arXiv:2603.03322v1 Announce Type: cross Abstract: Recent advancements in Large Language Model (LLM) agents have demonstrated remarkable potential in automatic knowledge discovery. However, rigorously evaluating an AI's capacity for knowledge discovery remains a critical challenge. Existing benchmarks predominantly rely on static datasets, leading to inevitable data contamination where models have likely seen the evaluation knowledge during training. Furthermore, the rapid release cycles of modern LLMs render static benchmarks quickly outdated, failing to assess the ability to discover truly new knowledge. To address these limitations, we propose DBench-Bio, a dynamic and fully automated benchmark designed to evaluate AI's biological knowledge discovery ability. DBench-Bio employs a three-stage pipeline: (1) data acquisition of rigorous, authoritative paper abstracts; (2) QA extraction utilizing LLMs to synthesize scientific hypothesis questions and corresponding discovery answers; and (3) QA filter to ensure quality based on relevance, clarity, and centrality. We instantiate this pipeline to construct a monthly-updated benchmark covering 12 biomedical sub-domains. Extensive evaluations of SOTA models reveal current limitations in discovering new knowledge. Our work provides the first dynamic, automatic framework for assessing the new knowledge discovery capabilities of AI systems, establishing a living, evolving resource for AI research community to catalyze the development of knowledge discovery.

arXiv:2603.02262v1 Announce Type: cross Abstract: Supervised fine-tuning (SFT) is essential for the development of medical large language models (LLMs), yet prior poisoning studies have mainly focused on the detectable backdoor attacks. We propose a novel poisoning attack targeting the reasoning process of medical LLMs during SFT. Unlike backdoor attacks, our method injects poisoned rationales into few-shot training data, leading to stealthy degradation of model performance on targeted medical topics. Results showed that knowledge overwriting was ineffective, while rationale poisoning caused significant decline on the accuracy of the target subject, as long as no correct samples of the same subject appear in the dataset. A minimum number and ratio of poisoned samples was needed to carry out an effective and stealthy attack, which was more efficient and accurate than catastrophic forgetting. We demonstrate though this study the risk of SFT-stage poisoning, hoping to spur more studies of defense in the sensitive medical domain.

arXiv:2507.01352v3 Announce Type: replace-cross Abstract: Despite the critical role of reward models (RMs) in Reinforcement Learning from Human Feedback (RLHF), current state-of-the-art open RMs perform poorly on most existing evaluation benchmarks, failing to capture nuanced human preferences. We hypothesize that this brittleness stems primarily from limitations in preference datasets, which are often narrowly scoped, synthetically labeled, or lack rigorous quality control. To address these challenges, we present SynPref-40M, a large-scale preference dataset comprising 40 million preference pairs. To enable data curation at scale, we design a human-AI synergistic two-stage pipeline that leverages the complementary strengths of human annotation quality and AI scalability. In this pipeline, humans provide verified annotations, while LLMs perform automatic curation based on human guidance. Training on this preference mixture, we introduce Skywork-Reward-V2, a suite of eight reward models ranging from 0.6B to 8B parameters, trained on a carefully curated subset of 26 million preference pairs from SynPref-40M. We demonstrate that Skywork-Reward-V2 is versatile across a wide range of capabilities, including alignment with human preferences, objective correctness, safety, resistance to stylistic biases, and best-of-N scaling. These reward models achieve state-of-the-art performance across seven major reward model benchmarks, outperform generative reward models, and demonstrate strong downstream performance. Ablation studies confirm that effectiveness stems not only from data scale but also from high-quality curation. The Skywork-Reward-V2 series represents substantial progress in open reward models, demonstrating how human-AI curation synergy can unlock significantly higher data quality.

arXiv:2602.18495v1 Announce Type: cross Abstract: Recent advances in tabular in-context learning (ICL) show that a single pretrained model can adapt to new prediction tasks from a small set of labeled examples, avoiding per-task training and heavy tuning. However, many real-world tasks live in relational databases, where predictive signal is spread across multiple linked tables rather than a single flat table. We show that tabular ICL can be extended to relational prediction with a simple recipe: automatically featurize each target row using relational aggregations over its linked records, materialize the resulting augmented table, and run an off-the-shelf tabular foundation model on it. We package this approach in \textit{RDBLearn} (https://github.com/HKUSHXLab/rdblearn), an easy-to-use toolkit with a scikit-learn-style estimator interface that makes it straightforward to swap different tabular ICL backends; a complementary agent-specific interface is provided as well. Across a broad collection of RelBench and 4DBInfer datasets, RDBLearn is the best-performing foundation model approach we evaluate, at times even outperforming strong supervised baselines trained or fine-tuned on each dataset.

arXiv:2602.19326v1 Announce Type: cross Abstract: As cities evolve over time, challenges such as traffic congestion and functional imbalance increasingly necessitate urban renewal through efficient modification of existing plans, rather than complete re-planning. In practice, even minor urban changes require substantial manual effort to redraw geospatial layouts, slowing the iterative planning and decision-making procedure. Motivated by recent advances in agentic systems and multimodal reasoning, we formulate urban renewal as a machine-executable task that iteratively modifies existing urban plans represented in structured geospatial formats. More specifically, we represent urban layouts using GeoJSON and decompose natural-language editing instructions into hierarchical geometric intents spanning polygon-, line-, and point-level operations. To coordinate interdependent edits across spatial elements and abstraction levels, we propose a hierarchical agentic framework that jointly performs multi-level planning and execution with explicit propagation of intermediate spatial constraints. We further introduce an iterative execution-validation mechanism that mitigates error accumulation and enforces global spatial consistency during multi-step editing. Extensive experiments across diverse urban editing scenarios demonstrate significant improvements in efficiency, robustness, correctness, and spatial validity over existing baselines.

arXiv:2512.24943v2 Announce Type: replace-cross Abstract: Search relevance plays a central role in web e-commerce. While large language models (LLMs) have shown significant results on relevance task, existing benchmarks lack sufficient complexity for comprehensive model assessment, resulting in an absence of standardized relevance evaluation metrics across the industry. To address this limitation, we propose Rule-Aware benchmark with Image for Relevance assessment(RAIR), a Chinese dataset derived from real-world scenarios. RAIR established a standardized framework for relevance assessment and provides a set of universal rules, which forms the foundation for standardized evaluation. Additionally, RAIR analyzes essential capabilities required for current relevance models and introduces a comprehensive dataset consists of three subset: (1) a general subset with industry-balanced sampling to evaluate fundamental model competencies; (2) a long-tail hard subset focus on challenging cases to assess performance limits; (3) a visual salience subset for evaluating multimodal understanding capabilities. We conducted experiments on RAIR using 14 open and closed-source models. The results demonstrate that RAIR presents sufficient challenges even for GPT-5, which achieved the best performance. RAIR data are now available, serving as an industry benchmark for relevance assessment while providing new insights into general LLM and Visual Language Model(VLM) evaluation.

arXiv:2602.14364v1 Announce Type: cross Abstract: Clawdbot is a self-hosted, tool-using personal AI agent with a broad action space spanning local execution and web-mediated workflows, which raises heightened safety and security concerns under ambiguity and adversarial steering. We present a trajectory-centric evaluation of Clawdbot across six risk dimensions. Our test suite samples and lightly adapts scenarios from prior agent-safety benchmarks (including ATBench and LPS-Bench) and supplements them with hand-designed cases tailored to Clawdbot's tool surface. We log complete interaction trajectories (messages, actions, tool-call arguments/outputs) and assess safety using both an automated trajectory judge (AgentDoG-Qwen3-4B) and human review. Across 34 canonical cases, we find a non-uniform safety profile: performance is generally consistent on reliability-focused tasks, while most failures arise under underspecified intent, open-ended goals, or benign-seeming jailbreak prompts, where minor misinterpretations can escalate into higher-impact tool actions. We supplemented the overall results with representative case studies and summarized the commonalities of these cases, analyzing the security vulnerabilities and typical failure modes that Clawdbot is prone to trigger in practice.