The paper introduces Diffusion-Conditioned Representation Alignment (DCRA), a training framework that uses the forward diffusion process as a structured corruption scheduler for time‑series representation learning. DCRA aligns representations across noise levels with a feature‑level consistency objective, preserving class‑discriminative structure and enabling smooth, semantically coherent trajectories in latent space. Experiments on the CHB‑MIT EEG dataset demonstrate that DCRA improves seizure detection performance under various noise conditions, achieving higher sensitivity at low false‑positive rates and producing more balanced, structured representations than baseline methods.
By Wenrui Xu, Anas Enanaa, Keshab K. Parhi
The paper introduces GACA, a critic‑free reinforcement learning estimator that adapts credit assignment granularity based on a step‑level uncertainty proxy. GACA assigns higher weight to fine‑grained signals for steps with above‑average negative log‑likelihood, while relying on episode‑level signals for less uncertain steps, improving task success on ALFWorld and WebShop for 1.5B and 7B language models. The authors provide a risk decomposition, a conditional bound on action‑value variation, and an error‑projection analysis to justify the method’s effectiveness.
By Taoran Liang, Yang Liu, Shang Luo, Yingguang Yang, Rongrong Zhang, Yingzong Min, Yulin Huang, Jianshen Zhang, Yongzhi Qi, Kefu Xu, Congjing Ran, Bin Chong
The paper introduces SWARM, a multilingual dataset of 2,183 search engine results in nine languages, annotated for support of Russian propaganda narratives. It evaluates a source-based blocklist, supervised classifiers, and zero‑shot large language models, finding that blocklists miss most propaganda and that content‑level models vary in performance, with the best LLM achieving an F1 of 0.73. The study highlights the need for per‑language, content‑level detection of search‑borne propaganda.
By Manuel Tonneau, Abhinav Dubey, Farhan Shaikh, Ilaria Vitulano, Martha Stolze, Hale Dedeoglu, Clara Riechert, Ella Kuka, Maryna Sydorova, Mykola Makhortykh, Elizaveta Kuznetsova
Doc2FRC introduces Fixed-Range Chunking (FRC), a dynamic programming method that partitions documents into chunks of a predefined length interval, ensuring consistent length distributions during training and inference. This approach reduces train-test length mismatch, mitigates n-gram repetition, and improves translation quality for 7B LLMs compared to direct Doc2Doc fine-tuning. Experiments on IWSLT2017 and a new 10-language test set, GlobVDoc, demonstrate that FRC outperforms existing document-level machine translation methods and enhances out-of-distribution translation performance.
By Xiaotian Wang, Youyuan Lin, Zhan Shen, Hitomi Yanaka
The study evaluates a multi‑agent large language model system for forecasting outcomes of the 2026 FIFA World Cup. Two specialist agents—one quantitative and one news‑focused—produce forecasts that are then reviewed by a critic and combined by a meta‑agent. Results show the news specialist performs best, matching betting market accuracy, while the meta‑agent adds little beyond the specialists’ predictions.
By Julian Varghese, Lucas Bickmann, Sarah Sandmann
The paper introduces a long‑context recall technique that keeps GPU memory usage nearly constant regardless of context length, without requiring additional training. It reconstructs facts by leveraging residual vectors stored in the LLM’s feed‑forward layers, enabling deterministic retrieval of query‑relevant information without accessing the original document. Experiments demonstrate the method can answer single‑fact questions in two‑million‑token stories, outperforming prior approaches.
By MyungHoon Ryu, XinYu Piao, Jong-Kook Kim
PACIFIC is a framework that aligns large language model responses with user preferences by leveraging stable Big‑Five personality traits as a latent signal. The authors built a 1,200‑pair dataset covering diverse domains and trait directions, and found that trait‑aligned contexts enable LLMs to achieve near‑ceiling accuracy (up to 99%) in personalized QA. They also introduced a persona‑aware contrastive retriever (PiRAG) that improves label‑free accuracy from 30% to 43% over standard semantic retrieval, highlighting retrieval as the main bottleneck.
By Tianyu Zhao, Siqi Li, Yasser Shoukry, Salma Elmalaki
OphBiWSSD is a new framework for temporal action localization in ophthalmic surgeries that uses Bidirectional State Space Duality to avoid the quadratic memory cost of Transformers. It employs a weight‑tied selective scan that incorporates both past and future surgical context, enabling linear‑time global synthesis of non‑causal temporal cues. On the OphNet benchmark, OphBiWSSD achieves state‑of‑the‑art mean Average Precisions of 44.42 % for phases and 43.08 % for operations, outperforming baselines by 6.80 % and 6.66 % respectively.
By Yang Liu, Qionghong Ma, Joongwon Chae, Lihui Luo, Yibing Shen, Yulin Zhuo, Yingting Zhu, Jiashu Chang, Xiaoyun Zhong, Dongmei Yu, Peter E. Lobie, Peiwu Qin, Chengming Yang
The paper introduces a post‑training approach that enables a single inference process to transition from text reasoning to image synthesis, eliminating the need for explicit modality switching. Using the 14B BAGEL model, the authors demonstrate that targeted post‑training data and reward‑weighted training improve multimodal image generation across four independent T2I benchmarks. The study highlights the benefits of joint text‑image generation and strategic data selection for enhancing T2I performance.
By Jiahui Chen, Philippe Hansen-Estruch, Xiaochuang Han, Yushi Hu, Emily Dinan, Amita Kamath, Michal Drozdzal, Reyhane Askari-Hemmat, Luke Zettlemoyer, Marjan Ghazvininejad
EventMemAgent is an active online video agent that uses a hierarchical memory module to handle continuous perception and long‑range reasoning in streaming video. The framework employs a short‑term memory layer to detect event boundaries and sample frames within a fixed buffer, while a long‑term memory layer archives observations event‑by‑event. It also incorporates a multi‑granular perception toolkit and Agentic Reinforcement Learning to internalize reasoning and tool‑use strategies, achieving competitive results on online video benchmarks.
By Siwei Wen, Zhangcheng Wang, Xingjian Zhang, Lei Huang, Wenjun Wu
ReVisIT is a train‑free framework that turns retrieved image‑label pairs into units of visual thought, combining structured class definitions, multimodal retrieval, and alternating user/assistant injection before joint decoding. On several benchmarks—including Fast Open MiniImageNet, Bongard‑OpenWorld, and the newly released MAAC‑Bench—ReVisIT achieves performance comparable to or surpassing large, trained models while using far fewer parameters. The approach demonstrates that high‑quality retrieval and a simple turns layer can provide a universal performance boost across diverse multimodal tasks.
By Bingchen Huang, Zhiling Wang, Yifu Chen, Yuanchao Du
The paper examines binary-choice truthfulness benchmarks, showing that systematic differences in surface-level features between correct and incorrect answers allow models to perform well without genuine reasoning. Using a six-feature logistic classifier, the authors demonstrate that such leakage is detectable and exploitable, and they find similar artifacts in multiple benchmarks. To mitigate this, they propose a cleaning method called Audit‑Prune that removes the most leakage‑reinforcing answer pairs, releasing a revised TruthfulQA dataset with reduced surface‑feature leakage.
By Foad Namjoo, Remy Ogasawara, Amirali Abdullah, Cullen Anderson, Narmeen Fatimah Oozeer, Jeff M. Phillips
The paper introduces Disengagement-Aware Student Simulators (DAS2), a protocol that models five learner-engagement states—engaged, gaming, wheel-spinning, off-task, and mixed—to evaluate AI tutor performance before deployment. Using annotated tutoring sessions from ASSISTments09, DAS2’s rule-based labels matched human consensus in 81% of cases, and conditioning simulations on intended states narrowed the correctness-rate gap between simulated and authentic sessions for gaming and wheel-spinning behaviors. The study also compares five AI tutors across these states, finding stable relative rankings but state-specific performance differences, and notes that automated evaluation does not fully align with human judgment.
By Xianghui Meng, Jionghao Lin
arXiv:2609.12475v1 Announce Type: new
Abstract: Comprehensive benchmark suites are essential for improving large language models (LLMs), but many widely used benchmarks are redundant, making evaluati...
By Zhongzhan Huang, Junxin Li, Guoming Ling, Yupei Lin, Shanshan Zhong, Hefeng Wu
arXiv:2609.12575v1 Announce Type: new
Abstract: Ambiguity is often treated as a bug for AI systems to resolve---but in human communication and culture, ambiguity can also be a generative resource. Fr...
By Cody Kommers, Mingrui Ye, Evelyn Gius, Daniela Mihai, Hoyt Long, Zheng Yuan, Drew Hemment
arXiv:2609.12791v1 Announce Type: new
Abstract: Retrieval-Augmented Generation (RAG) has empowered Large Language Models (LLMs) to tackle knowledge-intensive tasks. However, navigating global, hetero...
By Gengxian Zhou, Jian Xu, Zichen Tang, Shiming Xiang, Haihong E, Cheng-Lin Liu
arXiv:2609.13045v1 Announce Type: new
Abstract: Speech-to-speech translation (S2ST) has advanced significantly with speech LLMs, offering the potential for joint optimization and preserving non-lingu...
By Hayato Futami, Hassan Shahmohammadi, Tushar Dhyani, Alkis Koudounas, Rapha\"el Lafargue, Yosuke Kashiwagi, Quentin Jodelet, Emiru Tsunoo
arXiv:2609.11987v1 Announce Type: cross
Abstract: An agentic coding system couples a language model to a harness: the tools, prompts and control flow that turn a chat model into an autonomous softwar...
By Mohsen Arjmandi
arXiv:2609.12444v1 Announce Type: cross
Abstract: Simulated societies of large language model agents are used to study online polarization, and separately to study collective intelligence, but the tw...
By Raad Bin Tareaf
arXiv:2512.14691v3 Announce Type: replace
Abstract: Modern multimodal generative models can synthesize visually compelling images and videos, but it remains unclear whether this visual fluency reflec...
By Zefan Cai, Haoyi Qiu, Tianyi Ma, Haozhe Zhao, Gengze Zhou, Tingting Liao, Xinyan Velocity Yu, Kung-Hsiang Huang, Ke Wan, Shawn Lin, Parisa Kordjamshidi, Minjia Zhang, Wen Xiao, Jiuxiang Gu, Nanyun Peng, Junjie Hu