR
Bytes to Threats: Byte-Sequence Transformer For Malware Detection
Abdelrahman Yassin Eljamal and Abdeltawab Hendawi
Existing malware detection approaches heavily rely on handcrafted features derived from static or dynamic analysis of the executable files. These methods require specialized domain expertise, dedicated analysis tools, and time-consuming analysis, which limits their scalability and robustness in the face of the increasing complexity and diversity of modern malware. Consequently, there is a growing interest in approaches that operate directly on raw bytes, which contain rich semantic and structural information. However, capturing long-range dependencies within executable files remains challenging for conventional static and CNN-based methods. This paper presents MalwareFormer, a
transformer-based, byte-level malware detection framework that operates directly on raw bytes, without requiring domain knowledge or feature extraction. MalwareFormer achieves an accuracy of 98.35%, a balanced accuracy of 98.32%, a macro-F1 of 98.32%, and a false positive rate of 1.94%, outperforming both baselines across all threshold-dependent classification metrics and demonstrating that
transformers provide a powerful and scalable approach for byte-level malware detection.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CL
Where Do LLM Ensemble Gains Come From? Decomposing Candidate Value, Diversity, and Aggregation
Donghwan Kim
LLM ensemble gains are often measured against a single baseline, but the value attributed to additional candidates can change with the comparison used. We study an LLM aggregator under four matched conditions: Direct (no candidates), Single-slot (one candidate in a multi-slot interface), Duplicate (one candidate repeated), and Multi-candidate (separately generated candidates). Under either non-diverse control, overall candidate gain (CIV) splits algebraically into an additional-candidate contrast (DSV) and a candidate-context contrast (CCV); the split, but not CIV, depends on the chosen control. On MultiHopRAG model diversity (k=2), the repetition gap Rep=-8.33 pp [-13.25, -3.75] quantifies the difference between gains measured against Duplicate and Single-slot. On a matched FRAMES stronger-panel extension (n=200, k=3, aggregator output limit 4096), paired Delta Rep=+9.67 pp [+6.00, +13.50]; within panels, Rep changes sign (original -6.33; stronger +3.33). In FRAMES evidence aggregation, candidate-context contrasts remain strongly negative under both non-diverse controls. Our results show why ensemble evaluations should state the reference control before attributing gains to additional candidates.
Read the full abstract ▾
Artificial Intelligence
Pack Light, CarryOn: Building Object-Oriented Context for Long-Horizon Agents
Yuchen Lu, Julian McAuley and Lei Zhang
Long-horizon
agents accumulate progress in persistent task state, yet repeatedly process interaction histories describing how that progress was achieved. This raises a natural question: can long-horizon context instead be organized around reusable progress rather than the history that produced it? In this paper, we introduce CarryOn, an object-oriented context runtime that organizes continuation around reusable task objects. Rather than treating compressed histories or reconstructed accounts of progress as the primary basis for continuation, CarryOn builds a context view from continuation carriers, complementary execution context, and a live tail of recent interactions. A core challenge is keeping this reusable basis valid after its surrounding history is retired: how can CarryOn update the basis without reconstructing the full representation at every fold? CarryOn maintains carrier records across folds, revising affected records against bound evidence and applicable environment checks while carrying forward unaffected records and jointly updating execution context. Experiments on OSWorld 2.0 and BrowseComp-Plus show improved gains of 5.8% and 7.0% while reducing total
token usage by 51.9% and 22.9%, respectively. Project page: https://ethannju.github.io/carryon
Read the full abstract ▾
Artificial Intelligence
BLADE: Auditing Information Use for Failure Attribution in LLM Multi-Agent Systems
Xiaoguang Guo, Yaohao Long and Chuxu Zhang
Large
language model multi-
agent systems extend individual
agents' capabilities through collaboration across specialized roles, tools, and workflows. Yet collaboration makes failures harder to diagnose: one
agent may misinterpret a correct result while others carry the mistake forward. Failure attribution therefore seeks to identify the responsible
agent and decisive error step. Recent methods trace dependencies, but knowing where information came from is not the same as knowing where it went wrong. We introduce BLADE (Blame Localization via Audited Dependency Evidence), a training-free framework that moves beyond tracing information to auditing its use. Using Jev for structured probabilistic judgments, BLADE uses dependency probabilities to select relevant source-decision handoffs for auditing. At these handoffs, it examines whether receiving decisions use their sources faithfully and within task requirements, distinguishing error introduction, propagation, and repair. These findings inform a global diagnosis that considers all execution steps. Without reference answers, BLADE achieves the best or tied-best step accuracy among eight strategies under a shared Jev decision interface on 1,228 failed runs across five Who&When and Who&When Pro subsets, outperforming All-at-once by 2.2-11.1 percentage points. Making responsibility evidence explicit offers a path toward more interpretable and reliable agentic systems.
Read the full abstract ▾
Artificial Intelligence
Jeva: System One Models from First Principles
Qi Li and Xinchao Wang
System one models enable direct, low-latency decision-making. Yet recent practice of predicting an answer from serialized text does not, by itself, respect the structure of a decision problem. We argue that current designs underemphasize four important structural requirements. (i) null-aware: Models should distinguish answer absence from uncertainty among candidates; (ii) permutation-invariant: Reordering alternatives should not change the decision; (iii) option-scalable: Growing candidate sets should not require redefining the output space; (iv) cacheable: Request-independent computation over recurring alternatives should be reusable across requests. We introduce Jeva, a decision-native architecture that addresses these requirements jointly. A shared encoder independently represents the request and each option, while a compact decision stack refines only the request state against immutable option
memory. Symmetric aggregation yields permutation-equivariant probabilities; shared scoring accommodates variable-size candidate sets. A separate coverage estimate distinguishes answer absence from candidate competition, enabling explicit null outcomes. Request-independent option representations and layer projections support reuse across newly assembled sets. These mechanisms unify Choice, Noul, and Score outputs. Across evaluations, Jeva combines strong typed-decision performance with near-perfect null detection and no decision flips under any tested permutations. It scales to thousands of candidates and substantially reduces latency through cross-request reuse without changing decisions.
Read the full abstract ▾
Artificial Intelligence
Beyond the Verdict: Agentic Code Judging with Verifiable Feedback
Zheye Deng, Sha Li, Changlong Yu et al. (+4 more)
Test-time scaling for large
language models (LLMs) can involve generating multiple candidate solutions for selection or revising an earlier attempt. To guide selection and revision, we need a verifier that can estimate candidate correctness and provide evidence of failure. In competitive programming, however, official judging systems provide authoritative verdicts under finite submission budgets, without revealing failure explanations or failing inputs. To provide informative feedback without consuming official submissions, we formulate a task in which a model judges and diagnoses a candidate program using only the problem statement and source code, with generated counterexamples serving as checkable evidence. To evaluate this task, we build USACO-JUDGE from the USACO 2026 contests, using accepted and unaccepted programs written by human experts and multiple LLMs. We then train ARBITER, a 35B mixture-of-experts (MoE) agentic verifier, with reinforcement learning in a generic execution sandbox. Training raises counterexample success on incorrect programs from 40.61% to 62.67% and mean binary verdict accuracy on USACO-JUDGE from 71.41% to 87.41%, approaching frontier verification performance. Beyond verification, ARBITER outperforms baselines in both code selection and refinement. In an IOI 2026 case study, ARBITER-guided refinement raises GPT-OSS-120B's expected score above the gold-medal threshold under the official submission limit.
Read the full abstract ▾
All authors Zheye Deng, Sha Li, Changlong Yu, Xin Liu, Qin Lu, Priyanka Nigam, Yangqiu Song
Artificial Intelligence · Cross-listed from cs.SE
PreciseCoder: Faithful and Generalizable Debugging Beyond Unit-Test Rewards
Yunkai Zhan, Miaosen Chai, Shangshang Wang et al. (+5 more)
Reinforcement learning (RL) from unit tests is the common recipe for post-training code models, but a unit test only asks whether the final program passes. We analyze what this unit-test-only reward teaches a model during debugging and find that, while it raises pass rates, it also teaches the model to edit more and to hallucinate bugs in code that is already correct. We introduce PreciseCoder, a training
algorithm that augments the unit-test reward with a signal for semantically verified, localized edits, which correlates with smaller edits and higher pass rates. On 4B, 9B, and 27B models, PreciseCoder reduces the average number of edit lines from 7-15to 3 at a pass rate on par with unit-test-only training. Trained only on single-line bugs, PreciseCoder can generalize to repository-level program repair, where it resolves more issues with smaller patches. Its precision also transfers from weaker to stronger models: given the
reasoning trace of a 9B PreciseCoder model, frontier models such as GPT-5.6 Sol and Qwen3.8-Max gain 4.6-12.4 points in edit precision with no additional training. Linguistic analysis of the thinking traces shows that our reward encourages the model to reason about individual lines, name the buggy ones, and form consistent bug hypotheses.
Read the full abstract ▾
All authors Yunkai Zhan, Miaosen Chai, Shangshang Wang, Jike Zhong, Yuqing Yang, Deqing Fu, Serina Chang, Wang Bill Zhu
Artificial Intelligence · Cross-listed from cs.SE
SolBehave: Evaluating the Behavioural Correctness of LLM-Generated Smart Contracts
Giovanni Pio Delvecchio, Stefano Fantazzini, Giacomo Frisoni et al. (+4 more)
Large
language models increasingly generate Solidity smart contracts that, once deployed, can never be patched, making a single latent fault permanent and exploitable. Yet current evaluations mostly ask only whether generated code compiles. We introduce SolBehave, a
benchmark for evaluating the correctness of model-generated contracts at scale, moving beyond ad hoc unit tests. Across 250 real, reproducible Ethereum contracts, we apply three complementary tests of behavioural fidelity: differential fuzzing, historical-transaction replay, and, as our core contribution, synthetic-transaction replay guided by a Dynamic Condition Response (DCR) graph that enumerates the space of legitimate function-call sequences. Differential fuzzing compares each generated function with the original on automatically generated inputs; historical replay compares the whole contract on the transactions it actually received on-chain; DCR-guided replay compares it on sequences of calls that follow its intended usage, including orderings that never occurred on-chain. We evaluate seven state-of-the-art open-weight models after one round of automated repair. Compilation success ranges from 26.4% to 92.7%, but under differential fuzzing only 31.4% of all generated function bodies that compile, pooled over models and samples, are equivalent to the original. Divergence at the contract level is the norm rather than the exception: for six of the seven models, between 13% and 31% of the generated contracts that compile diverge from the original under historical replay, and between 13% and 44% under DCR-guided replay; only Gemma-4-31B stays below 3% on both. Compilation is therefore a necessary condition for correctness and far from a sufficient one.
Read the full abstract ▾
All authors Giovanni Pio Delvecchio, Stefano Fantazzini, Giacomo Frisoni, Gianluca Moro, Lorenzo Bacchiani, Mario Bravetti, Thomas Troels Hildebrandt
Artificial Intelligence · Cross-listed from cs.LG
A Rate Transfer Method: Convergence Rates in Stochastic Gradient Descent via Stochastic Calculus
Neil Walton
We develop a rate-transfer method that converts finite-time Lyapunov estimates for an associated
stochastic differential equation (SDE) into finite-time convergence bounds for
stochastic gradient
algorithms. Under generic moment and smoothness conditions, the convergence analysis of the
stochastic gradient method reduces to a calculus argument verifying a Lyapunov generator inequality for the SDE. The method recovers established rates for classical
stochastic gradient methods and provides a way to establish convergence rates for new optimization
algorithms through calculus arguments applied to their associated SDEs.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.RO
Steertuning Policies from a Few Human Videos
Billy Yan, Isaac Tu, Kelly Lee et al. (+8 more)
Current recipes for
robot learning require massive amounts of
robot demonstrations to obtain robust and adaptive policies. In this work, we introduce Steertuning, a system that enables
robots to learn new tasks in new environments from 5 human demonstrations in under 5 Steertuning consists of two parts. (1) We pretrain a 400M-parameter
diffusion policy on our egocentric human
dataset, which consists of 100 hours of densely annotated random interactions. This pretraining makes Steertuning robust to background differences and enables spatial and object generalization when deployed on new tasks. (2) Given 5 demonstrations of a new task, we finetune our model for only 2 minutes (2000 training updates) and steer the policy's actions during reverse sampling using the demonstration whose observation is closest to that of the deployment environment. Together with a carefully designed data processing pipeline for embodiment transfer, this enables Steertuning to achieve a 71% success rate across 12 tasks in 12 new environments, ranging from lab settings to kitchens in new homes. The entire process, from data collection for a new task to
robot policy deployment, takes less than 5 minutes. Our full
dataset, model weights, codebase, and
robot videos will be made available at our website: https://steertuning.github.io.
Read the full abstract ▾
All authors Billy Yan, Isaac Tu, Kelly Lee, Blake Chang, Kenny Ma, Kanad Patel, Dhawal Kabra, Neer Patel, Kevin Wu, Lerrel Pinto, Irmak Guzey