Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
arXiv:2609.21293v1 Announce Type: new Abstract: Autonomous software generation (ASG) aims to turn human requirements into executable applications, but delivering these applications does not necessari...
arXiv:2609.21493v1 Announce Type: new Abstract: Multimodal large language models, or MLLMs, perform well at visual understanding and structured generation, yet these capabilities do not establish whe...
The paper introduces DSRec, a dual‑interest sequential recommendation model that separates item representations into long‑term and short‑term semantic contexts. Long‑term embeddings capture stable preferences through historical aggregation, while short‑term embeddings focus on local session intent modulated by inter‑click time intervals. Each branch is processed by a distinct State Space Model— a full‑sequence Mamba for long‑term modeling and a time‑modulated SSM for short‑term dynamics— and a residual cross‑fusion mechanism aligns the two granularities while preserving their independence. Experiments on public benchmarks show that DSRec outperforms state‑of‑the‑art methods.
The paper introduces Centroid‑Guided Contrastive Loss (CGCL), a novel loss function that combines classification with dense clustering to improve fraudulent job posting detection. CGCL reshapes the latent space using a centroid‑driven top‑k push‑and‑pull mechanism, ensuring accurate decision boundaries while preserving compact clustering of similar fake posts. Experiments on the EMSCAD benchmark show that CGCL achieves state‑of‑the‑art performance.
arXiv:2609.21626v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed for enterprise information extraction (IE), where the same document must be reorganized differen...
arXiv:2609.21801v1 Announce Type: new Abstract: We study how far a simple statistical pipeline can go on univariate time series anomaly detection under a strict selection protocol. The method extract...
arXiv:2609.21841v1 Announce Type: new Abstract: Frontier language models now produce professional deliverables that expert graders judge to match human work on a substantial share of economically val...
arXiv:2609.22068v1 Announce Type: new Abstract: Training capable coding agents via reinforcement learning (RL) requires diverse tasks with reliable verifiers. Open-source codebases offer a rich sourc...
Designer‑RSI presents a continual adaptation framework that lets a frozen frontier model operate professional design software while an external procedural memory learns natural‑language design skills from user traffic. Over five rounds on 1,406 real briefs and 1,869 graded trajectories, the memory grew from 76 to 139 skills, boosting execution success from 72.7% to 99.3% and improving win rates on four design benchmarks. The study shows that widening and deepening the memory, especially together, significantly outperforms a no‑skill baseline.
The paper demonstrates that test‑time communication among agents can significantly outperform independent parallel attempts on complex tasks. In experiments on the ARC‑AGI‑3 benchmark, a team of $k$ communicating agents matched the success rate of $4k$ independent agents, with the advantage growing as the team size increased. The study also shows that communication enables solving tasks that no single agent can solve, and that these benefits transfer to research‑oriented problems such as polyomino packing and MNIST classifier compression, where communicating agents surpassed prior best scores.
The paper introduces the COmmunity-centered Peer Engaged Support (COPES) dataset and a three‑axis evaluation framework to gauge how well Large Language Models (LLMs) align with community perspectives on mental‑health support queries. Experiments show that fine‑tuning LLMs on COPES improves strategy alignment and emotion‑tone alignment by over 50% for general‑purpose models, yet these gains are uneven across subreddits and coping strategies. The study also finds that post‑training shifts the model’s recommendations toward problem‑focused advice while reducing emotion‑focused responses, indicating persistent disparities in performance across different communities and needs.
arXiv:2609.21133v1 Announce Type: cross Abstract: SQL has been augmented with AI operators, enabling modern data analytics platforms to derive insights from both structured and unstructured data. We...
arXiv:2609.21151v1 Announce Type: cross Abstract: Molecular solubility directly affects key aspects of molecular development such as reaction feasibility, formulation performance, separation efficien...
FOCAL‑VLA is a framework that improves vision‑language‑action models by combining subtask‑guided geometry distillation with implicit world modeling. It transfers geometric knowledge from VGGT to focus on subtask‑relevant image regions and uses Track4World features to capture future 3D evolution, guiding action generation without running these models at inference time. Experiments demonstrate that FOCAL‑VLA outperforms baselines on both simulation benchmarks and real‑world manipulation tasks.
arXiv:2609.21334v1 Announce Type: cross Abstract: Effective evaluation of zero-day jamming detectors requires robust adversarial models. However, existing attack models often assume prior knowledge o...
CESBench is a new benchmark for evaluating large language models on cryptographic engineering security for IoT devices, comprising 380 expert‑written items across six sub‑domains such as side‑channel, fault injection, and implementation. The benchmark includes four task types—multiple‑choice, judgment, scenario, and code—each designed to test different competences, with automatic scoring for the first two and LLM‑based judging for the latter two. Evaluation of 11 open‑weight and proprietary LLMs shows strong performance on multiple‑choice and code tasks but weaker results on judgment and scenario tasks, highlighting gaps in justifying security verdicts.
The paper introduces a Situation–Internal state–Behavior Persona method to improve large language models’ ability to impersonate real individuals in social media contexts. It also proposes an evaluation protocol that supplies LLM evaluators with reference information about the target individual. Experiments on a new dataset of social media replies show the method surpasses state‑of‑the‑art in‑context learning baselines, and the protocol correlates moderately with human judgments, while additional tests on fictional characters confirm broader applicability.
The paper introduces KG‑Chronos‑2, a surrogate forecasting model that augments a frozen Chronos‑2 time‑series predictor with knowledge‑graph‑conditioned retrieval and correction mechanisms for HEC‑RAS water‑surface elevation (WSE) prediction. In a benchmark involving 64 24‑hour windows across 4,675 cross sections, KG‑Chronos‑2 outperforms persistence, residual LSTM, GeoFNO, and hydraulic DCRNN‑style models, achieving a 0.246970 event‑balanced RMSE and reducing error by 14.13% relative to frozen Chronos‑2. The model also attains the lowest active‑window and final‑lead RMSE among the evaluated systems, demonstrating the benefit of coupling a frozen temporal predictor with project‑specific knowledge for warm‑start HEC‑RAS forecasting.
AgentVidBench is a new multi‑hop video question‑answering benchmark designed to evaluate spatial, temporal, and causal reasoning in multimodal large language models (MLLMs). Unlike existing tests that focus on simple scene queries or global summaries, AgentVidBench includes step‑by‑step solution traces to assess whether agents gather the necessary evidence to justify their answers. Experiments with 12 MLLMs show limited single‑turn performance, but integrating these models into agentic workflows improves both accuracy and trajectory scores, establishing AgentVidBench as a comprehensive testbed for future research on agentic video understanding.
LoG-VGGT is a memory‑efficient framework for long‑sequence 3D reconstruction that balances local temporal modeling with global camera consistency. It uses cross‑window attention in a small subset of transformer blocks to propagate information across adjacent temporal windows while keeping memory usage bounded. A global camera consistency refinement module further improves long‑horizon pose stability by enforcing scene‑level constraints through cross‑attention between camera and compact register tokens, leading to better depth accuracy and robust camera pose estimation on multiple benchmarks.