The paper investigates whether incorporating an evidence-support signal into retrieval evaluation for retrieval‑augmented generation (RAG) improves downstream decision‑making. Across multiple benchmarks and a TREC RAG 2025 setting, the evidence signal alters retriever rankings but its benefits vary: it does not consistently enhance retriever training, its usefulness for system selection depends on generator instructions, and it does not reliably predict answer quality on unseen topics. Human filtering of evidence‑rich passages preserves useful content, yet evaluators disagree on whether this improves final answers, indicating that evidence‑aware evaluation alone does not guarantee better downstream outcomes.
By Utshab Kumar Ghosh, Debayan Mukhopadhyay, Shubham Chatterjee
arXiv:2608. 26552v1 Announce Type: cross Abstract: Randomized sketch-and-solve algorithms accelerate overconstrained $\ell_2$ regression by replacing the input with a smaller problem.
By Zhao Song, Lichen Zhang
The paper investigates how the order of candidate documents in a prompt affects the decisions made by large‑language‑model (LLM) scorers, even when their ranking quality is similar. It shows that five scorers with only a 0.010 nDCG@10 difference can produce retained‑set overlaps as low as 0.66–0.84, and that existing rerankers still exhibit significant order dependence. The authors propose Order‑Consistency SFT (OC‑SFT), a training method that reduces this dependence, maintaining ranking quality while improving decision stability across multiple tasks.
By Markus Frohmann, Mahdiyar Alavi, Elizabeth Lingg, Navid Rekabsaz
WikiSkill is a framework that separates raw execution experience, accumulated knowledge, and executable skills, continuously consolidating experience into a persistent knowledge base (wiki). By co‑evolving agent skills with this wiki, the method consistently outperforms state‑of‑the‑art skill‑evolution techniques across diverse benchmarks and models. The study shows that larger models benefit more from evolved skills, smaller models can outperform larger ones when equipped with skills, and that skills transfer effectively across model families, with the wiki’s persistent knowledge being critical for success.
By Liyan Tang, Cyrus Rashtchian, Chun-Sung Ferng, Andrew Tomkins, Da-Cheng Juan, Tu Vu
The paper introduces "complexity induction," a method that distorts training data in a structured way to promote compositional generalization in a standard CNN without changing its architecture. Using synthetic images of colored geometric shapes, the authors encode classes as flat string labels (e.g., "red‑circle") and deliberately omit certain color‑shape combinations from training. Two distortion techniques—mixed labels (soft target distributions based on Jaccard similarity) and expanded dataset (false samples with incorrect labels)—both enable the model to predict unseen class combinations, with mixed labels leveraging the CNN’s embedding structure and expanded training improving embedding factorization. A control experiment with random false labels shows that the effect relies on the structured nature of the distortion rather than noise alone.
By Aleksandr V. Abramov
CorporateBench (CB) is a large‑scale, human‑validated Q&A benchmark designed to evaluate large language models on enterprise‑scale document collections. It contains over 230,000 documents derived from four synthetically generated firms, each modeled with a temporally evolving knowledge base that ensures logical consistency across hundreds of thousands of documents. The benchmark tests LLMs on information extraction and knowledge‑base querying, revealing that performance degrades as input size approaches realistic corporate scales.
By Sil Hamilton, Albert Yu Sun, Oscar J. Romero, Carl-Leander Henneking, David Mimno, Bishan Yang, Igor Labutov
The paper introduces Automated Research Design Tracking and Assessment (ARDTrA), a system for automatically identifying and evaluating the quality of causal research designs in social science papers. An expert‑annotated dataset covering six counterfactual design families was created, and the task was tested with a multi‑turn Retrieval‑Augmented Generation pipeline across various retrieval, language‑model, and embedding configurations. Results show that passage length largely drives performance, and that the designs most challenging for machines differ from those where human experts disagree most, revealing two independent sources of difficulty.
By Marco Rovera, Sergiu Burlacu, Dominique Cappelletti, Alessio Tomelleri, Sonia Marzadro, Martina Bazzoli, Annalisa Tassi, Jessica Gagete-Miranda
The paper demonstrates that a self‑supervised Vision Transformer (ViT) pretrained on a fast, low‑cost semi‑numerical simulator can produce data summaries that transfer across different simulators without retraining. In 21cm cosmology, the ViT—named SKATR—pretrained on 67,000 21cmFAST lightcones is applied unchanged to hydrodynamical Loreli II lightcones, enabling accurate inference of five astrophysical parameters with fewer radiative‑transfer simulations than a fully‑supervised baseline. SKATR remains accurate, informative, and calibrated even under realistic SKA antenna array noise, outperforming supervised models retrained on noisy data.
By Yannic Pietschke, Caroline Heneka, Ayodele Ore, Romain Meriot
The paper proposes a scalable, automated method for auditing candidate‑job matching systems for demographic bias. It employs large‑language‑model agents to generate neutral resumes, injects controlled demographic variations, ranks candidates with a fine‑tuned embedding model, and evaluates nine fairness metrics across counterfactual, group‑fairness, and merit‑aware families, producing a composite risk report. Experiments on a small corpus show that single‑score audits miss nuanced issues, underscoring the need for multi‑metric evaluation and LLM‑generated audits as a low‑cost complement to human reviews.
By Sai Yashwant, Shruti Bansal, Anurag Dubey, Samaroha Chatterjee, Satyam Kumar, Shreyash Gupta, Gantala Thulsiram
The survey reviews how Large Language Models (LLMs) are being used in High‑Performance Computing (HPC) programming, covering code generation, parallelization, frameworks, evaluation, and broader challenges. It finds that general‑purpose LLMs perform adequately on serial and OpenMP‑style tasks but struggle with distributed MPI workloads, while domain‑specialized models achieve higher accuracy yet are limited in scope and evaluation. The authors argue that LLMs will not replace HPC experts soon but can act as powerful collaborators, provided richer datasets, integration with performance tools, rigorous evaluation, and governance are developed.
By Strahinja Ljaljevic, Josep Jorba, Sergio Iserte
The paper introduces FairGIN, a fairness-aware graph neural network designed to predict demand for expanding bike‑sharing systems while addressing cold‑start challenges and equity concerns. It combines expansion‑simulated incremental training, attention‑based knowledge transfer, and income‑stratified regularization to improve predictive accuracy and reduce income‑based disparities. Experiments on NYC and Seattle show that FairGIN outperforms existing methods and supports more inclusive station placement without sacrificing overall efficiency.
By Man Luo, Yixuan Zhao
The paper introduces the Rashomon set for dimension reduction, a collection of equally good embeddings that preserve high‑dimensional structure. It proposes PCA‑informed alignment to make axes interpretable, concept‑alignment regularization to incorporate external knowledge, and a method to extract trustworthy nearest‑neighbor relationships across the Rashomon set for refined embeddings. These techniques aim to produce interpretable, robust, and goal‑aligned visualizations by leveraging multiple valid embeddings instead of a single one.
By Yiyang Sun, Haiyang Huang, Gaurav Rajesh Parikh, Cynthia Rudin
The paper reports that existing knowledge‑editing benchmarks cannot evaluate the scope decision—whether a stored edit applies to a query—because they are counterfactual and lack negative examples. Using the gradient‑free editor INLAY, the authors exhaustively test every router action on 1,689 queries across three datasets and find that an oracle router achieves no gain over a static policy, and abstention never wins. The authors attribute this to the structural design of the benchmarks and demonstrate that adding a missing negative condition restores some headroom and allows abstention to win.
By Aditya Pratap Singh
The paper introduces a self‑supervised framework that maps text, audio, image, and video into a shared 256‑dimensional embedding space and uses iterative clustering to uncover aesthetic structure. It examines how AI’s cluster assignments diverge from human affective labels on a weakly supervised multimodal dataset. The study highlights implications for cross‑modal similarity, media organization for Retrieval‑Augmented Generation, and automated data labeling.
By Corey D. C. Heath
The paper introduces KBevo, a co‑evolving framework that simultaneously builds a structured knowledge base and performs reasoning over it for knowledge‑intensive question answering. By optimizing both components end‑to‑end with QA outcome rewards, the system improves the quality and connectivity of the knowledge base, leading to higher answer reachability and better compositional factual reasoning. Compared to standard retrieval baselines, KBevo offers greater controllability and improved factual accuracy.
By Ryan Thomas Noonan, Linxi Zhao, Menghan Xu, Akanksha Sarkar, Mihir Mishra, Dongyoung Go, Kilian Q. Weinberger, Yoav Artzi, Jennifer J. Sun
The paper introduces a composite metric for quantizing small language models that balances information retention and throughput gains, using a normalized SQNR-based coefficient and roofline-based latency analysis. Profiling Gemma 3 1B shows that Feed‑Forward Network blocks and the embedding matrix are prime candidates for acceleration, with the metric enabling tuning of speed‑quality trade‑offs without actual execution. The authors demonstrate that their estimates predict accelerated speedup within about 4% error and allocate resources more effectively than evolutionary search or Shapley‑value methods.
By Artem Safronov
The paper compares classical and hybrid quantum machine learning models for a trigger-like binary classification task using CMS open data. Eight classical models (SVM, ANN, CNN, LSTM) and eight quantum counterparts are evaluated under identical preprocessing, data splits, and decision thresholds, with performance measured by accuracy, ROC‑AUC, F1‑score, precision, and recall. The best classical model is an artificial neural network (93.53 % accuracy, 0.9819 ROC‑AUC), while the best quantum model is a quantum convolutional network (90.89 % accuracy, 0.9731 ROC‑AUC), indicating that within an eight‑qubit budget the quantum models do not surpass the classical ones.
By Tariq Mahmood, Muhammad Awais Rafique, Talab Hussain, Juan Pablo Perez Aguilar, Alfredo Raya, Muhammad Ahsan
TabuLM is a new language model pre‑trained on Kinyarwanda tabular data, extending KinyaBERT‑large with row, column, and cell‑type embeddings and a table‑structure attention bias. It introduces two pre‑training objectives—Masked Cell Recovery and Column Type Prediction—and is trained on 172 Rwandan government tables. On the TabQA‑kin benchmark, TabuLM achieves 62.0% exact match, outperforming KinyaBERT‑large and multilingual baselines by significant margins.
By Ireddi Rakshitha, Devavarapu Yashwanth, Ntakirutimana Pierre
The paper introduces GROUND, a framework that limits large language model (LLM) analytics to a governed semantic layer for enterprise data warehouses. GROUND supplies approved metrics, dimensions, join paths, filters, and security rules, then validates generated SQL against these constraints before execution, retrying or abstaining on violations. In benchmarks, GROUND eliminates hallucinations across all evaluated categories and prevents row‑level security breaches, outperforming schema‑only, schema‑RAG, and semantic‑only approaches.
By Aravind Sasidharan Pillai
The study investigates how large language models (LLMs) handle diverse Indian oral traditions, using the Rajasthani Pabuji epic, Tamil Sangam poetry, and Bengali folk tales as case studies. By prompting Claude Sonnet and Gemini with 54 generation requests across generic, culturally specific, and regional-language prompts, the authors measured reference drift and cross-tradition convergence using Sentence‑BERT embeddings. Results show that while outputs stay closer to their own tradition than to others, there is significant cross‑tradition similarity (0.52–0.66), indicating partial homogenisation; moreover, regional‑language prompting consistently reduced fidelity to authentic traditions.
By Paarth Singh Rathore