The paper evaluates recent large language models (LLMs) for knowledge graph‑to‑text generation against traditional fine‑tuned encoder‑decoder models. Using both reference‑based metrics and graph‑grounded, reference‑free LLM‑based evaluation, the study finds that LLMs largely close the performance gap on reference metrics and outperform fine‑tuned models on graph‑grounded metrics. The results also show that LLM‑based evaluation correlates weakly with reference metrics, indicating that each captures distinct aspects of generation quality, and that prompted LLMs produce fewer contradictions and omissions, especially as graph size grows.
MixPT is a point cloud transformer that interleaves window and global attention blocks while keeping a fixed window serialization pattern. It applies global attention to pooled tokens for large scenes, feeding residual updates back to the full-resolution stream. Experiments on indoor and outdoor benchmarks show MixPT matches or exceeds the accuracy of existing window or global attention models while achieving lower latency, and it can be scaled by repeating the interleaved pattern.
The paper investigates how performance predictors can improve model routing for large language models (LLMs). It shows that a predictor only outperforms a model’s self‑confidence when the candidate model is suboptimal or the predictor adds significant capacity. The authors provide theoretical bounds, extend the analysis to two‑model cascades, and confirm that predictors help most with suboptimal models, with gains diminishing after fine‑tuning.
SQuAT introduces a shape‑gain vector‑quantization‑aware training (VQ‑QAT) method that separates weight direction and magnitude, using an $E_8$‑derived spherical codebook for directions and low‑bit scalar quantizers for gains. The approach employs distinct straight‑through estimators for shape and gain, enabling tangent‑space updates that act as implicit spherical optimization. Distributed exact nearest‑neighbor assignment reduces training time by 12.8×, allowing scaling to 10 B tokens and yielding state‑of‑the‑art accuracy on several large language models.
RIDAR is a framework designed to detect and mitigate research integrity violations (RIVs) in autonomous research agents. It introduces RIVT, a taxonomy of 14 RIV types, and offers two components: RIDAR-post-hoc, which identifies RIVs in agent trajectories with 85.2% F1 on a benchmark, and RIDAR-runtime, which dynamically corrects RIVs during execution, reducing remaining violations by 36.8%. The authors applied RIDAR to large datasets, uncovering 3,964 RIV instances in 99.7% of 300 agent trajectories across 50 tasks, highlighting the prevalence of RIVs in current agents.
MVLoMa is a sparse star‑graph matcher that extends the LoMa local‑feature matching framework from image pairs to multi‑view settings. It uses a frozen multi‑view geometry model to jointly describe all images in a star, injecting cross‑view geometric context into local descriptors and performing centre‑to‑neighbour matching in a single forward pass with per‑view self‑attention and bidirectional centre‑neighbour cross‑attention. On 25‑view star graphs from HyperSim, ScanNet++, Mip‑NeRF 360, and Tanks and Temples, MVLoMa achieves the best macro‑average relative‑pose AUC at all thresholds and is 2.20 times faster than sequential pairwise evaluation, while supporting multiple frozen geometry encoders and demonstrating the benefits of geometry‑aware descriptors, visibility filtering, geometric negative supervision, and joint multi‑view inference.
SurveyGraph introduces a framework that uses a Methods-Entity Knowledge Graph (MEKG) to generate academic surveys, addressing limitations of flat document retrieval by encoding method-level follow‑on, competition, and shared‑resource relations. The graph enables core‑paper discovery, retrieval, hybrid outlining, and evidence‑gated narrative synthesis, and is evaluated on retrieval, semantic edges, citation grounding, coverage, and generation components. Experiments show MEKG‑anchored fusion balances recall and structural density, graph‑aware retrieval nearly doubles recall over flat baselines, and a semantic evidence gate achieves an AUC of 0.75, with audits indicating strong citation support and relation accuracy.
IndoNutriQA is a curated benchmark of 911 Indonesian clinical nutrition question‑answer pairs derived from real reports. Each pair includes a clinically relevant context, structured metadata, and answers in free‑form prose, with expert validation indicating 6‑10% of answers are contestable. The dataset supports generative, extractive, and retrieval‑augmented QA, and is used to evaluate fine‑tuned models and large language models in zero‑ and few‑shot settings.
Moving object segmentation is conventionally formulated at the sequence and object level, marking an entire instance as moving once it moves anywhere in a video. Recent instantaneous formulations dist...
The paper introduces RetinaGroundBench, a new benchmark for visual grounding of retinal OCT images, featuring 815 B-scans annotated with 4,537 bounding boxes for 13 biomarkers across 8 diseases. It evaluates 27 multimodal large language models (MLLMs) and finds that visual grounding remains largely unsolved, with the best model achieving only a 23.7% macro F1 score. To improve performance, the authors fine‑tune an 8B‑parameter MLLM, RetGrounder, using an OCT‑specific strategy that raises the grounding F1 to 33.9%, a 43% relative gain.
whyItMatters":"The benchmark establishes a concrete reference for measuring progress in OCT diagnosis with multimodal large language models, highlighting the current limitations and guiding future improvements in visual grounding for clinical applications."
The paper introduces the Post‑Audit Mirage, a framework showing that offline tests may not detect whether an agent update remains safe after deployment. It defines a benchmark that uses matched safe and harmful lifecycles to expose identical verifier‑visible evidence while an isolated scorer records opposite deployment truth, and presents a theorem on the error‑abstention tradeoff using observational equivalence. The IdentifiedRangeMonitor tool applies fresh group‑labelled harm outcomes to decide whether to deploy, hold, or remain uncertain, and experiments demonstrate that monitoring can recover missing distinctions in favorable settings but may stay undecided for rare groups or small effects.
The paper investigates why long image captions often fail, arguing that failure depends not just on what facts are mentioned but also on where they appear in the caption. It introduces a position‑aware rubric that breaks down dense human references into atomic visual claims, evaluates generated captions against these claims, and identifies two distinct failure modes: unsupported specificity in the middle and vague, non‑evidential language at the end. Building on this diagnosis, the authors propose PAGER, a reinforcement‑learning objective that provides separate rewards for supported middle evidence, evidence‑bearing tail continuation, claim coverage, and length control, leading to improved position‑wise grounding and caption preference on detailed‑caption benchmarks.
The paper introduces Active Preference Acquisition (APA) for long‑term personalized web agents, defining the Prospective Personalization Value (PPV) of a missing preference as its expected cumulative utility over future tasks minus the cost of asking. APA selects questions within a per‑interaction budget, prioritizing preferences that may not benefit the current task but can be reused across many future tasks. Experiments on the ActiveWebPref benchmark show APA acquiring 80.3% of amortization‑dependent preferences and achieving higher net utility than baseline methods, while limiting the number of questions asked.
SecLog-Bench is a new benchmark designed to evaluate large language models on cybersecurity log reasoning. It features an extensible simulation platform covering 29 attack scenarios, uses multi‑source logs (audit, application, network traffic), and provides dense annotations with step‑by‑step evidence chains. Evaluation of eight mainstream LLMs shows modest performance, with the best model achieving 44.44% accuracy and 45.07% Macro‑F1, and even correct predictions often accompanied by flawed reasoning.
The paper introduces a ground‑truth‑first benchmarking framework for reinforcement learning that uses converse optimality to construct tasks with certified optimal value functions and policies. It unifies two benchmarking modes: sensitivity testing with continuous‑time generators to study robustness to task variation, and oracle‑referenced comparison with discrete‑time generators to evaluate policies via optimality gap, regret, and policy error. The methodology is applied to planar arms, nonholonomic vehicles, and quadrotor tasks, and extends to non‑smooth systems, raw‑action interfaces, and partially observed settings, with code and diagnostics publicly available.
GateFormer is a dual‑decoder Vision Transformer that forecasts intra‑hour global horizontal irradiance (GHI) from sky images by embedding observed GHI into the visual token stream and predicting both a future GHI trajectory and the final‑horizon change. The two predictions are adaptively weighted in prediction space, with magnitude‑ and direction‑supervision to reduce remaining error. On multiple benchmarks, GateFormer achieves the lowest final‑horizon RMSE, outperforming state‑of‑the‑art transformers and showing significant gains especially during positive ramp events.
The paper investigates natural language autoencoders (NLAs) that convert large language model activations into readable text. It examines whether NLAs capture compositional structure, generalize to fine‑tuned models, recover fine‑tuning objectives, and can be specialized for model diffing. Experiments across multiple models show positive results: NLAs preserve compositionality, transfer to lightweight fine‑tunes, recover hidden objectives, and outperform a black‑box baseline on diffing tasks, though they lag behind state‑of‑the‑art methods for misalignment detection.
ReflectPO introduces a reinforcement learning framework that enables large language model agents to recover from errors during long-horizon tasks involving external tools. The method consists of Reflection-Guided Sampling, which uses branch differences and repeated visits in interaction trajectories to generate recovery hints, and Tree-Structured Credit Assignment, which organizes trajectories into trees and assigns action-level credit based on parent–child reward differences and recovery gains. Experiments on two long-horizon tool-use benchmarks demonstrate that ReflectPO outperforms existing baselines and improves agents’ ability to detect deviations, recover from mistakes, and continue progressing toward task completion.
The paper introduces two lightweight inference harnesses, OPTICYCLE and OPTISTATE, designed to enable large language models to translate natural-language optimization problems into solver programs without additional training. Evaluated across nine benchmarks and 11 Qwen variants (0.6B–27B parameters), the harnesses consistently improve macro accuracy, with execution-guided revision boosting performance by up to 49.46 percentage points. Despite their simplicity, both harnesses outperform six adapted agentic workflows and four post-trained optimization models of comparable size, though the benefit of explicit formulation state varies by model.